Ziv, Lehrer, Schmid (Eds.) (2011) - Self-Evaluation. Affective and Social Grounds of Intentionality

Self-Evaluation
PHILOSOPHICAL STUDIES SERIES

VOLUME 116
Founded by Wilfrid S. Sellars and Keith Lehrer
Editor
Stephen Hetherington, The University of New South Wales, Sydney, Australia
Senior Advisory Editor
Keith Lehrer, University of Arizona, Tucson, AZ, U.S.A.
Associate Editor
Stewart Cohen, University of Arizona, Tucson, AZ, U.S.A.
Board of Consulting Editors
Lynne Rudder Baker, University of Massachusetts, Amherst, MA, U.S.A.

Radu Bogdan, Tulane University, New Orleans, LA, U.S.A.
Marian David, University of Notre Dame, Notre Dame, IN, U.S.A.
John M. Fischer, University of California, Riverside, CA, U.S.A.
Allan Gibbard, University of Michigan, Ann Arbor, MI, U.S.A.
Denise Meyerson, Macquarie University, NSW, Australia
François Recanati, Institut Jean-Nicod, EHESS, Paris, France
Mark Sainsbury, University of Texas, Austin, TX, U.S.A.
Stuart Silvers, Clemson University, Clemson, SC, U.S.A.
Barry Smith, State University of New York, Buffalo, NY, U.S.A.
Nicholas D. Smith, Lewis & Clark College, Portland, OR, U.S.A.
Linda Zagzebski, University of Oklahoma, Norman, OK, U.S.A.
For further volumes:

http://www.springer.com/series/6459
Anita Konzelmann Ziv · Keith Lehrer ·
Hans Bernhard Schmid
Editors
Self-Evaluation
Affective and Social Grounds of Intentionality
123
Editors
Dr. Anita Konzelmann Ziv Prof. Keith Lehrer
University of Geneva University of Arizona
Department of Philosophy Department of Philosophy
2 Rue de Candolle 213 Social Sciences Building
CH-1211 Geneva Tucson, AZ 85721-0027
Switzerland USA
anita.konzelmann@unige.ch lehrer@email.arizona.edu
Prof. Dr. Hans Bernhard Schmid
University of Basel
Department of Philosophy
Nadelberg 6-8
CH-4051 Basel
Switzerland
Hans-Bernhard.Schmid@unibas.ch
ISBN 978-94-007-1265-2 e-ISBN 978-94-007-1266-9

DOI 10.1007/978-94-007-1266-9
Springer Dordrecht Heidelberg London New York
Library of Congress Control Number: 2011927921
© Springer Science+Business Media B.V. 2011

No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by
any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written
permission from the Publisher, with the exception of any material supplied specifically for the purpose
of being entered and executed on a computer system, for exclusive use by the purchaser of the work.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Editorial
Self-evaluation is a term recently reassessed in academic philosophy by the joint

efforts of two Swiss research teams. The topic was launched on the occasion of the
workshop “Self-Evaluation – Individual and Collective”, held in Basel in January
2009. The aim of the workshop was to open new approaches for investigating tradi-
tional questions such as the scope and purpose of self-knowledge, the interrelation
between the social and the individual person, and the significance of emotional
appraisal. The scientific added value created by a perspective that shifts the focus
from “self-knowledge” to “self-evaluation” is threefold: (i) the wider extension of
the concept of self-evaluation allows a broader perspective on the structure of per-
sonal reflexivity; (ii) the notion of self-evaluation implies a matrix of values, and
insofar as valuations imply a social context, the broadened perspective on personal
reflexivity incorporates social relations; (iii) since affective states and attitudes play
a crucial role in the detection and recognition of values, the broadened perspective
on personal reflexivity incorporates affective assessment. In short, self-evaluation
is a conception of personal reflexivity which incorporates sociality and affectivity.
That this approach is more than promising was confirmed by positive responses of
leading philosophers in the respective fields of research. The initial exchange of
ideas on the occasion of the workshop led to deepened discussions on the topic
which eventually materialized in the present volume.
The workshop in Basel was jointly organized by the research group
Collective Intentionality – Phenomenological Perspectives, hosted at the Philosophy
Department of the University of Basel, and the research group “Thumos” – Emotion,
Values and Norms, hosted at the Philosophy Department of the University of
Geneva. We would like to express our gratitude to the Swiss National Science
Foundation, the Swiss Center for Affective Sciences, the Freiwillige Akademische
Gesellschaft Basel and the University of Basel for their generous support of
the workshop and the research leading to this volume on the philosophy of
self-evaluation.
Basel and Geneva
October 2010
v
This is Blank Page Integra vi
Contents
1 Self-Evaluation – Philosophical Perspectives . . . . . . . . . . . . . 1

Anita Konzelmann Ziv
Part I Evaluative and Self-Evaluative Attitudes

2 How to Have Self-Directed Attitudes . . . . . . . . . . . . . . . . . 33
Lynne Rudder Baker
3 Interpretation, Cause, and Avowal: On the Evaluative
Dimension of Selfhood . . . . . . . . . . . . . . . . . . . . . . . . . 45
Axel Seemann
4 Who Do You Think You Are? The How–What Theory
of Character and Personality . . . . . . . . . . . . . . . . . . . . . 59
Federico Lauria and Alain Pé-Curto
Part II Self-Evaluation and Rationality

5 Self-Evaluation and the Ends of Existence . . . . . . . . . . . . . . 81
Carol Rovane
6 Self-Evaluation and Action . . . . . . . . . . . . . . . . . . . . . . 97
Juliette Gloor
7 Self-Trust and Social Truth . . . . . . . . . . . . . . . . . . . . . . 119
Keith Lehrer
Part III Self-Evaluative Emotions

8 Sentimentalism and Self-Directed Emotions . . . . . . . . . . . . . 135
Jesse Prinz
9 Psychopathic Resentment . . . . . . . . . . . . . . . . . . . . . . . 155
John Deigh
10 Self-Knowledge, Knowledge of Others, and “the thing called love” 171
Edward Harcourt
vii
viii Contents
11 Is Shame a Social Emotion? . . . . . . . . . . . . . . . . . . . . . . 193

Julien Deonna and Fabrice Teroni
Part IV Evaluating the Social Self

12 Feeling Up to It – The Sense of Ability in the Phenomenology
of Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
13 Self-Evaluation in Intention: Individual and Shared . . . . . . . . 237
Lilian O’Brien
14 Where Individuals Meet Society: The Collective Dimensions
of Self-Evaluation and Self-Knowledge . . . . . . . . . . . . . . . . 253
Ulla Schmid
About the Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Contributors
Lynne Rudder Baker Department of Philosophy, University of Massachusetts,

Amherst, MA, USA, lrbaker@philos.umass.edu
John Deigh Department of Philosophy, University of Texas, Austin, TX, USA,
jdeigh@law.utexas.edu
Julien Deonna Department of Philosophy, University of Geneva, Geneva,
Switzerland, Julien.Deonna@unige.ch
Juliette Gloor Department of Philosophy, University of Basel, Basel,
Switzerland, juliette.gloor@unibas.ch
Edward Harcourt Faculty of Philosophy, Oxford University, and Keble College,
Oxford, UK, edward.harcourt@philosophy.ox.ac.uk
Anita Konzelmann Ziv Department of Philosophy, University of Geneva,
Geneva, Switzerland, anita.konzelmann@unige.ch
Federico Lauria Department of Philosophy, University of Geneva, Geneva,
Switzerland; Department of Philosophy, University of Fribourg, Fribourg,
Switzerland, Federico.Lauria@unige.ch
Keith Lehrer Department of Philosophy, University of Arizona, Tucson, AZ,
USA; Department of Philosophy, University of Miami, Coral Gables, FL, USA,
lehrer@email.arizona.edu
Lilian O’Brien Department of Philosophy, University College Cork, Cork,
Ireland, l.obrien@ucc.ie
Alain Pé-Curto Department of Philosophy, University of Geneva, Geneva,
Switzerland; Department of Philosophy, University of Fribourg, Fribourg,
Switzerland, Alain.Pe-Curto@unige.ch
Jesse Prinz Department of Philosophy, The Graduate Center, City University,
New York, NY, USA; Department of Philosophy, The University of North
Carolina, Chapel Hill, NC, USA, jesse@subcortex.com
ix
x Contributors
Carol Rovane Department of Philosophy, Columbia University, New York, NY,

USA, cr260@columbia.edu
Hans Bernhard Schmid Department of Philosophy, University of Basel, Basel,
Switzerland, Hans-Bernhard.Schmid@unibas.ch
Ulla Schmid Department of Philosophy, University of Basel, Basel, Switzerland,
ulla.schmid@unibas.ch
Axel Seemann Department of Philosophy, Bentley University, Waltham, MA,
USA, ASEEMANN@bentley.edu
Fabrice Teroni Department of Philosophy, University of Berne, Berne,
Switzerland; Center for Affective Sciences, Geneva, Switzerland,
Fabrice.Teroni@philo.unibe.ch
Chapter 1
Self-Evaluation – Philosophical Perspectives
Anita Konzelmann Ziv
Self-evaluation is not a technical term in philosophy.1 Rather, it is common to

associate the expression “self-evaluation” with contexts of psychologically oriented
supervision where individuals or groups are encouraged to evaluate themselves for
the purpose of getting clear about their goals, their motivations and possibilities
of improving their performances.2 The present volume is an attempt to add philo-
sophical weight to the concept of self-evaluation. The philosophical perspective
associates “self-evaluation” instantly with long-standing key topics of philosoph-
ical research such as the metaphysics of the self, the nature of self-reference and
the nature of values. “Self-evaluation” seems to be a notion conceptually close to
Socrates’ seminal “Know thyself” which, down to the present day, has been induc-
ing an unceasing stream of reflection under the key notes of “self-consciousness”
and “self-knowledge”. One central claim pervading the reflections on the specific
kind of relationship that rational agents bear to themselves is that this relation to
oneself is requisite for having intentional relations to objects and events in the
world. As the title of the present volume indicates, this claim is, on the whole,
I am deeply indebted to Keith Lehrer and Fabrice Teroni for their detailed and insightful comments
that were of great value in developing the ideas presented in this text. I also wish to thank Cain
Todd for linguistic help and an anonymous referee for valuable remarks.
1 This statement is based on the fact that there are no recent philosophical books or articles attempt-
ing to provide a systematic account of self-evaluation. It is not intended to mean that self-evaluation
has not been treated in philosophy. A prominent positive example is Harry Frankfurt’s theory of
second-order desires: Frankfurt characterizes the formation of second-order desires as a manifes-
tation of “the capacity for reflective self-evaluation” that he takes to be a human-specific property
(Frankfurt 1971, 7). And Charles Taylor attempts to refine Frankfurt’s approach to self-evaluation
by elaborating a distinction between “strong” and “weak” self-evaluation: the latter is concerned
with potential outcomes of a desire’s satisfaction, while the former is concerned with a desire’s
internal value properties (Taylor 1977).
2 In scientific psychology, instances of self-evaluation are accounted for in a variety of terms, such
as “self-assessment”, “self-control”, “self-esteem”, “self-representation” etc. (Tesser 2003).
A. Konzelmann Ziv (B)

Department of Philosophy, University of Geneva, Geneva, Switzerland
e-mail: anita.konzelmann@unige.ch
A. Konzelmann Ziv et al. (eds.), Self-Evaluation, Philosophical Studies Series 116, 1

DOI 10.1007/978-94-007-1266-9_1, C Springer Science+Business Media B.V. 2011
2 A. Konzelmann Ziv
maintained, whereas the tradition of investing the self-relating capacity with con-
cepts suggesting its primarily doxastic nature is called into question. Labels such
as “self-knowledge” and “self-consciousness” explicitly draw on the concept of
knowing (scire), thus insinuating that the function of the self-relating capacity is
to generate true judgments about oneself.
It is, however, precisely this doxastic conception that is under attack in many
accounts of self-relating attitudes. Much work has been done to develop approaches
that focus on those features of self-reference that are more closely connected to
experience and agency than to knowledge and truth. Accounts such as, for example,
theories of self-constitution through agency, emphasize that agential and experien-
tial varieties of self-referential activities must not be considered as mere ancillary
conditions of doxastic self-knowledge, but as genuine ways of holding the rele-
vant relationship to oneself (Moran 2001, O’Brien 2007, Brandom 2007, Korsgaard
2009). In view of this shift of focus it is appropriate to look for a conception that can
accommodate not only the doxastically but also the experientially oriented accounts
of self-reference. Furthermore, it is desirable to explore conceptions that question
the traditional limitation of conceiving the capability of relating to and acting toward
oneself mainly in terms of belief and desire. “Self-evaluation” is the concept that
satisfies both these needs, its advantage over the conception of “self-knowledge”
being threefold: first, the wider extension of “self-evaluation” opens up a broader
perspective on the general structure of self-reference; second, implying a matrix
of values, “self-evaluation” broadens the perspective so as to include social rela-
tions constitutive of a great number of values; third, given that emotions play a
crucial role in the recognition of values and the motivation to act accordingly, “self-
evaluation” calls for the explicit inclusion of affectivity into accounts of human
reflexivity.
Therefore, the suggestion made here is to consider self-evaluation as the self-
relating capacity of rational beings into which sociality and affectivity are concep-
tually built. This amends the conception of the specific relationship to oneself in
that it takes into account not only theories of the self and of reference to the self, but
also theories of values, norms, and choice.
This introductory essay is divided into two heterogeneous parts:
Section 1.1 provides an overview of the contributed essays. It contains a brief
outline of the reasons for choosing the topical sections “Evaluative and self-
evaluative attitudes”, “Self-evaluation and rationality”, “Self-evaluative emotions”,
“Evaluating the social self”, followed by the summaries of the essays.
Section 1.2 assembles some of my own reflections on the subject. It frames the
suggestion made so far by means of three topical questions: (1) Is evaluation – gen-
erally speaking – a kind of knowledge? (2) What challenges arise from the reflexive
character of self-evaluation? (3) How can the affective and the social be accounted
for in self-evaluation? Or, to put the question differently, how can the affective and
the social amend accounts of self-reference that are typically cast in terms of a dox-
astic performance or a bi-modal belief-desire approach? The aim of my examination
of these questions is to fathom the possibility of a general account of self-evaluation,
rather than treating specific aspects of the phenomenon.
1 Self-Evaluation – Philosophical Perspectives 3
1.1
The four parts of the volume focus on (Part I) the nature of self-evaluative atti-
tudes, (Part II) the relation between self-evaluation and rationality, (Part III) the role
of self-evaluative emotions, and (Part IV) the role of social or collective factors
in self-evaluation. The first part examines various sorts of self-directed attitudes
and addresses the question of which self-directed attitudes are self-evaluative. An
important issue with regard to self-directed evaluative attitudes concerns the way
the first person perspective is realized in them. Some take an evaluator’s first per-
son view to be dependent on certain concepts, particularly the concept “I”, while
others do not. Some take it to be dependent on doxastic states, while others empha-
size its agential or practical nature. Both the first and the third part of the volume
take up questions of self-directed attitudes, the first in a more general way, the third
focusing on self-evaluative emotions. The second part centers on the question of the
relation between rationality – both practical and theoretical – and self-evaluation.
The topic is especially pregnant with regard to the strong involvement of emotions
and other affective attitudes in self-evaluation, which seems to point rather towards
its being explicable in terms of arational or even irrational processes than in terms
of rational performances. The third part sheds some light on the important role of
emotions in self-evaluation. Recent research on the emotions strongly supports the
assumption that the main function of emotions lies in their capacity for appraisal, i.e.
apprehension of values involved in the emotion eliciting situation. Specifically self-
directed emotions such as shame, remorse or guilt clearly are evaluative faculties
that reveal a person’s worth with regard to certain values. The fourth part focuses
on the question of what features of a personal unit are evaluated in the reflective
turn to oneself. Many self-directed attitudes – self-love, self-hate, self-loathing, self-
respect, self-esteem, self-approval, cura sui, self-concern, self-interest, self-trust,
belief in oneself, narcissism, self-admiration, amour propre. . . etc. – are closely
tied to the position one occupies within a network of social relations. Indeed some
attitudes may have as their object the so-called “social person” (as opposed to the
intimate or “private person”) and value properties such as fame, reputation, gloire.
The fourth part aims at discovering some of the dependence relations between indi-
vidual self-evaluation and social positioning, and ways in which these relations
inform our understanding of self-directed emotions. It further examines possibilities
of collective self-evaluation by means of genuinely collective attitudes.
1.1.1 Evaluative and Self-Evaluative Attitudes

The opening paper by Lynne R. Baker – Chapter 2 – brings to the fore the topic
of the basic requirements for self-evaluative attitudes. From a radically externalist
point of view Baker argues that having linguistic and social relations is the abso-
lute requirement for any self-directed attitude, given that the latter is dependent on
a robust first-person perspective that makes self-concepts requisite. Baker assumes
4 A. Konzelmann Ziv
the positive claims that having a self-directed attitude requires the subject (1) to
realize in a first-personal way that the object of her attitude is herself, and (2) to
manifest the attitude in actions and thoughts of which both self-concepts and empir-
ical concepts are constitutive. The negative assumption is that having a self-directed
attitude does not imply that the subject realizes that they have this attitude. Baker
explains the robust first-person perspective in terms of mastering complex uses of
the concept “I” such as anaphoric reference in the structure “I believe (fear, hope,
say, etc.) that I am F” where the second use of “I” identifies the reference of the
first without the help of names, descriptions or demonstratives. Although the robust
first-person perspective is “far from being a Cartesian perspective that isolates indi-
viduals from communities”, the self-concepts required in adopting this perspective
are not taken to belong to the class of the L&S (linguistic and social) concepts.
Defined as concepts which a subject S would fail to have if S had been in a commu-
nity with relevantly different social or linguistic practices, L&S concepts depend on
inter-individual relations within communities. Since Baker holds that having a self-
concept presupposes having empirical L&S concepts, she concludes that being a
person depends not only on being part of a community, but also that a universe con-
taining beings with self-directed attitudes must contain communities with linguistic
and social practices.
In Chapter 3, Axel Seemann defends the claim that selfhood is intrinsically
normative, i.e. presupposes taking an evaluative attitude towards oneself.
Seemann argues for this claim primarily on the basis of three considerations:
(1) Chisholm’s conception of the self in terms of “immanent causation”, (2)
Campbell’s conception of the self conceived as the common cause of many events
connected over time, and (3) the evaluative weight of phenomenal experience.
Accounting for selfhood in terms of immanent causation is giving a causal expla-
nation of subjectively endorsed mental states without resorting to external causes
of endorsement. A subject’s “avowal” of her mental state is understood as being
embedded in an interpretive endeavour which aims at the correct individuation of
the state. Interpretation of phenomenal experience goes beyond describing it and
beyond understanding why it obtains. Rather, it frames a state presented through
insight or reflection in ways that modify its experience. Interpretation reveals what
matters to the subject and forces integration of episodic findings into a compre-
hensive view of one’s mind. It is this indispensable evaluative dimension of being
involved in experience that prompts the subject to endorse a state as hers. On
the basis of what developmental psychologists call “primary intersubjectivity”,
Seemann argues that the interpretive evaluative endeavor is necessarily a matter of
interaction with other subjects. “Primary intersubjectivity” is intimately related to
the capacity for sharing embodied psychological attitudes – a capacity developing
early in an infant’s life and prior to the ability to distinguish between self and other.
Sharing embodied psychological attitudes enables a subject to experience herself as
causing mental states – in others as well as in herself. In this important sense, the
essence of selfhood – understood as being a common cause connected over time –
is anchored in intersubjective relations.
The first part closes with the co-authored contribution of Federico Lauria and
Alain Pé-Curto (Chapter 4).
The authors state that, as a matter of fact, we describe people by appealing to
traits (e.g. shyness, generosity, sincerity) in order to explain, predict and evalu-
ate their general behavior. They suggest that there is a sense in which some of
these characterizations are deeper than others: when we try to “understand people”,
including ourselves, such deep characterizations seem particularly relevant. Trying
to specify our understanding of persons in a finer grained way, the authors explore
the metaphor of personal layers by distinguishing two types of traits, which they
call behavioral folds and behavioral vectors. Behavioral folds are characterized as
mere behavioral dispositions which persons instantiate (e.g. shyness as a disposition
to remain in the background), while behavioral vectors are characterized in terms of
attachments to certain values (e.g. sincerity as attachment to truth or genuineness).
Behavioral folds are claimed to be global and optional, in the sense that they are
neither specific (like habits are) nor necessary (like breathing is). Behavioral vectors
are global traits as well, but differ from behavioral folds by having values as their
objects. Claiming that behavioral folds can be explained by appealing to behav-
ioral vectors, i.e. attachments to values, the authors suggest that folds capture more
peripheral personal features, while the values one is attached to draw the boundaries
of the very person one is.
1.1.2 Self-Evaluation and Rationality

In Chapter 5, Rovane argues that self-evaluation – understood in the strict sense
of weighing and ranking on a scale from worst to best – plays an essential
role in the exercise of reflective rational agency. Her basic assumptions are (1)
that the existence of reflective rational agents is a product of effort and will,
and (2) that the rational agents created are of different sizes, being either agents
of human size, multiple agents within a single human being, or group agents.
Seeing the overarching goal of rational agency as lying in the achievement of all-
things-considered-judgments upon deliberation, Rovane claims that the frame of
deliberation is established by a unifying project which determines the scope of
“all-things” by practical requirements. The essential tie between projects and ratio-
nal agency makes rational agents existentially dependent on the commitment to
a project. And it makes internal evaluations of the project, its goals and means,
instances of self-evaluation. Since the ends of existence of rational agents are deter-
mined by the ends of the project, the agents exist for the sake of those ends.
Therefore, the rationally required evaluation of their projects unavoidably amounts
to an evaluation of the worth of their very existence. According to Rovane, this
idea can be best understood by considering rational group agents which do not exist
before the commitment to a unifying project. The group point of view established
by the commitment sets the boundaries for rational agency, without being immune
6 A. Konzelmann Ziv
to evaluative questions regarding the justification and quality of the initiated project.
Since the group-agent owes its existence to the commitment to the project, consider-
ations of this kind are not merely project-evaluative but also self-evaluative. Rovane
sees good reasons to generalize this model to all kinds of agents.
In Chapter 6, Juliette Gloor starts from Christine Korsgaard’s claim that the
function of action is self-constitution, arguing that if the function of action is self-
constitution then self-evaluation is evaluation as to how well or how badly actions
constitute their agents as unified authors of what they do. First it is discussed
what self-evaluation in non-reflective animals might amount to. Granted that non-
reflective animals are conscious animals, there is something that it is like for the
animal to hunt or to protect its offspring. But such an animal, although not capable
of reflective self-evaluation, experiences defective actions – actions which constitute
the animal’s nature or form badly; for example in that the animal fails to do what
it desires to do – as qualitatively different from her successful actions. Along these
lines, purely causal accounts of action are then criticized for their failure to take
seriously an animal’s center of subjectivity and the intimate relationship between
action and desire. In a second step, Gloor focuses on the self-evaluation involved
in reflective agency, claiming that the form of human action, unlike that of animal
action, is not desire but practical reason: human agents have to decide whether to
act on their desires as reasons for action. While all conscious animals have in com-
mon a desire for consciousness as a desire for his or her own consciousness, only
reflective animals are aware of the grounds for some action as his or her own. Thus,
self-evaluation in reflective animals addresses self-determination in a deeper sense:
the very activity of reflective deliberation directed at the goodness or rationality of
the action under consideration is self-consciously evaluative. Finally, Gloor consid-
ers how self-determination might be understood in cases where two or more subjects
constitute a single conscious community of shared activity.
Keith Lehrer’s contribution (Chapter 7) closes the part on rationality by dis-
cussing the relations between self-evaluation in the form of self-trust, collective
wisdom and truth. Lehrer claims that reasonableness requires assuming that one
is trustworthy in the evaluation of one’s intentional states. Positive evaluation of
beliefs in terms of background information yields acceptance of what one believes,
and positive evaluation of desires yields the preference to obtain what is desired.
The combination of acceptance of one’s trustworthiness and preference for being
trustworthy yields the conclusion that one is reasonable in what one accepts and
prefers. Lehrer then defends the obvious explanatory loop of self-trust by means of
a conception of sensory experience he calls “exemplarization”. The term denotes
a psychological process that generalizes from a sensory exemplar to represent a
class of experiences, while the exemplar serves to exhibit the properties exemplified
by all members of the class. Providing the security of one-to-one mapping, self-
representation so conceived makes the exemplar as its own symbol “true of itself”,
and, therefore, connects self-trust and truth. On the other hand, the ways represen-
tational systems generalize reveals cognitive diversity. The problem to be solved is
how to insure truth at the cost of such idiosyncrasies. Lehrer suggests a solution that
notably amends the individualism of the self-trust theory: being trustworthy requires
that individuals evaluate the claims of others, adapting what they accept and prefer
in the light of such evaluation. Idealizing acceptances and preferences as probabil-
ities and utilities allows mathematical modeling of inter-individual evaluation. The
model shows how iterated inter-individual evaluation merges individual self-trust
and social conformity into aggregates of consensual weight. On the basis of fur-
ther mathematical work (S. Page et al.) it is shown how cognitive diversity resulting
from self-trust makes consensual evaluation trump individual error and outperform
simple averaging.
1.1.3 Self-Evaluative Emotions

The part on self-evaluative emotions is opened by Jesse Prinz (Chapter 8). On
empirical grounds, Prinz defends moral sentimentalism and shows how serious
philosophical objections to the theory can be addressed by drawing on self-
directed emotions. Moral sentimentalists take emotions to be proper parts of moral
judgments, whereas opponents of this view think that morality and emotion are dis-
sociable. To move beyond this seemingly insuperable impasse calls for taking an
empirical turn. Empirical evidence in the form of neuro-imaging, behavioral meth-
ods and studies on psychopathologies suggest that emotions are in fact genuine
components of empirical judgments. Emotional constituents of token moral judg-
ments can be specified, for example, as tokens of anger, disgust or contempt. Prinz
discusses four objections to moral sentimentalism, namely emotions without judg-
ments, hidden circularity, disavowed emotions, and emotional intensity. All these
objections suggest that sentiments and moral judgments come apart. To deal with
them, sentimentalists are tempted to resort to emotional cognitivism in order to save
sentimentalism. It turns out, however, that it is not possible to identify the cogni-
tive construals that make moral judgments distinctly moral. Since this problem is
exactly what makes the reductive move of sentimentalism so attractive, the strat-
egy of saving sentimentalism by means of cognitivism is counterproductive. The
alternative strategy Prinz proposes consists in defining moral judgments in terms
of moral values and moral values in terms of sentiments. Sentiments are explained
as standing dispositions that essentially involve self-directed emotions. The respon-
siveness of shame and guilt to moral norms can be explained non-cognitively by
specific ways of socialization (e.g. withdrawal of affection, social ostracizing). Prinz
shows how self-directed emotions are essential to distinguishing non-moral emo-
tions from moral emotions, to explaining asymmetries between conflicting values,
and to providing an adequate measure of evaluative intensity.
John Deigh’s contribution (Chapter 9) elaborates the topic of self-evaluative emo-
tions with regard to moral responsibility. On the basis of Peter Strawson’s account
of “reactive attitudes” Deigh explores the twofold thesis that (1) not all reactive atti-
tudes are moral in nature, and (2) morality is not built on personal relations. With
this in mind, he discusses three different kinds of attitudes responding to injustice:
resentment, indignation and the self-reactive attitude of guilt (or remorse). They all
8 A. Konzelmann Ziv
presuppose specific relations between individuals, qualified by specific expectations

and demands concerning the display of goodwill toward each other. Accordingly,
they can be ranged with regard to the point of view from which unjustness is con-
ceived. While resentment implies a personal point of view, both indignation and
guilt take an impartial or disinterested point of view. Against the common assump-
tion that all reactive attitudes are properly moral, Deigh argues that only those taking
an impersonal view express a moral stance. Given that the moral stance presup-
poses an understanding of oneself and others as members of a moral community
with reciprocal rights and obligations, resentment alone cannot count as moral. As
a personally inspired attitude, resentment is bound to personal interests rather than
to general rights and obligations. This is why psychopaths can feel genuine resent-
ment although they lack moral understanding and attitudes. Contrary to Strawson,
who sees the proper moral attitudes as generalizations of personal reactive attitudes,
Deigh works out a Humean-inspired account of proper moral attitudes. According
to this approach, the impersonal view requisite for morality is acquired by means
of self-reactive attitudes. It is through feelings of compunction, regret or guilt that
one apprehends that others have equal rights to good-willed action, and that, conse-
quently, rights and obligations are equally distributed amongst all members of the
community.
Edward Harcourt’s paper (Chapter 10) asks how we acquire mental concepts
applicable both to ourselves and to others. In particular, Harcourt explores the role
of love in the development of self-knowledge, which is broadly understood as the
capacity to have thoughts about oneself. Drawing on recent work on child-caregiver
interaction, Harcourt argues for an account of mental concept acquisition inspired
by Wittgenstein that prioritizes the conceptual problem over the epistemic problem
of other minds. He assumes that it is possible to express at least some experienced
mental states before acquiring any mental concepts. Acquiring person-neutral men-
tal concepts requires, however, the presence of others who recognize the mental
states expressed and reflect them back in a way that reveals them to be representa-
tions of the original states. The thesis, roughly, is that self-knowledge depends on
the knowledge others have of us. Now loving nurture, as Harcourt argues on the
basis of recent empirical studies, plays an important role in the development of such
conceptually structured self-knowledge. Consequently, receiving the right kind of
loving nurture seems to promote one’s capacity to be a self-knower. It further fol-
lows that in order to explain how the capacity for self-knowledge develops via the
knowledge others have of us, the other must be thought of as a loving other. This
is in line with philosophical analyses of love according to which loving another
involves, among many other things, seeing the other as they really are.
The paper by Julien Deonna and Fabrice Teroni (Chapter 11) lines up with the
topic of self-evaluative emotions, and, tackling the “social claim” of shame theories,
builds the bridge to the fourth part. Shame is the paradigmatic case of an affective
state evaluating its bearer on the basis of some values held. The authors query the
common assumption that the self-evaluative function of shame is due to its being an
essentially social emotion, involving the regard of others. They first discuss differ-
ent understandings of the term “being a social emotion”, depending on whether the
emphasis is on the objects of emotions, their genesis, on the types and properties
of their cognitive antecedents or the typical contexts that elicit them. Based on this
analysis, they assess and give reasons to reject the popular claim that shame is an
essentially social emotion. The “social claim” with regard to shame has been made
in a variety of forms, among which figures the thesis of the heteronomy of shame, or
the claim that shame essentially relates to our “social selves”. Furthermore, accounts
focusing on the context of shame elicitation claim that shame occurs only in front
of real or imagined audiences. Deonna and Teroni argue that while all the differ-
ent claims are true of some important families of shame episodes, none of them
generalize so as to motivate the conclusion that shame is an essentially social emo-
tion. They proceed by explaining the connections between different versions of the
“social claim” and the constraints on a theory of shame they help uncover. Finally,
they show how a non-social picture of shame is not only capable of meeting these
constraints, but has the further virtue of being apt to shed light on those cases where
others seem to play no role in why we feel shame.
1.1.4 Evaluating the Social Self

The fourth part opens with Hans Bernhard Schmid’s contribution (Chapter 12). The
author points out that the sense of ability is an important self-evaluative state that
has hardly received any attention in recent phenomenology of action. Being a form
of awareness that agents unquestionably have of whatever they take to be their abil-
ities, the sense of ability seems to fulfill the function of securing a certain level of
success and minimizing failure. The sense of ability is presented as essentially self-
evaluative in that it determines the degree of involvement people take themselves to
have in their lives, the degree to which they rely on themselves rather than on other
agents or external forces. Consequently, it determines the degree to which they see
themselves as responsible for their lives, the degree to which they see their lives
as being up to themselves, instead of as a matter of fate or destiny. H.B. Schmid
discusses a couple of conflicting intuitions concerning the structure and role of this
self-evaluative faculty: some think that the sense of ability must be self-confirming,
others claim that it is subject to rational constraints; some take it is a matter of con-
vention, others insist that it must be informed by experience. Particular attention is
then devoted to the question of whether or not the sense of ability limits the class of
possible objects of intention. Showing that with regard to this issue, the case of both
sides is equally strong, H.B. Schmid argues that the resulting conceptual dilemma
can be resolved if the sense of ability is conceived of in affective rather than in
purely cognitive terms. The paper concludes with a consideration of the bearings of
this result on the phenomenology of joint action.
In Chapter 13, Lilian O’Brien attempts to understand necessary features of the
psychology of rational practical agents. One of her main claims concerning the
philosophical psychology of intentional action is that in forming an intention to act,
the agent takes a particular kind of attitude towards herself. Specifically, the agent
10 A. Konzelmann Ziv
regards herself as subject to evaluation, as a failure in certain circumstances and a

success in others. O’Brien calls this the “Self-Evaluation Condition” that subjects
the agent to comply with the content of the intention she formed or, alternatively,
to engage in rational revision of an intention before abandoning it. Pointing to the
relevance of the self-evaluation condition to understanding reasons for action, the
difference between desire and intention, and the difference between alienation from
and identification with the will, O’Brien addresses the question of the specific rela-
tion between the self-evaluation condition and the nature of shared intention. One
of the most disputed claims in this field is Margaret Gilbert’s claim that shared
intention grounds by itself a non-conditional obligation to the sharers not to defect.
Those that deny such independent grounding think that the obligation not to defect
essentially involves reasonable expectations generated by the defector in the relevant
social context. On the basis of an example discussed, O’Brien argues for a neces-
sary condition on shared intention that draws on her Self-Evaluation Condition. It
grounds a normative requirement on the part of participants to take their partners’
interests into account when deliberating about the shared intention, and it derives
from the very nature of shared intention. Although not equivalent to Gilbert’s non-
conditional obligation, this normative requirement might explain some aspects of
shared intention that Gilbert cites to motivate her view.
Ulla Schmid’s paper (Chapter 14) closes the volume by directly relating self-
evaluation and self-knowledge. Self-evaluation is understood as self-referential
feedback activities by which agents constantly survey the course of their actions
and, consequently, are able to intervene and adjust their behavior. Self-evaluation
ranges from basic bodily processes serving the adjustment of our movements and
bodily position, via self-evaluative attitudes such as shame or pride, to theoretical
judgments of one’s agency. The author concentrates on reflective self-evaluation,
arguing that it involves a certain kind of self-knowledge, which does not originate
from observation. Rather, it consists in the intentional activity itself as related to cer-
tain standards, which determine whether an action counts as deficient or successful.
The account thus dispenses with the tendency to take self-evaluation and a fortiori
self-knowledge as a primarily epistemic relation an agent holds towards herself. As
U. Schmid argues, the primary form of gaining knowledge of one’s mind is by mak-
ing it up, by taking a stance towards the world – towards what one is concerned with
in her respective situation. This corresponds to what is called the “non-observational
agential” form of first-personal knowledge, in contrast to its “observational empiri-
cal” form. While the former is manifested in exclusively self-referential evaluative
attitudes, e.g. shame or guilt, the latter is manifested in evaluations that can be
applied both to oneself and to others, e.g. anger or approval. To explain the norma-
tive system enabling an agent to take a value-oriented position towards herself, U.
Schmid refers to a “collective context” that delivers the standards of appropriateness
for value judgments, governing the range of possible interpretations and narratives
chosen to describe and explain one’s actions. On the other hand, the “collective con-
text” delivers intersubjective language systems that provide the concepts by means
of which self-knowledge is shaped and communicated.
Herewith I conclude the first part of this introduction to philosophical per-

spectives on self-evaluation which summarizes the content of the volume. After
having considered the research of others, I shall now turn to my own reflections
on the subject. They are guided by the question of how to give a general account
of self-evaluation. As the previously discussed philosophical perspectives suggest,
a general account of self-evaluation requires elaborating the following questions:
(1) Is evaluation – generally speaking – a kind of knowledge? (2) What chal-
lenges arise from the reflexive character of self-evaluation? (3) To what extent
can reflections on affectivity and sociality complement accounts of evaluative
self-reference?
1.2
The term “self-evaluation” combines the meaning of “evaluation” with the structural
feature of reflexivity. Generally speaking, evaluation involves a set of values that
establishes the standards against which the merit, worth, or significance of an object
is determined. Evaluation is an activity that characterizes and appraises objects of
interest in a wide range of human enterprises. Often, evaluation is effected to deter-
mine whether an object suffices to accomplish a certain task or whether it needs to
be improved. Or it is effected to determine which candidate is the best to accom-
plish a certain task. Aside from these instrumental applications, evaluation can also
be self-contained in the sense that it brings to the fore whether an object is valu-
able in a specific sense, or how different objects are to be ranged in an order of
worth. Self-contained evaluation is apparent in many cases of aesthetic evaluation,
for instance when the literary significance of a novel is evaluated in the process of
reading, or when the flavors of Whisky are evaluated in the process of tasting and
drinking. While “value” and “comparison” are primary meaning components of the
concept “evaluation”, “attunement” and “improvement” seem to be secondary. We
are able to evaluate the literary worth of a novel or the flavors of Whisky without
the further purpose of ranging the item in some charts, and without the intention of
improving the evaluated item. Evaluation is not essentially consequentialist in the
sense of bringing about optimal results, even if the term “evaluation” is often related
to instrumental purposes.
With regard to the reflexive structure invoked by the term “self-evaluation”, it
is appropriate to associate philosophical perspectives on self-evaluative attitudes
with long-standing key topics of philosophical reflection, such as Socrates’ seminal
“Know thyself!”. Socrates’ appeal seems to suggest that the relevant task of rational
beings consists in taking a doxastic attitude to oneself, resulting in the state of
self-knowledge. In order to assess self-evaluation we should ask, therefore, whether
self-evaluation is identical with self-knowledge, or whether, perhaps, it is a specific
kind of self-knowledge. Investigating these questions presupposes some reflections
on the relation between evaluation and knowledge in general. I will explore this
relation in Section (1.2.1), and proceed in Section (1.2.2) to the question to what
extent the relation is affected by reflexivity. Section (1.2.3) summarizes the attempt
to show how affectivity and sociality are built into the conception of self-evaluation.
1.2.1 Evaluation and Knowledge

A fundamental similarity between the terms “evaluation” and “knowledge” consists
in their being terms invoking two different applications: they can be used, on the
one hand, as process terms, and, on the other hand, as achievement terms. Consider
first the case of “knowledge”. In epistemological discourse, the term “knowledge”
is often used to describe the momentary state of an epistemic subject S who fulfills
conditions such as holding a true belief that p, being rationally justified in holding
this belief, and/or being causally driven to holding the belief that p. The leading idea
is that S arrived in a specific non-contingent way at a doxastic state which, conse-
quently, deserves the label “knowledge”. This is the strong meaning of “knowledge”
when the term is applied as achievement term. Alternatively, the term is also used
for a whole body of beliefs acquired by S in the appropriate way (consider, for
example, the expression: “mathematical knowledge”). It has been argued, however,
that there are other important “kinds of knowledge”, such as “coming to know that
p or apprehending that p (erkennen, dass p)” and “coming to be acquainted with
(kennenlernen, Kenntnisnahme) someone or something” (Mulligan 2007, 213).3 In
contrast to knowledge as achievement these assumed kinds of knowledge are rather
process-like than state-like. “Coming to know”, “apprehending”, and “coming to
be acquainted with” are terms denoting not the result of epistemic endeavor but
processes and activities which aim at knowledge (in the aforementioned sense of
achievement) and often may achieve their goal.
I take it that knowledge (as achievement) necessarily involves judgment. The
term “judgment” stands here for “asserted belief”, i.e. for the result of an assertive
act concerning a specified propositional thought.4 While propositional thoughts “p”
can be “entertained” or “held” without a commitment to their truth-conformity (e.g.
3 Admittedly, there is the problem that “there is no established or happy English translation of
‘erkennen’ unless, like some Anglophone epistemologists long ago, we talk of an ‘act of knowing
that’” (ibid.).
4 I use both the notions “judgment” and “doxastic” in a narrow epistemologically inspired sense:
while “judgment” stands for “asserted propositional thought”, “doxastic” stands for “concerning
knowledge understood as entailing correct judgings”. Differing uses of the notions “judgment”
or “judging” may yield different views on the interdependence of evaluation and judgment (and,
a fortiori, knowledge). K. Lehrer and N. Smith work with a notion of judgment meaning both
“acceptance” in the epistemological sense of “evaluated” or “justified” belief, and “preference”
understood as evaluated desires (Lehrer and Smith 1996, 10). Alternatively, R. Solomon charac-
terizes feelings as “judgments of the body”: “There are feelings, “affects” if you like, critical to
emotion. But they are not distinct from cognition or judgment and they are not mere “read-outs” of
processes going on in the body. They are judgments of the body, and this is the “missing” element
in the cognitive theory of emotions” (Solomon 2003a, 16).
using “p” as a hypothesis), judging is the act of asserting “p”, thereby committing
oneself to taking p to be true.5 My assumption is that “judgment” is the core notion
by which “knowledge” as an achievement term is to be distinguished from proces-
sual terms such as “coming to know”, “apprehending” and “coming to be acquainted
with”. The latter refer to a variety of processes, mental modes and activities implied
in the endeavor of achieving knowledge. Kevin Mulligan has argued, for instance,
that perceiving a fact in one way and identifying it with a fact perceived in another
way is constitutive of apprehending, and that apprehension has the potential to bring
knowledge (as achievement) into being. Mulligan emphasizes, however, that even if
identifying a fact perceived in one way with a fact perceived in another way “may
and often does take the form of a judgment”, apprehending “need not involve judg-
ment” (op. cit., 219). Taking this claim for granted gives us a means of severing
knowledge – accounted for in terms of having made judgments conforming to truth
in appropriate ways – from knowing – accounted for in terms of processes conducive
to knowledge in the aforementioned sense.
Before discussing the terminological analogy between “knowledge” and “eval-
uation”, I wish to sketch two more points that might be important in assessing the
question of whether the human self-relating capacity is essentially doxastic. Firstly,
epistemological analyses of knowledge (as achievement) invoke processes and
activities – e.g. perceptual processes or reasoning – subsumable under terms such
as “coming to know”, “apprehending” and “coming to be acquainted with”. Hence,
considering as “kinds of knowledge” both knowledge as achievement and knowing
as potentially knowledge-conducive processes makes analyses of knowledge appear
circular. To deal with this problem, two alternatives can be envisaged: either it has
to be shown that this circularity is not vicious, or the label “knowledge”/“knowing”
has to be abandoned for either of the two sides of the analysis. It is not my inten-
tion to approach this problem here, but the question of how extendible the notion
of knowledge is without losing its epistemological significance is important and
should be kept in sight. My second point is not unrelated to the first. It concerns the
assumed dispositionality of knowledge, that is the claim that we tacitly “hold” many
true beliefs without ever having thought or approved of the content of these beliefs
(Lycan 1986). It is argued, for example, that a subject S’s behavior and actions indi-
cate that S “implicitly” holds a belief that p, to the effect that, if confronted with
the explicit content p, S would immediately and justifiably assent to p. Therefore,
S might be said to “dispositionally know” that p. The possibility of implicit knowl-
edge is explicable in terms of “epistemic closure”, that is the view that whatever
logically follows from a true propositional thought explicitly asserted by S inher-
its the state of knowledge for S (Feldman 1995). Accounting for knowledge (as
achievement) in terms of judgments seems to forbid such a dispositional account of
5 My view of judging as asserting propositionally structured “thoughts” or “contents” is not com-

mitted to the view that assertion is a linguistic act manifested in intersubjectively perceivable signs.
It is inspired by the traditional intentionalist view that judging is first of all an essentially mental
act. I use quotation marks “p” to indicate subjective instantiations of p, whether they be purely
mental or linguistically shaped.
knowledge, unless one holds that judging that p factually is judging that q, given that
q logically follows from p. But if judging that q is supposedly identical with actu-
ally judging that p, the term “dispositional” would not apply any longer to knowing
that q. Talking about dispositional judgments is not only counterintuitive but seems
to be conceptually false. To be disposed to judge that p might be analyzed in terms
of perceiving or apprehending that p, or in terms of being acquainted with p, but
not in terms of making a dispositional judgment that p. Even a high probability of
being asserted seems too weak for a merely apprehended “p” to satisfy a concept of
knowledge which involves judgment. Therefore, the fact that evaluation and exper-
tise are eligible for articulation in judgment does not seem to provide a strong reason
to subsume them under the concept of knowledge. Accounting for knowledge (as
achievement) in terms of judgments implies that knowledge is actual in the sense of
involving concrete assent to specified propositional thoughts. This approach is com-
patible with a view of knowledge as a body of asserted propositions of which not
each and everyone is continually present in the mind. But it excludes from knowl-
edge (as achievement) propositions that have never been thought or asserted in a
judgment.
In many contexts of use, evaluating is put on a level with judging, which sug-
gests that evaluation belongs to the domain of knowledge. To be sure, “evaluation”
and “knowledge” share the feature of being terms that can be applied both to an
achievement and to the processes enabling the achievement. Considered under its
predicative form “evaluating”, “evaluation” is a process term denoting an activity
extended in time. Considered as an achievement term, “evaluation” might be ren-
dered in the noun form “evaluative judgment”, denoting the articulated result of
an evaluative process. In everyday language the word “evaluation” is often used
to refer to the results of evaluative processes articulated in evaluative judgments.
Think, for example, of expressions such as “She got a positive evaluation” where
“evaluation” is used in the sense of an achievement term. The context of use in most
cases enables us to specify whether the term denotes evaluation in the sense of a
process or, rather, in the sense of the stated achievement of such a process. For ana-
lytic purposes, however, it is preferable to mark the distinction between evaluative
processes and their results by using the terms “evaluating” and “evaluative judg-
ment” respectively. Without this terminological distinction we could too easily be
led to assuming right away that evaluation is a kind of knowledge (as achievement).
Although this seems to be a plausible assumption at first glance, it is not as obvi-
ous as we might think. Assuming, as I did before, that knowledge (as achievement)
implies judgment, is not assuming the converse. Judgment does not imply knowl-
edge. Judgments can be mistaken, and even if they conform to truth, the credit of
instantiating knowledge might be denied to them (for instance, when conformity to
truth is a matter of sheer luck). Hence, the fact that evaluative processes may and
often do bottom out in evaluative judgments does not justify the view that evaluation
instantiates knowledge. On the other hand, it could be stated that every judgment is
an evaluation, namely an act of evaluating a propositional thought with regard to its
truth value. Knowledge, then, appears to instantiate evaluation rather than the oppo-
site. This suggestion gains some strength if we take “knowledge” in the even more
specified sense of reflective knowledge, or discursive knowledge, referring not only

to a subject S’s evaluating a propositional thought “p” with regard to the truth value
of p, but, in addition, to S’s assessing the validity of her coming to evaluate “p”
positively with regard to truth (Sosa 2009, part II; Lehrer 2000). While evaluation
can be and often is manifest as knowledge, it is not the case that all evaluation is
doxastic. Evaluation may focus on other values than truth.
Elucidating the interplay between the terms “evaluation” and “knowledge”
requires considering their applications to processes as well. In traditional epistemol-
ogy, it is commonly held that reliable knowledge-conducive processes are bound to
cognitive or intellectual faculties, including perception, memory, inference, delib-
eration, etc. In contrast, processes bound to affective and volitional faculties are
often seen as negatively interfering with the purpose of knowledge. Notions such as
“wishful thinking” or “He argued emotionally” are mostly used to refer to processes
which are supposed to prevent subject S from staying in the fast lane to knowledge.
Whether and to what degree the segregation of volitional and affective processes
from processes that are reliably knowledge-conducive is justified is a question that
is widely discussed, but will not concern me here. What is more important in the
present context is to see that volitional and affective processes are clearly evaluative
processes. Volitional processes evaluate a certain x with regard to its desirability,
which might be interpreted as a hedonic value, or an adaptive value, or a “nobler”
value involving perhaps justice or beauty. Affective processes are widely recognized
as being the means of a subject S’s evaluating or assessing or appraising a certain
x, particularly with regard to x’s importance for S.6 Evaluative processes, however,
are certainly not restricted to volitional and affective processes. Being conducive
to knowledge, cognitive or intellectual processes such as inferring and deliberating
are proper kinds of evaluative processes. They evaluate sequences of propositional
thought with regard to truth preservation, or with regard to the profitability of actions
described by them.
This brief outlook on evaluation and knowledge understood in terms of pro-
cesses seems to match the previous findings concerning evaluation and knowledge
understood in terms of achievement: on both levels of analysis, knowing is quite
easily accommodated in evaluating, while the converse is much more contro-
versial. Evaluative judgments are incontestably part of our epistemic fabric and
eligible to count as constituents of knowledge, but they need not amount to knowl-
edge. Evaluation appears not to be constrained by the doxastic purpose of yielding
justified assent to properly acquired true beliefs.
Instead of being qualified in doxastic terms, many faculties involved in evaluative
processes can be qualified as experiential. I take it that the term “experiential” con-
notes, on the one hand, the synchronic aspect of experience, i.e. the way in which a
situation is phenomenally grasped at the very moment of its occurrence. On the other
hand, “experiential” connotes the diachronic aspect of experience, i.e. expertise in
6 See, for example, “Appraisal Theories” of emotions (e.g. R. Lazarus, K. Scherer) or neo-Stoic
“Cognitivist Theories” of emotions (e.g. M. Nussbaum, R. Solomon).
the sense of being adept at handling certain kinds of situations. Expertise and knowl-
edge are certainly closely related concepts, but it is widely held that expertise differs
from knowledge in that it is not necessarily propositionally structured. Furthermore,
expertise does not seem to be subject to any requirement excluding sheer luck in its
achievement. The alleged difference between being doxastic and being experiential
is often spelled out in Gilbert Ryle’s terminology that opposes “knowing that” to
“knowing how”. While the former is accounted for in terms of judgment, truth and
justification, the latter is accounted for in terms of abilities and skills. It is contended
that, contrary to the propositionally formed contents of knowing that, knowing how
(i.e. expertise) could not be fully captured in the algorithmic procedures of a com-
putational mind (Dreyfus 1992). It is, however, a matter of ongoing debate whether
a principled distinction between expertise and knowledge is sound, and what rela-
tions hold between skills and propositional thought (Stanley and Williamson 2001,
Sgaravatti and Zardini 2008). Let us admit for the present purpose that certain kinds
of expertise, say expertise in reasoning, directly depend on propositionally struc-
tured thought, while other kinds of expertise are not so dependent. Examples of the
latter kind might be found in expert perceptual discrimination, such as the phenom-
ena of sexing chickens or indicating the key of a melody (Biederman and Shiffrar
1987, Harnad 1996). Expertise of these and similar kinds requires being familiar
with certain scenarios, viz. the specific patterns of action they call for. This familiar-
ity in turn must be accounted for in terms of conceptual capacities, enabling an agent
to conceive of certain patterns as patterns, i.e. as configurations of things and proce-
dures to be ranged in comprehensive conceptual orderings.7 The kind of conceptual
“grasp” I have in mind overlaps with what G.E. Moore called “direct apprehension”,
a cognitive relation, which, contrary to knowledge, can hold between intentional
subjects and non-propositional entities.8 Both for analytic and empirical reasons it
is advantageous to uphold such a distinction between, on the one hand, cognitive
faculties that are doxastic in the sense of being manifest in propositional thought
bound to truth conditions and, on the other hand, cognitive faculties that are not
doxastic in this sense. One way to mark this distinction is to subsume the former
7 It has been argued that having no inferential structure, “scenario content” is not conceptual:
“Although a scenario content does have a structure, it does not have inferential structure – and so
does not have conceptual constituents” (Crane 1992, 155). It seems to me, however, that concepts
representing a certain order of things allow being arranged in patterns that need not be proposition-
ally structured. For a discussion of the problem of minimal recognitional capacities see Bermúdez
1998, chs. 3 and 4, and Tye 2009.
8 “[D]irect apprehension [. . . ] is a relation which you may have to a proposition, equally when you
believe it and when you do not, and equally when it is true and when it is false; whereas immediate
knowledge is one form of the relation which I called knowledge proper [which] is a relation which
you never have to a proposition, unless, besides directly apprehending it, you also believe it; and
unless, besides this, the proposition itself is true [. . . ]; direct apprehension is a relation which you
may have to things which are not propositions, whereas immediate knowledge [. . . ] is a relation
which you can only have to propositions” (Moore 1993, 75 f.).
but not the latter under the term “knowledge/knowing”.9 Cognitive faculties that
are not manifest in propositional thought do not necessarily bring about judgment,
but may bring about skilful behavior that manifests familiarity with situations and
things already encountered. However, in virtue of their seizing features such as situ-
ational similarity, difference and recurrence, and categorizing them into conceptual
units, these faculties are properly considered cognitive in nature.
The fact that expertise – accounted for in terms of activities which align with
conceptual patterns of familiarity – has repeatedly been given citizenship in the
epistemic realm under labels such as “knowing how” or “practical knowledge” may
be partly motivated by Aristotle’s distinction between practical and theoretical ratio-
nality. Since the concept “rational” is often used in close relation to concepts such as
“cognitive”, “deliberating” and “knowing”, the distinction between theoretical and
practical rationality has the tendency to dissolve into the distinction between the-
oretical and practical knowledge. Yet the latter systematically blurs the distinction
envisaged, since it conceptually fuses knowledge (the doxastic) with experience.
Another factor motivating the subsumption of expertise under knowledge may be
seen in the potentially doxastic character of expertise: the fact that expertise can –
in principle – be spelled out in judgment may be conducive to considering it as
doxastic and hence as a kind of knowledge, or proto-knowledge.10
It seems appropriate to close this section on the interface of evaluation and
knowledge by briefly mentioning a further problem arising from identifying evalu-
ation with knowledge. Contrary to factual judgments that either conform to truth or
not, evaluative judgments are not easily accounted for in terms of truth-conformity.
Whether and how truth conditions for propositions asserted in evaluative judg-
ments can be established is a thorny question. For our present purpose it suffices
to emphasize that there is a large class of evaluative judgments for which clear-cut
truth conditions cannot be given. Evaluation, as has been pointed out, encompasses
processes of value weighing that are not instrumental in the sense that the item eval-
uated is measured with regard to its conformity to a threshold value. Evaluation can
consist in simply apprehending the position a thing takes on a value scale. Consider,
for example, a literary club discussing a novel. Although the discussants gener-
ally are all experts in the field of literature and apply standard evaluative tools to
the products of literary effort, the evaluative judgments they articulate can differ
massively. Sometimes discussion of controversial issues and a closer look at certain
details related to them can lead indeed to a more unified set of evaluative judgments.
Evaluative judgments are not subject to complete arbitrariness. They are, however,
9 This move is quite natural in languages such as German where “wissen”, “erkennen”, “kennen”
and “können” are semantically distinguished. The distinction between “knowledge” and “expert-
ness” outlined above approximately captures the distinction between “wissen” (knowing) on the
one hand, and “kennen” (being acquainted) and “können” (being capable, knowing how to do) on
the other hand.
10 Notions such as “tacit knowledge” and “implicit knowledge” belong to the same family of
notions of potential knowledge.
neither subject to tests of truth conformity as a purely factual judgment such as “At
this very moment, it rains in New York” is. Due to the fact that a great number of
values are hardly eligible for determinate specifications, evaluative processes allow
for variances of possible responses, so that the evaluative judgments articulating
them often display indeterminacy with regard to univocal truth-conditions. There
is hardly a viable way to determine, say, the truth conditions of the content of a
judgment on the complex taste of Burgundy, or the rules in accordance to which
agents could be held epistemically liable for delivering such judgments.11 This is
particularly apparent in the case of so called “thin” values such as the good or the
beautiful. As long as values are not maximally “thickened” by specifying determi-
nate criteria for their obtaining, they seem too vague to ground the truth conditions
of evaluative judgments. This makes evaluative judgments tricky candidates for the
epistemological concern of determining the conditions under which they instantiate
knowledge.
To summarize: Evaluation is not exhausted by knowledge if the latter is under-
stood in the epistemologically relevant sense of asserting rationally justifiable true
beliefs about facts. “Evaluation” is a term whose primary meaning concerns a range
of processes which aim at comparing an object to some value. Given different abil-
ities involved in evaluative activities, as well as varieties of value specifications,
evaluation involves variances of possible responses, i.e. a range of options in assign-
ing values to the evaluated object. The notion of evaluation is in many respects
equivalent to the notions of appraisal and assessment, as well as to the notions of
weighing and balancing.
1.2.2 The Reflexive Character of Self-Evaluation

The topic of this book is self-evaluation, which is the reflexive variant of the relation:
x evaluates y. For simplicity’s sake, I shall consider evaluation as a dyadic relation
R(x,y) where either x = y or x = y holds.12 Grammatically, the seme “self” is not
a noun term bound to denote some kind of specific entity – the Self – but rather
an indicator of reflexivity, i.e. it reveals the obtaining of the fact that x’s R-ing is
directed to x.13 The semantics of the term “reflexive” pictures an item bent back on
11 Ithas been argued, however, that not even subjectivists about values deny there to be “objective
evaluations relative to standards” of quality or merit that determine the “justice” of value judg-
ments. The non-arbitrariness of the relevant standards is explained, for instance, in terms of the
purposes the evaluated item serves (Mackie 1977, 16 f.).
12 More correctly, evaluation is triadic R(x, y, V): x evaluates y with regard to value V.
13 In the following, I will use the notions “reflexive” and “reflexivity” for R(x, y) featuring x = y,
without specifying whether x = y holds necessarily or contingently for R. I take it that, logically,
evaluation is a nonreflexive relation. One could argue, however, that if a mental state ψ’ is directed
to its own subject S it is a state sui generis and cannot be modeled on mental state ψ directed to an
object different from S. It is a highly controversial issue whether, for instance, there is a difference
of kind or in kind between interpersonal deception and self-deception (Dupuy 1998, Mele 2009).
itself in a loop-like movement. Self-repairing flight control or self-cleaning ovens

are examples of systems displaying a reflexive structure. In linguistics, reflexivity is
realized, for example, in self-referential sentences such as “This sentence has five
words”. In philosophy, a main interest in reflexivity bears on the specific faculty of
human beings to reflect their own mental attitudes and states. The specific reflexivity
of intentional agents is often called “reflective” in order to distinguish it from mere
reflexivity such as self-locomotion or self-reproduction. “Reflective” in this under-
standing means that a thing is “bent back” on itself not only in reflexively structured
activity – like in moving or reproducing itself – but in a way that allows (1) taking the
ongoing activity as its own intentional object, and (2) apprehending that the source
of the ongoing activity and the objectualizing subject are identical. Often, the term
“reflective” is also meant to qualify the cognitive or, more specifically, the doxastic
nature that the reflexivity of intentional agents is supposed to exhibit (Dohrn 2008).
Generally, theories of “self-awareness”, “self-consciousness” and “self-knowledge”
are intended to explain what forms or modes such reflective reflexivity takes, and
to what extent it requires intentionality. A further central interest of these theories
bears on the role of first person concepts – usually limited to the singular first person
pronoun “I” – in conditioning reflective reflexivity.
The human capacity for reflectively apprehending oneself has led not only to a
wide variety of conceptions of this capacity but also to the reification of the struc-
tural concept “self”. “The Self” has been lexicalized as a noun term and is commonly
used to denote the core of an individual, the distinctive and non-repeatable collec-
tion of properties making up a person.14 In many contexts, “the Self” seems to figure
as the secularized substitute of “the soul”, suggesting a more scientific account of
what makes humans so remarkable a species. The tendency to denote the capacity
for reflective reflexivity in terms of a doxastic ability, such as “knowing” oneself,
suggests that this capacity is to be conceived as providing knowledge in the sense of
evidence-based true beliefs about a specific object in the world. This in turn strength-
ens the case of a reified self toward which a subject turns her epistemic attention so
that it becomes a proper object of research, just as the cosmos or the foundations
of logic become objects of her epistemic endeavour. I will not take issue here with
the question of the appropriateness of the reification of functional terms such as
“self”, “I”, or “identical”. For the present purpose, it is sufficient to mention that the
concept of self-evaluation does not imply a specific ontology of the human mind,
and in particular no reified self. For pragmatic reasons, we may agree that using the
14 A systematic and historically-informed discussion of views on the reference of “self” can be

found in Diggory 1966, chap. 1. Diggory argues against a reified self, but for an objective refer-
ence of the word “self”. According to him, the lowest common denominator of the meaning of
“self” is “the relation between the agent and the object of his action in which agent and object are
the same organism. The aim of this definition is to scrape away from the word “self” the confusing
encrustation of affective connotations” (op. cit., vii). On the other hand, Diggory holds that “with-
out an objective definition we could hardly hope to study self by the most powerful investigative
methods [. . . ] of experiment” (ibid.).
noun term “(the) self” knowingly to abbreviate the phrase “person x as reflectively
apprehended by x” is justifiable in a number of theoretical contexts.15
Notwithstanding the formal difficulties the term “self” implicates, it is commonly
agreed that reflective reflexivity confronts analysis with serious problems such as,
for example, the paradox of self-defeating intentions in purposeful self-deception.
On the other hand, the specificity of reflective reflexivity is often spelled out in the
positive term of a “privileged access” which yields unquestionable knowledge of
oneself. Obviously, there are serious objections to the claim of an unquestionably
privileged epistemic mode and its consequences for a coherent epistemology. And
there are, in response, various sophisticated trials to save the intuition of a privileged
access. The discussion and the literature on these topics are proliferating.16
In line with these observations I propose to reconsider reflective reflexivity in
terms of evaluative rather than doxastic abilities. Doxastic activity is knowledge-
conducive and, as such, evaluative when it assesses the truth value of propositional
thought. Given that doxastic activity is an instance of evaluation, the turn from a
doxastic to an evaluative approach of reflective reflexivity consists not so much in
shifting but rather in widening the focus – from explicitly doxastic evaluation to
other forms of evaluation. While in doxastic evaluation the truth of p is assessed
and asserted in a judgment “p” there are many varieties of evaluation that assess
values other than truth.17 Contrary to truth, values such as the beautiful or the good
are not bound to be borne by propositionally structured objects. Consequently, it
is at least plausible to hold that assessing these values does not necessarily involve
assessing semantic properties and syntactic structures. If this is true, aesthetical and
moral evaluations seem better modeled in terms of perception than in terms of belief
and judgment. In perception, we assess the natural properties of natural things and
artefacts, not the semantic and syntactic properties of propositions about them.18
Similarly, what is assessed in aesthetical evaluation are the aesthetical properties of
natural things and artefacts, and in moral evaluation the moral properties of actions
and persons. Furthermore, the truism that perceptual episodes are much richer in
information than the perceptual judgments they can yield seems to have its anal-
ogous counterpart in evaluative episodes. For instance, the aesthetical evaluation
proceeding when one is listening to a symphony involves not only directly auditory
perception, but also processes that seem best accounted for in the metaphoric terms
of haptic perception. The listener is “touched” by the sublime, the glorious, etc. and,
15 “For a time during this century, attempts were made to remove the word self from the vocabulary
of serious scientific researchers because of its inferential and unmeasurable status. [. . . ] For the
tough-minded [. . . ] it was observable behavior that really mattered. But the concept of self would
not vanish, and it now appears – usually in hyphenated form with some other word such as esteem,
efficacy, regulation, regard – in psychological literature with greater frequency than ever before”
(Ogilvie and Clark 1992, 186 f.).
16 Gertler (2003) provides a good overview.
17 The questions of value-rating or value-monism shall not concern us here.
18 This claim is neutral with regard to the question whether perceptual episodes are propositionally
structured. It sides, however, with ontological realism against constructivism.
accordingly, “grasps” these values or gets “in touch” with them. And like a percep-
tual episode, her evaluative experience of the music’s subtleties is by far richer in
content than what ensuing evaluative judgments can bring to the fore. These charac-
teristics of aesthetical evaluation make it at least plausible to think that the ongoing
evaluative activity is different from the properly doxastic activity that makes value
judgments such as “This music is jubilation” or “This wine blends the flavors of
dark chocolate and raspberry with the sensation of a soft blanket”. Similar con-
siderations apply to moral evaluation. In line with the foregoing, many theories of
aesthetical and moral evaluation account for the capacity to assess values in largely
manifold situations by drawing on the role of affective abilities, modeling the latter
on perception rather than on belief (Tappolet 2000, Prinz 2004, Döring 2007).19 One
variety of theories of this kind holds that “feeling” – as distinct from “emoting” –
is the basic “active” evaluative capacity, while a full blown emotion with its doxas-
tic concomitants is “reactive” to evaluative feeling (Mulligan 2009). Accounting for
evaluation in terms of coupling non-doxastic action and doxastically informed reac-
tion suggests that evaluation indeed requires resulting in an evaluative judgment. It
suggests, on the other hand, that evaluation is not exhausted in doxastic episodes.
The underlying view then seems to be that it is an essential feature of some non-
doxastic states either to yield doxastic states, or to endow ensuing judgments with
some sort of justification. Applied to self-evaluation, the former option allows con-
sidering reflective reflexive states as evaluative in a non-doxastic mode, claiming at
the same time that they are fully satisfied only after being articulated in the value
judgments of explicit self-knowledge.
If reflective reflexivity is genuinely evaluative, it is a process dependent on cer-
tain abilities by which the evaluator positions herself on a value scale. As in the case
of external object evaluation, self-evaluation can bottom out in value-judgments,
and, provided value judgments are eligible for truth-value assessment, thereby man-
ifest doxastic self-evaluation, i.e. self-knowledge. Now what exactly is at trial when
agent A evaluates A, i.e. herself? Three answers might be given to this question.
Firstly, evaluation may pertain to actions which A performed, or to thoughts she
entertained. Then, so it seems, A takes an observer’s perspective on episodic facts,
positions them on a value scale – e.g. purposefulness, originality – and eventually
charts the findings into her self-worth matrix. Similarly, A might evaluate her phys-
ical appearance by looking into a mirror and positioning the mirror image on a scale
of female beauty or attractiveness. This kind of self-evaluation pertains to “parts” of
the person that are, to some extent, external to her.20 Secondly, self-evaluation may
be immediate and non-intentional in that A’s experiencing her grim mood posi-
tions the state she finds herself in on the dark side of the mood scale. Similarly
for certain of A’s bodily states whose experience alone positions A on a scale of
19 In contrast, doxastic theories of emotions hold that emotions are “judgments of value” (Solomon
2003a, b, Nussbaum 2001).
20 The term “part” does not involve a commitment to an explicit mereology of the personal unit. It
could be replaced by terms such as “aspect”, “constituent”, “moment” etc., each of them trailing
its own theoretical load.
illness. Contrary to completed actions, propositional thought or objective images,

ongoing states such as a grim mood or stabbing pain are not likely to be detached
as “parts” the person can reflect upon. Registering one’s ongoing states might be
termed “weakly evaluative” since it resists a clear-cut distinction between perceiving
a fact and apprehending a value. This kind of self-evaluation pertains to immediately
experienced internal “parts” of the person.21 Finally, the self-evaluator might take
her entire person as the intentional object of evaluation. This version of the scope
or target of self-evaluation is at the core of most understandings of Socrates’ appeal
gnōthi seautón (“Know thyself!”). The appeal here is not so much for factual knowl-
edge in the epistemological sense of justified true beliefs about an item, its traits and
episodic states, as for apprehending what kind of person one is (consider the differ-
ent meanings of connaître vs savoir, of kennen vs wissen). This certainly includes
judgments about oneself, one’s actions and one’s thoughts, but it is not exhausted
by such judgments. Apprehension of the person one is involves evaluation in the
modes of emotional and conative appraisal. Furthermore, the appeal suggests that
evaluating oneself does not stop at mere comparison to values and registering the
findings. Rather, the appeal to the self-evaluator is to make herself fit to certain ide-
als on a value scale. Thus, Socratic self-evaluation includes both apprehending the
person one is and aiming at being the person one feels one ought to be.
The taxonomy of self-evaluation drafted here is admittedly quite sketchy and in
dire need of further elaboration. It brings to the fore, however, that self-evaluative
processes involve a range of different abilities and targets. In line with the fore-
going, it seems appropriate to consider the Socratic gnōthi seautón the most
relevant variety of self-evaluation, and therefore its philosophically central case.
Socratic self-evaluation reveals particularly well the main functional features by
which self-evaluation broadens the scope of reflective reflexivity over and above
self-knowledge:
i. Externality: To the extent that self-evaluation is reflective reflexivity that

involves the comparison of a person with a set of values, self-evaluation is
“external”: appropriate self-attribution of many value properties partly depends
on external factors in the evaluator’s environment;
ii. Globality: Contrary to reflecting subjective episodes such as first-order states –
a traditional role ascribed to self-knowledge – reflective reflexivity in terms of
self-evaluation evaluates primarily the person. Evaluation of episodes – be it
doxastic or not – is subject to the telos of person (or personality) evaluation;
iii. Motivation: Self-evaluation consists in the interplay of reflecting the actual per-
son and the intended person. It is not self-contained but aims at continually
evening out the present findings with some model pattern of values. Self-
evaluation is not in the first place a doxastically guided undertaking but a quest
for wisdom.
21 It
seems plausible to hold, however, that in virtue of the internal relations between such “parts”
and the personal “whole” immediate self-evaluation a fortiori pertains to the entire person.
Invoking externalism is a rough-and-ready way to express the view that self-

evaluation is not to be restrained to self-knowledge, as long as the latter is
understood in terms of true second-order beliefs about first-order mental states. The
externalism claim is not supposed to rule out a priori classical features of internal-
ism, such as the specificity of the first person perspective, or the possibility of a
contemplative “look” into oneself.22 Rather, it emphasizes the fact that reflective
reflexivity understood as self-evaluation is bound to involve subject-independent
entities, to the effect that the activity focuses on the relation between a person and
a value. In other words, however we conceive of the specificity of a person’s rela-
tion to herself, the person objectualizing herself in self-evaluative activity is doing
so by apprehending the relation between two distinct entities, a person and a set of
values (see also note 11 about evaluation being, in principle, a triadic relation).23
Being central to all three claims, the concepts “person” and “value” definitely call
for a deepened analysis that exceeds the limits of this essay. So far, all three claims
provide a provisional framework that manifestly needs further elaboration and crit-
icism. With respect to the concept of person, it is especially the assumption of a
coherent global unit – implicitly adopted by the philosophical perspectives consid-
ered so far – that requires being critically assessed. The challenge raised by some
empirical studies concerns self-evaluation in terms of attributing a stable personal-
ity which determines a person’s behavior and acting. Apparently, optimism about
the success of such self-evaluation is precipitate, since a person S’s relevant behav-
iors and actions are often determined by trivial situational features – such as being
late – which outweigh the influence of the presumed personal traits of S (Darley and
Batson 1973). The skepticism raised by so-called “situationalism” with regard to
the relevance and success of self-evaluation is a challenge to be met by a systematic
theory of self-evaluation (Doris 1998, Harman 2000, Sabini and Silver 2005).24 For
the time being, I shall now provide a brief overview of the role of affective states, as
well as of social implications in self-evaluation.
22 For a recent discussion of (non)compatibility between externalism and knowledge internalism

see, for instance, Pritchard and Kallestrup (2004) and Gerken (2007).
23 I do not consider here the possibility that the self-evaluator is a bundle of tropes – including
value tropes – reflecting on one of its constituents. A bundle view of persons plausibly rebuts the
externalist claim, but it carries along other problems such as the indeterminacy of the subject. I
admit that relating the main target of self-evaluation to the Socratic appeal “Know thyself!” is
somehow tendentious, since the account might seem implicitly committed to a classical substance-
attribute ontology of persons.
24 I guess that the conceptual framework of self-evaluation is better suited to meet the challenge
of situationalism than the conceptual framework of self-knowledge. Contrary to “evaluation”,
“knowledge” is not only an achievement term but also an explicit success term. Knowing oneself
implies that one is successful in making true judgments about oneself, while evaluating oneself
is not as tightly bound to success conditions. Given that evaluation is accounted for in terms of
experiential faculties and processes, it involves error and failure and can accommodate the latter in
the conception of evaluating oneself.
1.2.3 Affective and Social Dimensions of Self-Evaluation

Self-evaluating processes involve a range of faculties on a continuum from low-
level self-monitoring faculties up to sophisticated epistemic abilities providing
well-articulated self-knowledge. Affective states constitute an important part of self-
evaluating faculties. Think, for example, of essentially self-directed emotions such
as shame, remorse and guilt. Not only are these emotions self-directed but also
paradigmatically evaluative.25 In feeling guilt a person A evaluates herself with
regard to the justness of her acting towards person B, while in feeling shame she
evaluates herself with regard to certain other values she holds. Although emotional
evaluation is plausibly considered as resting on some stable cognitive attitudes –
e.g. conceptions of values, or beliefs about their range and importance – it is not
required to conceive of affective evaluation in general in terms of the doxastic
activity of judging. It has been argued that emotions, desires and some bodily sen-
sations share the common feature of being “felt evaluations”, i.e. “commitments to
the import of their foci” which precisely do not adopt a doxastic but an affective,
volitional or perceptive mode (Helm 2002, 19). Bennett Helm defines “felt evalu-
ations” in terms of “passive responses to attend to and be motivated by the import
of something impressing itself on us, responses that are nonetheless simultaneously
constitutive of that import by virtue of the broader rational patterns of which they
are a part and that they serve to define” (passim). The evaluation felt in an emotion
or in a desire is not manifest in an evaluative judgment about its focal object, but in
having “one’s attention gripped by the goodness or badness of something in such a
way that one thereby feels the pull to act appropriately” (op. cit., 20).
In a similar vein, various accounts of feeling underline the strength of primary
affective evaluation to reveal a specific value relation by the very nature of its
mental mode. One such account characterizes feeling as “being struck by value”,
and accordingly as a “type of episodic acquaintance [. . . ] although it is not any
type of knowledge that” (Mulligan 2009, 143). Contrary to full blown emotional
states – as conceived in most accounts of emotions – feeling in this sense of
affective acquaintance is not considered to be “reactive”, but, rather, to ground
“reactive attitudes” such as one’s dislike of a person, or the desire to avoid her
(op. cit., 144). Applying this view to self-evaluative affective states, one might
assume that A feels the indignity of her behavior toward B prior to (and/or inde-
pendently of) her judging that her behavior wronged B. Modeled on non-doxastic
accounts of perception, feeling can be construed as the means that immediately
acquaints its subject with properties of a certain kind. Whether and how such
25 Gabriele Taylor labels pride, shame and guilt as “emotions of self-assessment”. She defends,
however, a cognitivist account of self-assessing emotions, holding that the label applies to these
emotions because they “resemble each other in requiring the same sorts of beliefs”, i.e. the feeling
subject’s belief that s/he “has deviated from some norm” (Taylor 1985, 1). Hence, the evaluative
function of these emotions is clearly attributed to the doxastic element involved in the emotional
experience: “What is believed amounts to an assessment of that self” (passim).
acquaintance is conveyed into judgment and other components of full blown emo-
tions are questions that allow for different answers. One option is to say that value
judgments are reactions to feeling values. Another option is to say that a feeling
of guilt and a judgment of wrongdoing occur simultaneously as a matter of neces-
sity, perhaps mere nomological necessity (Lyons 1980, Ben-Ze’ev 2000).26 And still
other perceptual theories of emotions do not maintain a distinction between feeling
and emoting, but identify emotions with value perceptions (Tappolet 2000, Deonna
2006, Döring 2007).
The value revealing role of affective abilities figures among the most promi-
nent topics of recent research – both empirical and philosophical – on affectiv-
ity (Tappolet 2000, Charland and Zachar 2008, Goldie 2010, Mulligan 2010).
Investigations on the specificity of reflective reflexivity that take this aspect into
account will profit from the work done in these domains up to this day. This is
even more important with regard to the second characteristic role of affective states:
their motivating function (Brewer and Hewstone 2004, Gorman 2004). As outlined
before, the philosophically most relevant type of self-evaluation – Socratic gnōthi
seautón – assumes the double task of apprehending the person one is – up to a
given moment – and aiming at the person one evaluates one ought to be. Again,
this character of self-evaluation is realized in an exemplary way in the case of guilt.
Feeling guilt fulfils the double function of revealing to its subject the wrongfulness
of her behavior, and motivating her to engage in reparative action. Reparative action
motivated by guilt feeling is not exhausted by indemnifying the victim in the way
an insurance company pays compensation. Although this kind of making amends
is properly included, reparative action motivated by guilt feeling primarily aims at
restoring the disturbed relation between the wrongdoer and the wronged, as well as
re-attuning the wrongdoer to the kind of person she ought and wants to be. Both
their value-revealing and their motivating role make emotions valid candidates of
self-evaluative abilities in their own right. Including them in accounts of reflective
reflexivity enriches the latter by shedding light on structural and functional features
that cannot be accounted for in terms of mainly epistemic faculties.27
Previously I argued that self-evaluation is an externalist conception, provided
it assesses a relation between a person and a set of values, and the latter is
not a private entity. This second assumption, however, is not uncontroversial.
26 Ben Ze’ev defends a compositional theory of emotions according to which evaluation, motiva-
tion and cognition are the three components of the “intentional dimension” of emotions (Ben-Ze’ev
2000, 50). While the cognitive component “consists of information about given circumstances”, the
evaluative component “assesses the personal significance of this information”, and the motivational
component “addresses our desires, or readiness to act, in these circumstances” (passim). Ben Ze’ev
emphasizes, however, that he considers neither of these three components nor the feeling dimen-
sion of emotion as a “separate entity or state”: “Emotions do not entail the separate performance
of four varieties of activity” (passim).
27 Bennett Helm emphasizes that the evaluation felt in emotion and desire cannot be separated from
the motivation implicit in these states (“to evaluate in the way characteristic of desire just is to be
motivated to pursue it”), whereas “evaluative judgment can be dissociated from motivation” (Helm
2002, 15).
According to axiological emotivism, a variety of subjectivism, value statements do

not describe external facts but “express” the emotional attitudes of those making
them. Consequently, A’s value statement “Bashing people at random is mean”
expresses A’s negative feelings (or alternatively A’s believing that she has neg-
ative feelings) – e.g. resentment, indignation, guilt – in the face of X’s bashing
Y randomly. In this guise, emotivism is not only an internalist conception but seems
to be an instance of self-knowledge understood in terms of second-order beliefs
about one’s first-order states. One serious drawback of emotivism is the difficulty of
accounting for the fact that value statements can be wrong. Another is the apparent
non-existence of axiological disagreement. A brutal rowdy’s statement that there is
nothing foul with randomly bashing people strikes us as being wrong, even if bash-
ing people makes the rowdy indeed feel good and he cannot be mistaken about his
feeling. Similarly, it is difficult to accept that our disagreement about bashing people
is just a difference between our feelings in the face of it. To be sure, this is not a con-
clusive argument against axiological subjectivism in general. The debate between
subjectivism and realism about values is part of any basic attempt to account for
evaluation and deserves serious discussion. For the present purpose, however, I tend
to adopt value objectivism on the mere assumption that philosophically relevant
self-evaluation involves not only the diagnostic feature of locating oneself in a value
matrix, but also the attempt to make oneself fit the person one ought to be. And this
second feature seems to boil down to the first when value statements are made true
by the internal states of their subjects alone.28
Externalism of self-evaluation can be accounted for on the basis of either abso-
lute value realism, or of moderate realism taking personal values as supervenient
on intersubjective relations and practices. For the present purpose, I will assume
moderate value realism. Obviously, this assumption makes self-evaluation depen-
dent on social relations, even if only in an indirect way. Whether such dependence
should count as trivial or not is a subsidiary question for the present considerations,
since I do not intend to draw on this aspect of social involvement in self-evaluation.
What seems more promising is investigating the role of sociality in actual processes
of self-evaluation, to the effect that systematic dependences between social related-
ness and self-evaluative practices can be worked out, e.g. with regard to developing
and refining evaluative faculties, or abandoning certain self-evaluative activities.
One important claim concerning the role of the social in accounts of reflective
reflexivity is the general communitarian claim that the social is prior to the indi-
vidual. In this view, whatever makes up an individual person is conditioned by
the social relations holding in her immediate and broader environment. If this is
on the right track, self-evaluative activities must be internally related to the social,
especially when they are taken to be not only diagnostic of actual states but also
effective in adjusting the individual to the person one ought to be. Given the general
communitarian claim, the ideal person clearly is a result of social construction. Less
28 Objective self-evaluation is not to be confused with “gratuitous self-assertions” (Diggory 1966,

viii).
general claims concerning the role of the social pertain to more local phenomena,
for example to situations that typically elicit certain emotions. Think, again, of the
emotions of shame and guilt. Both are generally considered to be not only self-
directed but also social in nature. Feeling guilt responds to having harmed someone,
so it reflectively evaluates its subject in its role as one term of an interpersonal rela-
tionship. Feeling shame is most often conceptualized as essentially involving the
real or imagined “look of another” on oneself.29 In this conception of shame, the
feeling equally evaluates its bearer as part of an interpersonal relationship. Many
other evaluative abilities are essentially person-related attitudes – think, for exam-
ple, of love, contempt or respect – and, as such, social in nature. Often, these abilities
are also exerted reflexively to evaluate their own bearer. To what extent self-love,
self-contempt or self-respect carry over the social nature of their non-reflexive cog-
nates into self-evaluation is a question that cannot be answered by merely stating
that they are “derivative” forms of social attitudes. Investigating systematic relations
between reflexive and non-reflexive varieties of person related attitudes might yield
interesting insights into the structure and functions of reflective reflexivity, shifting
the focus of inquiry from epistemology (How can we know about ourselves?) to the
concerns of Philosophical Anthropology (Who are we as human beings?).
There is, in addition, another line to considering the role of the social for self-
evaluation. It takes issue with the ontology of the social, suggesting entities such as
“group agents” and “plural persons”. Collective agents or persons are supposed to
be constituted in virtue of specifically social commitments made in view of projects
requiring the joint efforts of individuals.30 Accounts that personalize group agents
take the latter as instantiating a kind of individual that has the property of reflective
reflexivity.31 Accordingly, group agents are thought to display self-evaluative atti-
tudes such as guilt (Gilbert 2002), or, by evaluation of the project the commitment
to which grounds their existence, to perform a genuine form of self-evaluation that
aims at justifying their very existence (Rovane 1998; Chapter 5 by Rovane, this vol-
ume). Obviously, one central question here is whether these forms of evaluations
can count as reflexive in the strong sense, i.e. as relations in which the evaluator and
29 For an account of what is constitutive and what is typical of shame and guilt see Deonna and
Teroni (2008). For a critical assessment of the claim that shame is essentially social see Deonna
and Teroni (Chapter 11, this volume).
30 Margaret Gilbert holds that a distribution of individuals will become a “plural subject” in virtue
of a specific normative act of “joint commitment” (Gilbert 2006, 144 f.). Being the ruling principle
of plural subjecthood, joint commitment requires that each member “must openly express his or
her readiness to be jointly committed with the relevant others”, and that all members “understand
themselves to have a special standing in relation to one another” such that “no individual party
[. . .] can rescind it unilaterally” (Gilbert 2002, 126 f.).
31 Philip Pettit explicitly augments the claim that “social integrates” are rational and intentional
entities to the claim that they exhibit institutional personhood, arguing that “rational unification
is a project for which persons must be taken to assume responsibility” (Pettit 2003, 185). Since
social integrates do display the behavior required for personhood – i.e. make avowals of intentional
states and actions and acknowledge them as their own – Pettit concludes “that they are institutional
persons, not just institutional subjects or agents” (ibid.).
the evaluated are strictly identical. Does a group agent have the abilities requisite
for self-evaluation? Or are these abilities restricted to the individuals that constitute
the group agent? If so, the reflexivity in question seems to be the relative reflexivity
of parts evaluating a whole they help constituting.
Many of the accounts emphasizing the role of the social in self-evaluation draw
on the observation that the first person perspective implied in reflective reflexivity is
expressible both in form of the singular pronoun “I” and the plural pronoun “we”. It
seems quite natural, then, to claim that self-evaluation can be exerted in either mode
of reflexivity, the singular or the plural. Self-evaluators are therefore often con-
ceived as beings that – while biologically individual – are endowed with genuinely
collective forms of intentionality (Searle 1995, 23–26; Tuomela 2006). Collective
intentionality so considered allows that individuals in self-evaluative activities eval-
uate either their “personal” or, directly, their “social” self. Different models of
collective intentionality promise to be fruitfully applied in explaining the interre-
lations between individually and socially held values implied in self-evaluation, as
well as between individual and social worth thereby apprehended.
References
Ben-Ze’ev, A. 2000. The Subtlety of Emotions. Cambridge, MA; London: MIT Press.
Bermúdez, J. L. 1998. The Paradox of Self-Consciousness. Cambridge, MA: MIT Press.
Biederman, I., and M. M. Shiffrar. 1987. “Sexing Day-Old Chicks: A Case Study and
Expert Systems Analysis of a Difficult Perceptual-Learning Task.” Journal of Experimental
Psychology: Learning, Memory, & Cognition 13:640–45.
Brandom, R. 2007. “The Structure of Desire and Recognition: Self-Consciousness and Self-
Constitution.” Philosophy & Social Criticism 33 (1):127–50.
Brewer, M., and M. Hewstone. eds. 2004. Emotion and Motivation. Malden, MA: Blackwell.
Charland, L. C., and P. Zachar. 2008. Fact and Value in Emotion. Philadelphia, PA: John
Benjamins.
Crane, T. 1992. “The Nonconceptual Content of Experience.” In The Contents of Experience:
Essays on Perception, edited by T. Crane, 136–57. Cambridge: Cambridge University Press.
Darley, J. M., and C. D. Batson. 1973. “From Jerusalem to Jericho: A Study of Situational and
Dispositional Variables in Helping Behavior.” Journal of Personality and Social Psychology
27:100–08.
Deonna, J. A. 2006. “Emotion, Perception and Perspective.” Dialectica 60 (1):29–46.
Deonna, J., and F. Teroni. 2008. “Distinguishing Shame from Guilt.” Consciousness and Cognition
17 (3):725–40.
Diggory, J. C. 1966. Self-Evaluation: Concepts and Studies. New York: Wiley.
Dohrn, D. 2008. “Epistemic Immediacy and Reflection.” In Epistemology and Emotions, edited by
G. Brun, U. Doğuoğlu, and D. Kuenzle, 105–24. Aldershot: Ashgate.
Döring, S. 2007. “Seeing What To Do: Affective Perception and Rational Motivation.” Dialectica
61 (3):363–94.
Doris, J. M. 1998. “Persons, Situations, and Virtue Ethics.” Noûs 32:504–30.
Dreyfus, H. 1992. What Computers Still Can’t Do. Cambridge, MA: MIT Press.
Dupuy, J.-P. ed. 1998. Self-Deception and Paradoxes of Rationality. Stanford, CA: CSLI
Publications.
Feldman, R. 1995. “In Defense of Closure.” Philosophical Quarterly 45:487–94.
Frankfurt, H. 1971. “Freedom of the Will and the Concept of a Person.” The Journal of Philosophy
68 (1):5–20.
Gerken, M. 2007. “Is Internalism About Knowledge Consistent with Content Externalism?”
Philosophia 36 (1):87–96.
Gertler, B. ed. 2003. Privileged Access. Philosophical Accounts of Self-Knowledge. Aldershot:
Ashgate.
Gilbert, M. 2002. “Collective Guilt and Collective Guilt Feelings.” The Journal of Ethics 6
(2):115–43.
Gilbert, M. 2006. A Theory of Political Obligation: Membership, Commitment, and the Bonds of
Society. Oxford: Clarendon.
Goldie, P. ed. 2010. The Oxford Handbook of Philosophy of Emotion. Oxford: Oxford University
Press.
Gorman, P. 2004. Motivation and Emotion. London: Routledge.
Harman, G. 2000. “The Nonexistence of Character Traits.” Proceedings of the Aristotelian Society
100:223–26.
Harnad, S. 1996. “Experimental Analysis of Naming Behavior Cannot Explain Naming Capacity.”
Journal of the Experimental Analysis of Behavior 65:262–64.
Helm, B. 2002. “Felt Evaluations: A Theory of Pleasure and Pain.” American Philosophical
Quarterly 39 (1):13–30.
Korsgaard, Ch. 2009. Self-Constitution: Agency, Identity, and Integrity. Oxford: Oxford University
Press.
Leary, M. R., and J. P. Tangney. 2003. Handbook of Self and Identity. New York: Guilford Press.
Lehrer, K. 2000. “Discursive knowledge.” Philosophy and Phenomenological Research 60
(3):637–53.
Lehrer, K., B. J. Lum, B. A. Slichta, and N. D. Smith eds. 1996. Knowledge, Teaching, and Wisdom.
Dordrecht: Kluwer Academic Publishers.
Lehrer, K., and N. D. Smith. 1996. “Knowledge, Teaching and Wisdom.” In: Lehrer et al. 1996,
3–17.
Lycan, W. G. 1986. “Tacit Belief.” In Belief: Form, Content, and Function, edited by R. J. Bogdan,
61–82. Oxford: Clarendon.
Lyons, W. 1980. Emotion. Cambridge: Cambridge University Press.
Mackie, J. L. 1977. Ethics. Inventing Right and Wrong. New York: Penguin.
Mele, A. R. ed. 2009. “Self-Deception.” Social Theory and Practice (special issue) 35 (1).
Mischel, T. ed. 1977. The Self. Psychological and Philosophical Issues. Oxford: Blackwell.
Moore, G. E. 1993. “Hume’s Theory Examined.” In Selected Writings, edited by T. Baldwin,
59–78. London–New York: Routledge.
Moran, R. 2001. Authority and Estrangement: An essay on Self-Knowledge. Princeton, NJ:
Princeton University Press.
Mulligan, K. 2007. “Intentionality, Knowledge and Formal Objects.” Disputatio II (23):205–28.
Mulligan, K. 2009. “On Being Struck by Value – Exclamations, Motivations and Vocations.”
In Leben mit Gefühlen. Emotionen, Werte und ihre Kritik, edited by B. Merker, 141–61.
Paderborn: Mentis.
Mulligan, K. 2010. Emotions and Values. In: Goldie 2010, 475–500.
Nussbaum, M. 2001. Upheavals of Thought: The Intelligence of Emotions. Cambridge: Cambridge
University Press.
O’Brien, L. 2007. Self-Knowing Agents. Oxford; New York: Oxford University Press.
Ogilvie, D. M., and M. D. Clark. 1992. “The Best and Worst of It: Age and Sex Difference in
Self-Discrepanies Research.” In Self-Perspectives Across the Life Span, edited by R. Lipka and
T. Brinthaupt, 186–223. Albany: SUNY Press.
Pettit, P. 2003. “Groups with Minds of Their Own.” In Socializing Metaphysics, edited by
F. Schmitt, 167–95. New York: Rowman & Littlefield.
Prinz, J. 2004. Gut Reactions: A Perceptual Theory of Emotions. Oxford: Oxford University
Press.
Pritchard, D., and J. Kallestrup. 2004. “An Argument for the Inconsistency of Content Externalism
and Epistemic Internalism.” Philosophia 31 (3–4):345–54.
Rovane, C. A. 1998. The Bounds of Agency: An Essay in Revisionary Metaphysics. Princeton, NJ:
Sabini, J., and M. Silver. 2005. “Lack of Character? Situationism Critiqued.” Ethics 115:535–62.
Searle, J. 1995. The Construction of Social Reality. New York: Free Press.
Solomon, R. 2003a. “Emotions, Thoughts and Feelings: What Is a ‘Cognitivist Theory’ of the
Emotions and Does It Neglect Affectivity? In The Philosophy of the Emotions, edited by
A. Hatzymoisis, 1–18. Cambridge: Cambridge University Press.
Solomon, R. 2003b. Not Passion’s Slave: Emotions and Choice. Oxford; New York: Oxford
University Press.
Sosa, E. 2009. Reflective Knowledge (=Apt Belief and Reflective Knowledge, vol. II). Oxford:
Clarendon Press.
Sgaravatti, D., and E. Zardini. 2008. “Knowing How to Establish Intellectualism.” Grazer
Philosophische Studien 77:217–61.
Stanley, J., and T. Williamson. 2001. “Knowing How.” The Journal of Philosophy 98 (8):411–44.
Tappolet, C. 2000. Emotions et valeurs. Paris: PUF.
Taylor, C. 1977. What Is Human Agency? In: Mischel, 103–35.
Taylor, G. 1985. Pride, Shame and Guilt – Emotions of Self-Assessment. Oxford: Clarendon Press.
Tesser, A. 2003. Self-Evaluation. In: Leary & Tangney, 275–90.
Tuomela, R. 2006. “Joint Intention, We-Mode and I-Mode.” Midwest Studies in Philosophy 30:
35–58.
Tye, M. 2009. Consciousness Revisited: Materialism Without Phenomenal Concepts. Cambridge,
MA: MIT Press.
Part I
Evaluative and Self-Evaluative Attitudes
Chapter 2
How to Have Self-Directed Attitudes
Lynne Rudder Baker
Self-directed and self-evaluative attitudes are often connected to one’s social

position. Before investigating the dependence relations between individual self-
evaluation and social positioning, however, we should ask—What is required to
have self-directed attitudes in the first place? In order to be the subject of self-
directed or self-evaluative attitudes, I shall argue, an individual must have linguistic
and social relations. I’ll discuss the first-person perspective, self-concepts and their
acquisition – all from a radically nonCartesian, externalist point of view. This paper
will combine my work on first-person perspectives with my work on “content exter-
nalism” in the philosophy of mind in order to understand how someone can have
self-directed attitudes at all.
Having self-directed attitudes – attitudes about oneself – is a precondition for
making any individual self-evaluation. Self-directed attitudes and self-evaluative
attitudes – such as self-love, self-esteem and self-loathing – “are [in the words of
our organizers] often closely tied to the position one occupies within a network
of social relations.”1 Quite so. But before investigating the particular dependence
relations between individual self-evaluation and social positioning, there is a prior
question to answer: What are the conditions under which an individual can have any
self-directed attitudes at all? That is the question that I want to address here. Then,
I want to draw a moral about what it is to be a human person.
In order to be the subject of self-directed or self-evaluative attitudes, I shall argue,
an individual must have linguistic and social relations. Some self-directed attitudes
obviously require the subject to have linguistic and social relations. For example,
pride in one’s class rank requires comparison between oneself and others (as well as
having the concept ∗ class rank∗ .) By contrast, other attitudes of self-satisfaction
(such as one’s self-satisfaction in sticking to a healthful diet) do not obviously
1 In the words of the organizers of the conference on Self-Evaluation – Individual and Collective,
held at the University of Basel, Switzerland, January 16–18, 2009.
L.R. Baker (B)
Department of Philosophy, University of Massachusetts, Amherst, MA, USA
e-mail: lrbaker@philos.umass.edu

34 L.R. Baker
require one to have linguistic and social relations. However – as I shall argue –
obvious or not, self-satisfaction in anything (even in having a healthful diet) does
require that one have linguistic and social relations in a less specific way. Indeed,
anyone who has any self-directed attitudes has linguistic and social relations.
All self-directed attitudes are reflexive in that their object is always the person
who has the attitude. But some self-directed attitudes are individuated not only by
object (the person who has the attitude) but also by content. What makes pride in
one’s class rank and pride in one’s social standing different attitudes is that they have
different contents. Other self-directed attitudes, such as narcissism and self-esteem,
do not seem to require such specific content. But, as I said, my aim here is to show
that every self-directed or self-evaluative attitude requires that its subject have lin-
guistic and social relations – regardless of whether or not the attitude is individuated
by specific content (like pride in one’s social standing) or is more generalized (like
narcissism) and regardless of whether or not the attitude is tied to the position that
the subject occupies within a network of social relations.
2.1 Self-Directed Attitudes

I shall begin with some preliminary remarks about self-directed attitudes. First, by
“self-directed attitudes”, I mean attitudes toward oneself as oneself, not just attitudes
directed toward someone who happens to be oneself. Let me adapt an example from
Ernst Mach. Mach got on a bus at the back and saw at the other end an unkempt
fellow getting on the bus.2 Mach thought to himself, “How disgusting of that fellow
to be so unkempt and shabby.” Mach did not realize that he was seeing himself
in a large mirror that buses have to keep up with the passengers. His disgust was,
in fact, directed toward himself. But, as I am using the term, this is not a case of
self-disgust; Mach didn’t realize that he himself was the object of his disgust. His
disgust was directed toward someone who happened to be himself – but who he did
not recognize as being himself. So, his disgust was not self-disgust. To have a self-
directed attitude, one must realize in a first-personal way that the object of one’s
attitude is oneself.
One may think of oneself under any number of concepts expressed by, e.g., “the
person in the mirror,” or “the winner of the lottery”. But from a first-person per-
spective, one thinks of oneself as oneself, under a special first-personal concept, a
self-concept. A self-concept is the vehicle by means of which one thinks of oneself
from a first-person perspective. This does not imply that there is a special entity “the
self.” What one thinks of from a first-person perspective is oneself, a person. It is
the ability to think of oneself under a self-concept that makes self-directed attitudes
2 This example is adapted from one cited by John Perry (2002, 192–93). Perry used the example to
argue for an essential indexical generally. My interest is narrower. Nonliguistic animals have index-
ical knowledge, but if I’m right, they do not have self-directed attitudes. My interest is specifically
in the self-directed attitudes.
2 How to Have Self-Directed Attitudes 35
possible. There is an enormous range of self-directed attitudes – from self-disgust

to self-love, self-admiration, self-loathing, amour propre, and so on. Among these
attitudes, there may be complicated relations – as when, say, recognition of one’s
narcissism gives rise to self-loathing. To have a self-directed attitude of any sort,
one must be able to conceive of oneself from the first-person. To be able to conceive
of oneself from the first person is to have a self-concept that can be a constituent of
thoughts about oneself as oneself.
Second, self-directed attitudes are manifested in thoughts and actions. Let me say
a word about how I construe thoughts. Thoughts are individuated by content, and
their content is determined by the concepts that are deployed in the “that”-clauses
of their attributions. Concepts are individuated by their application conditions.
Application conditions determine what falls under the concept. If there is some x
such that concept C applies to x and concept C’ does not apply to x, then C and C’
are distinct concepts. So, application conditions determine the identity of concepts,
and the identity of concepts, in turn, determines the identity of thoughts of which
the concepts are constituents.
My self-concept applies to me and only to me, and can only be used by (as
opposed to being attributed to) me. Thoughts that manifest self-directed attitudes
contain, not only a self-concept, but also various empirical concepts. A man going
to a white-tie affair glances in the mirror and thinks, “Wow! I look great in formal
clothes.” He thereby expresses pride that requires that he have a concept of ∗ formal
clothes∗ . If he hadn’t had such a concept, he could not have had the attitude that he
had. This point generalizes to all self-directed attitudes that are individuated partly
by specific content. You take pride in something (such as your appearance in formal
clothes). You are disgusted with yourself about something (such as a mean-spirited
comment you made about a colleague). In order to take pride in something, you
have to have a concept of what you take pride in; in order to be disgusted with
yourself about something, you have to have a concept of what you are disgusted
about.
Even the global, generalized self-directed attitudes that seem to lack specific
content presuppose that their subject has other attitudes about herself that do have
specific content. In order to be self-interested, for example, one has to be able to
locate oneself, in a first-personal way, in a group of people with competing interests.
In order to have self-esteem, one must be able to recognize and conceive of a variety
of situations with oneself in them. For example, a person invited to a formal dinner
at the White House manifests her polished self-control only if she recognizes the
situation she is in – only, that is, if she has concepts that apply to political elites like
∗ president∗ , or maybe ∗ crème de la crème∗ . Any self-directed attitude requires that
the subject have a battery of empirical concepts. So, self-directed attitudes require
not only that the subject have a self-concept, but also that the subject have requisite
empirical concepts to give content to the attitude.
Third, although there are no self-directed attitudes without a self-concept, one
may have a self-directed attitude without recognizing that one has it – just as some-
one may be stingy, yet sincerely assert that she is generous. You do not have to be in
a position to assert that you love yourself in order to have the attitude of self-love.
36 L.R. Baker
One’s self-love may be manifest in one’s actions (e.g., always putting oneself first)
and attitudes (e.g., believing that one is more important than anyone else). One may
have the attitude of self-love without believing that one has it.
To summarize: (i) Self-directed attitudes are attitudes that require the subject to
have a self-concept, which refers to the subject in the first-person. An attitude that
just happens to have oneself as object (as in the case of Ernst Mach on the bus) is not
a self-directed attitude. (ii) A person’s self-directed attitudes are manifested in the
person’s thoughts and actions. Thoughts that manifest self-directed attitudes have as
constituents empirical concepts, as well as self-concepts. (iii) A person may have a
self-directed attitude without realizing that she has it.
My main argument for the conclusion that all self-directed attitudes require lin-
guistic and social relations is simple: Having self-directed attitudes at any time t
requires having a robust first-person perspective at t. Having a robust first-person
perspective at t requires having social and linguistic relations at t or earlier than t.
Therefore, having self-directed attitudes at any time requires having social and lin-
guistic relations at that time or earlier. (I’ll drop the references to time from now on.)
What I need to do is to defend the two premises of the following one-step
argument – I’ll call it “the simple argument”.
2.1.1 The Simple Argument

(1) If x has self-directed attitudes, then x has a robust first-person perspective.
(2) If x has a robust first-person perspective, then x has social and linguistic
relations.
∴ (3) If x has self-directed attitudes, then x has social and linguistic relations.
2.2 Defense of First Premise of the Simple Argument

According to the first premise, anything that has self-directed attitudes has a robust
first-person perspective. Spelling out what I mean by a “robust first-person perspec-
tive” will show that the first premise is a conceptual truth. A robust first-person
perspective is the ability to conceive of oneself as oneself – from the first person,
without any third-personal way of identifying oneself.
If one can entertain thoughts like “I fear that I’ll live to see the melting of
the polar ice caps”, then one has a robust first-person perspective. This is not
just a simple first-person thought. It is a complex thought: the subject refers to
herself in the first-person twice. The main verb is a psychological or linguistic
verb (“fear”, “believe”, “wish”, “say”, “wonder”, etc.) Following the main verb
is a subordinate clause in indirect discourse that makes a second first-person ref-
erence. This second occurrence of “I” expresses a self-concept by which one
conceives of oneself without having to identify oneself by a name, description or
demonstrative.
Any thought expressible by a sentence of the form “I believe (fear, hope, say,
etc.) that I am F” requires the thinker to have a self-concept and thus manifests a
robust first-person perspective. Since to have a self-directed attitude requires that
one have a self-concept and to have a self-concept is to have a robust first-person
perspective, it trivially follows that anyone with a self-directed attitude has a robust
first-person perspective.
The argument for the first premise of the Simple Argument thus boils down to a
hypothetical syllogism:
(1a) If x has any thought expressible by a sentence of the form “I believe (fear,
hope, say, etc.) that I am F”, then x has a robust first-person perspective.
(1b) If x has any self-directed attitudes, then x has thoughts expressible by a
sentence of the form “I believe (fear, hope, say, etc.) that I am F”.
∴ (1c) If x has any self-directed attitudes, then x has a robust first-person
perspective.
Thus, the first premise of the Simple Argument is true.
2.3 Defense of the Second Premise of the Simple Argument

According to the second premise of the Simple Argument, anything that has a robust
first-person perspective has social and linguistic relations. How does an entity with
only a rudimentary first-person perspective (a human infant) develop a robust first-
person perspective? As the infant learns to speak, she acquires more and more
concepts. Her repertoire of concepts increases dramatically as the toddler learns to
talk, and her expanded repertoire of concepts makes possible an expanded range of
thoughts. As the child moves from verbal behavior like “Want water” to the ability
to say, “I wish that I had some water”, she develops a robust first-person perspec-
tive. The ability to assert such things, and more generally to entertain thoughts of
the form “I believe/hope/fear/desire (etc.) that I am F,” requires not only that one
have a self-concept, but also that one have a store of empirical concepts that can be
used to think about oneself.
And by the time a toddler acquires a self-concept (between 18 months and 2
years), she already has a battery of empirical concepts that can be the constituents of
thoughts.3 By “empirical concepts,” I mean concepts that purport to apply to objects
and properties in the world. I have in mind everyday concepts like wet, fun, supper,
sleep, dog, home, water, star, hungry – concepts expressed by ordinary predicates.
Here is my argument for the second premise of the Simple Argument:
3 A self-concept is not acquired until after a toddler is able to recognize oneself in the mirror
(at about 18 months). This is supported by empirical evidence from developmental psychology,
together with casual observation.
38 L.R. Baker
(2a) x has a robust first-person perspective only if x has a self-concept.

(2b) x has a self-concept only if x has many empirical concepts.
(2c) x has many empirical concepts only if x has social and linguistic relations.
∴ (2d) x has a robust first-person perspective only if x has social and linguistic
relations.
By “having linguistic and social relations,” I mean being in a community with cer-
tain linguistic and social practices; what makes it the case that one is in a certain
community just are one’s linguistic and social relations to others who make up the
community.
The first premise, (2a), is a conceptual truth that I have already argued for when
spelling out the notion of a robust first-person perspective.
The second premise, (2b) is likewise conceptual: A self-concept is not a stand-
alone item. One cannot have a self-concept unless one can use it in thought; if one
can use a self-concept in thought, it follows that one is able to have thoughts about
oneself. Thoughts about oneself have empirical concepts as constituents. Consider
the thoughts expressed by “I’m afraid that I’ll be hungry” or “I wish that I could
see the stars tonight”. One cannot think of oneself from the first-person perspective
unless one can entertain thoughts with concepts that one applies to oneself as one-
self. To put it another way, one does not have a self-concept unless one has empirical
concepts to apply to oneself. So, the second premise, 2b, is a conceptual truth.
The third premise, (2c) is the one that I take to be controversial. I’ll say that
one “has” a concept if one is able to entertain thoughts that have the concept as a
constituent.4 If one is able to entertain the thought, ∗ Grass is green∗ , one has the con-
cepts ∗ grass∗ and ∗ green∗ . Admittedly, ∗ grass∗ and ∗ green∗ are both vague concepts,
in that the application conditions for each leave borderline cases. Nevertheless, there
are some conditions under which each definitely applies and some conditions under
which each definitely does not apply. To have a concept in the relevant sense, one
must be able to apply it correctly in a variety of contexts.
But, as we know from Tyler Burge’s work,5 one need not have complete mastery
of the conditions under which a concept is correctly applied in order to have the
concept. If having a concept required complete mastery of its application conditions,
most of us would have very few concepts: If you know that Bach wrote a lot of
fugues, then you have the concept ∗ fugue∗ even if you don’t know a fugue from
a toccata. If you believe that Smith died of a brain aneurysm, then you have the
concept ∗ aneurysm∗ even if you don’t know an aneurysm from an embolism. If you
do not have the concept ∗ fugue∗ , then you do not know that Bach wrote a lot of
fugues; and if you do not have the concept ∗ aneurysm∗ , then you do not know that
4 My notion of having a concept deliberately collapses Peacocke’s distinction between attribution

conditions and possession conditions of a concept: I hold that an attribution of a concept to x is
correct if and only if x satisfies its possession conditions. See Peacocke (1992).
5 Burge (1979).
Smith died of an aneurysm. Indeed, without the relevant concepts you do not believe
or even wonder whether Bach wrote fugues.6
So, one has a concept C if and only if one can entertain thoughts with C as a
constituent, and one can entertain thoughts with C as a constituent if and only if,
first, one is able to apply C correctly in a significant range of cases, and, second,
one has at least partial understanding of the application conditions of C. Of course,
there is vagueness in these latter constraints, but (as I have argued elsewhere) there
is vagueness in almost everything in the natural world.7
What I want to show is this: There is a large class of concepts, such that having
any of them requires that one be in a linguistic and social community – and hence
that one have linguistic and social relations. I’ll call such concepts “L&S concepts”
for “linguistic and social concepts.” (The notion of an L&S concept concerns the
having of a concept, not its individuation. A concept must already be individuated
before one can have it.)
Although most, perhaps all, concepts may be had without complete under-
standing, partial understanding is not crucial to L&S concepts. As Burge has
argued:
[E]ven those propositional attitudes not infected by incomplete understanding depend for
their content on social factors that are independent of the individual, asocially and non-
intentionally described. For if the social environment had been appropriately different, the
contents of those attitudes would have been different.8
Here is a somewhat crisper characterization of L&S concepts:

Concept C is an L&S concept if and only if: for any S, if S has concept C, then
holding constant S’s internal physical states and physical history, if S had been in
a community with relevantly different social or linguistic practices, S would have
failed to have had C.9
A person’s internal physical states are nonintentional states, and her physical his-
tory is the history of her nonintentional states and their nonintentional interactions
with entities in the environment.10
L&S concepts include not only the obvious social and legal concepts like
∗ passport∗ or ∗ taxation∗ , and not only the concepts that Burge discusses –
∗ arthritis∗ , ∗ brisket∗ , ∗ clavichord∗ , and ∗ contract∗ – but also concepts, the having
of which is not usually thought of as dependent on social and linguistic prac-

tices – in particular, natural-kind concepts. (I am not making the stronger claim
6 A fugue is a polyphonic composition with a short subject harmonized by rules of counterpoint. A

toccata is a composition for piano or organ characterized by full chords and running passages, and
often used as a prelude to a fugue. An aneurysm is a pulsating tumor caused by dilation or rupture
of the wall of an artery. An embolism is foreign matter (an air bubble or blood clot) carried in the
blood stream.
7 Baker (2007c, chap. 6), “Metaphysical Vagueness.”
8 Burge (1979, 84–85).
9 I discussed this notion under the label “social externalism” in Baker (2007a, b).
10 Say that a state is nonintentional if it can obtain in a world without entities with propositional
attitudes.
40 L.R. Baker
that all concepts are L&S concepts, although I suspect that it’s true.11 I believe that
Wittgenstein succeeded in showing that ∗ pain∗ is an L&S concept; if he did, then
it’s difficult to see what concepts would fail to be L&S concepts. But this larger
claim is beyond the scope of this paper.) I take it that Burge and others have shown
that many ordinary empirical concepts are L&S concepts; what I want to add is that
natural-kind concepts are L&S concepts.
Concepts, as I mentioned, are individuated by their application conditions.
Application conditions determine what the concept correctly applies to. The con-
cept diamond and the concept cubic-zirconium apply to different natural kinds.
Despite their superficial resemblance, diamonds and cubic-zirconia have different
chemical compositions – diamonds are composed of pure carbon, cubic z’s are com-
posed of zirconium oxide plus yttrium or calcium structures, and hence the concepts
∗ diamond∗ and ∗ cubic-zirconium∗ are different concepts.12 Although application
conditions are independent of social and linguistic practices, which concepts one
has is not independent of social and linguistic practices.
To show that ∗ diamond∗ is an L&S concept, I need to show that a person who
has the concept ∗ diamond∗ , and whose internal physical states and physical history
are held constant, would not have the concept ∗ diamond∗ in a community with rele-
vantly different linguistic and social practices. Suppose that Jill lives in a community
in which there are lots of diamonds and no cubic-z’s. Although she has not seen any
diamonds, she knows a great deal about them. She has read about diamonds and
talked with her schoolmates about them. She knows, for example, that diamonds are
minerals that are often used in engagement rings and in decorative jewelry, that they
are often given as gifts, and that they sparkle. Indeed, she knows that she wants a
diamond. Jill knows how to apply the concept ∗ diamond∗ in a significant range of
cases, and she has (perhaps partial) understanding of its application conditions. Jill
thus has the concept ∗ diamond∗ .
Now, without changing any of her internal physical states or any of the non-
intentional ways that Jill interacted with her schoolmates or with anything else in
her environment (including books and magazines), suppose that Jill had lived in
an alternative community in which there were no diamonds, but a lot of cubic-z’s.
Suppose that the word used for cubic z’s in the alternative community is a homonym
of the world used for diamonds in the original community. (So, it sounds as if the
same word is used in both communities.) In this alternative community, Jill has not
seen any cubic z’s, but she knows a great deal about them. She knows, for example,
that cubic z’s are minerals that are often used in engagement rings and in decorative
jewelry, that they are often given as gifts, and that they sparkle. Indeed, she knows
that she wants a cubic-z. Jill knows how to apply the concept ∗ cubic zirconium∗ in
11 Carol Rovane brought up the question of whether self-concepts are L&S concepts. I do not think
that self-concepts nonvacuously satisfy the constraint on L&S concepts. However, the having of a
self-concept does presuppose the having of empirical L&S concepts – even though a self-concept
is not itself an L&S concept.
12 On the other hand, sameness of chemical composition does not suffice for sameness of natural
kind. Graphite is also pure carbon, but with a different arrangement of atoms from diamonds.
a significant range of cases, and she has (perhaps partial) understanding of its appli-
cation conditions. In the alternative community, Jill thus has the concept ∗ cubic
zirconium∗ .
In the original community, Jill had the concept ∗ diamond∗ but not the concept
∗ cubic zirconium∗ ; in the alternative community, with no nonintentional changes
in Jill or in her interactions with entities in her environment, Jill had the concept
∗ cubic zirconium∗ but not the concept ∗ diamond∗ .13 Thus, the natural-kind con-
cepts, ∗ diamond∗ and ∗ cubic zirconium∗ , are L&S concepts. Similar stories can be
told for other natural-kind concepts.
Many, if not all, of the store of ordinary empirical concepts needed in order to
have a self-concept – and hence to have a robust first-person perspective and self-
directed attitudes – are L&S concepts; and by definition, L&S concepts require that
anyone who has them has linguistic and social relations. So, the third premise – that
x has many empirical concepts only if x has social and linguistic relations – is also
true. Since the argument for the second premise is valid and has true premises,
its conclusion is also true: If x has a robust first-person perspective, then x has
social and linguistic relations. So, the argument for the second premise of the Simple
Argument is sound, and the second premise of the Simple Argument, like its first
premise, is true.
Hence, the Simple Argument is sound as well. The Simple Argument is valid
and both its premises are true. The first premise of the Simple Argument is sup-
ported by an account of the robust first-person perspective: Anyone who has any
self-directed attitude has a self-concept, and to have a robust first-person perspec-
tive just is to have a self-concept. The second premise of the Simple Argument is
supported by an argument backed by a thought experiment. Anyone who has a self-
concept has a battery of empirical concepts; and anyone with the requisite empirical
concepts is in a community with certain linguistic and social practices. Since both
premises are true and the argument is valid, the conclusion of the Simple Argument
is true. Therefore, any entity that has self-directed attitudes has social and linguistic
relations.
2.4 A Moral About Individuals

Individual human beings – persons – are social entities in an ontological sense. In the
absence of communities, there would be no persons: Human organisms, perhaps, but
no persons, no individuals who could reflect on themselves as themselves. So, the
fact that there are individuals who can have self-directed attitudes at all presupposes
that there are social and linguistic communities.14
13 Noticethat my use of this example has nothing to do with linguistic deference to authorities, or
even with error. Jill is using the words properly in both communities; there is not mistake.
14 The question of the conditions under which a social and linguistic community exists and who
its members are is a thorny one that I cannot pursue here. I simply want to say that a community
42 L.R. Baker
Let me illustrate this moral by recalling a famous scene from Descartes’ first
Meditation. If my argument has been correct, then, if Descartes had been alone in
the world – as he imagined – then he could not have had the thoughts he had in his
first meditation. He would have lacked the linguistic and social relations required
to have the concepts – e.g., ∗ I∗ , ∗ fire∗ , ∗ dressing gown∗ – that were constituents
of those thoughts. My argument has the consequence that if Descartes’ had been
alone in the world, he would have lacked the concepts to entertain any self-directed
attitudes. He could not have wondered whether or not he was dreaming. He could
not have entertained the hypothesis that he was not sitting in front of the fire in his
dressing gown. A Cartesian entity – one that was alone in the world – could not have
self-directed attitudes at all.
The notion of a robust first-person perspective that is required for self-directed
attitudes and for self-evaluation that I have discussed is clearly nonCartesian.
Entities cannot even have robust first-person perspectives unless they have numer-
ous linguistic and social relations by which to acquire a store of empirical
concepts to apply to themselves. Many of these concepts, I have argued, are
L&S concepts. Consequently, I suggest that we dissociate the idea of the first-
person perspective from the Cartesian ideas of transparency, infallibility and logical
privacy.
If my argument is correct, then it is metaphysically impossible for any entity
that was truly alone in the world to have a robust first-person perspective or self-
directed attitudes. And nothing that lacked a robust first-person perspective could
have thoughts about himself as himself. So, Descartes’s resolution to regard himself
as having “no hands, no eyes, no flesh, no blood, no senses” is not a thought that
Descartes could have had if it had been true: The very fact that he had that thought
guaranteed that it was false. Solipsism is a philosopher’s fantasy.
Someone may object that all I have shown is that empirical concepts are, in fact,
social, not that they must be social. The best way for a pragmatist like me to show
that they empirical concepts must be social is to give a range of examples of empir-
ical concepts that surprisingly turn out to be social. For such a battery of examples,
see Baker (2007a, b). Of course, it is possible that there is a counterexample, but I do
not know of any. So, the conclusion about self-directed attitudes seems to have rather
global consequences. A universe that contains beings like us, with our self-directed
attitudes, must contain communities with linguistic and social practices.
is not a mere aggregation of independent atoms; nor does a community supervene on an aggrega-
tion of independent atoms. Persons are not independent atoms. The collectives that make possible
phenomena like nationalism, religious solidarity or schism, “ethnic cleansing,” and so on are like-
wise not aggregations of independent atoms. So, the moral is only a negative one: however we
understand collective self-evaluation, we cannot understand the relevant collectives in terms of
independent atoms.
2.5 Conclusion
What I have tried to do in this paper is to bring together two strands of thought that
I have developed separately over the past two decades – concerning the first-person
perspective and content externalism – and I have applied the combined result to
understanding the conditions for self-directed attitudes. I have aimed to make the
modest point that self-directed attitudes require robust first-person perspectives. A
robust first-person perspective, far from being a Cartesian perspective that isolates
individuals from communities, instead implicates communities.
The neuroscientist who discovered mirror neurons, Giacomo Rizzolatti, com-
mented that mirror neurons show “how bizarre it would be to conceive of an I
without an us.”15 Perhaps we can take this as empirical support for my thesis.
References
Baker, L. R. 2007a. “Social Externalism and First-Person Authority.” Erkenntnis 67:287–300.
Baker, L. R. 2007b. “First-Person Externalism.” The Modern Schoolman 84:155–70.
Baker, L. R. 2007c. The Metaphysics of Everyday Life: An Essay of Practical Realism. Cambridge:
Cambridge University Press.
Burge, T. 1979. “Individualism and the Mental.” Studies in Metaphysics (Midwest Studies in
Philosophy, 4), 73–121. Minneapolis: University of Minnesota Press.
Peacocke, Ch. 1992. A Study of Concepts. Cambridge, MA: MIT; Bradford.
Perry, J. 2002. “Self, Self-Knowledge and Self-Notions.” In Identity, Personal Identity, and the
Self, edited by J. Perry. Indianapolis, IN: Hackett Publishing Company.
15 Quoted in The New York Review of Books LX, no. 11, June 26, 2008, p. 65.
Chapter 3
Interpretation, Cause, and Avowal:
On the Evaluative Dimension of Selfhood
Axel Seemann
Implicit in Roderick Chisholm’s (1964/2002) argument in favor of the idea of

an “immanent cause” is a certain conception of self: as a cause of a particular
kind, a source of change whose powers are not themselves dependent upon other,
antecedent events. Chisholm presents this concept of causation and the experience
of ourselves as agents as intimately intertwined, and he draws on Reid to substan-
tiate the thought that our general grasp of causation is constitutively tied to the
experience of ourselves as having the power to bring about a variety of events. On
this account, we could not operate with the concept of a cause quite generally if we
weren’t in a position to enjoy the experience of making things go otherwise.
This view is as alluring as it is problematic. It is alluring because it satisfies a
deep intuition: that we can indeed make a difference, that we are not just cogwheels
in the vast machinery of a mechanistic universe. And it substantiates this intuition
by means of reference to an experience that, once again, we really do enjoy, and that
we are loath to classify as illusionary: the experience of agency. On the other hand,
the metaphysical implications of this view don’t come cheaply. One might suspect
that the conception of the self as an immanent cause is to be had only if one thinks of
selfhood along Cartesian lines: we may be forced to accept that agents in a genuine
sense stand outside the cause-effect relationships that, on one level of description
at least, govern the natural world. Faced with the stark choice between paying this
price and giving up on the belief that our agency is genuine, in Chisholm’s immanent
sense, naturalists will deem the price too high to pay: if a naturalistic worldview
entails the view that there is no immanent agency, most present-day philosophers
will, perhaps motivated by a strand of empirical evidence that goes back to Benjamin
Libet’s well-known experiments, reluctantly give up on both the idea that we can
“make things go otherwise” in Chisholm’s sense, and on the conception of self that
flows from it.
In this paper, I will suggest that this choice is a false alternative: that we are
not, in fact, always forced to choose between a view of ourselves as independent
A. Seemann (B)
Department of Philosophy, Bentley University, Waltham, MA, USA
e-mail: ASEEMANN@bentley.edu

46 A. Seemann
causes of events and a naturalistic worldview. It is important to take seriously the

qualification; I am not, in what follows, going to propose a grand universal solution
to the problem of agency. My claim is much more modest: there is one kind of case
in which we are, in fact, genuine agents – in which we exercise, through our doings,
causal control over a class of events without them being causally dependent upon
other events – even though accepting that we are such agents does not, in this case,
commit us to metaphysical extravagance. Considering this kind of case leads us to a
certain conception of self, which I shall develop in some detail. On this conception,
being a self amounts to more than just being the (passive) bearer of psychological
predicates; it amounts to being the author of some of one’s psychological states.
It requires us, in other words, to see ourselves as able to endorse aspects of our
psychological lives: to see ourselves not just as the epistemic subjects of our expe-
riences, but as standing in a relation to these experiences that allows us to modify
them. On such a view, we are indeed agents in a real sense – we have causal power
over some (though certainly not all) aspects of our mental lives. We accomplish this
by framing these aspects in ways that modify our experience of them. By present-
ing particular mental states to ourselves – sometimes through a flash of spontaneous
insight, sometimes through reflection, sometimes through external guidance – in
ways that allow us to integrate them into a larger view of ourselves, our past history
and our future projects, we change the experiential character of these states; and we
thereby change our view of ourselves. This conceptual activity, which we may, fol-
lowing a term of Richard Moran’s (2001), call an “avowal” of aspects of our mental
lives, is evaluative, as I shall suggest. And I will conclude that there hence is an
evaluative element to selfhood itself.
To develop this idea, I will introduce a number of scenarios in which persons
do modify their mental lives in the suggested way, and will interpret this activ-
ity by drawing on Moran’s notion of “avowal” (Section 3.1). I will argue that the
possibility of avowal depends on the correct individuation of a mental state, and
will claim that this individuation involves an interpretive process (Section 3.2). This
will require me to say more about the conception of self that has to be in place
for avowal to be possible; the suggestion will be that it has to involve, first, an
awareness of oneself as causally connected over time and, secondly, a conception
of oneself as a common cause (Section 3.3). I then turn to a body of work from
developmental psychology that, on my reading, supports Chisholm’s idea that the
capacity of common-cause thinking is tied to the availability of particular experi-
ences (Section 3.4). I will argue that the capacity for individuation which makes
avowal possible has an evaluative dimension (Section 3.5) and will end by sug-
gesting that the account presented in this paper supports a view of the self as
an independent cause of a variety of events that is metaphysically unproblematic
(Section 3.6).
3.1
Your psychoanalyst tells you (or perhaps leads you to discover) that you are angry at
the way in which you have been treated in your childhood. You had never entertained
the thought that you may be angry at your parents; it had simply never occurred to
3 Interpretation, Cause, and Avowal: On the Evaluative Dimension of Selfhood 47
you. Yet now that you have been made aware of your anger, you see that if you can
accept, as a fact about yourself, that you are angry at your parents, a lot of things will
begin to make sense: you are able to explain now why visits at their house always
disintegrate into a fight; you finally understand why telling your own children about
how you grew up is so difficult. But in order for the psychoanalyst’s assertion to
have this explanatory power, simply being told is not enough. Even your being able
to see that the psychoanalyst’s claim is based on good evidence is not enough. You
may be convinced that the psychoanalyst’s explanation of your behavior is correct,
that he is right in saying that you are angry, without being able to feel your anger.
The explanation doesn’t seem to pertain to you, in the way in which it would have
to if you were to fully acknowledge its force; it may as well be about someone else.
What supplies it with this kind of force is that you begin to experience the emotion:
that you actually feel it. Once there is this experiential dimension to the mental state
in question, you are able to integrate it into your psychological life. Now it is really
your anger that is at stake. It couldn’t be anyone else’s. And only if that is the case
are you in a position to (perhaps) do something about it.
It isn’t just through psychoanalysis that we can become authoritatively aware of
states we didn’t know we were in. Consider a case of David Velleman’s (2000) in
which you are meeting up with a friend. You haven’t seen each other for a while; it
should be a fun occasion. But the meeting starts going awry very soon; you find that
you are not treating your friend as friends should be treated. You are rude, incon-
siderate, and just not your usual self. All of a sudden it occurs to you that you have
come to the meeting with the intention to terminate the friendship: that’s why you
are behaving so disgracefully. You now see that you had planned to end the friend-
ship all along; you agreed to meet up with this end in mind. The intention now is
present to you from a first-person perspective, and it is this mode of presentation that
enables you to explain why you are in this state in a way that pertains, specifically,
to you. You may now see that you have felt exploited by the other person for a long
time; or that you are spending time with her out of a mere sense of obligation, that
you are not drawn to her at all. As before, what matters crucially for this kind of
explanation to become available is that the mental state in question is presented to
you as yours: that you actually experience it.
In both of these cases, different though they are in various respects, the same
question arises: what is it that enables you to not only see the reason why you should
be in the state in question, but to experience it as yours? This is not a trivial concern,
since much about your further attitude towards the state, including what (if any-
thing) you are going to do about it, depends on this. Reasons do matter, of course:
understanding the reason why you should be in a state may put you in a position
to endorse it. But it doesn’t cause such an endorsement. You can see that there are
perfectly good reasons for your anger and still remain unable to experience it in a
first-person way.
Richard Moran pursues something like this question when he introduces the
notion of an “avowal” of a mental state: the idea that one’s endorsement of a state of
mind makes one authoritatively aware of that state. Thus he writes, about a person
who is informed that she harbors a feeling of betrayal towards someone else:
48 A. Seemann
The person might be told of her feeling of betrayal, and she may not doubt this. But without
her capacity to endorse or withhold endorsement from that attitude, and without the exercise
of that capacity making a difference to what she feels, this information may as well be
about some other person, or about the voices in her head. From within a purely attributional
awareness of herself, she is no more in a position to speak for her feelings than she was
before, for she admits no authority over them. It is because her awareness of her sense of
betrayal is detached from her sense of the reasons, if any, supporting it that she cannot
become aware of it by reflecting on that very person, the one by whom she feels betrayed.
The rationality of her response requires that she be in a position to avow her attitude toward
him, and not just describe or report on it, however accurately, for it is only from the position
of avowal that she is necessarily acknowledging facts about him as internally relevant to
that attitude (say, as justifying or undermining it), and thereby (also) as relevant to the fully
empirical question of whether it remains true that she indeed has this sense of being betrayed
by him (Moran 2001, 93).
So there is, on Moran’s view, a connection between a person’s feeling of betrayal

towards someone else and the reasons that explain this feeling; only if she sees the
connection, will reflection on those reasons enable her to become authoritatively
aware of her state of mind. Her sense that there are good reasons to feel betrayed
puts her in a position to endorse that state. The avowal of a mental state can thus
be facilitated by an explanation of the occurrence of the state; by reflecting on a
range of facts, including, in the case at issue, facts about the person by whom she
feels betrayed. This explanation really is a causal explanation of sorts: to say what
caused your anger at how you were treated in childhood is to provide the right
kinds of reasons for experiencing that anger. Spelling out what caused your anger
provides you with an explanation of it: you can now see that it makes perfect sense
for you to be angry. If all goes well, if your sense of anger is not “detached” from
your sense of this explanation, the explanation has the power to change the way
in which the anger is presented to you. You finally feel the anger that, perhaps, had
been simmering undetected for decades. And I already said that what is so important
about this enlargement of the scope of your reflective awareness is that it puts you
in a position to take a stance towards your emotional state. You may now be able
to integrate it into your larger psychology. Your anger may motivate you to finally
discuss what happened in your childhood with your parents. Or it might lead you to
conclude that being in a permanent state of anger really isn’t helpful, and you may
try to overcome it. Hence we can say that connecting the explanation for being in a
mental state to one’s sense of the state itself puts you in a position to endorse it, or
not: it puts you in a position of authority vis-à-vis the state. And this, in turn, is what
enables you to become an author of a given aspect of your mental life; it makes it
possible for you to integrate the state into your larger psychological economy. But it
is not clear, as yet, what it is about a particular such explanation that makes avowal
possible. As we have seen, it isn’t the force of the reasons itself; you may see that
they perfectly explain why you should be angry and still not be in a position to
endorse your anger.
I think that in order to get a grip on this issue, one has to think about the indi-
viduation of the mental state in question. The peculiar thing about one’s capacity
for avowal is that it can pertain to states of which you didn’t initially have reflective
awareness. And it just isn’t clear what individuates such states. We cannot sim-
ply assume that this non-reflective part of our psyche contains a range of readily
conceptualized states that can, with or without psychoanalytic help, somehow be
summoned to reflective awareness. In the next section, I will be concerned with the
question of how to think about the individuation of pre-reflective aspects of one’s
mental life.
3.2
Sense-making begins in response to questions: about what one is doing or feeling, or
why one is doing or feeling it. You ask yourself why you are behaving so shamefully
towards your friend; the psychoanalyst tries to find out why it is impossible for you
to talk to your parents. You may make some sense of your doings by attempting
to trace your state of mind back to its causes, even if this doesn’t enable you to
endorse the state. Explaining your behavior from a third-person perspective – the
kind of perspective that you adopt when searching for explanations of someone
else’s behavior – may still be illuminating. But, on the account presented here, it
doesn’t enable you to engage in the kind of activity whose pursuit provides one
of the biggest motivations for asking a “Why”- or a “What”-question: it doesn’t
enable you to adopt a position of authorship over relevant bits of your mental life. So
avowal is sense-making of a very particular kind: it’s the endorsement of a state that
enables you to integrate it into your larger psychology; that puts you in a position
to create a psychological life which is well attuned to both your history and your
future projects.
One will, of course, not always be able to so integrate an avowed state of mind.
You may endorse the feeling of betrayal that your psychoanalyst pointed out you
harbored, or the intention to terminate your friendship that you discovered you enter-
tained – you may connect it to the events that brought it about – and yet not be able
to put it in a constructive perspective. Your anger might keep consuming you, and
the termination of the friendship might turn out to be unavoidable. This can be so
even if you are fully convinced that you should give up on the intention (your friend,
though not flawless, doesn’t deserve so drastic a treatment), or that your anger isn’t
helpful (it has been troubling you for years and really your own interests would be
much better served if you were able to move on). So it is important to stress that
“authorship”, in the sense I have been employing the term, doesn’t always result
from avowal; and even where it does, it won’t amount to total control over one’s
mental life – if “control” is the right term here at all. But we are sometimes able
to work on the mental states we avow; you may be able to put your anger to rest or
abandon your intention in the end. One can, on occasion, influence one’s mental life;
and this ability, when you have it, is contingent on your authoritative, first-person
awareness of the state at issue. No one else can overcome your sense of betrayal,
your disappointment, anger or pain for you.
How can we substantiate the idea of avowal – of making the explanation of being
in a given state matter? I started this section by pointing out that the endorsement
50 A. Seemann
of a mental state happens in response to particular kinds of questions about one’s

psychological life. I think it absolutely crucial to pay attention to this fact when
considering how one makes sense of the events in one’s mind, and indeed one’s
actions quite generally. There are a great many ways in which you may formulate
a “Why”- or a “What”-question about your thoughts, feelings, or intentions; and
your response will naturally be shaped by the question. Your psychoanalyst, in her
attempt at getting to the roots of your strained relationship with your parents, may
enquire about events in your childhood, or she may ask why some seemingly inno-
cent habit of your mother’s irritates you so much. She may, through asking these
sorts of questions, direct your attention towards a whole complex of psychological
facts about yourself that you hitherto hadn’t seen; you begin to develop a dim sense
of your psychological state, and what has brought it about. It may only be a very
dim sense initially, and in that case, much work will have to be done until you really
have a grasp of what went on all these years ago. Perhaps you vaguely sense, after
a while, an attitude you may describe as a feeling of resentment: but is that really
the best term? Mightn’t it be a kind of frustration, or is it, perhaps, anger? You may,
after much probing, find that it is indeed anger: and this is the moment when the
relevant part of your psyche comes to life. It is the correct individuation of the state
that puts you in a position to experience it.
A great deal depends, then, on the adequacy of your individuating concepts.
A close enough description may make available an explanation of your state that
strikes you as plausible. But for avowal to be possible, for the connection between
your sense of the state and the explanation of it to become vivid, you have to get
things exactly right. Only if the individuation is correct, does it pick up on your
psychology in such a way that you become authoritatively aware of your state. Only
then do you have genuine first-person access to your mental life.
Determining how to correctly individuate an aspect of your psychology is not
an easy task; sometimes it may not be possible at all. Describing one’s mental life
accurately is a complex process, which I think is best understood as an undertaking
that is essentially hermeneutic in character. You cannot understand your own, or
another’s, psyche (or at least not those aspects which aren’t in plain sight) without
engaging in an interpretation of sorts; a process of going back and forth between
behavior, history, and phenomenology. It really is a process of what Max Weber
(1904/2001) called “empathic understanding”, an Einfühlung, a “feeling-into” your
own or another’s mind. It is this process that the psychoanalyst tries to engage you
in when asking what a particular event in your past felt like, or why you think you
react in the way you do to certain triggers. What these questions aim at is to involve
you in a process geared towards arriving at the correct description of your state.
Readers familiar with the debate about the methodological roles of the notions
of “understanding” and “explanation” in the philosophy of social science may
object that they are mutually exclusive: if you think that a hermeneutic way of
sense-making is correct, you are committed to a denial of the viability of causal
explanation, and vice versa. But on the account at issue here, this objection is
untenable, at least as long as what we want to account for is a mental event. The
possibility of endorsing such events requires both understanding and explanation,
and the two are intertwined: in your efforts to arrive at the correct description of
your mental state, you will try out a variety of possible causal explanations; having
tried such an explanation, and found that it isn’t the right one (it doesn’t provide
you with the experience of a relevant bit of your mental life from a first-person
perspective), you will modify your description of your state. The point is that the
interpretive exercise by means of which you arrive at the correct individuation of
your mental state involves causal explanations. On this account, there is no opposi-
tion between interpretation and causal explanation: the correct individuation of parts
of your psychological life – the description that makes endorsement possible – rests
on both.
3.3
In the previous section, I elaborated on what is involved in the individuation of a par-
ticular aspect of one’s psychological life. In the present section, I will be concerned
with the conception of self that has to be in place for avowal to be possible. It should
be clear from what has been said so far that this self is not just the bearer of psycho-
logical predicates. To have self-knowledge of the kind that allows you to endorse a
mental state is to be operating with a far more substantive conception of self. It also
includes, minimally, your ability to highlight particular aspects of your psychology,
with all that involves: you need to be able to consider which way of describing your
state best matches your experience (if any), and which relates it best to antecedent
external events and your larger psychological life. That is the interpretive process I
have been concerned with. So you have to be operating with a rich understanding
of your past: you have to possess a grasp of yourself as connected over time, of
your later states as being causally dependent on your earlier ones, and of the way
in which these states in turn depend on external events. You also have to possess
a sophisticated understanding of your mental life, of how things currently are with
you and how you would like them to develop: the better your grasp of the bit of your
psychology you are seeking to individuate, the better your chances of arriving at an
adequate description of it. So a cross-temporal sense of identity is one prerequisite
of the kind of self-awareness that makes avowal possible.
But mere grasp of yourself as connected over time, however sophisticated, is not
enough. You could, in principle, possess an excellent understanding of how your
later psychological states are causally dependent on your earlier ones without being
able to endorse any particular such state. For avowal to be possible, I said, you
have to be in a position to interpret your mental life. And taking an interpretive
stance towards one’s psychology isn’t a spectator sport. It requires an active step; the
correct description of the state is not, at least not in the kind of case where avowal
matters for therapeutic or other purposes, going to be automatically available. It
really is you who has to do the interpretive work whose result, if all goes well, is the
correct description of your state. You have to be able to put yourself at arm’s length
from your psyche, so to speak: you have to take a perspective on it.
52 A. Seemann
What is the appropriate perspective to take towards one’s mental life? It matters
crucially, in settling this question, that perspective-taking, and the interpretative
work it involves, serves a specific practical purpose. Sense-making of the relevant
kind is something you do because there is a need for it: because you are unwell,
perhaps, or because you find that you are not in control of your actions; and because
avowal of your state is necessary for you to be able to do something about it. So the
appropriate perspective, the one you are striving for when approaching particular
questions about yourself, is not that of a detached observer: it is the conception of
yourself as a cause rather than a mere bearer (or, if things go really badly, victim)
of your mental states. This is what the interpretive perspective amounts to: in order
to be able to individuate relevant aspects of your psyche in a way that makes avowal
possible, you already have to take it that in so doing you will affect how things are
with you.
So the concept of self that has to be in place for avowal to be possible has
two dimensions: it includes an understanding of oneself as causally connected over
time, and as a common cause of many events. It is a picture that picks up on John
Campbell’s (1995) work on self-awareness, which takes these two dimensions to
be constitutive of selfhood. He thinks that to have an understanding of oneself as
causally connected over time is to understand, counterfactually, what the impact
would have been of various interactions with one’s surroundings (148). A parallel
consideration holds for the idea of oneself as a common cause: to be able to engage
in common-cause thinking is to be able to understand what the consequence will be
of a number of interactions with one’s environment. Campbell ties his discussion of
the common-cause dimension of self-awareness to Andrew Meltzoff’s (Meltzoff and
Moore 1977; for a more recent account see Meltzoff 2005) work on infant imitation.
Meltzoff argues that what enables infants to take up a caregiver’s observed facial
expression more or less immediately after birth is the capacity for cross-modal rep-
resentation of both the observed person’s and their own facial expression: the infant
is operating with a cross-modal body image. Further, he suggests that the adoption
of a facial expression will result in the experience of something like the mental
state associated with the expression; and since the infant’s adoption of the care-
giver’s facial expression will often result in a modification of that expression, it is
tempting to conclude that the infant begins to understand herself as a psychological
cause in early interactions with her caregiver. This aspect of her self-understanding
would then be dependent on a parallel understanding of others as such a cause: self-
understanding could thus not be solipsistic. And it would, in line with Chisholm’s
thought, be grounded in experience. In the next section, I am going to marshal devel-
opmental support for the idea that the conception of oneself as connected over time
and as a common cause of many events is, in fact, so grounded.
3.4
In the 1970s and 1980s, Colwyn Trevarthen (1980) argued, against mainstream
opinion in developmental psychology rooted in the work of Freud and Piaget, that
young infants had mental lives which qualified as independent, and that they could
be said to possess perhaps not yet a sense of themselves as psychological subjects,

but at any rate as being connected to the world through a particular body and hence
as occupying a particular position in space. At the core of his account are the notions
of “primary” and “secondary intersubjectivity” that are prominently defended today
by Vasudevi Reddy and Peter Hobson. Their work draws on the observation that
infants are able, almost immediately after birth, to engage with the facial expres-
sions of their caregivers. It is a matter of some dispute whether to conceive of these
early interactions as imitative, as Meltzoff suggests, or whether one ought to take it
that they already presuppose some understanding of the other’s embodied mind, as
Reddy argues. How one settles this question is not of immediate importance here,
since the one thing that is crucial in the present context is uncontroversial: infants of
a few months of age engage with their caregivers in a way that involves the sharing
of feelings. The kind of face-to-face smiling, chatting, and vocalizing which occurs
between young infants and their mothers expresses an attunement of their respective
mental lives. Such young infants are clearly capable of distinguishing between non-
human objects and other people (Gallagher and Hutto 2008); what is less clear is to
what extent one should take it that they are in a position to distinguish between self
and other. I am going to follow suggestions by Hobson (2005) and Reddy (2008),
who take it that the experiential dimension is of key relevance here: that the ability
to develop a conception of self depends on the availability of particular experiences.
Reddy writes:
It seems that the emotions typically called self-conscious are rooted in perceptions of the
other’s attention and emotion rather than in thoughts about the self. Since there is no evi-
dence that conceptualization of the self develops any earlier than 18 months, the direction of
effect proposed by cognitive-developmental theory must be challenged. The case made by
some developmental psychologists such as Carroll Izard and Peter Hobson that the so-called
secondary emotions are in fact the primary movers in the development of conceptualization
of the self seems more plausible (Reddy, 148).
So the idea is that perceptions of a particular kind – person perceptions, as one

may call them – enable infants to develop an understanding of self. I have pur-
sued this suggestion elsewhere at greater length (Seemann 2008); here, the very
broadest sketch will have to suffice. My account starts from the consideration that
certain ways of perceiving another person’s expressed feelings are available early in
infancy (and continue to be so available throughout life), and that they can be seen
as the ability to directly, non-inferentially access another’s bodily expressed subjec-
tive attitude (a smile, a frown, a coy avoidance of eye contact, a defiant look). This
capacity is bound to the particular epistemic conditions of mutual or joint atten-
tion: it is restricted to constellations in which the individuals in question stand in
a second-person-relation to one another. In such constellations, so the broad idea,
what I call “simple feelings” such as pain, pleasure, anger, or joy have an experien-
tial meaning that remains constant across a variety of modes of presentation. Thus,
a token instance of a feeling of the type “pain” may be experienced perceptually,
perhaps by seeing someone in pain: if you are standing in a second-person relation
to another person, her perceived expression of pain is going to be painful in its own
right. Support for this idea may be garnered from functional considerations: what
54 A. Seemann
makes a particular feeling an instance of pain is that it has particular behavioral char-
acteristics. Pain is awful, so you will want it to stop; and in order to do that, you will
engage in the appropriate kind of behavior – you will adopt what one may call an
“avoidance attitude” towards the experience. And this attitude is of the same kind
regardless of whether you experience an episode of pain yourself or whether you
perceptually experience another’s pain. This particular way of accessing another’s
subjective attitudes, then, is available without presupposing any conceptual grasp of
the self-other distinction. All it takes to get the account off the ground is the experi-
ential distinction between a perceptually and an interoceptively presented token of
a feeling that is nevertheless of the same type.
Of course infants, even very young ones, are not just static bearers of percep-
tual states: their relation with their caregivers is very much of an interactive kind.
Infants of a few months of age will engage playfully with their caregivers in ways
that suggest they are actually experimenting with the relation between their own and
the other person’s expressed mental states – they seem to begin, as Reddy puts it, to
“play with intentions” (150). And my suggestion is that we conceive of these playful
ways of engaging the other person’s attention, of trying out what reaction a particu-
lar act will provoke, as the beginnings of the grasp of the distinction between self and
other. The crucial observation here is that the infant is engaged in an interaction –
that she does things that affect her social environment. She smiles in order to see
whether this will elicit a smile in return: she is actively manipulating her sur-
roundings, which in this case include another person. She is establishing a grasp of
regularities in the impact of her doings on the other person: and I think it is in these
instances, early in life, that she is beginning to develop a grasp of herself (and of
the other person) as a common cause, as a subject of this interoceptively presented,
causally effective feeling. The important thing to see here is that this account of
the conceptual development of self-awareness involves an agentive dimension: you
could not, on this account, possess a concept of self if you didn’t have a develop-
mentally earlier sense of doing, of testing the environment’s reaction to your actions
and thus of establishing a first grasp of the causal impact of your behavior on the
other person.
This account, which bears some similarity to (though it is also in important
respects different from) considerations put forward by José Luis Bermudez (2000),
lends empirical support to Chisholm’s claim that the concept of oneself as a cause
is grounded in particular kinds of experiences: what allows you to think of yourself
as a cause is the experience of being able to manipulate your (social) environment
with a certain kind of regularity. Further, it is in tune with, though it also modi-
fies, the other dimension that I, following Campbell, took as a necessary component
of a grasp of oneself – the understanding of oneself as causally connected over
time. For it is only if you have a sense of your own and the other person’s states
as so connected that you can begin to make sense of yourself as a common cause:
understanding of self as a common cause involves a grasp of oneself as temporally
extended. So the developmental account of self-awareness lends empirical support
to the idea that to think of oneself as a common cause is based on the experience
of oneself as an immanent cause in Chisholm’s sense: the experience of affecting,
causally, how things are with one’s environment (which includes other people) and,
eventually, with oneself.
3.5
In this section, I am going to say more about the connection between the individ-
uation of a mental state and its bearer’s ability to avow it. In doing so, I will draw
on the key consequence of the developmental account of self-awareness presented
above: that experience is at the heart of the acquisition of self-understanding as a
particular kind of cause. I suggested, in Section 3.2, that the correct individuation
of a state involved seeing its causal history as pertaining, personally, to you. It was
this feature, I claimed, that supplied you with first-person access to relevant aspects
of your psychology. But what is it about the individuation of a particular bit of psy-
chology that makes this possible? To get a grip on this question, it will be necessary
to consider, once again, the general view of selfhood recommended here. To be a
self, I said, is to be able to think of one’s various states in terms of the two causal
dimensions mentioned earlier. Further, being a self in a full sense means to be in a
position to endorse aspects of one’s psychological life: to single out such aspects in
ways that allow one to see both how they are connected to their causal antecedents
and how they might be integrated into one’s psyche. To be a self, in this full-fledged
sense, therefore is to have the ability to take a reflective stance towards one’s own
psychological life and to impose a causal structure on it, with the aim of thus con-
necting it to the external environment and of creating internal coherence. You could
hence see the self, in line with a tradition that goes back to Aristotle, as a system
whose function it is to maintain a certain overall state of internal and external inte-
gration. To have a well-ordered psychological economy is to be in a state in which
the various aspects of one’s psychological life are arranged in a way that allows
the system to absorb both external impacts and internal modifications while main-
taining stability over time. If you are stable in this sense, external or psychological
events may affect you, but they won’t throw you off kilter; if you are maximally
stable, no such event will lead to your psychological disintegration. In other words,
if you are stable, you will be well-equipped to cope with demanding events in your
environment (deaths, job losses, breaks with people you care about) and with emo-
tional and other states (fears, desires, pains, intentions) that themselves are causally
related to such external events. And the requirement that this system be stable, so the
suggestion, equips candidates for avowal (be they emotions, intentions, or beliefs)
with their evaluative dimension. The stability of your psychological economy is one
individuation criterion for states that are candidates for endorsement: for avowal to
be possible, such states have to be presented to you in a way that makes vivid, in
their mode of presentation, their relevance for your stability. Stability of the system
of which the state is part is one yardstick, if you want, for its individuation. It is this
dimension that makes the thus individuated state a candidate for endorsement: it is
this feature of your awareness of such a state that results in its presentation to you
as yours.
56 A. Seemann
Take the example of your anger that you are told about by your psychoanalyst.
Anger is a state that quite clearly has an evaluative dimension, in the sense outlined
above: if you are angry, this will have an impact on your mental balance. It can
simmer, for years and years, without being singled out as such, and affect you in
all kinds of ways; it can make you feel restive, it can cause sleeplessness; it may,
depending on your character and your circumstances, provoke you to act in ways
you will later regret. To see the anger as yours, as a state whose existence you
acknowledge in more than an impersonal way, is to see precisely that: it is to see
it as a psychological item that has the potential to affect the stability of your psy-
chological economy. To see it in this way is to see it in an evaluative light. To be
angry may be a bad thing if it threatens to destabilize you; but it may also be a good
thing if it enables you to finally do certain things such as, perhaps, to approach your
parents about how you’ve been treated in childhood. And what enables you to see
it in this light is that you are able to individuate it in the correct way. If your anger
is presented to you as yours, as the state that has an impact on your psyche and is
hence a candidate for endorsement, you will be able to see it as the consequence of
what happened all these years ago: you will then be able to feel that what happened
did happen to you.
The implication of this account of self, as a system geared towards maintaining
balance by means of a causal structuring of its own internal states both with regard to
external and internal demands, is that you could not be in a position to avow a mental
state if there wasn’t an evaluative dimension to your thinking about yourself. So
much is implicit in the view that the self is a system of sorts. I am not claiming that
the account presented here amounts to more than a rough sketch. Still, it provides
an argument in support of the view that selfhood, developed along the lines of the
question of what it takes to avow a mental state, is inherently evaluative.
3.6
I started this paper with a reflection on Chisholm’s notion of an immanent cause,
and pointed out both what was appealing about it and what possible implications of
his train of thought had to be rejected from a naturalistic perspective. I said that we
could think of at least one kind of case in which we ought to conceive of ourselves
as genuine agents, as being in a position to causally affect how things are, without
having to adopt a Cartesian view of selfhood. The suggestion is that we do act as
immanent causes in Chisholm’s sense when we avow a mental state – when we
impose a particular kind of causal order on aspects of our psychologies, with the
aim of maintaining systemic balance. I said that to avow a mental state was to act
as a particular kind of cause: it was to affect how things are. To avow a mental state
was to bring about a difference in the way it is presented to you; avowal puts you
in a position of authority and, potentially, authorship with regard to the state. To
avow a state thus is to make a difference, and it really is making a difference: the
possibility of avowal requires an active step of the self.
On the account presented here, the possibility of avowal depends upon a suitable
interpretation of relevant aspect of one’s mental life: the correct individuation of
the mental state at issue enables you to see its causal connection to what happened
earlier. An adequate description, I said, was of an evaluative kind: it was of the kind
that made available to you an experience of the state as yours, and of that being a
good or a bad thing. It was this framing of the state that allowed you to see its causal
history in a way that really made its endorsement possible: it enabled you to see the
state, and its antecedents, as relevant to you – as relevant for the stability of your
psychological economy. I suggested that the activity of the self that makes avowal
possible is interpretive – it consists in singling out relevant bits of one’s psychology
in the right way. This interpretive effort is not caused by anything; it changes how
things are not through making things go otherwise (through bringing about a state of
affairs that would not have pertained had one not interfered) but through structuring
one’s psyche in a way that allows for the integration of the individuated state. The
self’s role, on this picture, is not to create events; it is to present them in a particu-
lar way. That is its genuine contribution. It is an uncaused contribution, and it can
be uncaused in a metaphysically harmless way: after all, what is at issue is not a
change in state of affairs but a change in how they are framed. But it is nevertheless
a causally effective contribution, since it has consequences, often very substantial
ones, for how things will go.
Acknowledgements I would like to thank the organisers of the conference “Self-Evaluation –
Individual and Collective” at which this paper was presented for a stimulating and thought-
provoking event. Thanks are due to Juliette Gloor in particular for her detailed and insightful
comments on an earlier draft.
References
Bermudez, J. L. 2000. The Paradox of Self-Consciousness. Cambridge, MA: MIT Press.
Campbell, J. 1995. Past, Space, and Self. Cambridge, MA: MIT Press.
Chisholm, R. 1964/2002. “Human Freedom and the Self.” Reprinted In Free Will, edited by
R. Kane, 47–58. Oxford: Blackwell.
Gallagher, S., and D. D. Hutto 2008. “Understanding Others Through Primary Interaction and
Social Practice.” In The Shared Mind: Perspectives on Intersubjectivity, edited by J. Zlatev
et al., 17–38. Amsterdam; Philadelphia, PA: John Benjamins.
Hobson, P. 2005. “What Puts the Jointness into Joint Attention?” In Joint Attention:
Communication and Other Minds, edited by N. Eilan, C. Hoerl, T. McCormack, and J. Roessler,
185–204. Oxford: Oxford University Press.
Meltzoff, A. N. 2005. “Imitation and Other Minds: The “Like Me” Hypothesis.” In Perspectives on
Imitation: From Cognitive Neuroscience to Social Science, edited by S. Hurley and N. Chater,
55–77. Cambridge, MA: MIT Press.
Meltzoff, A. N., and M. K. Moore 1977. “Imitation of Facial and Manual Gestures by Human
Neonates.” Science 198:75–78.
Moran, R. 2001. Authority and Estrangement: An Essay on Self-Knowledge. Princeton, NJ:
Reddy, V. 2008. How Infants Know Minds. Cambridge, MA: Harvard University Press.
Seemann, A. 2008. “Person Perception.” Philosophical Explorations 11 (3):245–62.
58 A. Seemann
Trevarthen, C. 1980. “The Foundations of Intersubjectivity: Development of Interpersonal and

Cooperative Understanding in Infants.” In The Social Foundations of Language and Thought,
edited by D. Olson . New York: W.W. Norton & Co.
Velleman, D. 2000. The Possibility of Practical Reason. Oxford: Oxford University Press.
Weber, M. 1904/2001. “‘Objectivity’ in Social Science and Social Policy.” In Readings in the
Philosophy of Social Science, edited by M. Martin and L. McIntyre, 535–46. Cambridge, MA:
MIT Press.
Chapter 4
Who Do You Think You Are? The How–What
Theory of Character and Personality
Federico Lauria and Alain Pé-Curto
As a matter of fact, we describe people by appealing to traits (e.g. shyness,

generosity, sincerity, and so on) in order to explain, predict and evaluate their gen-
eral behaviour. Moreover, there is a sense in which some of these characterizations
are deeper than others. Intuitively, when we try to “understand people”, including
when one tries to understand oneself, such deep characterisations seem particularly
relevant.
In this paper, we seek to explore this metaphor of personal layers by distin-
guishing between two types of traits, which we will call behavioural folds and
behavioural vectors. Behavioural folds are mere behavioural dispositions that per-
sons can instantiate (e.g. shyness as a disposition to remain in the background),
while behavioural vectors consist in attachments to certain values (e.g. sincerity as
attachment to truth or genuineness). On the one hand, it is claimed that behavioural
folds are global and optional, in the sense that they are neither specific (like habits
are) nor necessary (like breathing is). On the other hand, behavioural vectors are
global as well, but have values as their objects. It is suggested then that behavioural
folds capture more peripheral personal features, because they can be explained by
appealing to attachments to values, i.e. behavioural vectors. If this account is on the
right lines, the values one is attached to draw the boundaries of the very person one
is. If this is so, our understanding of persons is here specified in a fine-grained and
interesting way.
4.1 Three Stories, a Moral and a Proposal

As a starting point, let us begin with three stories illustrating distinct ways of
describing persons.
F. Lauria (B) · A. Pé-Curto

Department of Philosophy, University of Geneva, Geneva, Switzerland; Department
of Philosophy, University of Fribourg, Fribourg, Switzerland
e-mail: Federico.Lauria@unige.ch, Alain.Pe-Curto@unige.ch

60 F. Lauria and A. Pé-Curto
4.1.1 Isn’t She Lovely?

Here is a fairly common experience. Imagine that a gentleman called Sam is fond
of Maria. Maria is at first glance not the best person for a gentleman like him: she
speaks loudly, has no sense of politeness or etiquette; she is blunt and irritable, not
to say testy. Despite all of this Sam likes Maria very much. When asked why, he
invariably replies: “I know she is not easy to get along with, but when you get to
know her better, you realise she is one of the most lovable people on earth.” We
might have told the reader an overly sentimental story here and we have done so
probably under the influence of works of fiction: the rough person on the outside
with a sensitive heart on the inside is a movie cliché. However, we all seem to know
people like this in real life too, and we all meet people that are the reverse of Maria:
nice at first, but not such good people when you get to know them better.
4.1.2 The Capricorn Cup

One morning, one of the authors of this paper (let’s call him George) was waiting for
his coffee in front of the office espresso machine. An abandoned cup stood on the
machine, decorated with a stylized jumping goat. Underneath the picture, George
read the word “Capricorn” and the following list of traits: dynamic, persevering,
quiet, intelligent, pessimistic, stubborn and ruthless.
By the time his own cup of coffee was ready, George was rather puzzled. What
puzzled him was that despite this honourably long list, he still did not know what
kind of a person a Capricorn was supposed to be. He knew (according to astrology
at least) how a Capricorn was supposed to act and do things in certain kinds of
circumstances; but beyond that, he did not know much. What are Capricorns aiming
for in their lives? What do they like? What do they dream of? More importantly:
are the answers to these questions what George felt was missing in the description?
Would these answers help George get to know what kind of people Capricorns are?
4.1.3 Definitely Bovaresque

“Definitely bovaresque, I would say.” This is what Maria said, when her beloved
Sam one evening asked her to describe him in only one word. “So, you think I am
a bovaresque person?” says Sam, puzzled and slightly disappointed by this answer.
What did Maria intend to mean? Sam thought about Emma’s “heart” as depicted in
the story: the way Emma was charmed by Charles’ incanto and the way disappoint-
ment, contempt and infidelity succeeded love, until Emma committed suicide. “In
which sense am I like Emma?” thought Sam. Does that mean that Sam despises his
lover Maria? Or is “bovaresque” intended to mean that he has fallen in love with
another woman or has decided to commit suicide, like Emma did? Since Sam does
not entertain thoughts like these, Maria’s answer remains mysterious to him. “But
4 Who Do You Think You Are? The How–What Theory of Character and Personality 61
think about Emma’s inconstancy and capriciousness” says Maria laughing. “Think
about Emma’s slight dose of naivety or Emma’s day-dreaming episodes” says Maria
smiling. “Think above all of her strong interest in literature and her deep trust in
romantic love” Maria goes on to say. “Are you not really like her, Sam?” “Am I?”
answers Sam.
4.1.4 Moral
What do these stories reveal about persons and the way we describe them? In these
stories, we are provided with traits describing people (e.g. rudeness, pessimism, per-
severance, naivety). But intuitively some of these descriptions appear more relevant
than others in the task of understanding people. Consider the Isn’t she Lovely? and
the Capricorn Cup stories. Despite the appeal to usual traits, some other kind of
important information about the people described in these stories seems to be miss-
ing. For instance, when Sam contrasts some traits of Maria with some (unspecified)
features of hers, he feels that the latter trump the former. On the contrary, when Sam
is described as being bovaresque in the last story, we are provided with a descrip-
tion of him that constitutes at least a first step in the task of grasping who Sam
really is. Put metaphorically, we can say that some traits are deeper than others, in
the sense that they capture a person’s deep self, and that these traits contrast with
more superficial or, to use a term which is not negatively laden, external traits.1
Intuitively, when we try to really understand persons, including ourselves, we are
looking for the deeper characterizations. The aim of this paper is to explore this
metaphor of persons’ layers, by trying to distinguish between types of traits in a
non-metaphorical way.
In order to examine the distinction we have put forward, it would be handy to
have two terms distinguishing the group of the more superficial features from the
group of the deeper features. It may be that no common language terms perfectly
correspond to what we have in mind here. The most relevant words we can think of
are “character” and “personality”. However, intuitions about these terms vary across
languages and across speakers of the same language, the variety of intuitions of
English speakers on the matter being one notable example. So, at first glance, there
is no conclusive reason to use one word rather than the other for the set of super-
ficial traits and the same goes for the set of deeper traits. At the end of this article
(Section 4.4), we will suggest a matching between our terminology and common
language, but for now we will rely on terms of art. We will speak of behavioural
folds for the more superficial traits and of behavioural vectors for the deeper traits.
If behavioural vectors are to be more important than behavioural folds, no matter
how good or bad the latter are, then there must be some difference in kind at stake.
If we succeed in finding an interesting distinction between behavioural vectors and
behavioural folds, this assumption will be at least partly vindicated.
1 For a similar geometrical metaphor, see Scheler (1973).

4.1.5 A Proposal: The How vs. What Theory

In order to flesh out the distinction, we suggest that behavioural folds are about
how one does things, while behavioural vectors are about what things one does. A
metaphor may help here: think of two roads, which lead to a city, one that runs along
the coast and is bendy, and another road, which runs inland and is straighter. The
two roads lead to the same place, but they lead there in different ways: transposing
our terminology, we could say that the two roads share the same behavioural
vector but have different behavioural folds. A second metaphor would appeal to the
familiar distinction between style and content in writing. There is what one writes
and how one writes it. Our basic idea can then be summed up as follows. The
metaphor of depth at stake when distinguishing between more superficial features
and deeper features can be explained in terms of the distinction between the traits
determining how one behaves (behavioural folds) and the traits which determine
what one does or, so to say, drive one’s behaviour (behavioural vectors).2 Turning to
our first example, we can put our initial intuition as follows: Sam’s naivety describes
the way he behaves, how easily he trusts people; but it remains silent on what Sam
is really after, on what makes him this particular person he is. Pointing to his strong
interest in literature and fantasy or his deep trust in romantic love, the description
is deepened. In what follows, we shall thus try to spell out what behavioural folds
(Section 4.2) and behavioural vectors (Section 4.3) are. Finally, we shall address
the question of the relation between the two types of traits (Section 4.4).
4.2 Behavioural Folds

In both the Isn’t she Lovely? and Capricorn Cup stories, some behavioural folds
are suggested: Maria speaks loudly, has no sense of politeness or etiquette, she is
blunt, irritable and testy. A capricorn is supposedly dynamic, persevering, quiet,
intelligent, pessimistic, stubborn and ruthless. In the same spirit, Emma appeared
naive, capricious and inconstant. This provides us with a starting point, which we
will now use in order to clarify the nature of these traits. In a nutshell, we will
suggest that behavioural folds are behavioural dispositions (Section 4.2.1) which
are optional (Section 4.2.2) and global (Section 4.2.3).
2 For the sake of our exposition, we will use the phrase “determine what one does” as a
short-cut. This should not obscure the fact that behavioural vectors determine what one does in
a broad sense, including intentional actions, but also the adopting of mental states. See our notion
of behaviour in Section 4.2.1.
4.2.1 Behavioural Folds Are Dispositional

First of all, behavioural folds are stable features: we typically attribute traits such
as loudness, impoliteness, or pessimism to subjects who manifest these regularly
and through different contexts. A straightforward way to account for the stability
of behavioural folds is to conceive of them as dispositions. While many analyses
of dispositions have been suggested, we will, for the sake of simplicity, consider a
conditional analysis3 of them along the following lines:
S instantiates the disposition D to F if, and only if, S would F, under

conditions C.
This of course is only a first step in finding out what a behavioural fold might be.
It is obvious that although it might be necessary for a behavioural fold that it is
a disposition, it is in no way a sufficient condition: there are many dispositions
(e.g. our disposition to breathe) which are clearly not behavioural folds. Moreover,
the distinction between behavioural folds and other dispositions cannot be only a
difference in the kind of subject S that has the disposition (say a human being like
Maria) since the same kind of subject can have both behavioural folds (loud, blunt
etc.) and dispositions which are not behavioural folds (the disposition to breathe).
So what makes a certain disposition D a behavioural fold is not the fact that it is
possessed by a certain subject (i.e. S in the formulation above), but what it is a
disposition to do (i.e. F, the manifestation of the disposition). It is then by looking at
the kind of F’s at stake that we may discover what makes a disposition a behavioural
fold.
A first important point to emphasise about the manifestation of the disposition
is that it consists in behaviour, where by “behaviour” we mean something broad
enough to include intentional and non-intentional actions, but also certain attitudes,
emotions and feelings for instance. Recall that we said that Maria was irascible.
Yet when she is angry, she is not (at least not always) acting intentionally or even
acting at all. She sometimes is having (only) an emotional reaction which does not
constitute intentional action and may not qualify as an action at all. Along the same
lines, describing a Capricorn as being pessimistic might just amount to saying that
this person is prone to have certain beliefs (beliefs that things will go badly) rather
than others. Again, it is not clear that believing is an intentional action or even an
action at all. Behavioural folds thus seem to describe ways of behaving or types of
behaviour in a wide sense. But can we say more?
3 We ignore the well-known problem of such analyses, since we believe that it does not affect the
claims defended here.
4.2.2 Behavioural Folds Are Optional

One difference between our disposition to breathe and Maria’s irascibility is that
breathing is a natural disposition, one that, contrary to irascibility, all humans share.
Stated otherwise, necessary or almost inevitable global behavioural dispositions do
not count as behavioural folds; only optional dispositions do.
In order to flesh out the criterion of optionality, let us consider behaviour in which
optionality is absent. Think of sunflowers for instance. They have the disposition
to respond to the movement of the sun. Does this mean that sunflowers have the
behavioural fold of “heliotropism”? If the intuitive reply is negative, this is due to
the fact that all sunflowers follow the sun (unless the flower is sick or some other
circumstance prevents it from doing so) and each and every sunflower seems to be
programmed to do so just because of what it is to be a sunflower. One conclusion we
would like to draw is that behavioural folds are contrastive qualifications of one as a
member of a certain class of objects. To the extent that heliotropism is not a differ-
entiating feature within the class of sunflowers, it does not make sense to consider
that heliotropism is a behavioural fold. By contrast, in the three earlier stories and
in most everyday cases, behavioural folds are attributed to human beings or persons
as members of the class of human beings or persons. Thus conceived, attributing
behavioural folds to someone is to describe the way she behaves in domains where
she could have behaved differently as a human or person. By now, we can sum up
our account as follows:
BF1 : an S of the kind/type y has a behavioural fold x if, and only if, S has
∗
a disposition to behave in a certain way, say F, under conditions C, where
the disposition x is optional or not implied by the fact that S belongs to the
∗
kind/type y .
This result could be related with ease to the way we attribute behavioural folds:
often, they are used to explain or predict some person’s behaviour by contrast with
someone else’s behaviour or the most expected behaviour. For instance, if we are
aware of Maria’s irascibility we might predict or understand why she got very mad
at Sam for being ten minutes late rather than being slightly annoyed by the wait.
This also connects with the differentiating function of behavioural folds attribu-
tion. A behavioural disposition of a human being which is shared by (almost) all
human beings will not give us a differentiating feature, but specifying an optional
trait will differentiate the subject at stake and may help predict and explain her
behaviour.
There is a problem however with this picture, since there are other kinds
of non-necessary behavioural dispositions, for instance habits, addictions and
temperaments, which seem to be distinct from behavioural folds.
4.2.3 Behavioural Folds Are Global

Consider Maria’s habit of having a cup of coffee first thing in the morning. If
all behavioural dispositions were behavioural folds, this habit of hers would be a
behavioural fold. But drinking a coffee first thing in the morning is not the same
kind of property as being blunt and irritable. What is the difference between a habit
and a behavioural fold then?
The answer seems to lie in the distinction between a local and a global disposi-
tion. Contrary to Maria’s irascibility, Maria’s habit of drinking a cup of coffee first
thing in the morning is a very local behavioural disposition: it is a behavioural dis-
position to do a relatively limited variety of actions (those constituting the action of
drinking coffee and the various ways in which it can be performed), whereas Maria’s
irascibility is best described as a more global disposition.4 This suggests that
behavioural folds are essentially global dispositions. But what does this globality
consist in precisely?
We can distinguish three features of globality: a disposition is global if it is fairly
stable (1) through times, (2) through contexts, and if it is (3) multi-track, i.e. it
can lead to various kinds of outputs.5 Stability over time (2) is already granted by
the fact that behavioural folds and habits are both dispositional. Temporal stability
however is not sufficient for the criterion of globality that we are trying to cash out,
since there are temporally stable but very local dispositions, which do not seem to be
behavioural folds: for instance, Maria’s habit of drinking a cup of coffee first thing
in the morning when arriving at her office.6 There are also habits which are stable
across both times and contexts: for instance, Maria’s habit of having a coffee first
thing in the morning (independently of whether she is at the office or at home). This
suggests that behavioural folds are global, in the further sense of being multi-track
dispositions (3). It is reasonable to think that what distinguishes Maria’s irascibility
from her context-independent coffee-drinking habit is that the former is multi-track,
while the latter is not. There are many ways in which Maria’s irascibility can be
manifested, be it in her emotional life, her actions or her mere physical movements,
but Maria’s habit of drinking coffee first in the morning can only manifest itself by
Maria drinking coffee first thing in the morning. The multi-track dimension involves
a certain degree of indeterminacy: for a disposition to be multi-track, it must indeed
be determinable (cf. Ryle 1949 on determinable dispositions). This indeterminacy is
revealed through the fact that we can always ask how one exemplifies a behavioural
fold. If you are told that Maria is blunt, you can always ask in which ways she is
blunt or manifests her bluntness. If you are told that Maria has the habit of drinking
a coffee first thing in the morning, it makes no sense (or at least less sense) to
4 It may be that habits are not only local but also concern only actions, or even intentional actions.
The question of the nature of habits goes beyond the scope of this paper.
5 In contrast to single-track dispositions; for this terminology, see Molnar (1999, 198).
6 The contrast appears when we compare for instance the global trait of honesty with the local trait
of honesty-in-the-presence-of-lie-detectors or honesty-in-courtrooms. See Doris (2002).
ask in which way she manifests this habit.7 This clarification suggests thus that
behavioural folds are more global than the other kinds of behavioural dispositions
which we have considered above. If this is so, the following picture is on the right
lines:
BF2 : an S of the kind/type y has a behavioural fold x if, and only if, S has a
∗ ∗
global disposition to behave in a certain way, say F, under conditions C,
where the disposition x is optional or not implied by the fact that S belongs to
the kind/type y.
It should be noted though that a similar line of argumentation could be sustained

about addictions, jobs and hobbies. If someone is addicted to nicotine, is a surgeon
or plays the piano, this person has a behavioural disposition to F, where F is optional
(see Section 4.2.2). Does this mean that dispositions like these are behavioural
folds? We do not think so, because the relevant behavioural dispositions are far
more determinate than behavioural folds. This is revealed by the oddity of asking
in which way one instantiates the relevant addiction, job and hobby (e.g. “How are
you addicted to x?”).8 These distinctions among the degrees of globality, although
fuzzy and not clear-cut, seem to be important in our way of describing persons.
So far, we have described behavioural folds as optional and global behavioural
dispositions. In keeping with the how–what distinction outlined in the introduc-
tion, we can sum up our previous requirements by saying that behavioural folds are
optional and global behavioural dispositions to do things in certain ways. They are
optional and global adverbial traits. By specifying a Capricorn’s behavioural folds,
we have not said anything about what things a Capricorn will endeavour to do, but
7 Surely there are ways of instantiating the same habit (“drink coffee for breakfast at home/in a
café”, “drink coffee for breakfast listening to the radio/silently” and so on), but this variation in
style does not amount to a difference in the kind of behaviour.
8 Strictly speaking, it is always possible to ask how one instantiates an addiction, but in this case the
question is different in nature from the specification of the way one instantiates generosity or other
behavioural folds. For instance, one may consider that a person is slightly or badly addicted to nico-
tine. But in that case, the manifestation of the disposition is still restricted to the action of smoking
and there are no varieties concerning the very type of action at stake, since it remains smoking.
The variety is thus far more restricted in the case of addiction than in the case of behavioural folds.
One way to put this difference is by contrasting between, on the one hand, cases where varieties
concern merely the instantiation of a single-track disposition, and, on the other hand, cases where
the variety can concern the disposition as well as the manifestations of it, since in some cases the
disposition is multi-track. Addictions are dispositions of the first kind, while behavioural folds are
of the second. There is a last difficulty though. One could indeed think that it is possible to specify
an addiction by describing the manifestation of the disposition: Sam may be addicted to nicotine
in such a way that he never leaves his apartment without ensuring that a packet of cigarettes lies
in his pockets, and so on. However, in this case, the specification does not intrinsically concern the
addiction itself, but only specifies the necessary means that the action triggered by the addiction
requires, because the action in question is not basic.
we have an idea of how a Capricorn will do most of the yet unspecified things she
will do: she will do it with dynamism, perseverance and so on. This contrast will
become clearer if we turn to what drives behaviour, namely what we have called
behavioural vectors.
4.3 Behavioural Vectors

Our provisional conclusion about behavioural folds is that they are stable optional
and global ways of behaving. Now, if we go back to the three stories with which we
started our paper and think about our intuitions concerning the deeper traits i.e. the
behavioural vectors at stake, do we arrive at a different conclusion? Unfortunately
not. Remember the Isn’t she Lovely? story. Even though we do not know what Sam
actually means by Maria being a lovable person when you get to know her, it seems
that if the relevant behavioural vectors are to trump the negative behavioural folds
at stake, behavioural vectors must be at least as stable as behavioural folds.
Moreover, if we recall our suggestion that behavioural folds are about how one
does things whereas behavioural vectors are about what one does, behavioural vec-
tors seem to share the relevant globality by being stable across time and contexts as
well as multi-track. There is indeed a strong presumption that a behavioural vector
is the kind of trait that manifests itself in a variety of behaviours. The reason for this
presumption is that the idea of a deep self involves a limited amount of traits which
impact a broad area of one’s behaviour; a very local disposition, such as a habit,
would not qualify because it does not have this broad impact on one’s behaviour.
We will say more on this later.
Finally, if her behavioural vectors are what Sam likes about Maria and what
makes her one of the most lovable people on earth, behavioural vectors are clearly
optional; if the traits at stake were not optional and hence shared by all people, we
could not distinguish Maria from others in virtue of being lovable.
To sum up, all the features of behavioural folds which have lead us to conceive
of them as optional stable global ways of behaving are also features of behavioural
vectors. So we have not yet pointed out any distinction between the former and
the latter traits. What is missing is a clear specification of what it means to say
that behavioural folds are about how one does things, while behavioural vectors are
about what one does.
In the following section we shall try to make sense of behavioural vectors by
exploring various kinds of psychological states and will have a look at what we
deem to be one of the most influential theories of identity, namely Frankfurt’s. It
should be stressed upfront that the theory is not clearly tagged as a theory of identity.
It nonetheless constitutes a promising starting point for our purpose.9
9 Seeespecially Frankfurt 1971, 1976, 1992 and 2006. For interesting outlooks on Frankfurt
and his influence on theories of personality, cf. e.g. Schechtman (2004) and Velleman (2002).
On authenticity cf. Schechtman (2004), on identification cf. Moran (2002).
4.3.1 The Conative Theory of Identity

While Frankfurt’s account of personhood has developed over time, his proposal (or,
at least, one form of it) boils down to the following: the desires constitutive of one’s
identity are the desires with which one identifies through higher-order attitudes with
which the subject is wholehearted. Stated in more general fashion, there are certain
psychological states that constitute one’s identity. The criterion for selecting these is
the existence of higher-order attitudes which must in turn satisfy the further criterion
of wholeheartedness.10
This picture involving a first-order inner life which undergoes a process of selec-
tion has been very influential. For the purposes of the present paper, we would like
to focus on one assumption common to most of the relevant literature inspired by
Frankfurt, namely the strong emphasis on conative states as constituents of our
identity. The idea of seeing identity as constituted by a certain class of desires is
interesting because it points to a simple but insightful idea: identity is about what
one really desires or what one wants in a strong sense. If this is true, we might have
found here a substantial difference between behavioural vectors and the account of
behavioural folds we suggested above. If we define behavioural vectors in the way
Frankfurt sees identity, we may flesh out the distinction between behavioural folds
and behavioural vectors by pointing to the idea that behavioural folds drive one’s
behaviour because they would be identical or connected to one’s wholehearted cona-
tions, understood here as what one really wants or wishes. For instance, Sam can be
viewed as bovaresque since what he really desires is ultimately love and romance.
In brief, we find Frankfurt’s insight that one’s identity is to be found in what
one really desires, what one deeply wants, interesting. We nevertheless disagree
with the details of his proposal. Why focus solely on conative states in the first
place? Other kinds of elements come to mind as equally important to one’s identity:
affective states such as love of the beautiful and disgust of the ugly, strong interest
in literature, or even epistemic states such as believing that solidarity is important
all seem central. That being said, one might agree with the idea that non-conative
states matter for one’s identity, but still consider that conations are the kernel of
identity, because they somehow encapsulate or entail these other states. In the next
section, we will be considering an issue with this conative conception of identity
and our discussion will lead us to present progressively four important constraints
for states to qualify as components of one’s identity (Sections 4.3.2, 4.3.3, 4.3.4,
4.3.5, and 4.3.6.).
4.3.2 The Fragility of Desire and the Criterion of Resistance
One important problem with relying exclusively on desires11 is revealed by the

following reductio.
10 Onwholeheartedness cf. e.g. Frankfurt (1987).

11 We will speak of “desires” in very broad manner so it should be read as synonymous with
“conations”.
Death of Desire Argument
(i) For any desire D, if D is represented as satisfied (i.e. the subject represents that its content
obtains), then D ceases to exist12 . [death of desire assumption]
(ii) Suppose identity is essentially constituted by and only by desires.

[assumption about identity à la Frankfurt]
(iii) Suppose a powerful demon makes it apparent to Sam that all his desires are satisfied
[reductio hypothesis]
∴ Sam’s desires cease to exist. [given the death of desire assumption]
(C) Hence Sam’s identity ceases to exist. [given the Frankfurtian assumption about identity]
The argument leads to a consequence which seems absurd. Consequently, if the

death of desire assumption is true, then we should deny the assumption about
identity à la Frankfurt. Therefore, identity is not only constituted by desires.13
The Death of Desire argument hinges on the fact that one’s conative states dis-
appear if one believes them to be satisfied. In order to avoid this problem, it is thus
necessary to focus on states that do not cease to exist as soon as the subject believes
them to be satisfied (i.e. represents the obtaining of their content). Amongst these,
we find mental states of the imaginative, affective (e.g. emotions and sentiments)
and epistemic (e.g. beliefs, knowledge, memories) types. We take it that all these
states are non-conative mental states. To use another metaphor, states of the latter
kind enjoy a resistance that is absent in the case of conative states. Note however
that we do not intend to say that desires are irrelevant for one’s identity: on the con-
trary, they might be very important. Our point is only that they cannot constitute the
kernel of one’s identity and might be important in virtue of non-conative attitudes.
4.3.3 The Playfulness of Imagination and the Criterion

of Seriousness
What about imaginary states as essential features of one’s identity? Although imag-
inative states are not prone to the demon reductio, (because one’s imaginative states
do not cease to exist, if one believes their contents are actualised), the view that
imaginative states are essential components of one’s identity suffers from another
problem. Imaginative states are said to be “non-serious”, contrary to epistemic and
affective states. In other words, when you believe or feel something, you believe or
feel something about the world, and your belief and feeling can be adequate or not
12 Cf.Smith (1994), Humberstone (1992) and Kenny (2003).

13 Note that the point does not only apply to desires (in the limited sense of “desires”) and could
be extended to other conations since supposedly all conative states “die” when they are believed to
be satisfied.
to the actual state of the world. On the contrary, when you imagine, there is no such
constraint: as logic says, you are allowed to imagine anything including the very
opposite of what is the case.14 Because of that, we might speak of playfulness as
characterizing the non-seriousness of imagination.
Let us turn to our issue. If all you know about someone is that he imagines the
end of the world twice per day, you do not yet know anything about this person’s
identity from this very fact. You might certainly draw inferences from this fact: you
might guess that this person fears such an end-of-the world will soon take place.
But note that by doing so your attention is moving away from non-serious states to
serious states. In the same vein, knowing that someone very often imagines being
someone else might lead you to think that this person is somehow unsatisfied with
some features of hers; but again we have moved from imaginative states to serious
ones. Unless the subject’s imaginative states are accompanied by serious states, such
as fears or feelings of unsatisfaction, your beliefs are not warranted. If we try to
imagine a subject that has no serious states but only non-serious ones, it is indeed
hard to assign her any kind of identity: we do not know what the subject really thinks
is the case and how she really feels about it. This is why we believe identity cannot
be constituted only by non-serious states.15 Since non-serious states do not seem
to qualify as components of identity, we should then turn to serious states, except
conations which have already been put aside for the moment (cf. Section 4.3.2).
4.3.4 Emotions, Beliefs, and the Criterion of Normativity

This leaves us with epistemic states, like beliefs, on the one hand and affective
states16 or dispositions like emotions, moods and sentiments on the other. How are
we to choose between the two? There is a strong intuition that affective states are
more relevant to identity than epistemic states. Consider Sam’s passion for litera-
ture and contrast it with his belief that it is raining. Intuitively, the former states an
important fact about his identity, while the latter is silent about it. Does this amount
to saying that epistemic states are to be excluded from identity?
No. Although the intuition at work has some truth in it, we believe epistemic
states should not all be excluded from identity. Suppose Maria believes that solidar-
ity is a good thing. It seems that, contrary to the previous belief, this belief can count
as a component of Maria’s identity contrary to the previous one. What is the relevant
difference between the two? The difference between the belief that it is raining and
14 According to standard accounts of imagination, this is because imagination is a case of pretence

or make-believe (-desires, and so on), as in some childrens’ games, and hence is not a serious
attitude. In this paper, we will assume this view of imagination and conceive of its non-seriousness
along these lines. We owe terminology to Kevin Mulligan.
15 This might be related to one of our initial intuitions: remember that Emma’s tendency to imag-
ine or daydream appeared as an outer feature of hers, at least taken as such. If our treatment of
imaginative states is on the right lines, this might give an explanation of this intuition.
16 For the sake of this paper, we assume affective states are resistant in the sense specified above.
the belief that solidarity is a good thing is that the former belief has non-normative
or purely factual content while the latter has normative content.17 This is why the
second belief is a better candidate for identity than the first. If this is so, what does
the job here is not affectivity itself but the normativity implied by it.18
Suppose someone, let’s call him “Spock”, is capable of having beliefs with
non-normative content but is totally blind to values and norms: does Spock have
an identity of its own? Confronted with a splendid landscape, Spock could know
everything factual there is to know about it, he could even learn to mimic admi-
ration, but he would not be able to understand or have direct access to the beauty
of this landscape. More importantly, he would not be able to prioritise values and
norms: he could not personally prefer liberty over equality and beauty over charm
(or the other way around).
Unless one allows oneself to guess and draw conclusions about one’s relation
to the normative, about what one values and disvalues, about which norms one
sees as more and less important, it seems to us that epistemic states with non-
normative content do not give substantial information about one’s identity. If this is
so, it appears that beliefs with non-normative content are not part of one’s identity,
although beliefs with normative content are.
So far, we have accepted that identity is related to conative states, affective states
as well as epistemic states with normative content. This detour about the concept
of identity taught us a few things: desires are not resistant enough to changes in
the world to constitute alone identity, imaginative states are not serious enough, and
epistemic states with non-normative content do not seem to be relevant to identity.
If we add the idea that affective states are themselves normative in a sense,19 we end
up with an interesting set of mental states which all involve a normative dimension:
desires (even though they cannot be the whole story), affective states or the heart
(e.g. emotions, feelings or sentiments) and epistemic states with normative content
all seem relevant to one’s identity.
4.3.5 Indeterminateness of Content and the Criterion of Generality

Does this amount to saying that any attitude with normative content is part of the ker-
nel of one’s identity? Is any instance of an affective state, for instance, part of one’s
identity? We do not think so. In order to assess this claim, consider the following
statements:
17 We use the phrase “normative content” as an umbrella phrase for axiological (good, bad, neutral;
elegant, inelegant, etc.) and deontic content (obligatory, forbidden, optional, etc.).
18 The same argument could be replicated with other epistemic states such as knowledge states and
memories.
19 A lot more should be said on this point, we have to refer the reader here to e.g. Mulligan (1998),
Tappolet (2000), Deonna and Teroni (2009).
1. Maria likes eating sushi.

2. Maria believes that eating sushi is healthy.
Example 1 involves an affective state and example 2 involves an epistemic state

with normative content. Consequently, they involve the kind of states we have so far
isolated as being relevant to one’s identity. But are these states really part of one’s
identity? We already have a reason to think that the answer to these questions is
negative.20 We have argued that behavioural vectors have to be at least as global
as behavioural folds, globality being defined in terms of stability through times and
contexts as well as the potential multiplicity of the manifestation of the traits; the
last condition at least does not seem to be fulfilled in these statements.
The way out seems to move to a further level of generality, as exhibited by the
following statements:
3. Maria likes eating healthy food.

4. Maria believes eating healthy food is important.
Indeed, 3 and 4 reveal more to us of Maria’s identity than the mere fact that Maria
likes sushi, for instance. And this seems to be due to the fact that the content of
Maria’s attitudes has gained in generality. To obtain more generality we have moved
from more determinate content to more indeterminate content; in Aristotelian terms,
we have moved up in the genus-species tree. Can we move still further up the
tree? If we do, perhaps we will come to the idea that Maria likes being healthy
and Maria believes her being healthy is important, her preference for healthy food
being derivative of these. In short, we will come to describe Maria’s relation to the
value of health. It seems that by so doing we will reach an intuitively interesting
level of generality in order to describe Maria’s identity. Imagine that you can give
one single piece of information about Maria to someone who does not know her at
all and wishes to have a grasp on the very person Maria is: it seems more promis-
ing to say that Maria cares about health rather than point to the fact that she likes
sushi.
Besides, it should be noted that by considering Maria’s relation to (her) health, we
moved from determinate to indeterminate content but also from objects which exem-
plify the corresponding values (e.g. sushi, healthy food) to abstract values (general:
health; or particular: Maria’s health). We used the word “care” to describe the rela-
tion between Maria and the value of health. But there are many others words we
20 This assertion might seem shocking to some. If Maria really likes sushi, why could it not be a
part of her identity? We will see how we can still account for Maria’s liking sushi being relevant to
her identity, while denying that this state is a behavioural vector.
could have used: valuing, being attached to, or preferring x to y can all be used to
express one’s attitude to values. In order not to multiply terminology, we choose to
speak here of “attachment to values” or “axiological attachment”.21 After all, this
seems to be what one of our examples suggested: to be a bovaresque person seems
to lie in one’s attachment to romantic love or to aesthetic values like charm and
fantasy.
4.3.6 Attachments to Values and Their Manifestations

Axiological attachments seem to have the right kind of globality required of
behavioural vectors. However two questions remain: What is an axiological attach-
ment? What are we to do with states such as desires, emotions, feelings and
normative beliefs which seemed to be relevant to identity: what is the relation
between the latter and attachments to values?
The dispositional nature of behavioural vectors might provide an answer to both
questions. It makes sense indeed to consider that axiological attachments are dis-
positions, which manifest themselves through desires, affective states and epistemic
states with normative content. The fact that Maria likes eating healthy food such
as sushi is a manifestation (F, in the schema above) of her broader attachment
to (her) health, an attachment which is itself the relevant behavioural vector (the
disposition D, in the schema above).
This gives us an insight into a broader problem for the conative theory of iden-
tity we have considered: it mistook manifestations of one’s identity, i.e. conations,
for components of it (and asserted that these were the only components). This also
provides a way to keep both the intuition that a variety of states are relevant to one’s
identity and the intuition that, for instance, the fact that Maria likes eating healthy
food is not properly part of her identity: it is not a behavioural vector of Maria,
because a certain level of generality is required.
Now let us return to our initial metaphors of behavioural folds and behavioural
vectors. Since attachments to values specify one’s stance towards values and provide
one with directions of behaviour,22 it seems that they fit nicely with our initial intu-
ition that behavioural vectors drive one’s behaviour, in contrast to behavioural folds.
For this reason, and given the development above, we suggest that one’s attachments
to values are one’s behavioural vectors.
21 We should say more about the relation between preferences and the monadic states mentioned
above, but, for reasons of space, we will not do so. We talk of axiological attachments, although it
might turn out that axiological attachments are actually based on, or reduce to value-preferences or
sentiments, the latter being conceived as long-standing axiological dispositions. On these issues,
we are particularly indebted to Kevin Mulligan (personal communications as well as e.g. Mulligan
2007 and 2009 on preferences, sensitivity to values and the so-called “tyranny of values”).
22 See Taylor (1989) for the metaphor of orientation.
4.4 Conclusion
We started our paper with the following idea: there is an intuitive distinction between
behavioural folds and behavioural vectors that needs conceptual clarification. On
the one hand, we have suggested that behavioural folds such as shyness or pes-
simism are global dispositions to behave in certain ways, while these dispositions
are not implied by the kind to which their subject belongs. On the other hand, we
have defended the idea that behavioural vectors (being attached to beauty, respect or
another value) amount to having certain axiological attachments, which are global
dispositions towards abstract values.
In the same vein, our three initial stories pointed to important tensions between
descriptions in terms of behavioural folds and behavioural vectors. The Capricorn
Cup story underlines the incompleteness of personal descriptions relying exclu-
sively on behavioural folds. In Isn’t she Lovely?, Maria’s axiological attachments
are suggested. Sam’s apparently contradictory evaluation of Maria can indeed be
interpreted along the following lines: under Maria’s harsh ways, Sam can see what
Maria really cares about. Finally, in Definitely Bovaresque, Maria’s description of
Sam as a bovaresque person takes us from superficial features to deeper ones, since it
states what Sam really cares about (beauty and love), and not only how he regularly
behaves.
If we had to choose non-technical terms, we would call behavioural folds “char-
acter traits” and behavioural vectors “personality (traits)”. But as we said, this
choice is controversial. In particular, it contrasts with Goldie’s recent treatment of
the terms “character” and “personality” (2004). Goldie argues indeed that character
traits are necessarily reason-responsive and because of that are to be distinguished
from personality traits which are not necessarily reason-responsive (for instance
punctuality). The author concludes from this consideration that character traits are
deeper than personality ones.23 This disanalogy between the account presented here
and Goldie’s may be confusing to the reader, since our treatment of personality traits
is closer to Goldie’s account of character traits, and vice-versa. This being said, we
will not argue here for our preferred tags for the two kinds of traits, since this might
turn on a merely terminological issue. We do however believe that both Goldie’s
account of character and personality traits as well as our own point to an important
and real distinction.24
More generally, it must be said that Goldie’s account is not the only one appeal-
ing to sensitivity to reasons in the understanding of character traits. In addition,
behavioural folds are in fact typically reason-responsive and therefore one might
legitimately wonder why the account on offer is silent about reasons. The answer
23 Goldie (2004, 13).

24 For a similar contrast with the same terminological disanalogy, see Deonna and Teroni (2009):
their account of character traits is similar in spirit to our account of personality, while our account
of character traits is closer to what they call stable behavioural outputs.
to this question lies in the fact that behavioural folds are not necessarily reason-
responsive. It is true that traits like “generous”, “pessimist”, “stubborn” and so on
are indeed reason-responsive dispositions.25 For instance, a generous person seems
to be globally responsive to the reasons speaking in favour of helping or giving
to others. On a similar vein, pessimism can be conceived of as the tendency to be
particularly struck by the reasons speaking in favour of the worse possibility; irri-
tability as an oversensitivity to reasons to get angry that is somehow inappropriate;
stubbornness as an enduring tendency to be unresponsive to epistemic reasons in
favour of changing one’s mind. Nevertheless, some behavioural folds do not fit this
picture as easily: curiosity, being a day-dreamer and charming, for instance. In par-
ticular, traits concerning physical behaviour like clumsiness, being energetic, quick
or slow and others, are hardly accountable for in terms of reason-sensitivity. If this
is so, it seems that reason-sensitivity is not all there is to behavioural folds. This is
why the present account does not appeal to reasons.26
We would like to conclude by alluding to the theoretical benefits and possible
uses of the distinction sketched in this paper. In a nutshell, we think that the distinc-
tion between behavioural folds and behavioural vectors might be useful in debates
concerning the way we understand, evaluate and interact with persons, including
oneself i.e. in a reflexive mode. Capturing a person’s set of global dispositions to
behave (behavioural folds) as well as her axiological profile (behavioural vectors),
while distinguishing between the two, is indeed a particularly useful step towards
the understanding of people. Indeed, describing someone’s stable ways of behaving
is quite a different business then describing someone’s attachments to values, or, to
re-use the same metaphor, someone’s heart. We might for example raise the ques-
tion whether one of the sides of the distinction is more important than the other.
Of course, the priority of one type of trait over the other depends on the type of
description or evaluation at stake. For instance, if the description concerns etiquette,
then behavioural folds are clearly central. But it is not clear that it is so as soon as
we try to deepen our description. This paper has in fact suggested that one’s attach-
ments to values matter more to who one is than one’s ways of behaving. After all,
this is what is assumed in the Isn’t she Lovely? story: Sam likes Maria because of
her heart (behavioural vectors) and despite her manners or lack thereof (behavioural
folds). This way of prioritising traits, which we would like to call the classical way
because it puts values at the centre, has a certain objectivist and moralistic flavour
to it. We are positing that what matters more objectively is someone’s attachment to
values and, to this extent, that someone who disliked Maria (overall) because of her
25 By reasons we mean entities that speak in favour of adopting a certain attitude and that might jus-
tify it or make it appropriate (on this standard conception of reasons, see Raz 1975, Scanlon 1998).
The link with rationality has been developed in recent accounts of character traits emphasizing the
role of character traits in justifying, predicting and evaluating behaviour (Goldie 2004, Schueler
2007, and Deonna and Teroni 2009 for a view in the same spirit, albeit referring to values).
26 It might be however that behavioural vectors are necessarily reason-sensitive, but this is another
issue.
bad manners (behavioural folds) and despite her good axiological profile would be
objectively wrong in doing so.
It is beyond the scope of this paper to settle the question of the priority of one
type of trait over the other in a conclusive way. The important point is that these
questions can now be spelled out more clearly. Nevertheless, let us say that we
favour the classical conception insofar as descriptions or evaluations concern the
core of persons. We do so for a series of reasons, which we will quickly mention.
(1) By definition, values matter. While ways of behaving might also be important
and some behavioural folds might involve an evaluative component,27 they are not
about values in the way attachments to values are. (2) According to some analyses
of moral values,28 one’s axiological profile determines one’s morality: being moral
is to have attitudes to non-moral values which fit the non-moral values in the world
and the hierarchy of these. For instance, preferring justice over one’s personal plea-
sure (justice and pleasantness being two non-moral values) is morally good, whereas
preferring one’s personal pleasure over justice is immoral. If we assume that their
morality is an important aspect of persons, then behavioural vectors seem crucial.
(3) Following Scheler’s conception of persons, we believe that one’s relation to val-
ues is a “deeper” feature than one’s ways of behaving, which are more dependent
on external circumstances. For instance, some physical behavioural folds clearly
depend on one’s physical constitution and, because of this reason, depend on cir-
cumstances external to the subject qua person. On the contrary, independently of
a person’s physical constitution or environment, it makes sense to pay attention to
her attachment to justice, for instance, even if these traits are not manifested in
her behaviour: if Maria is attached to justice, but has in fact no power to correct
injustices, it is still relevant to consider her attachment to justice when describing
her axiological profile – even though it is not manifested at the behavioural level.
(4) One’s behavioural folds can often be explained and justified in terms of one’s
attachment to values, while the converse does not seem to be true. For instance,
when asked why a person generally acts generously, one can naturally point to her
attachment to equality, respect and solidarity. By contrast, asking for the reasons for
Sam’s attachment to respect, things remain unexplained by mentioning Sam’s ten-
dency to be punctual. This priority in explanation is according to us symptomatic of
an underlying ontological priority.29 One promising thesis to explore would be that
certain personality types imply the presence or absence of certain character traits.
27 Some behavioural folds can indeed be seen as thick properties, be it negative (e.g. stubborn,
irritable), or positive (e.g. persevering, dynamic).
28 Cf. Scheler (1973).
29 This being said, even if we assume the classical thesis, we are not saying that behavioural folds
play no role in understanding persons. On the contrary, it might be of great importance, in particular
in cases where one’s attachment to values cannot settle all questions. For instance, where one’s
options are equally fitting in term of one’s axiological profile, behavioural folds might settle matters
in favour of one option rather than the other.
The presence of behavioural vectors seems to imply some patterns of behaviour,

while the reverse does not seem to be true.30
Finally, let us mention that each consideration presented so far can easily be mir-
rored in a self-reflexive perspective and shed light on the point of view a person
takes with regard to herself – and the distinct aspects of herself. If so, the distinc-
tion might help clarifying self-understanding, self-regulation, and self-evaluation.
For instance, assuming the classical thesis introduced above, the more one’s self-
evaluation focuses on attachment to values, the deeper one’s reflexive gaze. On the
contrary, if the classical thesis is denied, i.e. if behavioural vectors are not central to
who one is, self-evaluation will obviously take a very different form.
Besides the issues we have mentioned, we believe that we have sketched
an important contrast between ways of behaving (behavioural folds, how) and
attachments to values (behavioural vectors, what). By addressing the issue of the
boundaries of the self, we hope that the present inquiry has contributed to under-
stand better, at least partly, who we are. A problem which, we fully agree with
Bernard Williams,31 is of the uttermost importance.
Acknowledgements We wish to thank Jiri Benovsky, Otto Bruun, Tom Cochrane, Jack Darach,
Julien Deonna, Julien Dutant, Amanda Garcia, Philipp Keller, Anita Konzelmann Ziv, Olivier
Massin, Kevin Mulligan, Hans Bernhard Schmid, Gianfranco Soldati, Christine Tappolet, Fabrice
Teroni, Cain Todd, an anonymous referee, as well as the audiences of various presentations of this
paper for their helpful comments, support and friendship.
References
Deonna, J. A., and F. Teroni. 2009. “Taking Affective Explanations to Heart”. Social Science
Information 48 (3):359–77.
Doris, J. 2002. Lack of Character. Personality and Moral Behaviour. Cambridge: Cambridge
University Press.
Frankfurt, H. G. 1971. “Freedom of the Will and the Concept of a Person”. The Journal of
Philosophy 68 (1):5–20.
Frankfurt, H. G. 1976. “Identification and Externality”. In The Identities of Persons, edited by
A. Rorty. Berkeley, CA: University of California Press.
Frankfurt, H. G. 1987. “Identification and Wholeheartedness.” In The Importance of What We Care
About. Cambridge: Cambridge University Press.
Frankfurt, H. G. 2006. Taking Ourselves Seriously & Getting It Right. Stanford, CA: Stanford
University Press.
Goldie, P. 2004. On Personality. Thinking in Action. London: Routledge.
Humberstone, I. L. 1992. “Direction of Fit”. Mind 101:59–84.
Kenny, A. 2003. Action, Emotion, and Will. London: Routledge.
Molnar, G. 1999. Powers. Oxford: Oxford University Press.
Moran, R. 2002. “Frankfurt on Identification”. In Contours of Agency. Essays on Themes from
Harry Frankfurt, edited by S. Buss and L. Overton. Cambridge, MA: Bradford Book.
Mulligan, K. 1998. “From Appropriate Emotions to Values”. The Monist 81 (11):161–88.
30 Note that this might impinge on the very possibility of cooperation. Cooperation certainly
requires a grasp of persons’ dispositions to behave but also of their axiological attachments.
31 Williams (1976).
Mulligan, K. 2007. “Intentionality, Knowledge and Formal Objects.” Disputatio 2 (23):205–28.

Mulligan, K. 2009. “On Being Struck by Value: Exclamations, Motivations and Vocations”. In
Leben mit Gefühlen: Emotionen, Werte und ihre Kritik, edited by B. Merker. Paderborn: Mentis.
Raz, J. 1975. Practical Reasons and Norms. London: Hutchison.
Ryle, G. 1949. The Concept of Mind. London: Penguin.
Scanlon, T. M. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University Press.
Schechtman, M. 2004. “Self-Expression and Self-Control”. Ratio (new series) 17:409–27.
Scheler, M. 1973. Formalism in Ethics and Non-Formal Ethics of Value, trans. M. Frings and R.
Funk. Evanston, IL: Northwestern University Press.
Schueler, G. F. 2007. “Rationality and Character Traits”. In Moral Psychology, edited by S.
Tenenbaum. Vol. 94 of Poznan Studies in the Philosophy of Sciences and the Humanities.
Amsterdam; New York: Rodopi.
Smith, M. 1994. The Moral Problem. Oxford: Blackwell Publishings.
Tappolet, C. 2000. Emotions et valeurs. Paris: PUF.
Taylor, C. 1989. Sources of the Self. The Making of the Modern Identity. Cambridge, MA: Harvard
University Press.
Velleman, D. 2002. “Identification and Identity”. In Contours of Agency. Essays on Themes from
Harry Frankfurt, edited by S. Buss and L. Overton. Cambridge, MA: Bradford Book.
Williams, B. 1976. “Persons, Character and Morality”. In The Identities of Persons, edited by A.
Rorty. Berkeley, CA: University of California Press.
Part II
Self-Evaluation and Rationality
Chapter 5
Self-Evaluation and the Ends of Existence
Carol Rovane
Reasons are irreducibly normative, in the sense that there is a distinction between
having reasons and acting upon them. This distinction is sometimes characterized in
terms of the difference between what we ought to do vs. what we are actually moved
to do in fact, though when we characterize it in this way, we must take care not to
confuse the “ought” of irreducible normativity with the “ought” of moral obligation.
The former is a wider notion than the latter, covering all of the practical “oughts”
on which we act in connection with our nonmoral ends as well as our moral ends,
and also logical “oughts” and epistemic “oughts”. Being thus wider than the latter,
the former is also a weaker notion, insofar as it registers a normative force that may
fall short of a binding obligation.
We cannot either embrace or act on reasons in this irreducibly normative sense
unless we possess a self-critical perspective from which it is possible to detect a gap
in our own lives between what we ought to think and do and what we actually think
and do. It is natural to view the self-critical activities through which we recognize
such “ought-is” gaps in our own lives as forms of self-evaluation. We are evaluating
our own thoughts and actions as either measuring up to or falling short of various
normative standards. However, these self-critical activities that go together with the
irreducible normativity of reasons need not involve an evaluation of one’s self in the
strict and literal sense, by which I mean, the sort of comparative evaluation that we
carry out in the context of ordinary deliberation, in which we rank and weigh the
values of disparate things on a single scale from worst to best.
The difference between these two forms of self-evaluation – the self-critical
activities that are presupposed by the irreducible normativity of reasons vs. self-
evaluation in the strict and literal sense – will become much clearer as I proceed.
But here is one quick route by which we may become convinced that there must
be a difference. Many rationalist moral philosophers who work in the Kantian tra-
dition hold that reasons are irreducibly normative, and they regard the self-critical
activities through which these gaps come to light as important undertakings for all
C. Rovane (B)
Department of Philosophy, Columbia University, New York, NY, USA
e-mail: cr260@columbia.edu

82 C. Rovane
rational agents. Nevertheless, they refuse to view the value of an agent as something
that can be ranked or weighed against the value of another agent and, accordingly,
they reject self-evaluation in the strict and literal sense. This goes to the heart of their
opposition to utilitarian morality, for when utilitarians recommend that we conceive
the moral good in an aggregative way, as the greatest good of the greatest number,
they are asking us to compare the relative worth of different agents in terms of the
different contributions that their existence would make to aggregate worth.
I don’t wish to defend a utilitarian or, indeed, any other more broadly consequen-
tialist conception of the moral good. But all the same, I will be arguing against the
Kantian stricture against the comparative evaluation of rational agents. Or, to state
my thesis more plainly, I will be arguing that self-evaluation in a strict and literal
sense plays an essential role in the exercise of reflective rational agency.
The essential role I have in mind is not a causal role. That is, I am not claiming
that reflective rational agents always do, in fact, evaluate their own worth in compar-
ison with the worth of others. That claim is obviously false, since avowed Kantians
are committed to refraining from such evaluation. What I am claiming is that they
ought to do it nevertheless, and that when they fail to do it they fall short of being
fully rational.
Thus, the essential role that I have in mind is a normative role. But even on a
normative reading, my claim isn’t obviously true. And this very fact might seem to
suggest that it couldn’t be true. After all, it makes sense to suppose that reflective
rational agents are committed to being rational, since otherwise they wouldn’t have
any grounds on which to adopt a self-critical perspective from which to evaluate
their own rational shortcomings. But this requires that they have a positive concep-
tion of what they are committed to – what being fully rational would consist in. So
it is indeed hard to see how it could be true that being fully rational would require
engaging in self-evaluation in a strict and literal sense without it being obviously
true.
There is, however, a good explanation of why my claim isn’t obviously true:
reflective rational agents are prone to make a false assumption about their own
nature, which has the effect of obscuring its truth. The assumption I have in mind
is that the existence of a reflective rational agent is a metaphysical given. It is such
a natural assumption that it generally goes unnoticed, and as a result, it is generally
allowed to stand without being examined at all. Yet once we gain a correct under-
standing of the nature of reflective rational agency, we can see that it is false. What
is metaphysically given is not the existence of reflective rational agents per se, but
only the existence of human beings, along with their capacity for reflective ratio-
nal agency. And it requires the actual exercise of the human capacity for reflective
rational agency in order to bring reflective rational agents into existence. This can be
hard to see until we come to understand another, related aspect of human capacity
for reflective rational agency, which is that it can be exercised so as to create agents
of different sizes. In addition to agents of human size, it may also be exercised so
as to produce multiple agents within a single human being, and group agents com-
posed of many human beings. In each and every case, these agents of different sizes
are a product of effort and will. Not only does this mean that their existence is not
5 Self-Evaluation and the Ends of Existence 83
a metaphysical given; it also means that they ought to engage in self-evaluation

in strict and literal sense – a sense that goes beyond adopting the self-critical per-
spective of rational agency, so as to assess the very worth of their existence in a
comparative light.
In order to make my normative claim about self-evaluation fully plausible, I
would have to provide a much more elaborate defense of the metaphysical picture
of agent-identity from which it derives than I will be able to provide in this paper.1
I can only sketch that picture and outline the main considerations in favor of it.
I’ll start in Section 5.1 by setting out an initial account of the nature of reflective
rationality, in which I’ll confine myself to claims that most philosophers would pre-
sume they can accept without taking on board my more controversial claims about
agent-identity and self-evaluation, which I’ll then go on to defend in Section 5.2. We
shall see, however, that the relatively uncontroversial claims of Section 5.1 provide
crucial and powerful premises for the arguments to come in Section 5.2.
5.1 An Initial Account of Reflective Rational Agency

Let me begin by laying down two assumptions about the nature of reflective rational
agency that I take to be uncontroversial: (1) the distinguishing mark of reflec-
tive rational agency is the capacity to deliberate; (2) the goal of deliberation is to
determine what it would be best to think and do, all things-considered.
If these two assumptions seem controversial, then they are most likely being
interpreted in a stronger way than I intend. So for example, the first one might
seem controversial if it is interpreted as saying that reflective rational agents always
deliberate before they act. Of course, it wouldn’t be merely controversial to say that
but downright false, since there can be no denying that agents often act without
deliberating beforehand. But my first assumption doesn’t say that. What it says is
that the capacity for reflective rational agency goes in hand with the capacity for
deliberation – so if something lacked the capacity to deliberate, then although it
might qualify as an agent of some sort, it wouldn’t qualify as an agent who possesses
reflective rationality. Similarly, the second assumption might seem controversial if
we were to adopt an ambitious view of what rationality requires of agents, by which
I mean, a view that allows for so-called “external” reasons. But when we view the
requirements of rationality as supplying only internal reasons, it is not controversial
to suppose that the proper goal of deliberation is to arrive at all-things-considered
judgments.
In the rest of the section, I want to expand and clarify this point that underlies
my second assumption, concerning the internal character of the requirements of
rationality, because much of what follows in Section 5.2 depends specifically on
this.
1 Space doesn’t permit a full elaboration and defense of this metaphysical picture of agent-identity
in this paper. For a more complete discussion, see Rovane (1998).
84 C. Rovane
The internalist view of rationality emphasizes that deliberation is always carried

out from a first person point of view. This means that its topic is first personal: it
concerns what I ought to do, all things considered. It also means that its scope is
likewise first personal: the things to be considered when I consider all things are all
and only my own attitudes. It goes against ordinary language intuitions to say that
what I consider in deliberation are my own attitudes; it seems much more natural to
say that I consider the objects of those attitudes – facts about what the world is like,
about what options my abilities and circumstances afford, and about the relative
merits of those options. But as we shall see below, this more natural way of talking
lends itself to an externalist view of rationality. What I want to emphasize here is
that, on the internalist view, what an agent ought to think and do is a function of
the combined normative force of all of her attitudes taken together. That is what her
all-things-considered judgments aim to register, and that is why I say that what she
ought to consider are all and only her own attitudes.
It must be admitted that agents don’t often achieve this deliberative goal, or even
try to achieve it. Nevertheless, it is a goal that they must be committed to achieving,
at least insofar as they are committed to being fully rational; and they may have this
commitment even if they don’t always live up to it and, moreover, even if they don’t
always try to live up to it, so long as they regard their failures to live up to it as
grounds for self-criticism.
There are three main categories of such rational failures, in which agents may fail
to live up to their commitment to being rational: (1) They act without deliberating
at all (as in cases where they act directly from habit and inclination), with the result
that they fail to do what they would have judged to be best if they had taken the
time to deliberate. (2) They take the time to deliberate but fail to act in accord with
their all-things-considered judgment – in such cases, they suffer from weakness of
will. (3) They fail to deliberate well, in the sense that their all-things-considered
judgments do not accurately reflect the combined normative force of all of their
attitudes taken together; or to put it another way, they fail to identify what it really
would be best to do, as that would be judged by their own best lights.
When agents recognize their own rational failures, they are of course recognizing
“ought-is” gaps in their own lives, and this is why the internalist view can claim to
respect the irreducible normativity of reasons. Advocates of external reasons never-
theless object that internalism cannot do full justice to such irreducible normativity.
It seems to me that these objections have been adequately addressed by others in
the philosophical literature.2 So instead of rehearsing that whole literature, I will
confine my attention to what I see as the two most important objections that have
been raised against internalism. My purpose in doing so is not to refute externalism,
in the sense of showing that there are no such things as external reasons; my pur-
pose is merely to clarify what speaks in favor of an internalist view of rationality in
particular, and what the internalist view of rationality really amounts to.
2 See for example, Christine Korsgaard (1986), and Akeel Bilgrami (2006).
The first objection that I want to consider is that the internalist is wrong to say
that what we ought to take into account is all and only our own attitudes, because
what we really ought to take into account are objective matters of fact and value.
The second objection is that internalism amounts to an implicit egoism that dis-
counts some important reasons for altruism – why we ought to take into account the
attitudes of others.
The main impetus for the first objection is an undeniable fact, which is that we
are subject to various epistemic limitations. These epistemic limitations keep us
from recognizing many considerations that bear on our deliberations – not because
we don’t bother to deliberate, or don’t deliberate well, but because we either don’t
know about the considerations, or are mistaken about them. We find it natural to
say of such considerations that we ought to take them into account nonetheless, in
spite of our ignorance and error; and when we say this we are certainly referring
to external reasons. Insofar as I am advocating an internalist view of rationality, I
don’t mean to suggest that such references to external reasons are unintelligible. But
I do mean to suggest that failures to respond to such external reasons aren’t really
failures of rationality. When we say that an agent ought to take into account various
considerations in the rational sense of “ought”, we are not referring to such external
reasons, but to considerations that are internally available to the agent. Recall, for
example, the famous example of a worker at a gas refinery who saw a co-worker fall
into a tank. He knew perfectly well that gas is highly combustible, but in his rush to
save his co-worker he took a kerosene lamp into the tank and set off an explosion
that killed them both. It is clear that he failed to take into account all of the things
that he rationally ought to have taken into account in the internal sense – all and
only his own attitudes. In contrast, consider the case where a passer-by happens to
see a worker fall into the tank, but doesn’t know that it is a gas tank, or doesn’t
know that gas is combustible. If such a passer-by were to rush in to save that worker
with kerosene lamp in hand, she would end up killing them both. Although this
would definitely be a mistake of some kind, it would not be a failure of rationality.
For it cannot be said that the passer-by would have failed to act in accord with her
own best lights; she would simply have been ignorant of some crucial relevant facts.
Perhaps there is room for an advocate of external reasons to insist that there is still
some sense in which she ought to take those facts into account – a sense that goes
together with the thought that if she doesn’t know about them then she ought to find
out about them. But in response, the internalist will insist that this external “ought”
cannot be a rational “ought” unless she has relevant attitudes, in the form of beliefs
about her own ignorance. If she believes that she is ignorant about relevant matters
of fact, then she has internal reasons to undertake the relevant inquiries in order to
rectify her epistemic shortcoming, and in such a case it would be a rational failure
on her part not to undertake those inquiries.
To sum up my response to the first objection against internalism: the sense in
which we ought to take into account all and only our own attitudes is that our own
attitudes are the proper basis of our deliberations – they are what we ought to delib-
erate from; there is no such thing as deliberating from the facts, but only from our
attitudes about them; and there is no rational requirement to have attitudes other
86 C. Rovane
than the ones we have, except insofar as our attitudes themselves give us internal
reasons to think that this is so.
The internalist’s response to the externalist’s objection from altruism follows a
similar course. To make things vivid, let’s consider whether my daughter’s desire
to go to Disneyland should constitute a deliberative consideration for me, and if so,
what sort. As it happens, I believe that the only good reason to undertake leisure
travel is to gain an experience of other cultures that one could not otherwise have.
When I deliberate from this attitude, I can see that I have a prima facie reason to go
to India but not to Disneyland. But I also believe that my daughter would very much
like me to take her to Disneyland. And the question arises, in what way should
I take her desire into account? An externalist might insist that if we are to make
sense of the possibility of altruism, then we must suppose that I ought to take her
desire into account in a fairly direct way, as something to be satisfied along with
my own desires. In contrast, the internalist will insist that this is not so, unless I
have a specifically altruistic desire, namely, the desire to satisfy her desire to go to
Disneyland. The internalist will also insist that the normative force of such a higher
order altruistic desire is profoundly different from the normative force of a first
order desire to go to Disneyland. I cannot rationally take on board a first order desire
to go to Disneyland without revising my general beliefs about what makes leisure
travel worthwhile. But I can rationally take on board an altruistic desire to satisfy
my daughter’s desire to go to Disneyland without revising my general beliefs about
what makes leisure travel worthwhile. I may rationally conclude that, all things con-
sidered, I ought to take her to Disneyland, even though I continue to disagree with
her about whether fantasy vacations really are worthwhile – so long as I rank the
value that I place on ceding to her wishes ahead of the value I place on worthwhile
forms of travel.
So far, I’ve brought out that internalism doesn’t rule out altruism because it
doesn’t rule out altruistic desires, and also that deliberating from altruistic desires
is very different from taking the desires of others on board as one’s own (which
would place very different normative constraints on our deliberations). However,
moral rationalists may not be satisfied with this portrayal of our grounds for altru-
ism, because it makes altruism contingent on the presence of specifically altruistic
desires which agents may or may not have, whereas they want to see altruism as
rationally mandated. Kantians, in particular, are explicit that they wish to portray
the moral “ought” that is couched in the categorical imperative as a rational “ought”
as well. If they are right, then rationality itself requires us to take into account the
desires of others in a quite specific way, by confining ourselves to reasons that are
universalizable. But it is a confusion to think that this feature of the Kantian posi-
tion requires us to abandon the internalist view of rationality. On the contrary, one
of the main attractions of rationalist moral theories is that they portray the demands
of morality as being recognizable by anyone who is capable of deliberating at all
and, hence, as internally available.3
3 This is a central point in Korsgaard, op. cit.

The real source of tension between Kantian morality and the initial account of
reflective rationality that I have offered in this section isn’t its commitment to an
internalist view of rationality; the real source lies in its metaphysical implications
for agent-identity and self-evaluation. So let me now turn to those issues.
5.2 Agent-Identity and Self-Evaluation

As I indicated in my introductory remarks, my claim about the proper role of self-
evaluation in the lives of reflective rational agents is bound up with a metaphysical
picture on which the existence of such agents is not a metaphysical given but rather
a product of effort and will; and, in turn, this metaphysical picture requires us to
recognize the twin possibilities of group and multiple agents, as well as agents of
human size. It is only when we recognize these two possibilities that we can appre-
ciate the role of effort and will and, ultimately, self-evaluation, in the lives of agents.
I’ll now briefly outline the argument for those two possibilities.
In a nutshell, the argument says that groups of human beings, and also parts of
human beings, may be committed to meeting the requirements that define individual
rationality, and when they have this commitment they qualify as individual agents
in their own rights. What most fundamentally drives the argument is a conceptual
tie between rationality and agent-identity.
In order to clarify how this conceptual tie figures in my argument, I need to pro-
vide some background about the nature of deliberation. It is a process that involves
many component rational activities, such as the following: identifying one’s options;
embracing some options as ends that are prima facie worth pursuing; evaluating
ends in the light of the means necessary to achieving them; evaluating ends in the
light of the consequences of achieving them; working out many various implications
of one’s attitudes; removing inconsistencies among one’s attitudes – in the case of
beliefs this generally means rejecting some as false because they conflict with others
that one takes to be true, whereas in the case of evaluative attitudes this generally
involves arriving at a transitive preference ordering. There can be reasonable dis-
agreement among philosophers about exactly what should be included on a list of
such component rational activities of deliberation. But it doesn’t really matter for
my purposes in this paper what the list includes. What matters is that they all have
a similar role, which is to facilitate the larger goal of deliberation, of arriving at all-
things-considered judgments about what it would be best to do. That goal articulates
an overarching requirement of rationality, in the sense that any agent who is commit-
ted to being rational must be committed to realizing this goal. Alongside (or perhaps
underneath) this overarching requirement of rationality, there are other, more spe-
cific requirements that articulate the various goals of the various component rational
activities that go into the larger process of deliberation. For example, when we work
out various implications of our attitudes we are responding to the rational require-
ment of closure; when we rank our preferences we are responding to the rational
requirement of transitivity of preferences; and when we reject certain propositions
as false because they are inconsistent with what we take ourselves already to know,
we are responding to a rational requirement of consistency in belief.
88 C. Rovane
The conceptual tie between rationality and agent-identity is perhaps easiest to

see in connection with these more specific requirements of rationality that guide
the component rational activities of deliberation. (To reiterate: it doesn’t matter
to my argument in this section what the list of such component activities should
include, but only that all of them contribute to the larger goal of arriving at all-
things-considered judgments.) The reason why is that each of them clearly specifies
a condition that an individual agent would have to meet in order to be fully ratio-
nal – a point that I’ll now briefly illustrate in connection with the requirement of
consistency in belief. If an individual agent holds inconsistent beliefs, that is clearly
a failure of rationality. But one agent may hold beliefs that are inconsistent with the
beliefs of another without being guilty of any failure of rationality. In fact, this may
be so without it being the case that either agent is guilty of a failure of rationality.
The two agents may simply be working from different states of information, from
which it follows that each may be internally consistent – and therefore rational –
even though they disagree. What goes for the requirement of consistency in belief
goes for all requirements of rationality, at least on the internalist view: these require-
ments apply only within a point of view and not between points of view. And when
I say all requirements I don’t only mean the specific requirements that apply to the
component activities of deliberation but also its overarching requirement to arrive
at all-things-considered judgments about what it would be best to do. On the inter-
nalist view, this overarching requirement also applies only within a point of view.
That is why both the topic and the basis of deliberation are first personal – what I
ought to do in the light of all and only my own attitudes.
I hope it is now clear that that internalism is also a species of individualism, in
the sense that when we view the requirements of rationality as holding internally,
they define what it is for an individual agent to be fully or ideally rational. And that
is why there is a conceptual tie between rationality and agent-identity.
The next step in the argument for the possibility of group and multiple agents
focuses on the fact that human beings are capable of pursuing projects that require
sustained and coordinated effort – what I call “unifying projects.” It is a fairly
intuitive and uncontroversial point that different unifying projects require the coor-
dination of effort within different boundaries. The project of running a graduate
program in philosophy requires coordinated efforts within a group of human beings,
while writing a book in philosophy requires coordinated efforts within only a portion
of a single human life; other projects, such as life-long relationships and careers,
typically require coordinated efforts throughout the larger part of a whole human
life.
Another fairly intuitive point about unifying projects is that, insofar as they
involve coordinated effort, they also require deliberations that have as their aim
the planning and direction of such coordinated effort. These deliberations must take
into account any consideration that can be seen to bear on the planning and exe-
cution of the project, including the following: identifying various component steps
through which the larger project is to be carried out; remembering which component
steps have been carried out so far; bearing in mind which steps remain to be done;
determining which steps ought to be done now in the light of which have already
been done and which are yet to be done. (To take a simple example: each sentence
of this article must be written in the light of how it fits in to the whole article.) This
doesn’t necessarily mean that an agent who has a commitment to a unifying project
can rationally choose to perform only those actions that would directly contribute to
the larger project. What it means is that the question whether these other options are
worth pursuing should be judged by reference to how pursuing them would affect
the unifying project.
While I have been emphasizing that these points about unifying projects are quite
intuitive, it is not at all intuitive that they bring in train the possibilities of group and
multiple agents. So let me now explain why they do, by putting them together with
the conceptual tie between rationality and agent-identity.
I’ve said that unifying projects would require deliberations in which various
considerations would have to be taken into account together. We might think of
the conclusions of such deliberations, then, as all-things-considered judgments, in
which the scope of the “all” is determined by the nature and requirements of the
unifying project that occasions those deliberations. And we might ask: what is the
difference between these all-things-considered judgments that would be required
in the context of pursuing a unifying project, and the all-things-considered judg-
ments that are the proper goal of any rational agent’s individual deliberations? One
is tempted to answer that in the latter case the scope of the “all” is not determined
by any particular practical commitment, but rather by the agent’s identity – thereby
respecting the conceptual tie between rationality and agent-identity, on which an
agent ought to take into account all and only its own attitudes. But then we must
also ask: what makes attitudes an agent’s own? And when we answer, we must bear
in mind that whatever account we give of agent-identity, it had better preserve the
conceptual tie, so that it remains true that an agent ought to take into account all and
only its own attitudes. The one answer that is guaranteed to preserve the tie links
agent-identity with commitments to unifying projects. It allows us to say that when
a unifying project sets the scope of the “all” in such all-things-considered judg-
ments, it thereby determines the boundaries of a particular point of view from which
deliberations proceed and, therewith, the identity an individual rational agent who
deliberates and acts from that point of view – one who is committed to realizing the
overarching goal of individual rationality, which is to arrive at all-things-considered
judgments.
This way of individuating agents clearly entails that there can be agents of dif-
ferent sizes. Insofar as a unifying project requires the coordinated efforts of many
human beings together a group point of view will have to be forged, from which a
group may deliberate and act as one. Examples of such groups who could in princi-
ple qualify as individual agents in their own rights might include couples who have
particularly intense marriages, academic departments, sports teams and research
teams.4 Insofar as a unifying project requires coordinated efforts within a whole
4 The Manhattan Project is a particularly interesting real case to explore, which I briefly do in
Rovane (2004).
90 C. Rovane
human life, a human-size point of view will have to be forged, from which a human
being may come to deliberate and act as one agent. Most of the agents we actually
recognize fall into this category, and their unifying projects are typically life-long
relationships and careers. Finally, insofar as different unifying projects require coor-
dinated effort within different portions of the same human life, then multiple points
of view will have to be forged from which those various projects may be carried out
by smaller agents within that human being. The only commonly recognized exam-
ples of such multiple agents are pathological ones, namely, the alter personalities
that are exhibited by human beings with dissociative identity disorder. However, my
argument implies that there can be non-pathological cases as well. Consider, for
example, children of immigrant parents who maintain traditions at home that have
no meaning in the social space outside the home, in which the children must go to
school and eventually make their way as adults. Because such children live in such
fragmented social circumstances, there may be no single set of deliberative con-
siderations from which they could coherently aim to arrive at all-things-considered
judgments. When at home, it might be necessary to deliberate from one set consist-
ing mainly of traditional attitudes, which would make it possible to maintain family
relations and activities; in order to navigate the larger world outside the home it
might be necessary to deliberate from a quite different set of non-traditional atti-
tudes. If that is what happens, the children will come to house two distinct points of
view that belong to two distinct agents.5
It might objected that it cannot be the case that all agents come into existence
in the way that I’ve been describing, as a product of effort and will for the purpose
of pursuing a unifying project, because there must be some original agents whose
existence is a metaphysical given, through whose efforts such “willed” agents could
come into existence. The obvious candidates for the status of such metaphysically
given original agents are, of course, human beings. In order to respond to this objec-
tion, I need to say more about why the mere existence of human beings does not
suffice for the existence of an agent. I want to approach this task by exploring some
important points of contrast between the two metaphysical pictures of agent-identity
that have been at issue in this paper – the one I’m defending and the one we are prone
to take for granted on which the existence of an agent is a metaphysical given.
On the picture I’m defending, agents can exercise their rational capacities within
different boundaries, and so they need reasons to exercise them within one set of
boundaries rather than another, and what supplies these reasons are commitments
to unifying projects whose practical requirements set these boundaries, and thereby
5 It makes no odds against this point that such fragmentation might be a rational solution to a
problem. On the contrary, it is central to my account of multiple and group agency that agents
can have reasons to fragment into smaller agents or to integrate with others into larger agents.
My point is that when they act on such reasons, they bring about the existence of new agents of
different sizes. So if an immigrant child finds that the rational strategy of trying to maintain one
unified human-size point of view is unfeasible, and begins to deliberate from two different points
of view, that child will no longer be one agent but rather two – and this fact is not compromised by
the fact that the child arrived at this outcome by rational means.
delimit the separate points of view from which individual agents deliberate and act.
Implicit in this picture is a negative claim to the effect that in the absence of such
reasons to deliberate and act as one, which are supplied by unifying projects, there
simply aren’t any distinct points of view from which deliberations can or should
proceed. Here it might be useful to bring in Harry Frankfurt’s notion of a wanton.
As he defines it, a wanton simply doesn’t care what its desires are, but is prepared to
ride their motivational force wherever it leads.6 I take it to follow that the wanton has
no reason to deliberate. More controversially, I also take it to follow that the wanton
doesn’t actually have a deliberative point of view at all, since that is something
that would arise only insofar as there was a reason to deliberate – that is, to take
various considerations into account together in all things-considered judgments. In
the absence of such a reason, there is no point of view from which an agent ought
to arrive at such judgments, and hence no agent in that sense at all – no reflective
rational agent who ought to arrive at such judgments.
According to the contrasting metaphysical picture that almost everyone takes for
granted, the existence of reflective rational agents is a metaphysical given, in the
form of human beings who possess as part of their metaphysically given nature the
capacities that are required for reflective rational agency. This picture invites us to
view our reasons for deliberating on the model of other, non-rational capacities that
human beings also possess as part of the native endowment, such as their capacities
to walk and talk. It is fair to say that the most fundamental reason why human beings
walk and talk is simply that they can. This doesn’t mean that human beings don’t
generally have other reasons to use their capacities for walking and talking. On the
contrary, they generally do have such additional reasons, in the form of places to
6 See Harry Frankfurt (1971). As is clear from his title, his topic in that paper was freedom of the
will, which he proposed to model on freedom of action, understood as actions that are in accord
with an agent’s desires. On his model, freedom of the will arises when the desires on which an agent
acts are in accord with its higher order desires to have those first order desires (though the agent
must also desire that those first order desires be effective). Frankfurt’s whole approach is predicated
on a Humean conception of reasons as desires, which are construed as dispositions to act rather
than as irreducibly normative reasons. If we were to embrace the Humean conception, the self-
critical perspective that I’ve claimed follows upon the recognition of such irreducibly normative
reasons would have to be captured in a different way. Wherever we might posit a first order attitude
with irreducibly normative content, the Humean must posit a first order disposition to act, which
is accompanied by a second order desire from the perspective of which the first order disposition
gains a quasi-normative significance, as either desired or not desired. It seems to me that opponents
of the Humean conception have tended to underestimate how much of the self-critical perspective
of rational agency might be captured in this way, by exploiting Frankfurt’s model of free will. Yet at
the same time, it also seems to me that there is one major drawback to this whole approach, which
is that it forces us to see the self-critical perspective of rational agency as essentially hierarchical,
in the sense that the self-critical function is always located in higher order desires to have or not
have certain lower order attitudes. But it seems to me that all attitudes, even first order attitudes,
have normative force, and that lower order attitudes may well provide grounds on which an agent
ought to revise its higher order attitudes rather than vice versa. (For instance, if I have a higher
order desire to be the sort of person who has an effective first order desire to train for a triathlon, I
might draw reasons to revise that higher order desire from other of my first order desires, such as
my desire to not to ruin my joints in middle age, or my desire to read more books.)
92 C. Rovane
go and things to say. Without these further reasons for walking and talking, there
would be little point in walking or talking. But at the same time, if we ask why do
human beings engage in any activities that involve walking and talking at all, the true
answer seems to be that it belongs to their nature to walk and talk. Since the capacity
to deliberate also belongs to human nature, it might seem that the most fundamental
reason why human beings ever deliberate should be the same – because they can.
They may generally have other, more specific reasons to deliberate of the kind that
I’ve been discussing, that derive from commitments to unifying projects that cannot
be accomplished without significant planning and coordination of effort. But as in
the case of walking and talking, it may seem fair to say that the most fundamental
reason why human beings ever engage in activities that require deliberation at all is
that it belongs to their nature to do so. When we look at matters in this way, it seems
to follow that the wanton actually has the same fundamental reason to deliberate that
all human beings have, and that the wanton’s refusal to deliberate is a rejection of its
own nature – akin to Oblomov’s refusal to get out of bed,7 or the hermit’s refusal to
talk. It also seems to follow that the wanton already is a reflective rational agent just
by virtue of possessing the requisite capacities as part of its native endowment. And
this impression is reinforced when we try to imagine what it would be like if the
wanton came to have practical commitments that would give it reason to use those
capacities, i.e., reason to deliberate. We wouldn’t typically think of this as bringing
into existence an agent who wasn’t there before, but rather as a case of gaining a new
status that the wanton didn’t have before. It would have become a fully reflective
rational agent, in something like the sense that young children become walkers and
talkers – and, we generally assume, become reflective rational agents too.
I just characterized the metaphysical picture that I’m opposing as sympatheti-
cally as I can, in order to bring out why it is indeed tempting to view the existence
of a reflective rational agent as a metaphysical given. Let me now explain more
carefully why we should nevertheless resist this temptation. The reason why derives
from a profound difference between the nature of deliberation and the nature of
walking and talking. Human beings may come to walk and talk without framing
any conception of themselves as rational beings who are subject to rational require-
ments. But reflective rational agents cannot deliberate without doing this. They must
view themselves as subject to a normative requirement to take various attitudes into
account together, in order to arrive at all-things-considered judgments; and they
cannot view themselves in this way without regarding certain attitudes as their own
and, hence, as attitudes that they ought to take into account together. If the exis-
tence of a reflective rational agent were metaphysically given with the existence of
a human being, it would appear to follow that the attitudes that it ought to regard
as its own, and to deliberate from, are precisely those attitudes that arise within that
human being’s body and consciousness. However, the mere fact that certain atti-
tudes arise within the same human body and consciousness does not guarantee that
7 See Bilgrami, op. cit., for a full discussion of the problem that Oblomov presents and its
significance.
there is a reason to take them into account together. There must be something worth
doing that requires a coordination of effort within the boundaries of that body and
consciousness. Furthermore, even if there is something worth doing that requires
deliberating from just those attitudes that happen to fall within the same human body
and consciousness, those attitudes will not together comprise the point of view of an
agent until it is recognized that there is a reason to take them into account together.
Because agents of human size tend to make the metaphysical mistake of thinking
that their existence is a metaphysical given, they tend to interpret the requirements
of rationality as requiring them to take into account all and only such attitudes as
fall within their own human body and consciousness. But what they fail to see is
that when they interpret matters in this way, they are bringing into existence an
agent of human-size through effort and will, where the unifying project for the sake
of which they are doing so is the project of living a unified human life. The very
fact that a wanton can forego this project shows that it is indeed a project and that
it is optional. Without such a unifying project, the human being literally does not
have a point of view from which to deliberate as one. It does have a bodily point
of view from which it perceives and moves – just as many other animals do. It also
has a phenomenological point of view from which it has access to various episodes
in consciousness – and this too is a kind of point of view that many think other
animals may have. But what a reflective rational agent requires is another kind of
point of view, which is a rational point of view. My argument is that such a rational
point of view cannot exist until its boundaries exist, and its boundaries cannot exist
without a commitment to a particular unifying project that sets those boundaries, by
requiring that all of the attitudes that fall within them be taken into account together.
The argument for the possibility of group and multiple persons makes it particularly
vivid that commitments to unifying projects play an essential role in constituting
agent-identity, by drawing attention to the ways in which commitments to different
unifying projects would yield agents of different sizes.
It may still appear to be paradoxical to say that an agent with its own point of
view cannot exist until there is a commitment to deliberating from a given set of
attitudes together, which then comprise that agent’s point of view. Who, it might
be wondered, could possibly be there to undertake this commitment? Wouldn’t the
agent already have to exist in order to do so, and wouldn’t this belie my claim that the
commitment serves to constitute the agent’s identity? If these questions still seem
troubling, it may help to focus on the case of group agents. Sometimes, separate
human-size agents see the possibility of coordinating their efforts together for the
sake of something worth doing, which requires them to deliberate together rather
than separately. If they choose to do this, they will together bring into existence a
new agent with its own group point of view. The crucial thing to see is that they
needn’t already be a group agent before they do this. It is equally crucial to see
that, once the group agent is constituted, its human constituents will no longer be
reasoning from separate, human-size points of view. Yet from the group’s point of
view, it may eventually become clear that the ends for the sake of which it exists –
the projects that can be accomplished by coordinating effort at the level of the whole
group – are not better than the ends that could be pursued separately by smaller,
94 C. Rovane
human-size agents. This may lead the group agent to fragment back into human-
size agents. But it should not be thought that such human-size agents continued to
exist all along during the life of the group agent. The group agent did have human
constituents. But if the group succeeded in forging a group point of view for the sake
of group projects, then its human constituents did not, at the same time, deliberate
from separate human-size points of view. (I should clarify that a hybrid case of
group and multiple agency is also possible, in which the group does not absorb all
of the rational and practical energies of its human constituents but only some of
them, with a resulting fragmentation in the lives of its human constituents, leaving
some portion of those lives available to pursue smaller projects that do not absorb a
whole human life, that would be pursued by agents of less than human size.)
These reflections on the group case were meant to drive home that points of view
are, quite literally, brought into existence through commitments to unifying projects
that provide reasons to deliberate and act from a particular set of attitudes. But
they also serve to bring out why reflective rational agents ought to engage in self-
evaluation in the strict and literal sense. No one would be tempted to regard group
agents as ends in themselves in Kant’s sense, since they so clearly exist for the sake
of the further ends – the unifying projects – that their existence makes it possible
for them to pursue. It is not reasonable to suppose that once a commitment to such
a unifying project is in place, along with the existence of a particular group agent
whose commitment it is, this mere fact can or should foreclose the evaluative ques-
tion whether the project in question is really justified. Any such group agent ought
to consider what else might be accomplished in its stead through coordinated efforts
within larger boundaries or smaller boundaries. Whenever a group agent recognizes
that it ought to do this, it thereby recognizes that it ought to engage in self-evaluation
in the strict and literal sense. For to evaluate the ends for the sake of which it exists
is tantamount to evaluating the very worth of its existence.
Au fond, I am proposing to model all cases of agent-identity on the group case.
On this model, all agents, no matter what their size, exist for the sake of the ends
that their existence makes it possible to pursue – their unifying projects. It would
clearly be a failure of rationality on their part if they refused to regard the unifying
projects for the sake of which they exist as subject to evaluation in the strict and
literal sense; and such an evaluation of their projects would, unavoidably, amount to
an evaluation of the worth of their very existence.
Some will try to resist generalizing my conclusion about the proper role of
self-evaluation in the lives of group agents to all cases of reflective rational
agency – by setting the human individual apart as special. Their idea will be that
human beings have a status and integrity as individuals that must be respected
and, moreover, respected in a way that is in harmony with the Kantian view of
them as ends in themselves whose worth is not to be compared with anything
else. This idea runs through the entire liberal tradition that places the individ-
ual at the center of morals and politics. It is a nearly universal presumption that
the only individuals with which liberal individualism is properly concerned is the
human individual. They are regarded as the sole loci of individual rights, respon-
sibilities and freedoms. And note that just as human individuals are regarded as
standing apart from the scale on which the values of all other things may be
weighed and compared, so too are the individual rights for which liberal theo-
rists have argued. The whole point of attributing rights to individuals is to set
certain values apart as not to be weighed against and compared to other goods –
even when respecting those rights comes at the cost of other goods. Now, the mere
fact that the existence of human beings is a metaphysical given doesn’t suffice by
itself to set them morally apart as loci of individual rights, whose existence is not
to be viewed as having a worth that can be weighed against and compared to the
worth of anything else. Liberals recognize this of course. And when they try to clar-
ify what it is about human beings that sets them morally apart, they tend to invoke
the very sort of reflective rational agency that I’ve been discussing in this paper. It is
of course qua rational agents that human beings are supposed to qualify as ends in
themselves. Likewise, it is also qua rational agents that human beings are supposed
to be loci of specifically liberal rights. For it is central to the idea of such liberal
rights that their possessors can claim them on their own behalf, and this is obviously
something that only a reflective rational agent can do. The difficulty is, such agents
cannot coherently be viewed in the way that liberals would have us do, as things
that stand beyond or above all comparative evaluation. Because their existence is
a product of effort and will, they are rationally bound to raise questions about the
relative worth of their very existence, which I’ve claimed is inseparable from the
worth of the ends for the sake of which they exist.
References
Bilgrami, A. 2006. Self-Knowledge and Resentment. Oxford: Basil Blackwell.
Frankfurt, H. 1971. “Freedom of the Will and the Concept of a Person.” Journal of Philosophy 68
(1):5–20.
Korsgaard, Ch. 1986. “Skepticism About Practical Reason.” Journal of Philosophy 83 (1):5–25.
Rovane, C. 1998. The Bounds of Agency. Princeton, NJ: Princeton University Press.
Rovane, C. 2004. “What Is an Agent?” Synthese 140 (1–2):181–98.
Chapter 6
Self-Evaluation and Action
Juliette Gloor
6.1 Introduction
In this paper I will argue that to conceive of our desires as simply given to us (in
a special way), as instrumentalists do, is inconsistent on the following grounds:
unless we understand desire and action as being internally and necessarily connected
(thereby following Aristotle) instead of as two causally related entities that can be
identified independently of one another we cannot begin to make sense of the con-
cept of agency. More specifically, to understand animal agency is to conceive of
acting not merely as being causally affected by one’s desires to do a certain thing.
Rather, the active animal is being affected in a way that makes it subjectively aware
of the object of the corresponding activity by virtue of doing the activity. Purely
causal accounts of action, I will claim, cannot account for (the importance of) such
self-awareness since all that counts for them is that some action be taken to satisfy
the objects of one’s given desires; there is no subjective centre of action, and there-
fore no self. Accordingly, some versions of instrumentalism err in their paying too
much attention to the objective side of animal agency while remaining blind to the
subjective side (Section 6.2). For I believe we should follow Aristotle in thinking
that in action, these two sides are internally connected and are just two aspects of
the same thing: an actively conscious being or self. On the basis of this essential
relationship between non-reflective self-awareness and action, I will then offer a
specific understanding of self-evaluation: evaluation of self is nothing else but eval-
uation of the activity of the self, which can be understood as the central aspect
of Korsgaard’s Aristotelian claim that the function of action is self-constitution
(Sections 6.3 and 6.4). Here I shall discuss first self-evaluation as it can be under-
stood in non-human animals (Section 6.3), and then how we should conceive of it
in reflective animals (Section 6.4), while stressing that the difference between them
although crucial, is only one of degree and not one of kind. This is because the bare
self-consciousness in non-human animals is not something different in kind from
J. Gloor (B)
Department of Philosophy, University of Basel, Basel, Switzerland
e-mail: juliette.gloor@unibas.ch

98 J. Gloor
the self-consciousness that accompanies the thinking of human animals. Rather,

self-consciousness in thinking animals like us provides us with a further way of
expressing ourselves. Where self-presence is accessible to reflective self-awareness
(awareness that can take itself as object by being the object) the paradigmatic form
of action or what is expressed in the acting of animals capable of reflective thought
cannot be that of desires; rather it is that of practical reason and it is this that pro-
vides a standard of evaluation. Insofar the subject is guided by practical reason it
constitutes itself well. Against the Aristotelian background of self-constitution pre-
sented in Sections 6.2, 6.3, and 6.4 I will finally sketch the form of self-constitution
supposedly involved in cases where two or more selves constitute a single conscious
community of shared activity (Section 6.5).
6.2 Two Senses of Desires as “Given”

To say that an animal’s desires are given to her comes along rather innocently, one
might think. In this section I want to argue why this is not so by showing that there
are two senses in which one can speak of a desire as given to an animal; but only
one of these captures an essential idea about the conception of agency, namely, that
it must involve the special relation between desire and action. This idea of internal
relation between desire and action, where action is sometimes said to be the expres-
sion of desire, i.e. where desire is the formal cause or form of action, goes back
to Aristotle and has been explicitly defended by Charles Taylor (1979) and more
recently by Christine Korsgaard (2008, 2009a, b). Taylor argues that action is a sort
of “full expression” of animal desire. In cases of full expression what is expressed
(X, say) and its expressing (Y, say) cannot be separated: the expression of X must
manifest itself in Y and cannot be observed on its own or separate from Y. When
an animal eats an apple out of desire for eating and in her eating expresses such
desire it cannot be revealed or be manifested in a way other than through expres-
sion, although of course she could express her desire in other ways than eating the
apple, for example through speaking if she is an animal capable of speech.1 The
animal eating then not merely allows us to “see” through her eating the “happen-
ing” of the desire for eating. Rather she actively expresses it through the activity
of eating (Taylor 1979, 76). To better understand in what sense here exactly desire
is both inseparable from action and actively manifested through action, I consider
it helpful to look at Kosman’s (1975, 2004) interpretation of Aristotle’s concept of
bare self-consciousness.2 Kosman convincingly shows that much speaks in favor of
1 In the case of eating the expressing itself may be seen as source and that which realizes the desire.
Then the desire cannot only not be manifest in any other way than by expression but it cannot be
expressed but by eating.
2 So I distinguish between what has been called “faux reflexivity” or the basic non-intentional
self-consciousness any living animal is supposed to be endowed with and self-knowledge that
only reflective beings can have. As will become clear later, I think that the latter is in fact a
special expression of the former. I further distinguish within human self-knowledge between
6 Self-Evaluation and Action 99
understanding Aristotle to hold the following view concerning consciousness under-

stood as non-reflective consciousness of self: to be conscious of self or to be aware
of self is the same as to be conscious that one is conscious; so self-consciousness
is consciousness of consciousness. But this requires that the conscious subject is
affected or determined by what it is immediately conscious of in a way that makes
it aware of that object. Obviously this is not to say that the animal must have some
cognitively highly demanding awareness of self such as recognizing its own mental
attitudes or even recognizing them as its own if one agrees with Korsgaard (2009b)
that consciousness of self (just like intentionality) is a matter of degree. What is
important here is that since for animals, which are not capable of reflective thought,
their ends are determined by their affective attitudes, the objects of their essentially
receptive consciousness must be objects in the world.
I take this to help us understand the sense in which desire and action are insepa-
rable according to Taylor: animal action or the intentional engagement with objects
in the world is an expression of a living animal’s desire for self-consciousness (or, if
you like, for living as such), which is realized by taking an object outside itself.3 The
desire for consciousness is for each animal a desire for his or her own conscious-
ness and this “is a necessary part of the structure of that desire” (Kosman 2004,
140). It is in this sense that animal action must be self-conscious. However, desire
for such consciousness must be desire for active consciousness or an active exercise
of consciousness – in contrast to merely possessing the power or capability of such
exercise – insofar as it must be directed towards some object in the world through
which the subject’s open consciousness – which is “not in and of itself anything” –
is determined (ibid., 141).4 This is why receptive consciousness is not merely a mat-
ter of the subject being affected but rather a matter of its being affected in such a
way that makes it aware of the object it is affected by in virtue of its intentional
self-knowledge implied in practical reflectivity and theoretical or non-agential self-knowledge

(considering the former essential for self-evaluation). Practical self-knowledge includes both the
evaluation of actions that are completed and the evaluation of ongoing states; after all, self-
constitution is considered an ongoing activity. Self-evaluation in human animals is reflective in
the sense that in a person’s reasoning about what to do she is with herself as object but whose
reality springs from the thinking self and in this sense is spontaneous and not an empirical object
in the world. Of course this is not to say that such a reflective animal will not have to choose
certain empirical identities in virtue of its constituting herself. Acknowledging such choice impor-
tantly involves acknowledging the person one is while at the same time aiming at the ideal of
one’s particular identities (being a good mother, a loyal spouse, etc.). See also Anita Konzelmann’s
introductory remarks about Socrates’ appeal to “know thou self”.
3 Note that desire for self-consciousness is not part of an overtly intellectualistic account of animal
self-consciousness. It is rather a metaphorical way to describe the instinct or drive for self-
preservation that all living animals share. Moreover, “taking an object” just means to be directed
towards an object in one’s intentional activity. By “intentional” I do not mean a straightforward
propositional capacity, but a capacity that involves mind guidedness of a lesser degree. Cf. Gozzano
(2007, 271–82) who argues that just because animal thought is not as fine-grained as human
thought this does not mean that is not thought at all.
4 Activities such as eating, hunting, smelling, etc. are all examples of self-conscious activities that
take their object in the world.
100 J. Gloor
activity. I take this to be the central idea behind the Aristotelian dictum that every
action or activity5 is accompanied by an act of apperception: so to perceive is
to be aware that one perceives. In other words, non-positional self-consciousness,
also called subject-consciousness, is a necessary condition for conscious experience
(both physical and mental), which is supposed to be a hallmark of all living ani-
mals, or at least the higher animals,6 and not just of human beings. This is not quite
correct if one assumes that it is only the conceptual that transforms mere percep-
tions and intuitions into an organized pattern, an experience that is something for
the subject. When I say that animals can have experience I do not thereby mean
that they have full-blown concepts of the objects they represent in doing something;
rather I follow Alva Noë (2004, 183) who argues that animals possess something
like “proto-conceptual skills” (sensorimotor skills are often given as an example).
However this may be, I think what is clear is this: although the animal is not think-
ing about the purpose or object of its acting, which is given to the animal by its
affective states, there is nevertheless something what it is like for the animal to eat,
to hunt, or to look after its off-spring. Likewise, I will argue, it is something what it
is like for human beings to act for reasons. More precisely, I think that while all our
experiences, sensory and cognitive, are necessarily permeated with non-transferable
subjectivity, reflective animals qua reflective thinking are capable of a further way
of expressing their subjectivity.
I think I can best explain what I mean here by presenting an example for the
Aristotelian claim that action is accompanied by an act of apperception. The point
can be illustrated as follows: to perceive some object in the world, it is argued, is to
be affected in a way that makes one aware of it through perceiving it.7 In virtue of
perceiving the object the subject is being affected by a representation from the inten-
tional object of its act of perception. This is why the act of perceiving some object
(that is, positing the object) cannot be separate from the awareness that one perceives
the object. This is probably what Aristotle means when he makes the somewhat per-
plexing claim that the perceiver becomes one with the object perceived in the sense
that she takes on the sensible object’s form, its visibility. Let me try to put it less
metaphorically: in virtue of the structure of an animal’s self-consciousness (con-
sciousness of consciousness) she is drawn in her receptive engagement with the
world to unify herself with or to be determined by the object to which she sus-
tains an intentional relation in her acting. Although this “power of receptivity”,
“a power of openness to determination by the other”8 is an essential feature of
all self-conscious beings, I think it takes a special form of expression in reflective
animals, which are essentially determined not by objects whose reality is outside
5 I do not differentiate here between action (as full-blown) and mere activity but use the terms
“action” and “activity” interchangeably.
6 I will say more below about how we might distinguish between higher and lower animals and
why we can speak of the movements of the former as actions.
7 See for example Kosman (1975, 2004).
8 Kosman (2004, 144).
themselves – objects they perceive as other – but by objects whose reality is included
in the self as reflective agent. Therefore, I would like to suggest that human animals
are essentially determined by selves – either by themselves or by other selves – in a
way non-human animals are not. At this point of the argument I confine myself to
a sketchy summary of what I mean by this. Due to the openness of consciousness –
which has no determinate nature of its own apart from the capacity to become what-
ever it takes as object – the desire for one’s own consciousness must be a desire for
further determination either expressed in the animal’s non-reflective action or, if it
is a reflective animal, in its reflective action formed by its reasoning activity. While
in the former it is a desire to actively exercise consciousness’ receptive power with
respect to an object in the world, this power finds a further and more active expres-
sion in reflective animals since such an animal is self-aware of what it is doing not
so much from perceiving, or if you like, receiving, some object outside itself in
virtue of its active engagement with the world, but from reasoning about what to do
thereby sustaining a relation of self-reference to its object of reasoning.
The more general point here is that Taylor’s and Aristotle’s claims concerning
the internal relation between desire and action help us see in what sense animal
action is truly intentional: animals who are conscious of what they are doing in the
sense described exhibit intentionality insofar as they entertain a subjective point of
view. For this reason it would be confusing at best to substitute intentional talk with
functional talk, where the latter is understood as making no reference to an agent
or subject with that intention or purpose. In other words, to describe some action
not merely as functional but as intentional is to attribute it to an agent, that is, to
an animal that in a certain way knows what she is doing when she acts. This might
involve in animals incapable of (proto-)conceptual thought, as Korsgaard (2009b)
notes, the ability to categorize objects as objects of a certain purposive type: the
predator as danger or the apple as food, say. There is clearly a purpose involved
in the fact that an animal represents a predator as something “to be fled” or an
apple as something “to be eaten” even though the animal may not be able to think
about the purpose itself. This describes a kind of activity, which we are probably not
ready to attribute to lower animals such as spiders. Although insects exhibit reactive
attitudes towards their environment, these attitudes are probably not necessarily tied
to conscious awareness of their environment but rather are mere perceptual instincts
that cause the insect to adequately respond to its environment (so here one might
talk of an organism’s functioning).
Once we attribute to an animal a centre of subjectivity and thereby the capability
of intentional action, however, there is also room for meaningfully enquiring as to
what the animal is really doing or trying to do, which rules out that such animal
behavior is described as mere causation.9 I think this is one way towards under-
standing Taylor’s claim that desire is actively manifested through action. Although
a non-reflective animal’s ends are determined or given by the animal’s desires and
9 Gozzano (2007, 275) discusses an example for the latter along the lines of a water tank indicating
the presence of water.
102 J. Gloor
instincts, in the sense that the animal does not have the choice to act or not to act on
(the basis of) them, the animal in virtue of instinct and learning directs her actions
towards these objects, she is not merely affected by them. When such behavior is
flexible in this sense we may not always be able to simply see or predict from observ-
ing the animal what it is really doing. This may also account for our believing that
higher non-reflective animals differ from lower animals with respect to their ability
to perceive objects in the environment as “options” or means for goal satisfaction.
For it seems that consciousness of something as a goal or end requires consciousness
of means that satisfy the goal.10 So we may follow Korsgaard (2009a) in under-
standing an animal’s desires as that through which she sees or conceptualizes the
world where this conceptualization of the world is not open to interpretation by
the animal but is determined by her affective states. Note that to think of higher
animal desires as given in this way is compatible with claiming that animals expe-
rience the objects and purposes of their desires as desirable or to be done. After all,
an animal’s desires function like a perceptual instrument through which the animal
can represent and structure its environment.11 But from the fact that animal desires
determine their goals it does neither follow that (i) the desires for some goals are
themselves mere objects for the animal – that is, are given to the animal as mere
objects – nor that (ii) the desires determine the animal such that the animal just is her
desires.
If (i) were correct, desires would be given to the animal as objects, i.e. as merely
desired and all that counted would be that some action be taken to satisfy the objects
of one’s given desires. That this is not merely a matter of having or lacking metarep-
resentational abilities becomes clear once we recognize that creatures whose desires
or mental states are given to them as objects lack a self regardless of whether they
have sophisticated metarepresentational abilities or not. An overtly contemplative
creature who is such that it has self-knowledge only in virtue of taking its own men-
tal states as objects of theoretical investigation (as result of brain manipulation, say)
would not differ from a higher animal whose desires are given to her as objects with
respect to the following fact: neither of them would have a subjective centre, a cen-
tre in virtue of which it is something it is like to act and have desires and emotions.
Neither of these creatures could be said to have a self. The desires of each creature
would be perceived by the creature from the point of view of an observer as hap-
pening to someone who is identical with her (on what grounds, actually?), but not
as having anything to do with her.
If (ii) were correct we could not speak of an agent or subject of action either, for
all that we know, it is the desires or other mental attitudes that directly cause the
10 Thanks to Fabrice Teroni for pressing me to be clearer on this point.

11 To object that this is impossible given the fact that desires and perceptions have different direc-
tions of fit, I believe, is to fail to see that an animal’s desires and perceptions must be intimately
related for the simple fact that the way such an animal perceives the world must be related to the
animal’s needs and desires in order for the latter to serve as the animal’s guide to engage through
action effectively with the objects of this world. This just is how non-reflective animals that do not
have propositional attitudes differ from reflective animals like us.
action and then there is no room for an agent or self. It is hard to make sense of
such action as intentional at all, since the attribution of intentional action seems to
require some alignment between the animal and the object it is directed towards in
acting in such a way that the animal’s perception of the environment and its objects
be able to guide her movements in the environment. More precisely, this means that
for a movement to count as an action and not as a mere reaction to some stimuli
(which we might attribute to plants and lower animals) we have to think of animal
movement as responsive to the animal’s representation of her environment.12
However, to see that neither of the descriptions (i) and (ii) of our animal nature
can be adequate does not require much philosophical deliberating. It suffices to
recall that desires, emotions and sensations are something for us, something what
it is like for us to have – we are not indifferent to having them.13 Perhaps it also
helps to notice that this “what it is like” has in the philosophical literature first
been identified with bodily sensations, in particular, with bodily pain. The sensa-
tion of pain – what it is like to undergo the experience of pain – is inseparable
from pain: the existence of pain necessarily is conscious existence. But that which
makes the experience conscious I think is the fact that physical pain cannot be
grasped unless it is imagined together at the same time with our body’s condi-
tion (just as emotional pain cannot be grasped unless it is imagined together at
the same time with our mental or emotional condition). But this is just another
way of saying that to be self-consciously aware of something, pain sensations
included, just is to be aware that one is aware of that something. Through conscious-
ness the animal subjectively experiences the object of the corresponding activity
through doing it.14 A story about the phenomenology of action and agency therefore
cannot confine itself to a description of what happens on the level of neurophys-
iology. What we need is an adequate description of what the subject does when
she acts.
The problem of the defender of action causation I criticized above is that she
does not seem to be able to accommodate the assumption that the explanation of
action cannot dispense with the existence of a subject or centre of consciousness. For
without such a subject or self, one cannot refer to the animal’s representational and
structuring activity of perceiving certain aspects of its environment through desire as
“to be fled” or “to be eaten”. The point is that even though desires are given to non-
human animals in the sense of being learned or instinctual, the desires’ objects are
nevertheless experienced through the active form of desire, that is, as desirable (and
12 Cf. Korsgaard (2009a, 93–99).

13 While for non-reflective animals their mental states matter for them through their taking certain
objects and thereby connecting them to those objects in experience, for reflective animals the mere
having of certain mental states may be all that matters.
14 But surely, you might want to object, experiencing physical or emotional pain is not something
one experiences through doing it! To understand experiencing pain as activity, however, is sup-
ported by the consideration that by suffering from pain we are, as Korsgaard (1996, 148) puts
it, “in the grip of an overwhelming urge to do battle” either against our sensations if the pain is
physical or against the world if the pain is emotional.
104 J. Gloor
not merely as desired).15 And this means, as we have seen, that the agent cannot
be either reduced to her desires or be alienated from herself in such a way that she
merely is the observer of “her” actions and mental attitudes. Thus, if we want to save
the concept of agency and intentional action we must save the idea of the concept
of self or subject that is held together, so to speak, by the intimate relation between
desire and action. If we understand the concept of intentional action as necessarily
connected with the point of view of the agent to whom we attribute the action,
then it comes as no surprise that we should be interested in what it is for the agent
herself to act. This can be captured in different ways, but one especially illuminating
way of doing so consists in focusing on the internal relation between action and
desire: that is, by the animal’s characteristically subjective mode of access to its
desires in action (i.e. in taking the objects of her desires), which cannot be captured
even by the most sophisticated general account or category of cause (Taylor 1979,
84). From this it also follows that there must be a connection between the animal’s
mental attitudes (or her mind, if you like) and her actions while she is acting and
not just prior to acting when she starts desiring something.16 This is not to argue,
as Taylor (ibid., 84) emphasizes, that desire and action are not or even cannot be
cause and effect. Rather it is to claim that action and desire in the normal or basic
situation cannot be “identified as separable components in this situation” (ibid., 84).
The situation considered basic or normal is that of unconstrained action where the
animal does what she wants or desires. But in this situation, the desire cannot be
identified without reference to the action of which it is the formal cause. In other
words, the desire here can only be experienced in relation to the action in which
it is embodied as form. Therefore it makes no sense to analyze unrestrained action
(or “non-deviant” as such actions have been called in the Davidsonian tradition) in
terms of a causal relation between independently identifiable things.17
6.3 Self-Constitution as the Function of Action

and Non-reflective Evaluation of Self
What our discussion concerning animal action in general so far has shown is that ani-
mal action can be characterized as expressing the form of desire, or in other words,
that animal action is inherently connected with desires whose ultimate source is
15 This is of course not to say that when a non-human animal is conscious of some object as
desirable that she is conscious of it in the sense that she understands the object as worthy of desire.
Rather, what is meant is that the animal is able to categorize objects as objects of a certain purposive
type or whole: water as drink, a dangerous animal as to be avoided.
16 For a similar criticism see Mossel (2005, 130).
17 Korsgaard (2009b, 3, fn. 1) makes the same point with respect to the relation between intention
and action in arguing that it is a mistake to assume that just because intention and action are two
separately identifiable things in defective cases of action they must be separately identifiable in
non-defective cases of action as well.
the living animal’s desire for self-consciousness, a desire for his or her own con-
sciousness (Kosman 2004, 141). Therefore, such an animal can be characterized by
the activity of self-constitution through which she is constantly satisfying her desire
for self-consciousness: the animal maintains and reproduces itself through doing
things, i.e. it maintains and reproduces its form or what she is through eating, mov-
ing, sleeping, etc. without performing any reflective activity. So a living animal has
a self-maintaining form in virtue of which it is and continues to be what it is. A
living animal is therefore its own end to the realization of which all the animal’s
activities are directed (Korsgaard 2009a, 35 ff.). So what Korsgaard claims with
Aristotle is that the function of action is self-constitution through self-maintaining
activities.18 Recall that desire for consciousness according to Aristotle is desire to be
determined in consciousness through the activities of one’s consciousness’ objects.
This is just another way of saying that desire for consciousness is a continuous
activity, making oneself into that which one is. And that which one is, if we follow
Korsgaard and Aristotle, is one’s nature whose form is expressed in the form of the
animal’s actions. Along these lines of thought self-evaluation can be understood as
the evaluation or assessment of how well or how badly an animal constitutes its self,
its nature, through its activities. This provides us with the standards according to
which we can evaluate the movements of the animal.19 If the animal succeeds in
bringing about the end or state of affair she intends to bring about (for example,
catching her prey) in the way the animal is organized to do so (by running after
the prey and killing it, say) instead of by chance or some external determination
where the effects of her action are not her own, then the animal is guided by both
a “standard of efficacy”,20 a norm which she can fail to live up to (for example by
falling over and thereby letting the prey escape) and by a standard of autonomy.
When the animal fails with regard to the latter, then she fails to constitute herself as
the agent she is in a deeper sense than when she fails with regard to the standard of
efficacy. An animal’s action lacks autonomy if it is not merely ineffective but “the
wrong sort of thing altogether”.21 A non-human animal that falls short of autonomy
in her action may be an animal whose environment has changed so radically that the
animal’s form, her instincts lose their self-maintaining function. But under normal
circumstances the animal conforms to the standard of autonomy since in the normal
case she causes her own movements in acting, that is, she acts according to her own
nature: namely, according to the organization as the thing she is. She determines her-
self to be the cause of some end (although not self-consciously in a way that only
reflective animals are capable of). It is the animal that causes her movements, and
18 Note that to talk of the function of action here does not square with intentional talk. To refer to
the function of action as self-constitution just is to say that we cannot speak about the function of a
living thing (to constitute itself) without referring to it as a subject intentionally and continuously
constituting itself through acting.
19 Mossel (2005, 143–51) discusses similar issues of standards of action under the concept of what
he calls “agent control”.
20 Korsgaard (2009a, 82).
21 Ibid., 102.
106 J. Gloor
not, as some philosophers seem to think, the animals representations and desires.
Nevertheless it is true of course that for such an animal since “the laws of its own
causality”, its “will”, are her instincts and her desires, it is they that determine the
animal’s ends.22
Now we can also understand why for animals, which are not capable of reflec-
tive thought, their ends, which are determined by their affective attitudes or their
instincts must be objects in the world, that is, objects in the animal’s environ-
ment: for the animal conceptualizes her environment only through her instincts (and
not also through concepts and reasons). This can be of help to her only if those
instincts can be satisfied by that very environment and its objects: the animal’s self-
constitution is therefore essentially dependent on the world it lives in.23 Moreover, if
acting just is making oneself into oneself presupposing a standard of autonomy, then
means-end behavior receives a new meaning, too: an animal that takes the means
to her end determines herself to bring about the end. It is not just something that
happens to her even though the ends are given to her through the animal’s instincts.
I have said that self-evaluation can be understood as the evaluation or assessment
of how well or how badly an animal constitutes itself, its nature, through its activ-
ities. This needs some clarification whose need arises from the following objection
based on the thought that perceptions (or perceptual beliefs) and intentions have
different directions of fit.24 The objection is that we cannot here talk about evalua-
tion at all (not to say about self-evaluation) since an animal’s instincts and affective
perceptions, which provide the animal with empirical information about the world
are very much unlike anything, which can be subject to genuine evaluation, sub-
ject to considerations of worthiness. This objection makes sense only, however,
if one understands evaluation as conceptually linked to moral responsibility and
rational obligations. But I don’t think we should reserve the concept of evaluation
only for the assessment of human action, which is indeed characterized by moral
responsibility. Once we agree to ascribe agency to animals in the way I have sug-
gested we should do following Aristotle and Korsgaard, then we do have standards
by which we can assess their movements as actions, namely in terms of how well
they contribute to the animal’s self-constitution, that is, how efficacious and how
autonomous the animal’s movements are. Just because the nature of the animal does
not involve morality and full-blown rationality, that is, just because animal action
has a different form from human action, this does not mean that it cannot be sub-
ject to (self-)evaluation. Moreover, the objection is misplaced in a second way if it
implies that intentions unlike perceptions are subject to evaluation. For as we shall
see more clearly below, strictly speaking, the description of the intention or purpose
22 Ibid.,104. Note that this does not confine us to argue that an animal is nothing but her desires
and instincts. Much rather, the animal’s desires and instincts are her will, that which is constituted
by her actions (ibid., 100).
23 See Korsgaard (2009a) on this point, 102.
24 See also fn. 12 in this paper.
on its own, i.e. without involving the corresponding means or act, cannot be subject
to evaluation.
But the objection is helpful in its rightly pointing to the essential difference
between human action and non-human animal action: only the former can be subject
to self-evaluation in the sense that only human animals are able to evaluate them-
selves and their actions reflectively, by reasoning about the value of their choices,
that is, the value or worthiness of their whole actions (I will say more about what
this means in the next section). A non-reflective animal does not choose her action
in this way, as something that involves both means and end; her ends are given to
her by her instincts. This is why she is not free in the way a human agent is who
chooses not merely the means but a maxim that entails a choice of means to ends.
Nevertheless, a non-reflective higher animal, as we have seen, is self-conscious in
the sense that she is aware of or subjectively experiences her bringing about the end,
which her consciousness is directed towards. For a non-human animal to be success-
fully the cause of her ends, she must perceive or represent her environment in such
a way that it contributes to the animal’s self-constitution. The way it achieves this is
through being efficacious and autonomous. Moreover, this suggests that animals do
evaluate their own activities insofar as they flexibly react to failures to conform to
the standards of self-constitution. So I agree with Taylor (1979, 86) that an animal
which succeeds at self-constitution experiences her actions as qualitatively different
from defective actions in which she fails to do what she desires to do, be it due to a
failure25 in efficacy or autonomy.
6.4 Reflective Agency and Self-Evaluation

In this section I am going to elucidate what it means to say that only human animals,
reflective animals that is, choose which incentive to act on. The upshot of this is that
the form of action of human animals is reason and not desire or instinct. So for
reflective animals the subjective mode of access to their own reasons for action is
self-conscious reflectivity. But this is not to say that our desires and instincts do
not play any important role for our agency; after all, we are animals, too. It just
means that there is a sense in which human animals are given their desires in a way
essentially different from how desires are given to non-human animals. Only the
latter are capable of acknowledging their incentives as reasons. That is, only they
are capably of choosing or deciding to act. But before arguing for these claims in
more detail I would like to briefly reconsider a version of the second of the two
causal views about action and agency, which I have criticized in Section 6.2 on
general grounds, that is, not specifically with regard to human action and agency.
So I will now briefly look at a version of this second view, which I think to be still
25 Itshould be clear by now that I do not intend to mean by “failure” the kind of failure we
employ when talking about moral or rational failure for which human beings are supposed to
be responsible.
108 J. Gloor
very prominent among many philosophers today, and I will do so with reference to
human action and agency in particular.
The view I argued against holds that desires determine the animal such that the
animal just is her desires, which in turn cause the action. A less extreme version
thereof holds that it is the essential part of an agent, which causes action. Note that
the fundamental claim of this view does not relevantly change relative to what we
insert for “essential part”. One might for example argue that with regard to human
agents that there is a special desire, such as a desire for self-understanding that
causes the action.26 But this changes nothing as to the fact that the agent as a whole –
unlike her mind or the centre of control, if we want to call it that – is not given any
role in action causation. For once the intention or desire to do something is in place,
the causal chain of events is released, where the person “serves only as the arena
of these events: he takes no active part”.27 But we shall see that to conceive of the
connection between the person and her action in this causal way, that is, in terms
of the causal connection between the person’s mental attitudes (i.e. that which is
supposed to actually amount to being the essential part of the person) and her action
is to be unable to account for what makes a person’s movements her own. But if this
is true, it will turn out, the view cannot defend the essentially instrumental nature of
human action either.
Donald Davidson is one of the most prominent defenders of a causal theory of
action and his model of intentional action and instrumental rationality has been most
influential. The model is supposed to explain how in the right way a desire-belief
pair, the essential part of the agent, causes the agent to act. This involves giving
the primary reason, the reason for which the agent acts. A primary reason for an
action is a belief-desire pair that is both the action’s cause and that which motivates
the agent in his acting. Intentional action on this view is a bodily movement that
is caused in the right way through belief-desire pairs, which rationalize it: primary
reasons explain actions in just the way in which they are causes for those actions.28
So action-causation works qua reasons. Admittedly, this rendering of the standard
model of intentional action is much too brief and fragmentary and thus fails to do
justice to all the intricate and twisted arguments that are dominating discussions
until today. However, I think it does show the main problem with which such a
model of intentional action is confronted: namely that of explaining how the agent
participates in action, that is, how it is that she and not merely her belief-desire pair
produces the action.29
26 Velleman (2000).
27 Velleman (1992, 461).
28 See Davidson (1963). Note that on such a view explaining or rationalizing is a form of causal
explaining while the reverse is not true: not all causal processes are rational processes.
29 Of course I am not the first to note these difficulties and the literature here is simply too vast for
me to give an overview. My contribution here is to highlight one specific but fundamental difficulty
for views of action causation in general, which also bears on the very possibility of self-evaluation.
Although the model is perfectly right in assuming that desires and incentives
operate causally on human agents the conclusion drawn from this is defective. After
all, we are part of the causal world, and to say what human action is and explain it
just is the endeavor to insert ourselves into this causal picture meaningfully. But it
does not follow that desires together with the right beliefs – something within the
agent and not the agent herself – cause the action. Although I have said that the
same is true for non-human animals it is perhaps understandable why one might
grant this, if one does, only with regard to human beings: only human beings are
aware of how their mental attitudes operate on them and how these attitudes tend to
move them towards acting in a certain way. Since we are aware of the fact that our
desires operate on us causally, we are in a position to influence our actions by way of
accepting or denying our desires to be reasons for action. In following the Kantian
picture I have ascribed to Korsgaard, this is just another way of saying that the form
of human action is reason, i.e. that we must decide whether we want to go along with
some desire or not based on a universal law: we have to act for reasons or universal
principles. This is also why what is relevant in the explanation of human action is
not the desire (as cause) itself but the representation of a desire as a potential ground
for action. But such activity needs an agent. What I mean by this is that the agent
although she is presented with desires for action that arise from her animal nature, it
is she who decides whether for her this desire is a ground for action or not. Of course
I may be deluded in thinking that I have a certain desire and then consequently I
would not really act on that desire if I decided to act on it but on another; but this
changes nothing about the fact that my human nature requires me to act on the basis
of reasons. Moreover, this is not to say that I always have to ask myself before acting
whether this desire really is a reason for action. Normally, I just act without explic-
itly reflecting on what I ought to do or what I am doing. Rather, when I do ask myself
what I ought to do or what I am doing here, this indicates that there are grounds for
doubt that the desire is a reason for action. It is especially in situations as these that I
experience the force of a reason as something what it is like to be me, to be a human
being.30 Surely, this would not be the case if action were simply caused by the essen-
tial part of me. This would also make superfluous the concept of motivation for this
requires an agent. But if we want to make sense of motivation as essential to our
action (explanation) – which probably few action theorists would deny – and if we
believe that there is such a thing as motivated action, our actions cannot be caused
within us without our own contribution. The Kantian answer is that it is the agent
who causes herself to act in that she accepts or rejects her desires as grounds or rea-
sons for action. This is obviously the agent’s own contribution. But if this is right,
then motivation cannot exist without self-determination. If the defender of the stan-
dard model of intentional action agrees that action explanation essentially involves
30 Of course I am not only that (but also a daughter, a philosopher, a cook, etc.) although my
practical identity as a human being is the most fundamental. Cf. Korsgaard’s discussion of practical
identity in her Sources, pp. 120–30.
110 J. Gloor
motivation, then it seems she also has to explain the possibility of self-determination
in action.31
Now we are in a position to appreciate the special autonomy of human agency,
which is expressed in the form human action takes. I have said with Korsgaard that
non-human animals act according to their own nature, according to the organization
as the thing they are. They are autonomous insofar as they are the cause of their ends
even though they do not choose their ends. A non-human animal, as we have seen in
Section 6.2, is aware of what she is doing, although not reflectively so: she is aware
of her doings by being affected in her acting by the sensible objects of her acting.
But with respect to reflective animals “the laws of its own causality”, its “will”, is
not her instincts but the principles of practical reason of which the reflective animal
is aware self-consciously in the sense that she is aware of the grounds for action
as her own. This is another way of saying that the activity of reflective thinking
expressed in human action is in a sense deeper than the activity of desiring expressed
in animal action. Although the human agent engages with the world and its empirical
objects, the objects of her reflective thought are not drawn from perception but from
reasoning itself, which is the agent’s own doing. This is another way of making
my point that animal action takes the form of desire while human action takes the
form of reason. This has an important implication: non-human animals only choose
the means to their ends but never the means and ends together. This is because
their instincts and desires determine the ends for them. But since human animals
have to decide what to do based on a principle or law, since they have to act on
reasons, human animals do not choose mere act types but whole actions, whose
description have the form of a maxim like “I shall do Y in order to do X”. Therefore,
the form of action or what is expressed in the acting of animals capable of practical
reasoning cannot be that of desire; rather the actions of human animals are guided
by normative principles of practical reason. This means that the laws of such an
animal’s causality are not her instincts but rational principles. Along the Kantian
lines here presented, this means that a human animal’s movements are her own
when she wills her intentions or commitments to do the act A for the sake of the end
B to be universal laws. I here follow Korsgaard (2008, 207–29) in distinguishing act
and action as follows: an action, unlike an act type, is something that belongs to a
class; an action is that which combines acts or act types – like fishing or killing –
with an end or purpose. To fish in order to prepare supper with the fish would be the
action, while “fishing” is the act type and “in order to prepare supper” is the purpose
or end. Autonomous action is good action insofar its purpose and act are related in
the right way. Because there is a gap between what the human animal receives as
a “proposal” for action by her senses and environmental surroundings on the one
hand and her response to that proposal on the other hand, she has to bridge that
gap, which she does by acting for reasons. An autonomous human agent reflectively
31 This is not uncontroversial. Many deny that a motive is like an evaluative judgment. Setiya (2007,
93) for example argues that one must not conflate “reflection as reflexivity – as in the knowledge
of what one is doing and why – with reflection as evaluative thought”.
evaluates whether she should take some desire as a reason for action, which just is
to judge that some action is good or bad.
Recall that the traditional model of intentional action does not hold that act and
purpose are related in this way. “Why did you go to the groceries? – I want to make
an aubergine soufflé and I believe that going to the grocery is the best way to get
good aubergines. Therefore I went to the groceries.” If this primary reason is also
what caused the agent to get up from her chair and actually go to the groceries, we
have an instance of intentional action explanation. Philosophers commonly distin-
guish such instrumental explanation of action from action evaluation, where only the
latter is supposed to involve some judgment as to the goodness of the whole action.
Although I cannot argue for this here, I suspect that this has to do with simply assum-
ing that the only possible explanations for action are instrumental explanations. To
exclude deliberation and evaluation from the context of action explanation is widely
accepted and motivated by distinguishing between justifying reasons and motivating
or explanatory reasons.32 In practical philosophy, motivating reasons are considered
to be those reasons that provide an agent’s motives for action from the agent’s per-
spective (or from the perspective one attributes to her). In giving explanatory or
motivating reasons for action, it is not necessarily to give the good reasons there
were for doing the action. Of course, this is not to say that an agent’s motivational
reasons cannot be at the same time good reasons, i.e. reasons that justify her action
from a more objective point of view, which need not be the agent’s own. So in the
ideal case motivational reasons just are justifying reasons. From this it is often con-
cluded that motivational reasons are more fundamental than justificatory reasons. I
have doubts as to whether this is correct. For if the aim of action explanation is to
make the action intelligible we may have at least sometimes to do more than just
asking the why-question (the question for motivating reasons) because this question
by itself may not be enough in our quest for gaining intelligibility. Consider the fol-
lowing example: your friend tells you that she went on a three-hour drive to buy a
box of chocolates (which she could have easily had within a short distance from her
flat). Now you may ask why she drove such a long way and she answers that she
did do so in order to get a box of chocolate, fully aware of the fact that she could
have had some chocolate just around the corner of where she lives. This is why it
seems not enough to say that the action is defective with regard to the standard of
efficacy without at the same time assuming the action to be subject to the standard
of autonomy. The agent’s failure as it is described here, is not intelligible for us as
mere failure to maximize the adequacy of her means precisely insofar as we have
difficulties to regard the action as autonomous in the first place; that is, the action as
a whole, the relation of its parts – act and purpose – cannot be understood as valu-
able or justified without further explanation. In our search for explaining the action
we are looking for a purpose that makes sense of the whole action, i.e. a reason.33
In other words, it is not enough that the purpose supports the act, it must support the
32 See e.g. Wallace (2006, 68–70).

33 Korsgaard (2009a, 14).
112 J. Gloor
action: “to do A for the sake of B” as a whole must be intelligible. So a good or full
action explanation requires more than just providing the (in)adequate means or act
to some purpose. The standard of efficacy only really gets hold if in principle one
can make intelligible the goodness of the relation between the act (or means) and
the end (or purpose). But if one agrees that the standard of efficacy is dependent on
the standard of autonomy, then the former standard, the norm of instrumental rea-
soning, cannot be the only standard of practical reasoning, something, which most
instrumentalists would want to deny.34
I think that these observations speak against regarding the relation between rea-
son and action as analogous to how a purpose is related to an act, that is, as separate
from or outside of the act.35 Recall that this is the same idea we have discussed with
respect to the relation between desire and action relevant with respect to living ani-
mals in general. We have noticed that the desire for consciousness is for each animal
a desire for his or her own consciousness and this “is a necessary part of the structure
of that desire”. Likewise we can now say that it is a necessary part of the structure
of a reason as the form of human action that the agent is reflectively aware of what
she is doing as being good as a whole: the agent thereby acknowledges or judges
that the action is good. And she is doing it because she judges it to be good. To be
motivated to act is not something over and above or different from being motivated
by the awareness that the particular purpose or end is one that justifies the particular
act in the circumstances, that act and purpose are related in the right way expressing
the value of the whole action.36 It is to be motivated by the particular value of the
object of such choice, i.e. by the value of the whole action. Pace the standard model
of intentional action, we don’t choose acts. We choose act-purpose pairs.37 And if
34 Of course this would need a much more elaborate justification which cannot be done here.
Korsgaard (2008) (especially pp. 27–68, a reprint of her seminal paper “The Normativity of
Instrumental Reason”) has done most I think to defend this claim, namely that instrumental rea-
soning cannot be the only standard of practical reasoning, or more strongly, that the hypothetical
imperative depends on the categorical imperative.
35 See Korsgaard (2008, 226–29).
36 More precisely, it is to be motivated by the fact that to do act A for the sake of purpose B is
a description of an action that is worth doing for its own sake. Of course this is not to say that
motivating reasons cannot come apart form justifying ones, otherwise defective action would not
be possible.
37 This does not imply that all purposes are chosen, that is, choice is possible even when the pur-
pose is fixed, but not the act. Imagine that you are being married at young age because tradition
prescribes it. It seems that the purpose, your marriage is fixed. Once you can see something good
about doing certain acts for the sake of your being married (and perhaps later, for the sake of
marriage) as a whole, then although you have been forced to marry, your choice is free. But this
is not to say that you must absolutely see something good about the whole action. Perhaps your
husband-to-be is known as a rough and condescending man and you just cannot bring yourself to
accept this marriage for yourself whatever you imagine as related means. Your choice then lies is
in your reflectively judging that the whole action is not good and therefore cannot be willed by
you as a universal law. So even if you are forced to marry your will is free (although this sounds
terribly cynical and so a lot more would have to be said).
act and purpose are related in the right way we choose objectively good actions. An
action is objectively good, according to Kant, if its description is given by a maxim
that one can will as a universal law. If this is correct, then practical reasoning con-
stitutive of action is an activity where the agent herself must combine the belief and
the desire (relating to the purpose and the act) in the right way. But then explaining
action cannot imply that reason and action are separate things.
I think that there are mainly two considerations for adherents of the standard
model to think that reason and action must be separate things. First the primary rea-
son as cause of the action must be separate from the action. This separateness is then
interpreted in terms of action and reason identifying separate things. Such a picture
fits well with the belief that reasons and some forms of intentions are mental states
that agents have prior to acting without saying anything as to the mental activity of
an agent while she is acting. Second, recall that the core elements of instrumental
rationality, as understood by the standard model, are given ends or purposes, which
are presented to us as objects of our desires and which are supposed to be satisfied
by way of taking the necessary means. This is I think, why philosophers such as
Broome (1999) argue that intending or willing something does not give us a reason
to take the means to it (we might not have a reason to will the end in the first place),
but only a rational requirement. So once I will some end (even though I have no rea-
son to will it, say), I am under rational pressure to take the means to it. Rationality
and reason are on these accounts thought to be two different things – and it is this
that philosophers of the view I am advocating deny.
I have tried to show on mainly Neo-Kantian grounds why the traditional instru-
mentalist model of action faces difficulties to account for the existence of an agent
or self and why this is a problem, not least because these views as a consequence
also have difficulties in accounting for self-evaluation. In fact, one result of my
argument is that if one agrees with Korsgaard that the function of action is self-
constitution, then agents must be capable of self-evaluation since this is what helps
them constitute themselves well in the first place. If self-evaluation is evaluation of
that which is constituted by one’s actions, then self-evaluation must be evaluation
of what makes an agent what she is. Moreover, I have tried to account for a specific
way in which only human animals are capable of self-evaluation. A human animal’s
action is a movement caused by her reflectively responding to her desires or moti-
vationally loaded representations put before her in treating a certain consideration
as counting in favor of a certain act. To say that a human agent chooses actions
(instead of acts) is just to say that she acts for reasons, the latter being the form of
her actions. Moreover, we have seen that since the will of human agents is not deter-
mined by instincts but by normative principles of practical reason, she constitutes
herself well by following these principles. But then it follows that guiding oneself
in acting by these principles self-consciously (or, in reflective self-awareness) is a
form of constant self-evaluation. And this is only possible if the action is not merely
the (causal) product of the essential part of the agent, but what constitutes the author
of the action. In a lose way of speaking, we can say that an agent is her doings: she
114 J. Gloor
is what she does and decides to do. So the evaluation of herself must be based on
the evaluation of her doings.38
Note finally that the capacity of being self-reflectively aware of one’s own atti-
tudes – for it to be useful for the animal – must come along with a change in the
kind of control such an animal exhibits over her actions.39 Reflective self-evaluation
can thus be considered a new kind of control constitutive of animals, which are able
to look at themselves and their actions as open to evaluation on the basis of the
goodness of their doings as actions, that is, of the way the act and purpose are
related. This openness to reflective self-evaluation, I would like to suggest, is an
openness to determination by self, instead of by something outside the self. This
also means that such an animal is capable of expressing her bare self-consciousness
in a further way: the awareness she has of herself is not one informed by an object
of perception, whose reality is outside herself. Self-determination of the latter ani-
mals involves determination by objects perceived as other, as separate from self. A
reflective animal, however, is aware of herself by being aware of what she is doing
from reasoning about what to do, from being aware of the grounds of her action as
her own. Reflective animals therefore are not determined by objects that have their
reality outside themselves but by objects springing from themselves. This is why
reflective animals or human animals essentially determine themselves.
6.5 Thinking Together and Interacting as Reciprocal

Determination Between Selves
In the previous section I have tried to show in what sense the openness of a
reflective animal’s consciousness is determined by herself and how this relates to
self-evaluation and self-constitution as the function of action. In this last section
I want to briefly look at the related idea according to which thinking and acting
together with others must essentially involve determination by other selves. Kosman
38 Fabrice Teroni has pointed out to me that this sounds like an implausibly strong claim, and
that one might want to object that surely not all self-evaluation is connected to action the way I
described it. I grant that there may be other forms of self-evaluation than the one I describe here.
However, if one agrees with Korsgaard (as I do) that the function of action is self-constitution
and that the standards for how well this is achieved is given by what is constitutive of agency
(claims I here simply assume), then self-evaluation, the evaluation of what is constituted through
action must naturally be intimately related to action. It is the second part of this claim that I try
to make plausible in this paper. Those who are not convinced of the first part of the claim will
not agree with the second part either. But surely criticism of my argument is possible independent
of the assumptions it rests on. What I have tried to do here is merely to make plausible the more
general idea that action and self are intimately related. If this is accepted then at least some form
of self-evaluation must be intimately related to action.
39 Human animals are aware of their own mental states or attitudes as influencing them towards
acting. But this does not imply that human animals never act against their human nature, i.e. that
they never fail to be efficacious or autonomous agents. In fact it seems that they must be capable
of violating these standards in order for these standards to be normative.
(2004) has illuminatingly outlined such an idea on the basis of Aristotelian thought.
Without being in the position here to give an adequate presentation of his view, what
I want to do is draw attention to the special phenomenology of shared conscious-
ness suggested by it, a kind of sharedness that seems notoriously difficult to grasp
for most accounts of shared intentionality.
One such account is Velleman’s insightful analysis of “How to share an intention”
(name of article). His analysis of shared intention is inspired by Gilbert’s (1990)
solution to the problem of interdependent intentions. The problem, which has found
much recent attention, not least in the social sciences, is that of determining what
we are going to do together without invoking an endless regress that springs from
the nature of instrumental rationality. For what I am going to do with you depends
on what you are going to do with me and vice versa. To jointly intend to A seems
to imply that each of us intends to A only if the other intends or wills to A. But
then we get an exchange of commitments of the form “I will if you will”.40 If my
intention or willing to A is conditional on your intention or willing to A and vice
versa then it is not clear how action, let alone joint action, can get off the ground
to begin with since neither of us will act unless the other acts first.41 Velleman
(1997, 39) thinks that the difficulty lies in the fact that through such an exchange of
conditional commitments, the speakers make their action conditional on the action
of the other. Neither settles on the relevant course of action since neither is confident
of what the other intends to be doing. His solution, inspired by Gilbert is this: we
have to understand the speaker’s conditional commitment “I will if you will” not
as making the speaker’s action conditional on the other speaker’s action but on the
other speaker’s commitment. The central idea is that then each of the statements
expressed “I will if you will” together can constitute the categorical willing or joint
commitment.
But it is not altogether easy to see in what sense exactly the commitment
is supposed to be shared. For consider that the first speaker has to express her
unconditional willing in response to the second speaker’s expression of conditional
commitment, which will then bring her to act. Only then will the condition of the
second speaker’s statement be satisfied who then will join the first speaker in bring-
ing himself to act. So the jointly intending to A requires that one of the two speakers
commits herself first in this categorical way, and thus is also the first to act. Note
that the first speaker will have been doing the “joint” action by the time the second
speaker joins in. So it is no accident that Velleman’s (1997, 46–48) description of
the structural exchange of conditional commitments ends with an individual cate-
gorical commitment and not with a joint commitment with the content “Then we
40 Note that Velleman’s individual conditional intentions strongly resemble Bratman’s (1999)
intentions of the form “I intend that we J”.
41 The groundlessness implied by double-bindedness may be expressed as follows: “I will if you
will but since if you will depends on if I will, that is, you will if I will, we get the not very helpful
result that I will if I will”.
116 J. Gloor
will”. The successive build up of these commitments to joint commitment can be

represented as follows:
First speaker (“You”): “I will if you will.”
Second speaker (“I”): “I will if you will.”
First speaker (“You”): “Then I will.”
That is:
A: If B then A
B: If A then B
A: Then A.
But why, we might want to ask, would the first speaker actually commit herself as
she does in taking step three, saying: “Then I will”? Because she recognizes the
other speaker by what she says as truly willing (1997, 47). We find between the two
speakers a “tacit conceding” of mutually recognizing each other as willing, which
itself is based on the tacit assumption of mutual understanding and crediting in com-
munication. What Velleman must mean by this is that the content of the speakers’
conditional statements straightforwardly expresses readiness to cooperate. But such
“mutual understanding and crediting”, it seems, is really based on individual self-
commitment. For speaker A can trust speaker B only insofar as he believes that she
is a lover of truth and abhors making wrong statements. In fact each must hope that
the other is as much a lover of truth, and thus committed to him- or herself, as one
is to oneself. So, the individual commitments adding up to a shared commitment,
although uttered in public or social space, are individual psychological commit-
ments that the speakers, as we might put it, have primarily to themselves and only
derivatively to each other.
Being conscious together in this sense seems to be merely consciousness side
by side, which at some point collapses into shared consciousness, that is, when an
unconditional commitment has been made by one party and then (automatically)
fulfills the condition and therefore the commitment of the other party. So what seems
peculiar about such an analysis of shared intention is that the formation of shared
intention is a result of successive acts of individual intentions, and it is not clear why
the phenomenon thus described is not merely one of a “concomitant propinquity of
two instances of consciousness”.42 Let us try to capture this criticism by recalling
the Aristotelian idea of open consciousness that needs determination by some object.
We have seen that insofar an animal’s instincts and desires determine the objects of
her consciousness these objects must be perceived by the animal as other or separate
from itself. But we have also seen that reflective animals although self-conscious
in this way too, have yet another way of expressing such self-presence, namely
through reflective thinking. A thinking animal is with herself or aware of herself
not from perceiving, which is an activity that takes an object separate from the self.
It is aware of herself from reasoning about what to do whose object is the self.
Similarly, we might now argue in the spirit of Kosman that acting and thinking
42 Kosman (2004, 150).

together means that the selves involved are determined at the same time reciprocally
by each other, that is, by another self. Each active consciousness becomes its object,
that is, another self. Shared consciousness so understood means that the other’s
consciousness becomes in a sense one’s own and the other way round. There is
one single shared consciousness of which each partakes and which is larger than
each party’s consciousness on its own. On such an account of shared consciousness
I have to trust that the other self regards me in the same way as he regards himself,
whereas on Velleman’s account I have merely to trust that the other self regards
himself in the same way as I regard myself. It is not clear to me how consciousness
in the latter account can be shared in the sense of really overcoming the boundaries
between persons or selves, being a reciprocal determination between selves.
On accounts such as Velleman’s it seems that the selves partaking in the shared
consciousness always remain as other to the other party in a way similar that an ani-
mal regard’s the intentional object of its consciousness separate from itself. Shared
consciousness, however, seems to exhibit the same structure as basic self-presence
when expressed by reflective animals: the object of thought brings an animal back
to itself, so to speak. The object of shared thought or shared consciousness then
brings selves back together, into a single communal subjectivity of we-ness instead
of otherness. Interestingly, often such enlarged selves themselves are defined by the
positive evaluation of the very actions by which they are constituted – such as is the
case with friendship.
Acknowledgements I would like to thank Fabrice Teroni and Julien Deonna for their both
insightful and helpful comments to an earlier draft.
References
Bratman, M. 1999. “I Intend That We J”. In Faces of Intention, 142–161. Cambridge: Cambridge
University Press.
Broome, J. 1999. “Normative Requirements.” Ratio 12:398–419.
Davidson, D. 1963. “Actions, Reasons, and Causes.” The Journal of Philosophy 60 (23):685–700.
Gilbert, M. 1990. “Walking Together.” Midwest Studies in Philosophy 15:1–14.
Gozzano, S. 2007. “The Beliefs of Mute Animals.” In Cartographies of the Mind, edited by
M. Marraffa, M. D. Caro, and F. Ferretti, 271–82. Dordrecht: Kluwer.
Korsgaard, C. 1996. The Sources of Normativity. Cambridge: Cambridge University Press.
Korsgaard, C. 2008. The Constitution of Agency – Essays on Practical Reason and Moral
Psychology. Oxford: Oxford University Press.
Korsgaard, C. 2009a. Self-Constitution – Agency, Identity and Integrity. Oxford: Oxford University
Press.
Korsgaard, C. 2009b. “The Activity of Reason.” Proceedings and Addresses of the American
Philosophical Association 83:2.
Kosman, A. 1975. “Perceiving that We Perceive: On the Soul III, 2.” The Philosophical Review 84
(4):499–519.
Kosman, A. 2004. “Aristotle on the Desirability of Friends.” Ancient Philosophy 24:135–54.
Mossel, B. 2005. “Action, Control and Sensations of Acting.” Philosophical Studies 124:129–80.
Noë, A. 2004. Action in Perception. Cambridge, MA; London: MIT Press.
Setiya, K. 2007. Reasons Without Rationalism. Princeton, NJ: Princeton University Press.
118 J. Gloor
Taylor, C. 1979. “Action as Expression.” In Intention and Intentionality: Essays in Honour

of G.E.M. Anscombe, edited by C. Diamond and J. Teichman, 73–89. Ithaca, NY: Cornell
University Press.
Velleman, D. 1992. “What Happens When Someone Acts?” Mind 101 (403): 461–481.
Velleman, D. 1997. “How to Share an Intention?” Philosophy and Phenomenological Research 57
(1):29–50.
Velleman, D. 2000. The Possibility of Practical Reason. Oxford, NY: Oxford University Press.
Wallace, J. 2006. Normativity and the Will. Oxford, NY: Clarendon Press.
Chapter 7
Self-Trust and Social Truth
Keith Lehrer
Self-trust (Lehrer 1997) is, I have argued, central in what one accepts and prefers.
The argument is that the evaluation of belief and desire and the assumption that
one is trustworthy in such evaluation is a condition of reasonableness. I should have
argued that the evaluation of emotion and feeling is equally a condition. The pos-
itive evaluation of belief in terms of background information yields acceptance of
what one believes. Similarly, the positive evaluation of desire yields the preference
to obtain what is desired. Moreover, such evaluation, notably when it is reflective,
involves trusting oneself and, implicitly, the acceptance of one’s trustworthiness
in such evaluation. If one fails to accept that one is trustworthy or fails to prefer
being trustworthy, the assumption that such evaluation is reasonable is defeated. On
the other hand, the combination of acceptance that one is trustworthy and prefer-
ence for being trustworthy, when successful yields the further conclusion that one is
reasonable in what one accepts and prefers.
The question arises of whether one is trustworthy as one accepts and prefers.
Moreover, the further questions arises concerning the success of acceptance in
obtaining truth and the success of preference in obtaining what has value. Now
suppose that one is in fact trustworthy and successful in the pursuit of the goals of
what one accepts and prefers. It is a consequence, that if one is trustworthy and suc-
cessful in what one accepts and prefers, then one is trustworthy and successful in
accepting that one is trustworthy, for that is one of the things one accepts. Similarly,
it is a consequence that, if one is trustworthy and successful in what one accepts
and prefers, then one is trustworthy and successful in preferring to be trustworthy in
what one accepts and prefers. The result is that the acceptance and preference con-
cerning one’s own trustworthiness and successfulness thereof has the feature that if
one accepts and prefers these conditions, then the satisfaction of the conditions, the
trustworthiness and successfulness, explains the trustworthiness and successfulness
of accepting and preferring what one does. That includes the explanation of what
one is trustworthy in accepting and preferring what one does.
K. Lehrer (B)
Department of Philosophy, University of Arizona, Tucson, AZ, USA; Department of Philosophy,
University of Miami, Coral Gables, FL, USA
e-mail: lehrer@email.arizona.edu
120 K. Lehrer
7.1 Trustworthiness: An Explanatory Loop

The loop in the explanation is transparent (Lehrer 2007). The argument could be
reiterated in terms of reasonableness or justification of acceptance and preference.
One defense of the loop is that a loop is a desideratum of the objective of maximizing
explanation. For suppose, putting the argument in terms of reasonableness, that one
seeks to explain why one is reasonable in accepting and preferring what one is
reasonable in accepting and preferring. Suppose further that to achieve that goal
of maximizing explanation one seeks to articulate a theory of reasonableness, ER,
to explain why one is reasonable in what one accepts and prefers. The need for the
loop is obvious when one considers oneself reasonable in preferring to be reasonable
in the way that ER articulates and accepts that one is reasonable in accepting ER.
If one is reasonable in accepting ER, and ER explains why one is reasonable in
accepting everything one is reasonable in accepting, then ER must explain why one
is reasonable in accepting ER itself. If ER fails to explain this, then either one is
not reasonable in accepting ER or ER fails to explain why one is reasonable in
accepting something one is reasonable in accepting. That suffices, perhaps, to give
you a sense of the importance and defense of the loop. I conclude by noting that if
you ask whether it is the acceptance or the truth of ER that affect the explanation,
then I answer it is the loop that is explanatory. The reason is that acceptance of a
theory that is not true fails to explain, and the truth of a theory we do not accept fails
to explain anything to us.
7.2 A Truth Loop of Exemplarization

A second defense of the loop of self-trust is a form of conceptualizing experience
in terms of sensory experience that is more directly connected with truth. This form
conceptualizes a sensory exemplar or sample in a process I have called exemplariza-
tion (Lehrer 1996, 2000, 2006). The process, suggested originally by Hume (1739)
involves generalizing from the sensory exemplar to let it stand for or represent a
class or plurality of sensory experiences, while the exemplar serves to exhibit what
the class of sensory experiences is like. The process resembles exemplification sug-
gested by Goodman (1968), and I am indebted to his account, but with the important
difference that I take the exemplar to be the representational symbol itself rather
than referring to a word denoting the exemplar to achieve the representation. I have
argued that the content of conscious states, most notably sensations to which one
attends, and the content of works of art, again to which one attends, is the result of
exemplarizing the sensory experience. The exemplar is used to represent or stand
for a class of states as result of exhibiting what they are like. As a consequence, the
exemplar represents or stands for itself as a result of the process of exemplarization
and is, therefore, true of itself.
There is a truth loop in exemplarization as the exemplar represents itself as it rep-
resents and exhibits what the class of represented objects is like. Exemplarization is
7 Self-Trust and Social Truth 121
a psychological process, one that is illustrated nicely in viewing of art objects, and
is subject to the fallibility of such processes. Nevertheless, when exemplarization
proceeds in the normal way by using the exemplar to exhibit what the represented
objects are like, it will represent itself. There is a security in such self-representation
beyond that achieved by linguistic representation in the style of Schlick (1959), for
example, because of the way in which the exemplar denotes without the interven-
tion of words. Extension of representation beyond the exemplar will not retain the
security of self-representation. Moreover, the exemplarized concept or content will
be altered and enriched as it becomes connected with a broader conceptual and
inferential network, especially when it connects with linguistic representation of
conception.
7.3 Self-Trust and Social Consensus

A problem arises concerning the individualism of the theory of self-trust.
Individualistic theories of acceptance and preference seem to ignore the social con-
straint of intersubjective sources of information and intersubjective testing. The
reply is that the individual, as a condition of being trustworthy, has to evaluate
the claims of others and change what one accepts and prefers in the light of such
evaluation. The evaluation is, of course, the task of the individual that must incor-
porate the condition of self-trust. Indeed, the appeal to the social constraint does not
enable one to escape the individual burden, privilege and responsibility of evaluat-
ing the reports and preferences of others. If you accept what another accepts or what
another prefers, that brings with it your acceptance of your trustworthiness in eval-
uating them. There is the exceptional case in which, though you lack information
enabling you to evaluate others, you must decide whether to accept or prefer what
they do, nonetheless. One might object to the role of self-trust here arguing that here
need drives you to accept or prefer what others do out of desperation or a lack of an
alternative. The answer to the objection is that need is not necessity, and there is no
way of abrogating responsibility and role of self-trust even in extreme cases.
The argument that I developed to articulate the role of social constraint and inter-
subjectivity is that the evaluation of others is equivalent to deciding what weights to
give to what they accept or prefer in the particular case and circumstances. I assumed
that the set of weights, including the weight one assigns to oneself, are nonnegative
and sum to unity. The simplest model of the role of such weights is obtained by
turning from qualitative acceptance and preference to degrees of acceptance and
preference, assuming the idealization of these degrees as probabilities and utilities.
This is an admitted idealization. Entering into the idealization of individuals assign-
ing weights to others, I offered, with Carl Wagner (Lehrer and Wagner 1981), the
further idealization of the aggregation of probabilities and utilities, in terms of the
weights. If we imagine that individuals average the probabilities and utilities of oth-
ers starting with an initial state applying the weights that they assign, and iterate
such averaging holding weights constant, the process will converge provided that
122 K. Lehrer
the members of the group are connected by a vector of positive weights assigned.
The transition to state x+1 probability, pm x+1 , for member m from state x for m,
pi x , aggregating with the weights, wmi , m assigns to each member i is the following
summation:

i=1
pm x+1 = wmi pi x
n
where the process of iteration converges for the various members aggregating
toward a consensual probability, pc , as x goes to infinity.
It suffices for the convergence that members assign positive weight to themselves
and that there is a sequence of all members of the group such that each member in
the sequence assigns a positive weight to the next member in the sequence. This con-
nectedness could result from each member in the sequence giving positive weight
to only one other member, the next one in the sequence. It can also result from one
member of the group, the central figure, assigning positive weight to everyone else
who in turn give positive weight to that figure.
The fiction of iterated aggregation is suggestive of the way in which radical shifts
occur, paradigm shifts, for example, resulting from members of a group of outsiders
becoming connected with the insiders by a member of the group of outsiders, a
central figure. If that central member of the outsiders is given positive weight by
a member of the insider group, a central figure of that group, to whom he gives
positive weight in turn, connection between the two groups results. The two central
figures of the two groups giving positive weight to each other connect the two groups
into a unified group. The result of unification is convergence toward a consensus for
the larger group.
However, convergence toward consensus overrides the contrast between indi-
vidual autonomy and self-trust, on the one hand, and social conformity and
intersubjectivity on the other. As the process converges, the individual assignments
of probability and utility, of acceptance and preference, converge toward the con-
sensual assignments. Individual self-trust and social intersubjectivity are fit together
by the mathematics. The individual incorporates the results of social factors and
information articulated in the weights assigned by individuals to other individuals,
and, indeed, the weights that they indirectly assign to the weights that individu-
als assign to other individuals including themselves. Similarly, the social consensus
incorporates the results of individual self-trust and evaluation. Evaluation indirectly
includes evaluation of the evaluations of others. The result is a unification of the
individual and society in the mathematical aggregate.
7.4 A Fixed Point of Consensus

It is, however, important to reflect on the degree of idealization of the iterated
averaging and to notice that it is a fiction that, once considered for any explana-
tory suggestiveness of social processes, may be replaced with a mathematically
equivalent representation that puts aside the iterated averaging. Instead, we may
think of the initial situation as one in which individuals assign diverse vectors of
weights to other individuals. This leaves us with the problem of finding the con-
sensual weights, wcm , to assign to each member in the group based on the diverse
weights initially assigned to find a consensual probability or utility assignment. The
need for such consensual weights is to summarize what the information contained in
the diverse weights initially assigned tells us about what weight is socially appropri-
ate on the basis of the total information contained in the social group. Fortunately,
there is an answer that is both natural and mathematically cogent. Suppose we
could find an original vector of weights to employ to average the diverse weights
assigned to a given individual that would yield back the weights in the original vec-
tor. Averaging by the original vector yields back the original vector in equilibrium.
Letting, wcm , represent the consensual weight assigned to m, it is computed from
the original weights members i assigned to m in the initial state 0, w0im , from the
vector of consensual weights, wci , by the summation:

i=1
wcm = wci w0im
n
The vector of consensual weights is a fixed point vector with a unique value given
connectedness. It is a kind of explanatory loop by which a vector of weights yields
itself back as a result. For, wcm , computed on the left must be the weight used
to find the product, wcm w0mm , in the summation. The mathematics of the matter is
that the fixed point vector used to average probabilities and utilities of members
of the group in the initial state yields the same result as convergence of iterated
weighted averaging of those probabilities and utilities described in the original
fiction.
The problem, having noted the convergence of the individual and social resulting
from iteration or application of the fixed point vector, is to offer some argument that
the consensual assignment is successful in reaching the goals of truth and value.
One convinced of the merits of the argument as providing a reasonable or trust-
worthy probability and/or utility assignment needs some further argument for the
conclusion that such assignments are successful in reaching external goals. There
are examples, one paradigm example, in which the weighted averaging is connected
with truth in statistical examples. Suppose that a group of people examine samples
S of a population P to determine the frequency of some characteristic C in the pop-
ulation. Suppose the members m1 , m2 to mn each project the frequencies of C in P,
each member mi records the frequency Fi of C in his sample, projecting it as his best
estimate of the frequency of C in P. Suppose further that the samples are disjoint, no
two samples sharing any members, and that the set of samples exhausts the popula-
tion so that all members of C have been examined by some member of the group.
Now if we assign weights w1 , w2 and so forth to wn that equal the percentage of P
examined by the members m1 , m2 to mn , the weighted average of the frequencies, fi ,
the sum of wi times fi , for each member mi , will give the correct frequency of C in P.
124 K. Lehrer
However, all the example shows is that there is a set of weights, which may be
unequal, that will yield the correct answer. The problem remains to provide some
argument that aggregation of the efforts of individuals to reach truth is effective.
Initially, the problem seemed to me to lack a solution. I remained satisfied with
noting that the fixed point vector aggregated the total information that members of
the group had about other members of the group and, therefore, constituted a rational
summary of such information. Applying the fixed point vector to find consensual
probabilities or utilities was justified by the principle of rationality that one should
use total information to determine what to accept or what to prefer.
7.5 Personalism and Truth
If one takes a personalist approach, the justification above may argue for the ade-
quacy of using weighted averaging because it is formally equivalent to Bayesian
methodology of using priors to average in a way that sustains coherence (Lehrer
1983). However, the desire remains to offer some argument that goes beyond nor-
mative constraints of coherence and rationality to assigning probabilities that are
successful estimates of frequencies. That departs from the personalist approach to
reach beyond coherence to truth. One motive for that desire is that personalist prob-
abilities, aimed at avoiding Dutch book, sometimes lead to assignments of personal
probabilities that one knows are incorrect estimates of frequencies. For example, if
one has two sources of information about the frequency of C in P from two individ-
uals who examined the total population, one reporting that the frequency of C in P
is .9 and the other that it is .1, Bayesian conditionalization, assuming that one assigns
equal priors to the correctness of the reports of the two sources, requires assigning
a probability of .5 to the hypothesis that a randomly selected member of r of P
is C. However, one may be very confident that one of the sources of information has
incorrectly reported, without having any idea which one, that the frequency of C in P
is either .9 or .1, and that it is not .5. So coherence requires an assignment of prob-
ability that the subject may be confident, or even certain, is not a good estimate of
frequency in the population. I do not propose this as a decisive objection to Bayesian
conditionalization, but it shows nicely why someone might desire a methodol-
ogy for what one accepts that has some connection with success in obtaining
truth.
We may formulate the problem about truth in another and more traditional way.
There is a liberal tradition that defends self-trust and individualism in the quest for
truth stemming from Mill (1869). But one is left with the desire to find some argu-
ment other than political liberalism for thinking that individuals following the path
of self-trust in what they accept will lead to social success in attaining truth, in short,
that social consensus will probably trump individual error whatever the vicissitudes
of the path. One might reformulate the famous defense of induction by Reichenbach
(1948) based on the use of the straight rule converging toward truth in the long
run as the rule of combining individual frequencies. That methodology would not,
however, deal with the further issue that self-trust exercised in what one accepts is
usually not based on counting the results in samples. Indeed, one justification for
personalism, an important one, is that our information about the world is usually
represented qualitatively, not quantitatively, and the achievement of personalism is
to provide a method for quantitative articulation of qualitative information in terms
of probabilities. The bottom line is that we have more information about the world
than is articulated as counting frequencies in samples. So, we still require some
methodology beyond the Reichenbach’s method restricted to sampling to provide a
truth connection between self-trust and social aggregation.
7.6 Diversity, Self-Trust and Truth

Recent work by Page (2007) contains an argument that may be expanded to create
the needed connection between self-trust and social success in reaching truth. The
argument Page has advanced is a mathematical proof that social diversity of rep-
resentation and cognitive method has a lower error rate than individual expertise.
To put the matter in the simplest terms, aggregating a smart diverse social group,
diverse in terms of how they articulate a problem and the methods they use to solve
it, will yield more correct results than appeal to a single expert. The implication
is that smart diverse individuals trusting themselves to articulate and solve a prob-
lem will have a higher success rate than a single expert in reaching truth, that is, a
correct answer to the problem. Note that this is not an empirical conjecture. It is a
mathematical truth (Hong and Page 2001, 2004, Page 2007).
The Page argument for social success is based on aggregation that gives equal
weight to each solution of the problem. Acceptance is here construed qualitatively
as acceptance of a claim or hypothesis. Another application would be the averag-
ing of degrees of acceptance wherein each member of the diverse group is given
equal weight. Starting with this theorem, we may ask the further question of what
the result would be of assigning weights in such a way that members of a group
who reflect the same methods of representation and methods for solving the prob-
lem be considered as a group. One modification of the equal weighting scheme, one
person, one vote, or alternatively, a weight of 1/n, where n is the number of mem-
bers in the group, used for averaging, is to assign a weight of 1/g to the acceptance
or probability of each group in the g number of groups. For aggregation purposes,
that weight could be equally divided among members of the group. Mathematically,
the result of clumping the members of the group into a single acceptance or degree
of acceptance, giving it the weight of 1/g in the averaging, or giving each mem-
ber a weight of 1/gc in the averaging, where c is the number of members in the
clump, is equivalent. The latter representation may have the benefit of symbol-
ically respecting individuality while accommodating the lack of diversity within
the group.
126 K. Lehrer
7.7 Aggregation and Diversity

This way of articulating the problem of aggregating diversity raises the question
of whether averaging undoes the benefits of diversity. The alternative to averaging,
even in the quantitative case of degrees of probability, is to count the votes and
choose the winning degree of probability instead of averaging. The advantage of the
winner takes all strategy is that it represents the work accomplished by diversity.
The disadvantage is that it ignores all the information contained in the losing group.
Which is more effective for avoiding error?
Here is a further complication of the issue. Suppose that members of the group
evaluate other members of the group with respect to their chances of success by dis-
tributing a unit vote among all the members of the group, including themselves. Now
the set of vectors can be aggregated, that is, averaged as the fixed point theorem tells
us above, to yield a set of weights for averaging that summarizes the information
that individuals have about other individuals. The resulting fixed point vector sum-
marizes the diversity of ways of representing skills of others and cognitive tools used
to evaluate those skills. Thus, the benefits of diversity are summarized by the fixed
point vector. Again, one could use the fixed point vector as a winner take all choice
mechanism or as a means of aggregating the first level probabilities or degrees of
acceptance. The advantage of averaging is that the winner takes all strategy ignores
the diversity of information of other individuals or groups of individuals that are not
the winners.
The question remains, however, as to whether averaging by weights in the fixed
point vector to obtain a consensual or collective probability has a greater success
rate than a randomly selected individual. To put the matter in another way, we need
to answer the question of when consensual probability constitutes collective wis-
dom about truth. We have some results that bear on the problem from Page and his
collaborators (Hong and Lamberson, in manuscript). First of all, there is the result
that consensual weights are better than simple averaging if and only if the weights
reflect the accuracy rates of individuals and those rates differ among individuals.
Secondly, the consensual weights must reflect the accuracy rates of individuals as
opposed to their charisma. Thirdly, and most importantly, the accuracy of collective
wisdom is determined by individual expertise combined with diversity. Individual
expertise is a function of the method of representation and the models of prediction,
that is, the cognitive models of individuals.
The details of this work are worth the reflection of philosophers, but the results
are what concern us here in considering the role of consensual weights in collective
wisdom. Note that the models take some predicted value, which may be a success
rate of an individual, as the basis for collective wisdom when these values are aggre-
gated. Given this fact, we can see that there is a basis for a multilevel application of
the results concerning collective wisdom to the use of consensual weights. In short,
suppose that we think of the initial state probabilities as estimates of truth frequen-
cies. So let fi (A/P) being the estimated frequency of an attribute A in a population P
by individual i. Each member has an estimate, and the results of Page and collabora-
tors tell us that if the models they use are sophisticated and diverse, then collective
wisdom resulting from simple averaging will have a lower error rate than an average
member of the group. So collective wisdom outperforms individual expertise on the
average. What about diverse weights used for averaging? We can say this much.
Combining individual self-trust and social diversity in cognitive methods, collective
wisdom resulting from averaging with diverse weights that represent diverse success
rates among individuals will outperform simple averaging.
7.8 Consensual Weights and Diversity

The question that faces us now, given some encouraging results, is whether the con-
sensual weights may be used as a measure of diverse accuracy. We move toward
a positive answer by restricting the initial assignment of weights by individuals to
those assignments that are based on the normalization of meta-frequencies, mi (T/S),
being the estimated frequency of the attribute of truth T in the first level estimates of
subject S by individual i. These m estimates, like the original f estimates, may often
not be the result of statistical sampling. The projection of estimates should, after all,
embrace information about the accuracy of a subject and the cognitive models used
by the subject. It is tempting to require that a subject use the same cognitive model
to estimate the m function as to estimate the f function for the sake of uniformity.
But such a requirement appears unrealistic. One might use different models to make
predictions about the success of physicists than one would use to predict the correct-
ness of physical hypotheses. Nevertheless, averaging by using weights that estimate
the accuracy of subjects yields the same results in the aggregate as using weights
to estimate first level frequencies. So, the results concerning the collective wisdom
outperforming individual estimates should transfer to the metalevel. Reichenbach
proved a long time ago that higher order aggregation could, under specified con-
ditions, reduce to first level results. We can see something comparable occurring
here as meta-frequencies are used as estimates of accuracy of individuals in their
estimates of first level results.
The argument above is a blend of subjective factors, estimates, with objective
factors, frequencies. The defense of this has to be the one offered by personalists,
Bayesians, in using preferences for extrapolating to probabilities, namely, that we
have more information to use in estimation, knowledge and truth attainment, than
is represented in counting frequencies in samples. The same point is made at the
second level. We have more information to use in estimation, knowledge and truth
attainment about the accuracy of others than is represented in the counting we have
done in samples of their success. So the proposal here is that meta-frequencies,
though estimates of accuracy, will, when models are sophisticated and diverse, give
us estimates of accuracy that may be used to conclude that collective wisdom, con-
strued as consensual probability, will do better on the average than simple averaging.
One way to look at the estimates of accuracy, one that reaches back toward the orig-
inal result of combined wisdom outperforming individual expertise, is that simple
averaging is analogous to individual expertise, it is one way of estimating among
128 K. Lehrer
many. If the many are sophisticated and diverse, and the fixed point theorem rep-
resents that sophisticated diversity, then the diverse estimates of accuracy should
outperform the individual one.
7.9 Explanatory Circularity, Exemplarization and a Truth Loop

There is an appearance of circularity in the argument. Higher order estimation and
evaluation is used to estimate truth at the first level, and truth at the first level could
be represented without appeal to the collective wisdom grounded in it. In the final
result, the performance of the collective, of the method of aggregation whether
simple or weighted, has to be justified by the success of the method in attaining
truth. So one might ask whether it is the underlying truth or the collective wisdom
that explains the greater rate of success of the collective procedure. The answer is
that there is no explanation without a loop connecting the underlying truth with the
collective wisdom. If you ask whether it is the aggregation of individual expertise
combined with social diversity that is explanatory or the truth it finds, the answer
has to be that the question is like the question: what comes first the chicken or the
egg? As to the question of whether it is the underlying truth that explains why col-
lective wisdom and social diversity prevails over the individual or whether, on the
other hand, collective wisdom and social diversity explains why the truth is found,
one must answer that explanation here, as elsewhere, is to be found in the loop con-
necting the method of diversity with truth (Lehrer 2007). Without truth to be found,
diversity explains nothing. Without diversity, truth is not as likely to be found. It
is, however, crucial to notice that it is the connection between diversity of cogni-
tive models and the attainment of truth that is explanatory. Moreover, the role of
individual expertise, if outperformed by collective wisdom, is, together with diver-
sity, essential to the performance of the collective. Self-trust, collective wisdom, and
social truth are tied together in an explanatory loop.
I want to return to exemplarization to find a connection between representation
and experience, and one-to-one mapping combined with more general representa-
tion that strengthens the truth loop. Both science and art leave us with questions
concerning acceptance and preference. Consider looking at an electron photograph
in science or a painting in art. There is an initial response to a sensory exemplar. The
character of the exemplar becomes representational of what the experience is like.
Each observer receives a sensory exemplar in his or her experience and, as the exem-
plar becomes representational, by exemplarization (Lehrer 2004, 2005, 2006, 2007).
As noted above, the sensory exemplar is generalized, representing at the same time
what the conscious experiences is like, an experience of a color, a smell or a sensa-
tion, for example, that will enable one to identify other experiences that are like the
exemplar. Exemplarization is a kind of representation that provides some security
from error concerning what the experience is like as the result of self-representation
included in it. The exemplar is part and parcel of what it represents. It is part of
what it represents being one of the items represented at the same time that it is a
parcel representing things of that kind. As a result, the exemplar represents itself
because it is used as an exhibit of what the experiences are like. At the level of
exemplarization, there is an initial security from error in self-representation.
7.10 Limits to Security from Error

Security from error achieved by exemplarization does not extend beyond the exem-
plar, however. The exemplar represents a class by exhibiting what members of the
class are like, and, in the role of the exhibit of what is represented, represents itself.
There is safety in self-representation. Page (2007) notes that representation by sym-
bols consisting of a one-to-one mapping of symbols to objects, though it contains
security, fails to give us the cognitive advantages of representation that involves gen-
eralizing to a class of things represented by the symbol. Exemplarization gives us a
model of a cognitive process that achieves the security of a one-to-one mapping, the
exemplar represents itself in a way that is constitutive of the process, while going
beyond self-representation to generalized representation of a class of objects. At
the level of generalization, as Page notes, diversity makes its appearance. There are
some innate and social influences on how the individual generalizes, but there is also
individuality and diversity. At the level of generalizing, representational systems
reveal cognitive differences.
In summary, the security from error in exemplarization brings with it the costs
and advantages of idiosyncrasy and diversity. The exemplar as symbol is like the
use of a word in an idiolect. There may be common factors influencing how we
generalize, some of them may even be innate dispositions, but there are differ-
ences. Moreover, those differences become magnified in two ways. First of all,
the generalization goes beyond employing a personal experience to stand for per-
sonal experiences of the same kind to standing for interpersonally perceived external
objects. A great deal of discussion in philosophy has concentrated on whether the
content of representation is internal or external, that is, on whether sensory exem-
plars with all the idiosyncrasy thereof represent something external, red objects,
for example, or just internal experiences. The dichotomy is, of course, a false one.
Sometimes an exemplar, especially of some new experience, some new feeling or
sensation, may leave one puzzled as to the external sources, indeed, even clueless.
However, we know something about what it is like in itself and may be able to
identify experiences like it because of our memory of the original exemplar. This
identification does not depend on judgments of similarity. Generalizing is a more
primitive reaction than judging similarity and is the basis of the latter.
7.11 Externalism of Verbal Content
Externalists, provided they admit the existence of such internal states, will argue that
they represent external objects. The sensation of red, for example, will represent red
objects and, when no such object is present, they will be erroneous representations.
130 K. Lehrer
Sellars has suggested that the initial external representation is one that becomes
more sophisticated as we learn that the sensation may misrepresent the external
world. But the externalist will insist that, whether the representation is correct or
incorrect, it is a representation of something external. Moreover, the process of rep-
resentation will, in the normal case, connect the exemplar with a word representing
the same external objects. We quickly consider how to describe an experience, and,
even if the exemplar initially plays a unique representational role, it soon becomes
connected with a word. That is the standard picture.
In my opinion, the process of representation moving from exemplarization to
description is not automatic accumulation of content. The representational content
of the exemplar may not be something that can be simply added to the con-
tent of a word. There may be conflict in how the word would be applied and
how the exemplar would be applied, for example. The classification of a new
item may be different when you ask whether the word applies to the new item
than when you think of the exemplar and ask yourself whether the new item
is like the exemplar. However, the connection of exemplarized content with the
verbal content carries the former into a network of verbal and inferential con-
nections. Inferential connections may be added to the process of generalization
of the exemplar before the verbal connections are made. One may think of red
objects, apples and blood, for example, before describing what the exemplar rep-
resents, but the verbal connection is inferentially robust. Here it will suffice to
note that modification of the verbal content may result from the exemplarized con-
tent as well as the other way around. Inferential connections may be added, cut or
reconfigured as we combine sensory conception with verbal conception. Scientific
interpretation of data and critical interpretation of works of art depend on the recon-
ceptualization of experience in language and the reconceptualization of language in
experiences.
There may be cases in which the sensory exemplars become transparent sym-
bols of external qualities and objects. Some sensations of touch, as Reid (Lehrer
1989) noted, may pass through the mind to the representation of some quality, like
hardness, for example, without attention being focused on what the sensation is like.
Other sensations, may call attention to themselves, sensations of smell, for example.
In the latter case, the sensation is exemplarized, indeed, attending to a sensation may
be or essentially involve the process of exemplarization of sensation as an exemplar.
Attending to a sensation involves knowing what it is like, after all, and the account
proposed of such knowledge is exemplarization which yields truth as the exemplar
as symbol applies to itself as an exhibit of what is represented. In the case of smell,
if it is a distinctive smell of an object, the sensation of smell may represent both
what the sensation is like and, at the same time, represent the object giving rise to
the smell, a rose to use a polite example. The idea here is that exemplars may be
Janus faced symbols looping backward onto themselves and forward to an external
object. The conflict between internal and external accounts of content rests on a false
supposition that a sensory state must either represent something internal rather than
something external or vice versa. That conflict is a relic of the failure to recognize
the robust ambiguity of representation pulled toward giving us an account of our
internal life and the external world in the economy of a sensory exemplar.
We noted above that the process of exemplarization, providing it involves the

loop of the exemplar back onto itself as an exhibit of what it is like, brings a truth
loop into the loop of self-trust. If a philosopher asks whether self-trust is success-
fully truth connected, we can offer him exemplarization as providing a truth loop
in self-trust. I argue (Lehrer 1988) that the state of self-observation, of looking into
oneself as Hume (1739) described it, is a process of exemplarizing a sensory state
in such a way that it belongs to the bundle of states of the self at the same time that
it refers to itself as a state in the bundle. Exemplarization allows for a state be a
member of the class of states it represents as an exhibit of them. However, self-trust
incorporating the truth loop of exemplarization, leaves us with a form of represen-
tation that insures truth at the cost of idiosyncrasy. The exemplar is a symbol in the
idiolect of the individual who exemplarizes it.
It might be thought, and externalists have thought this, that the move from inter-
nal states to public language takes us from individuality to conventionality in one
fell swoop. Having swooped, we can then brush all the privacy and idiosyncrasy
of conception into the trash folder and enjoy the pleasures of external representa-
tion. That is a semantics that ignores psychology, including psychology of language.
Individuals differ in how they use words. We may suppose that there is a correct way
to use words and that the reference of the words of the language is determined by
that correct use, and with that the truth conditions, no matter what anyone thinks.
Putnam (1975) took us in that direction and others, having found the path, ran down
it to the coal pit of skepticism for a person about what he means. Starting with psy-
chology, we must acknowledge diversity in how people apply words and the verbal
connections between them. What we have noted above, however, is that cognitive
diversity, that includes diverse ways of representing the world, is an advantage in
avoiding error.
7.12 Communal Language as Logical Fiction

What are we to say, however, about the character of the communal language, the
communal system of conception, if all there exists is the idiolects and the idiosyn-
crasy of exemplars in the individuals? The answer is clear enough (Lehrer 1984,
Lehrer and Lehrer 1995). The communal language is like the average person, in this
case, a weighted average of consensus we have characterized above. It is a mathe-
matical fiction, but it is factually grounded in idiolects and weights that individuals
assign to each other, directly or indirectly, to find the fixed point of aggregation.
Moreover, as individuals interact, as they modify usage in terms of the weights they
give to each other, the communal consensus of sign and concept, is approached by
in the individuals who converge toward it as they aggregate with others. What we
have learned from Page and his colleagues is that the diversity resulting from self-
trust is an advantage rather than a disadvantage in obtaining truth and avoiding error.
Part of the reason is that self-trust contains a truth loop of exemplarization and the
application thereof. That security of truth does not take us far into the external world
132 K. Lehrer
of the interpretation of science and art, but, starting from an internal truth loop and
self-trust in what we accept and prefer in the external world, we may expect our
cognitive differences to be an asset in avoiding error. The truth loop of self-trust
expands beyond the egocentric to the cognitive diversity of enlightened inquiry.
References
Goodman, N. 1968. Languages of Art. Indianapolis, IN; New York: The Bobbs-Merrill Co.
Hong, L., and S. E. Page. 2001. “Problem Solving by Heterogeneous Agents.” Journal of Economic
Theory 97:123–63.
Hong, L., and S. E. Page. 2004. “Groups of Diverse Problem Solvers can Outperform Groups
of High-Ability Problem Solvers.” Proceedings of the National Academy of Sciences 101
(46):16385–89.
Hume, D. 1739. In A Treatise of Human Nature, edited by D. F. Norton and M. J. Norton. Oxford;
New York: Oxford University Press, 2000.
Lehrer, K. 1983. “Rationality as Weighted Averaging.” Synthese 57 (3):283–95.
Lehrer, K. 1984. “Coherence, Consensus and Language.” Linguistics and Philosophy 7:43–55.
Lehrer, K. 1988. “Coherence, Justification and Chisholm.” In Philosophical Perspectives, edited
by J. E. Tomberlin, 2, Epistemology, 125–38.
Lehrer, K. 1989. Thomas Reid. London; New York: Routledge.
Lehrer, K. 1996. “Skepticism, Lucid Content and the Metamental Loop.” In Philosophy and
Cognitive Science, edited by A. Clark et al., 73–93. Dordrecht: Kluwer.
Lehrer, K. 1997. Self-Trust. Oxford: Oxford University Press.
Lehrer, K. 2000. “Meaning, Exemplarization and Metarepresentation.” In Metarepresentation,
edited by Dan Sperber, a volume of Vancouver Studies in Cognitive Science, 299–310. Oxford:
Oxford University Press.
Lehrer, K. 2004. “Representation in Painting and Consciousness.” Philosophical Studies 117
(1–2):1–14.
Lehrer, K. 2005. “Analysis and Experience in Art.” In Experience and Analysis, edited by Maria E.
Reicher and Johann C. Marek, Proceedings of the 27th International Wittgenstein Symposium,
8th to 14th August 2004. Kirchberg am Wechsel (Austria) (Schriftenreihe der Wittgenstein-
Gesellschaft 34), 357–362. Wien: oebv & hpt.
Lehrer, K. 2006. “Consciousness, Representation and Knowledge.” In Self-Representational
Approaches to Consciousness, edited by U. Kriegel and K. Williford, 409–20. Cambridge, MA:
MIT Press.
Lehrer, K. 2007. “Loop Theory: Knowledge, Art and Autonomy (John Dewey Lecture).”
Proceedings and Addresses of the American Philosophical Association 81 (2):121–36.
Lehrer, A., and K. Lehrer. 1995. “Fields, Networks and Vectors.” In Grammar and Meaning, edited
by F. R. Palmer, 26–48. Cambridge, MA: Cambridge University Press.
Lehrer, K., and C. Wagner. 1981. Rational Consensus in Science and Society: A Philosophical and
Mathematical Study. Dordrecht: Reidel.
Mill, J. S. 1869. On Liberty. London: Longman, Roberts and Green.
Page, S. E. 2007. The Difference: How the Power of Diversity Creates Better Groups, Firms,
Schools, and Societies. Princeton, NJ: Princeton University Press.
Putnam, H. 1975. Mind, Language and Reality. Vol. 2 of Philosophical Papers. Cambridge:
Reichenbach, H. 1948. Theory of Probability. Berkeley; Los Angeles, CA: University of California
Press. First edited in German, 1935.
Schlick, M. 1959. “The Foundation of Knowledge.” In Logical Positivism, edited by A. J. Ayer,
209–27. Glencoe, IL: The Free Press.
Part III
Self-Evaluative Emotions
Chapter 8
Sentimentalism and Self-Directed Emotions
Jesse Prinz
There is a long-standing philosophical debate about the role of emotions in moral

judgment. Some argue that emotions are inessential; we can make moral judgments
without having an emotional response. Others, so-called sentimentalists, argue that
emotions play an essential role. Philosophers have debated these positions for ages
with no resolution. Intuitions vary as to whether emotions are essential to morality.
The stalemate can be broken by moving beyond intuitions an empirically investi-
gating the psychological processes that underlie moral judgment. Such methods are
delivering results that strongly favor sentimentalism. Healthy moral judgments seem
to depend on emotions. But sentimentalism faces a number of serious philosophical
objections, which have not been directly addressed in the scientific literature. This is
where self-directed emotions come in. Self-directed emotions have been compara-
tively neglected in the recent flurry of empirical research on moral judgment. That is
an unfortunate oversight. I will argue that they are crucially important for address-
ing some of the most damaging objections to sentimentalism. In what follows, I
will present the most publicized evidence for sentimentalism; survey four objec-
tions that threaten the theory; present a novel non-cognitive account of self-directed
moral emotions; and show how these can help defuse the four objections.
8.1 Empirical Evidence for Sentimentalism
8.1.1 Empirical Evidence

Sentimentalism takes various forms, but the version I will present has its origins in
Hutcheson (1738) and Hume (1739). The basic idea is that moral judgments have
emotions as constituent parts. To judge that something is wrong is to feel a sentiment
of disapprobation towards it. On this view, a token moral judgment consists in a non-
emotional mental representation of the thing being judged and an emotional state
J. Prinz (B)
Department of Philosophy, The Graduate Center, City University, New York,
NY, USA; Department of Philosophy, The University of North Carolina, Chapel Hill, NC, USA
e-mail: jesse@subcortex.com
136 J. Prinz
bound to that representation, such as a feeling of anger bound to a representation of

injustice. The binding here can be thought of as predication. The anger constitutes
a token of the concept bad, and it predicates badness of the thing being judged. For
example, the judgment Jane taking John’s property is bad is an angry attitude toward
Jane’s act of taking. Psychologically, this is implemented by a representation of the
taking and a state of anger, which together play the functional role that judgments
play. For example, this combination of emotion and representation disposes us to
assert, “Jane’s taking John’s property was morally bad.” We also have words – so
called “thick concepts” – that express the descriptive and emotional components in
one lexeme: “Jane stole John’s property.” Both sentences express the same thought,
and the emotion is an ineliminable part. If this brand of sentimentalism is true, then
every moral judgment contains an emotion.
Opponents of sentimentalism claim that we can make moral without emotions.
They admit that moral reflection often arouses emotional responses, but these are
dispensable. In principle, one could judge that injustice is wrong – not merely mouth
those words – in a completely dispassionate way. Philosophers typically push for
this possibility by pumping intuitions in thought experiments. They have us imagine
the possibility of a person who knows what’s right and wrong, but is completely
indifferent to such facts. This fictitious creature is known as an amoralist (Brink
1986).
The problem is that philosophers disagree about whether amoralism is conceptu-
ally possible. They have divergent intuitions. Some think it is obvious that morality
and emotion are dissociable, and some think a dispassionate judgment would not
deserve to be called a moral judgment. This impasse seems to be insuperable. If
thoughtful philosophers disagree about hypothetical cases, there is no way of telling
who is right. Perhaps our concept of what counts as a moral judgment is vague
or open-textured. Perhaps our intuitions are inescapably theory-laden. Perhaps our
capacity to introspect on what we are doing when we make moral judgments is lim-
ited or error-prone. To move beyond this impasse, some philosophers have taken
an empirical turn. We can replace the intractable question of whether dispassionate
moral judgments are conceivable with the empirical question of what goes on in
people’s minds when they judge that something is morally good or bad. The latter
question may seem like a feeble surrogate for the former, since it replaces a modal
claim (dispassionate moral judgments are im/possible) with an observed regularity
(dispassionate moral judgments do/don’t arise is us under certain test conditions).
But one can get modal traction out of observed regularities. If moral deliberation
sometimes takes place without emotions (i.e., if people sometimes have no emo-
tions when asked to make a moral evaluation), then that favors the conclusion that
emotions are not essential parts of moral judgments. If, on the other hand, we find
that emotions accompany such judgments under a wide range of test conditions,
then the conjecture that there could be moral judgments without emotions begins to
look like a possibility that should not be taken seriously, because it would rest on
frail intuitions and nothing more.
With respect to the empirical question, I think the evidence strongly favors sen-
timentalism. There are several lines of evidence to consider. First, neuroimaging
8 Sentimentalism and Self-Directed Emotions 137
studies show that areas of the brain that are associated with emotions become active
over a wide range of different tasks. Greene et al. (2001) find that emotions are active
when people think about moral dilemmas as opposed to factual dilemmas. Moll, De
Oliveirra-Souza, and Eslinger (2003) find that emotions are active when people have
to judge whether sentences say things that are morally wrong as opposed to factu-
ally wrong. Heekeren et al. (2003) find that emotions are active when people judge
moral incorrectness as compared to semantic incorrectness. Harenski and Hamann
(2006) find that emotions are active when people view photographs of immoral acts,
such as drinking while driving. And so on. In every neuroimaging study I know,
moral judgments are accompanied by emotions even though the task demands differ
widely.
These studies are powerful in establishing a correlation between emotions and
moral judgments, but they fall short of establishing the sentimentalists’ thesis that
emotions are components of moral judgments. After all, it could be that emotions
are effects of moral judgments that arise, invariably, after a judgment has already
been made. To overcome this limitation of neuroimaging research, behavioral meth-
ods are needed. One important class of studies shows that moral judgments can
be influenced by extraneously introduced emotions. For example, Schnall, Haidt,
and Clore (2008b) induced disgust using films, smells, and other methods and
found that people give more severe moral judgments when they are disgusted. If
emotions were merely effects of moral judgment, this shouldn’t happen. If emo-
tions are parts of moral judgment, Schnall’s results can be explained. We assess
something as wrong by introspecting on our emotions and, if those emotions have
been elevated by some extraneous source, we arrive at a harsher moral assessment.
Wheatley and Haidt (2005) obtained similar results using hypnotic induction of dis-
gust. Other researchers have found similar effects by inducing anxiety through films
(Blumenthal 2005) and anger (Lerner et al. 1998).
Further support for sentimentalism comes from the fact that a reduction in
emotions leads to corresponding reductions in strength of moral judgments. For
example, Valdesolo and De Steno (2006) have shown that, after watching comedy,
people are three times more likely to judge that it is permissible to kill a person in
order to save the lives of others. As they interpret the results, comedy reduces the
negative feelings associated with killing. In a similar spirit, Schnall, Benton, and
Harvey (2008a) asked subjects to make moral evaluations after washing their hands,
and found that cleanliness makes people more morally tolerant. The authors propose
that hand-washing reduces negative emotions.
The most dramatic findings concerning the relationship between emotion and
moral judgment involve criminal psychopaths. These individuals often seem indif-
ferent to the moral domain, carrying out a wide variety of anti-social behavior
with little remorse or regard for the well-being of others. This indifference seems
to stem from a general deficit in emotional responsiveness. Flattened affect is a
diagnostic condition on psychopathy, and it affects emotional response both within
and beyond the moral domain. Psychopaths are the best candidates for real-world
amoralists, because they are usually able to verbally acknowledge that their actions
are wrong. It appears, then, that they can make moral judgments without emotions.
138 J. Prinz
But close inspection tells a different story. Blair (1995) found that psychopaths
fail to distinguish moral and conventional rules. They do not recognize the seri-
ousness of moral violations or appeal to the suffering caused by immoral behavior
when trying to explain why their actions are wrong. Clinicians and law-enforcement
authorities note that psychopaths seem to have only the most superficial conception
of moral categories. They know that certain actions are condemned by the commu-
nity and can result in punishment, but they fail to see these actions as intrinsically
wrong. In Kohlberg’s language, psychopaths reside at a pre-conventional level of
moral development (Campagna and Harter 1975). In other words, psychopaths do
not understand moral concepts the way that healthy people do. Arguably, when a
psychopath says that killing is morally wrong, what she understands by those words
is that that killing violates local conventions and laws. If this interpretation is cor-
rect, then psychopaths show us that, without emotions, people fail to make genuine
moral judgments. This supports sentimentalism.
The empirical case for sentimentalism is quite strong, but not uncontroversial.
Some researchers claim that the evidence offers a mixed verdict for the sentimen-
talist. According to one popular alternative, there are two ways of making moral
judgments: we can rely on emotional intuitions or cool reason. Defenders of this
dual-process view often say that the emotional route to moral judgment is just an
unreliable shortcut, and that it is therefore prone to performance errors; we’d be
more accurate if we made moral judgments on the basis of reasoning alone.
The dual-process view is usually supported by appeal to empirical research on
trolley cases. Greene et al. (2001) argue that people use emotions when reflecting on
dilemmas in which they have to push a person into a trolley’s path in order to save
five lives, but not dilemmas in which they can save five lives by diverting a trolley
to another track where one person will be killed as a side-effect. They also suspect
that emotion-based judgments are sensitive to features that we would not reflectively
endorse; for example, we are more likely to resist a survival maximizing consequen-
tialist trade-off if we come into physical contact with those whose lives would be
lost as a result of the intervention. In a related line of research, Koenigs et al. (2007)
have shown that people with brain injuries in the ventromedial prefrontal cortex,
which effects emotional responsiveness, give the consequentialist response even in
cases where uninjured people are reluctant to (e.g., pushing someone into the trol-
ley’s path). These authors conclude that purely consequentialist moral decisions can
be made without emotion.
I think the evidence for dual-process theories is weaker than advertized. Greene
et al.’s neuroimaging results actually show that emotion areas of the brain are active
for both kinds of dilemmas, just to a different degree of intensity. And ventromedial
patients are not devoid of emotions. Their main difficulty involves using negative
feedback to curtail appetitive goals. Perhaps they have a desire to help people in
need, driven by strong emotions, and cannot curtail this desire when confronted
with the negative discovery that they can save five people only by killing another
person. On this interpretation, all moral dilemmas are decided by means of emo-
tional processing, but that decision process varies as a function of whether people
can adjust one emotional motivation in light of an anticipated emotional cost. This is
supported by the fact that ventromedial patients, unlike psychopaths, normally show
a normal level of moral development, and normal levels of emotional responsiveness
during autobiographical recall and imagination. This also undermines the claim that
emotion-based judgments are performance errors. Without emotions, we’d be blind
to the moral domain. Emotions can mislead us (e.g., they can be increased by filth
and reduced by cleanliness), but, when we control for extraneous factors, they issue
directly from our moral values. Those who refuse to push a man into a trolley’s path
are acting on deep convictions, not mere squeamishness about physical contact.
In conclusion, the empirical evidence suggests that emotions are present when-
ever people make moral judgments; that these emotions are components of
judgments; and that elimination of emotions prevents people from making moral
judgments. In other words, the evidence supports sentimentalism. This is not a
demonstrative proof, of course, but an argument to the best explanation.
8.1.2 Which Emotions?

When Hutcheson and Hume articulated the sentimentalist theory of moral judgment,
they tended to use the words approbation and disapprobation to characterize the
nature of our moral sentiments. Reading this, one might get the impression that
these words name two particular emotions that belong to our basic repertoire of
affective states along side joy, fear, surprise, and all the rest. But this turns out to
be an unsatisfying view, and recent empirical research has advanced beyond these
early formulations by delving into the nature of approbation and its opposite more
deeply.
The concern about Hutcheson and Hume’s formulations was actually articulated
in their lifetimes by the most important heir to the sentimentalist tradition of that
period, Adam Smith. In his Theory of the Moral Sentiments, he complains,
If we attend to what we really feel when upon different occasions we either approve or
disapprove, we shall find that our emotion in one case is often totally different from that in
another, and that no common features can possibly be discovered between them. Thus the
approbation with which we view a tender, delicate, and humane sentiment, is quite different
from that with which we are struck by one that appears great, daring, and magnanimous. . . .
The same thing holds true with regard to disapprobation. Our horror for cruelty has no sort
of resemblance to our contempt for mean–spiritedness. (1759: III.iii)
In other words, approbation and disapprobation are not names of basic emotion,
but rather broad labels for two classes of emotions that differ as a function of what
actions we happen to be assessing.
Recent research implicitly echoes Smith. No one in contemporary psychology
uses the broad emotional terms of Hume and Hutcheson. Instead, research has
focused in on more specific emotions. We have already seen examples of this, in
fact. Research on how emotions influence moral judgment has focused on emotions
such as disgust (Schnall) and anger (Lerner). Both of these two qualifies as a distinct
form of disapprobation.
140 J. Prinz
Psychologists have begun to explore how these different emotions operate. Anger
and disgust have different functional roles outside the moral domain. Anger is asso-
ciated with threat and frustration. An animal whose goals are thwarted or who is
aggressed against by a conspecific may react with a state that looks similar in key
respects to human anger. Animals bear their teeth, lower their eyes, and aggress
against those that prevent them from getting what they want. Disgust, in contrast,
has little to do with threat and frustration. Its most primitive forms have to do with
physical contaminants. Noxious food and rotting flesh are paradigm disgust elic-
itors, and the action tendency is not aggression, but rejection, expulsion from the
body, and avoidance. Within the moral domain, the roles of these emotions are cor-
respondingly varied. Rozin et al. (1999) show that anger is associated with crimes
against persons, such as physical harm, theft, and violations of rights. Disgust, in
contrast, is associated with “unnatural acts” such as violations of sexual mores or
other transgressions that are conceptualized as polluting the body. For example,
Haidt (2001) reports that disgust grounds moral judgment about consensual incest;
no one is harmed, but the behavior still meets with moral condemnation, which is
expressed both verbally and facially as disgust.
Anger and guilt are highly prevalent moral responses, but they are not the only
ones. Rozin et al. (1999) explore one other form of disapprobation: contempt. They
argue that this emotion is reserved for crimes against community, such as social
hierarchy violations and destruction of public goods. If someone does not show
appropriate respect for authorities, such as a selfish teenager who is rude to her par-
ents, we tend to feel contempt. This threatens the social order. We also feel contempt
for vandalizers and polluters, because they damage things that belong to everyone.
Similarly, contempt is directed at politicians who violate the public trust, because
the community is harmed. Smith’s example of mean-spiritedness as an elicitor of
contempt is interesting because it looks, at first, like a crime against individual
persons, and thus the kind of thing that should instill anger. But, on reflection,
mean-spiritedness is not an action, in which someone gets harmed, but a charac-
ter trait that disrupts social cohesion and cooperation. Thus, it can be construed as
a crime against community. The same may be said of hypocrisy. We feel contempt
for those who say one thing and do another. This does not look like a crime against
community at first, but it is a trait that undermines trust. Since cooperation depends
on trust, hypocrisy can be seen as a threat to cooperation, which means it threatens
social cohesion.
In reflecting on these contempt elicitors, one might worry that it is tricky to distin-
guish contempt from other forms of disapprobation. There is something disgusting
about the hypocrite, and something outrageous. The polluter introduces filth (hence
disgust), which cases harm (hence anger). The selfish teenager offends her parents
(hence anger) and also acts untamed, like an animal (hence disgust). For these rea-
sons, I have argued elsewhere that contempt is actually a blend of anger and disgust
(Prinz 2007). Crimes against community harm individuals, but they also seem like
unnatural acts since the social order can be seen as the natural order of persons.
These three emotions of disapprobation (or two and one blend) may not exhaust
the range. Smith mentions the feeling of horror at cruelty, which may blend
anger with surprise or fear. There may also be subspecies of anger. Inequitable
distributions tend to elicit indignation, whereas physical assault instills a desire for
revenge. Disapprobation takes subtly different forms.
Note that this analysis of the distinction between anger and disgust does not
presuppose that emotions are cognitive in nature. According to cognitive theories,
emotions contain judgments. I have argued against cognitivism elsewhere (Prinz
2004), and it is important for sentimentalists to resist cognitivism, for reasons that
will become evident below. On a non-cognitive theory, anger and disgust are per-
ceptions of bodily changes that correspond to distinct action tendencies, such as
aggression, in the case of anger, and physical rejection in the case of disgust.
Contempt may blend these action tendencies. In addition, each emotion is asso-
ciated with what I call a “calibration file,” which is a set of innate and acquired
mental representations that can trigger the action tendency. Calibration files can con-
tain cognitive judgments, but they also contain non-cognitive representations such
as sensory images. Anger can be triggered by a glare from a conspecific or a phys-
ical attack. Disgust can be triggered by the sight of rotting food, or a bad smell. As
we extend these emotions into the moral domain, calibration files expand to include
mental representations of misdeeds, such as stealing or bestiality. The things that
trigger moral anger and moral disgust will depend, developmentally, on similarity
to innate elicitors of these two emotions. Crimes against the body, for example,
are more likely to enter the calibration file for disgust because they are similar in
content to innate disgust elicitors, which include mutilated and deformed bodies.
Subtle differences in subspecies of a given emotion may reflect divisions within the
calibration file. For example, an anger response labeled by a representation of injus-
tice often gets labeled indignation. Ultimately, calibration files may come to contain
representations that are cognitive in nature (a mental description of tax evasion may
trigger anger, for example), but the emotions themselves are feelings of the body
preparing for behavioral responses.
So far I have been focusing on disapprobation, but similar work could be done to
distinguish forms of approbation and their corresponding action tendencies. Smith’s
examples of approbation towards humane acts and approbation towards bravery are
suggestive. Perhaps we revere the hero and feel gratitude towards the fundraiser.
Helping, saving, sharing, cooperating, and other acts of kindness may give rise to
different emotions. This possibility is ripe for investigation, but I will not pursue
that matter here. The main point for the moment is that an empirically informed
sentimentalist theory can advance beyond the formulations in Hutcheson and Hume
by specifying the emotions that constitute token moral judgments, following Smith’s
observation that these will vary from one judgment to the next.
8.2 Objections to Sentimentalism
8.2.1 Four Objections

I have been arguing that sentimentalism enjoys empirical support and that recent
research has allowed us to characterize the view at in greater and greater detail.
142 J. Prinz
Experiments reveal that emotions are integrally involved in moral judgments, and
the specific emotions that do this work include anger and disgust, as well as their
subspecies and blends. But the empirical elaboration of sentimentalism may not be
viewed as a worthy enterprise by all. Sentimentalism faces some objections, which
appear to be very damaging. If these are insuperable, all the enthusiasm for empirical
results may be misplaced. No studies can save a theory that cannot work in principle.
No oar can steer the course of a boat that will not stay afloat. In this section, I will
present four objections that demand attention.
(1) Emotions without Judgments The first objection concerns the possibility of
the aforementioned emotions arising in the absence of moral judgments. This
clearly happens in the case of disgust. We can feel disgust without thinking
anything is morally wrong. For example, we might feel disgusted when we see
people eat foods we detest, but we don’t think they are immoral. This suggests
that having a disgust response is not sufficient for moral judgment. The same
may be true for anger. Suppose you get mad at your computer because it crashes,
or irritated at a dripping sink. The emotions you feel in these situations could be
described as anger, but it doesn’t follow that you think your computer is morally
bad or that your sink has sinned.
(2) Hidden Circularity To deal with this last objection, one might try to distinguish
moral anger and disgust, on the one hand, and non-moral anger and disgust, on
the other. But doing so raises a question of what makes an emotion moral, and
the natural answer is that moral emotions contain moral judgments. But that
would undermine sentimentalism by exposing a vicious circularity: if moral
judgments are defined by appeal to moral emotions, then moral emotions cannot
be defined by appeal to moral judgments.
(3) Disavowed Emotions A third objection to sentimentalism draws attention to the
fact that we sometimes disown our emotional reactions in the moral domain. We
feel angry or disgusted by something but then insist that we do not think that
thing is morally wrong. This is clearest in the case of inculcated values, which
we later come to disavow. Suppose, for example, you are raised in a homopho-
bic society. You may end up feeling disgust when you see public displays of
affection between members of the same sex. But you have come to believe
that homosexuality is morally permissible. In fact, you think discrimination
against homosexuality is a violation of fundamental civil rights. So, you dis-
miss your disgust as the residue of a conservative education, and announce with
confidence and conviction that gay couples should be able to show affection in
public.
(4) Emotional Intensity The final objection I will consider concerns intensity.
Emotions can vary in strength. We can feel faintly queasy about something,
or intensely revolted. We can feel mildly irritated or intensely outraged. Moral
judgments also come in degrees. There are minor infractions and major iniq-
uities. The problem for sentimentalism is that emotional intensity and moral
intensity do not coincide. You can become irate at your spouse for some trifling
insensitivity (your husband left the toilet seat up), while feeling faint ire after
reading about a massacre in some distant country. We judge the latter to be more
wrong, but it excites weaker passions.
8.2.2 A Cognitive Solution?

All four of these objections suggest that sentiments and moral judgments come
apart. We can have emotions without making moral judgments, and the intensity
of our emotions does not reflect depth of moral conviction. It is tempting to deal
with these objections by offering a revised form of sentimentalism that combines
emotions with some cognitive machinery. One could propose that every moral judg-
ment contains an emotion plus a cool cognitive construal. These construals would
determine when an emotion qualified as moral, thus dealing with the first two objec-
tions. Anger at a computer may not qualify as moral because the construal is absent.
This could also handle the third objection. Disgust at homosexuality may lack the
requisite construal. And there is even hope of dealing with the fourth objection.
Intensity of judgment could depend on a cognitive rating of moral import rather
than emotional strength. Call this the cognitive strategy.
The cognitive strategy certainly seems attractive. It looks like a needed panacea
that can save sentimentalism from a battery of objections. But I think it should be
resisted. One problem is that it is very hard to figure out what the relevant construal
would be. To suppose that an explicit cognitive construal arises whenever people
make moral judgments is tantamount to offering a conceptual analysis of moral
concepts. Conceptual analyses have a long history of failure. For every proposal, it
seems to devise counter-examples. Consider anger. What might distinguish moral
anger from other forms of anger? One answer is that moral anger is accompanied
by a judgment that someone has intentionally harmed another person. Anger at a
computer is not moral, because we don’t suppose the computer acted intentionally
(likewise for irritation at a crying baby or barking dog). This looks plausible enough,
but it does not survive on scrutiny. For one think, it is overly demanding to presume
that moral anger requires ascriptions of intentions. Young children seem to show
moral outrage, but they lack the ability to readily attribute mental states (cf. Nichols
2004). One can also get morally outraged at people who cause harm unthinkingly
through negligence, such as drunk drivers. It’s not even clear that we construe every
cause of moral anger as harmful: we get mad at brazenly selfish behavior, like play-
ing loud music in a car, even if no one is hurt. I suspect that no cognitive judgment
can be identified to explain what makes moral anger distinctively moral.
The cognitive strategy also goes against the spirit of sentimentalism.
Sentimentalism tries to reduce moral judgment to emotion. That reductive move is
what makes the approach so attractive. If we could identify cognitive construals that
were present in every moral judgment, we might not need to develop an emotion-
based theory. Cool cognition might suffice. But identifying such cognitions has been
difficult, which is one thing that makes the emotional reduction of morals seem like a
promising strategy in the first place. And if the cognitive strategy is the only hope for
144 J. Prinz
dealing with the objections surveyed here, then it threatens to undermine sentimen-
talism rather than saving it. If moral emotions must be supplemented by cognitive
construals, then perhaps we should abandon sentimentalism and resume the search
for a purely cognitive theory of moral judgment.
Clearly, the sentimentalist needs another way to cope with the four objections. I
will now suggest that self-directed emotions are the key.
8.3 Self-Evaluation and Sentimentalism
8.3.1 Self-Directed Emotions

Most of the empirical research pertaining to sentimentalism has focused on other-
directed emotions. Anger, guilt, and contempt are characteristically emotions that
we feel towards other individuals, not the self. These emotions are clearly very
important in morality, as the aforementioned research has borne out. It’s natural
that they should dominate research, since moral norms are largely used to regu-
late other people’s behavior. The role of morality in social control is fundamental.
But, in focusing on other-directed emotions, recent research has left sentimentalism
incomplete. Self-directed emotions also play a role in moral life. This oversight is,
I believe, one of the main reasons why sentimentalism appears vulnerable to the
objections that I have been discussing. I will make that case below. But first, an
introduction to the self-directed emotions is an order.
We have many emotions that directly concern the self. Pride is a pleasure in one’s
own achievements, embarrassment is a discomfort that arises when the self becomes
an object of unwanted attention, certainty is a feeling associated with recognition
that one possesses good grounds for a particular belief. In morality, the most preva-
lent self-directed emotions are guilt and shame. These terms are sometimes used
interchangeably, but they are importantly different. Traditionally, the distinction is
drawn by focusing on the objects of these two emotions. Both arise when a person
transgresses against a norm, but they focus on different consequences of transgres-
sion. Guilt is said to concern actions, and shame is said to concern the global self
(Lewis 1971). That is, when you feel guilty, you feel as if there is something you
have done that is bad, but you don’t necessarily feel like a bad person. With shame,
you feel like you are somehow corrupted by your actions. Correspondingly, the two
emotions are associated with different action tendencies. Guilt promotes reparative
behaviors, such as apology or making up for damages that have been caused. The
guilty individual approaches those who have been affected by her actions. Shame, in
contrast, tends to promote concealment or social avoidance (Schmader and Lickel
2006). The canonical expression of shame is a lowered head.
These differences between guilt and shame are sometimes characterized in terms
of distinct appraisals: the judgment that my action was wrong or that I have a bad
self. But such a cognitive view could introduce a problem for the appraisal theorist
akin to the circularity worry raised above. If guilt and shame require judgments
that I or my actions are bad, then the project of reducing moral appraisals to
moral emotions could lead to a circle, because moral emotions will be defined as
containing the very moral appraisals that they are supposed to explain.
Fortunately, I think one can preserve the action/self distinction without assuming
that guilt and shame contain appraisal judgments. To begin with, it is plausible that
these two emotions are socialized differently. In both cases, the emotions might be
instilled in children by adults who are trying to shape their behavior. A child vio-
lates an adult rule, and the adult tries to make the child feel bad about it. Beyond that
similarity, the differences between the socialization of guilt and shame may be con-
siderable. In the case of guilt, socialization may hinge on sadness induction. Adults
can make children feel bad by expressing disappointment or cutting off affection.
These interventions make children feel sad. If a child sees her caregivers’ sadness,
she will become sad by emotional contagion. If a child experiences her caregivers’
withdrawal of affection, she will be saddened by the loss. In both cases, she will also
be motivated to try to re-establish the threatened relation to the caregiver. This pro-
motes approach behavior. The child will work to make her caregiver happy again. In
the language of developmental social psychology, this dynamic can be construed as
involving attachment. Children and caregivers have an attachment relation, which
is threatened when children make caregivers feel bad. Because they value attach-
ment, this will upset children and motivate reparation. At no point in the process
does a child need to think, “I did something bad,” much less, “I committed a moral
transgression.”
In the case of shame, socialization may hinge on the induction of aversive
embarrassment. Sometimes caregivers respond to misbehavior by ridiculing chil-
dren (“that’s no way for a proper young lady to behave!”) or, more frequently, by
ostracizing them (“go to your room!”). Such techniques draw attention to the fact
that children are being viewed in a bad light, or are not worthy of being viewed at all.
They encourage children to conceal themselves. The result is something like embar-
rassment, but of a particularly unpleasant kind, since the level of social rejection is
high. We come to call this feeling, “shame.” Shame is especially likely if a child’s
misdeed is witnessed by a group of individuals rather than a lone parent, because
collective criticism will exacerbate the desire for concealment. This unpleasant urge
can occur even if the child lacks the cognitive resources to explicitly judge that she
has broken a moral rule.
Notice that these explanations of how guilt and shame are socialized do not
presuppose that moral emotions are innate. Instead, they may emerge out of more
fundamental emotions, such as the distress of losing an attachment relationship and
embarrassment. But guilt and shame may nevertheless be nearly universal, given
the fact that all cultures punish misdeeds in ways that could lead to the formation of
these two self-directed emotions. Most cultures have words corresponding to at least
one of these emotions or a blend of the two (Casimir and Schnegg 2002). Prevailing
methods of socialization may determine which of the two is more prevalent and
whether they are fully differentiated.
If my developmental conjectures are right, this may shed new light on the
action/self distinction. It is usually presumed that guilt promotes reparation because
146 J. Prinz
of an action-oriented appraisal: I did this bad thing so I better make up for it. And
shame is said to promote concealment because of a self-oriented appraisal: I am a
bad person, so I better hide. But this may get things backwards. Guilt may count
as action-oriented in virtue of promoting reparation, and shame may count as self-
oriented in virtue of promoting concealment. In other words, if we suppose that
the action tendencies of these two emotions are inculcated without any cognitive
appraisals, we can interpret the action/self distinction non-cognitively as a descrip-
tion of the kinds of behaviors associate with each emotion. Guilt involves a tendency
to repair, and that generally requires compensating for some specific action, and
shame involves a tendency to conceal, which is an action directed toward the self.
This sketch of the distinction between guilt and shame is almost complete, but
there is still one crucial detail to fill in. I have described differences in how guilt and
shame are socially conditioned and the behaviors associated with each, but I have
not explained why we have these two different emotions. Why are there two self-
directed moral emotions? The answer may be that they are associated with different
kinds of moral norms. Recall that anger and disgust have a division of labor within
morality. Anger tends to arise when there has been a crime against a person, whereas
disgust tends to arise when there has been a crime against the body. Guilt and shame
may play corresponding roles in the first-person. If you perpetrate a crime against
another person, you may be more likely to feel guilt than shame, and if you violate
your body in some way, you may be more likely to feel shame than guilt.
The link between the guilt/shame dyad and the anger/disgust dyad may emerge
through socialization. Earlier I emphasized socialization techniques such as the
withdrawal of affection and social ostracizing. Anger and disgust may actually
serve as special tools for achieving these two effects. If a child harms someone,
e.g., pushes a playmate, a caregiver may get angry, thus communicating a lack of
affection. The child will then be motivated to repair the attachment relationship. In
contrast, if a child violates a norm pertaining to the body, such as taking clothes
off in public, a caregiver may express disgust, thus communicating social rejection.
If this happens, the child will be motivated to hide. Thus anger and disgust can
instantiate the socialization techniques associated with guilt and shame.
To test the hypothesis that guilt and shame play roles similar to anger and disgust,
I conducted a study that mirrored Rozin et al.’s (1999) investigation of moral norms.
Those authors gave subjects vignettes about actions performed by other people and
asked, what the subjects would feel if they witnessed another person performing
some morally questionable act. I devised similar vignettes stated in the first person,
and asked subjects what they would feel if they performed morally questionable
acts themselves. My vignettes included crimes against persons and crimes against
the body. In the first category, examples were: “Suppose you take something from
someone and never return it,” and, “Suppose you negligently leave out a hot fry-
ing pan, which burns your roommate’s hand.” In the second category, examples
were: “Suppose, at a moment of weakness, you allow a person who is really old
to kiss you romantically,” and, “Suppose your roommate catches you masturbat-
ing.” Given a choice between emotions, subjects overwhelmingly said that cases of
the first kind would make them feel guilty, cases of the second kind would make
them feel ashamed. In the study I also included some crimes against community,
and found no significant difference between subjects’ preference for shame or guilt,
suggesting that such transgressions may elicit a guilt-shame blend, akin to contempt,
which I described as a blend of anger and disgust.
This simple study provides preliminary support for my conjecture that guilt and
shame play different roles, and those roles mirror the roles played by anger and
disgust (and blends thereof). This completes my analysis of guilt and shame. The
emotions are socialized differently, associated with different action tendencies, and
used to regulate different kinds of norms. This analysis is consistent with the stan-
dard view that guilt relates to actions and shame relates to the self, but it puts
that distinction into a new light, relating to the non-cognitive functional roles that
characterize these two emotions.
8.3.2 From Self-Evaluation to Sentiments

The foregoing analysis of self-directed moral emotions is intended to compen-
sate for the disproportionate emphasis on other-directed emotions in recent moral
psychology. These emotions haven’t been completely ignored. Some major ethi-
cal theories in philosophy have drawn attention to shame and guilt (Gibbard 1990,
Greenspan 1995), and there is a massive research program on these two emotions
in developmental psychology (Tangney and Dearling 2002). But guilt and shame
have been neglected in the research program surveyed above, which has been used
to establish that emotions are constitutively linked to moral judgments. I think it is
crucially important to bring these two areas of research together. Guilt and shame
are essential for developing an adequate sentimentalist theory.
Here I will offer a suggestion about the role that self-directed emotions play in
moral judgment. Sentimentalism is the view that emotions are constituents of moral
judgment: to judge that something is wrong is to have an emotion of disapprobation
towards it. Above, following Adam Smith, I suggested that disapprobation is not
a single emotion, but a family of different emotions governing different kinds of
moral transgressions. That’s what brought us to anger and disgust. I suggested that
these are two forms of disapprobation. Now I’d like to suggest that the same is
true of guilt and shame. These are also species of disapprobation. In other words,
disapprobation is felt differently depending on both the type of transgression that
has been committed and who has committed it. Like anger and disgust, guilt and
shame can be components of moral judgment. If you commit a crime against nature
and come to recognize that your action was wrong, that judgment will consist in
shame directed at a representation of your action.
There is one more detail needed to complete this sentimentalist proposal. I have
been talking about moral judgments. Moral judgments are mental events. They are
brief episodes that occur when we assess the moral permissibility of some action.
In addition to such brief mental events, we have long-standing moral attitudes,
which we call moral values. Moral judgments reflect our moral values. We judge
148 J. Prinz
that something is wrong if it violates a moral value that we have internalized. So

a theory of moral judgment requires a theory of the moral values from which such
judgments emanate.
I think this missing ingredient is actually quite simple. Moral values can be
defined as dispositions to make moral judgments. More accurately, they can be
defined as the categorical bases of the psychological dispositions to make both other-
and self-directed moral judgments. If you think bestiality is wrong, then you have
some standing state (physically realized in your nervous system) that disposes you
to feel disgusted when you hear about someone who has had sex with animals and
to feel ashamed if you find yourself so inclined.
I call such dispositions to have emotions “sentiments.” An emotion is a short-
lived occurrent state, on this use of terms, and a sentiment is a standing state that
causes a person to have particular emotions in specific circumstances. Importantly,
the very same sentiment can cause a person to have different emotions on different
occasions. If you detest fast food, you will be horrified when a fast food restaurant
opens on your street and delighted when it closes. Moral sentiments can be defined
as the standing states that cause us to feel other- and self-directed emotions. A moral
value can be defined as a moral sentiment towards some specific thing.
Why think that moral sentiments are values? Notice, first, that moral sentiments
qualify as general behavioral rules. If I get angry at you for breaking a promise, that
anger serves as a form of punishment, which pressures you into keeping promises in
the future. But my anger towards you does not yet qualify as a rule, because it is not
sufficiently general. To internalize a rule against promise-breaking, I need to enforce
it more generally, including in the first person case when I myself break promises.
Having negative attitudes towards your broken promises and mine serves to enforce
a general prohibition against promise breaking. If I just got mad at you (or just felt
guilty when I broke promises myself), my response would be more like a personal
preference than a rule. Of course, there are ways of enforcing rules that don’t require
emotions. Legal rules, for example, work through systems of punishment; conven-
tional rules work by social conformity; and game rules work by explicit stipulation.
But we don’t call any of these rules values. Rules qualify as values only when they
are grounded in our emotional responses. To use a popular metaphor, values are
“hot.” Thus, sentiments qualify as values because they are hot rules. Values must be
rules and they must be hot; sentiments are both.
8.3.3 Sentiments Save Sentimentalism

It is now time to revisit the objections presented in Section 8.2.1. Collectively, these
objections threatened to show that emotions could not be sufficient for moral judg-
ments. The seemed to demand that moral judgments be supplemented with some
kind of cognitive foundation, which, in turn, threatened to show that emotions are
irrelevant. Our detour into the world of self-directed emotions has given us resources
for responding to this threat.
Let’s begin with the objection that there can be moral emotions without moral
judgment. For example, one can be angry at a malfunctioning computer or disgusted
watching people eat a detested food. Clearly mere anger and mere disgust do not
qualify as moral judgments. I think there is an obvious reason for this. Anger at
a computer does not reflect a moral value. We don’t have a value that says that
it’s morally wrong for a hard drive to crash. Likewise, we don’t have a value that
says it is morally wrong for one person to eat food that another person reviles.
Episodes of anger and disgust qualify as moral judgments only if they issue from our
values.
On its own, this response is not going to save sentimentalism. It implies that anger
must be supplemented by some other thing in order to play the predicative role in
a moral judgment, and this goes against the spirit of sentimentalism, which tries to
explain moral predications in terms of emotions. But we now have the machinery
needed to see how the sentimentalist can make this response without backing down
on the spirit of the program. Moral values are sentiments, and sentiments are dispo-
sitions to have self- and other-directed emotions. Thus, the fully elaborated response
goes as follows: anger at a malfunctioning computer qualifies as a moral judgment
only if it issues form a moral value, and it issues only from a moral value if the per-
son has a standing disposition to feel angry when another party malfunctions and to
feel guilty when she herself malfunctions. It is absurd to think any of us has a dispo-
sition to feel guilty when we malfunction in the way that a computer does, since we
are incapable of malfunctioning in that way. People don’t have hard-drive crashes.
So anger at a computer is not generally a moral judgment. I am not denying that
we can make moral judgments about our computers. If we anthropomorphize and
think that a computer intentionally thwarted our work, then the resulting anger will
qualify as moral, but such anger will rest on an erroneous attribution of intentions to
the computer. A parallel story can be told for the disgust case. If you eat something I
detest, say egg salad, I may feel disgusted, but my disgust will not qualify as a moral
judgment, because I have no disposition to feel ashamed if I eat egg salad myself.
In contrast, suppose you eat something I find morally questionable, such as monkey
meat. Then my disgust might qualify as moral condemnation, because I would feel
ashamed if I ate monkey meat.
If this response is right, no emotional episode is sufficient on its own to qual-
ify as a moral judgment. Moral judgments are identified by the values from which
they emanate, and those values are dispositions that include both other- and self-
directed emotions. Without self-directed emotions, there are no moral values and no
moral judgments. This conjecture can also be used to address the second objection
to sentimentalism. According to that objection, sentimentalism is secretly circular.
The theory says that moral judgments contain moral emotions, but usually offers
no account of what it is to be a moral emotion. On many theories of the emotions
(cognitive appraisal theories, especially), an emotion qualifies as moral if and only
if it contains a moral judgment. But this would introduce circularity. Sentimentalists
cannot define moral judgments as judgments containing moral emotions and then
define moral emotions and emotions containing moral judgments. Fortunately, there
is a way out of this circle.
150 J. Prinz
I suggested that disgust qualifies as a moral judgment when it issues from a

sentiment that also disposes us to feel shame. One can say that anger qualifies as
a moral judgment under just the same circumstances. What makes anger moral, on
any occasion, is its relationship to a sentiment that has both an other-directed and
self-directed component. The moral emotions are emotions linked together in such
sentiments. They qualify as moral, when so linked, because they constitute moral
values. And they qualify as moral values because they are hot rules. Thus, we do
not need to build moral judgments into moral emotions. We can say, instead, that an
emotion qualifies as moral by serving as a component of a hot rule.
The third objection to sentimentalism concerns disavowed emotions. Recall the
case of the person raised to feel disgust when witnessing homosexual affection. The
person may notice this feeling but insist that homosexuality is permissible, despite
the negative emotion. This looks like a counter-example to sentimentalism, but sen-
timentalists can actually offer an explanation. First, sentimentalists can say that such
a person has inconsistent values. They can say that the disgust response does qual-
ify as residual homophobia – this person can be accused of harboring a negative
moral attitude towards gay affection – but the person presumably has other attitudes
that go against this inculcated response. For example, the person may feel outraged
when reasoning about gay-bashing or laws that discriminate on the basis of sexual
preference. If so, we should say that this person both opposes homosexuality and
opposes opposition to homosexuality. This is inconsistent, of course, but there is
no reason why a person can’t end up with inconsistent values by being exposed to
different sources of social influence. Eventually, one of the two values might lead to
the elimination of the other, but that doesn’t always happen.
This reply strikes me as plausible in many cases, but incomplete. It does not
yet explain an important asymmetry. The person we are imaging does not testify to
believing that homosexuality is wrong. The person might admit residual homopho-
bia under cross examination, but won’t regard this as a considered moral opinion.
Why not? One answer may be that the disgust response is not a moral attitude at all.
We have already seen how this might be the cases. Suppose the person has no dis-
position to be ashamed of having homosexual fantasies. If so, there is no sentiment
against homosexuality. There are other-directed feelings of disapprobation, but no
self-directed feelings. Such a person would not qualify as believing homosexuality
is immoral, even implicitly.
Now suppose that the person does have a full-blooded sentiment against homo-
sexuality – a disposition to feel both disgust and shame. There still may be an
explanation of the asymmetry between this homophobic sentiment and the opposing
sentiment according to which homophobia is wrong. The sentiment against homo-
phobia will cause the person to feel ashamed of the homophobia. If the person
feels disgusted at a public display of gay affection, that disgust will elicit a shame
response. But the converse does not hold. The person who opposes homophobia
despite a homophobic upbringing may not feel ashamed after attending a gay rights
rally. Thus, the asymmetry in values consists in the fact that one value is used to
condemn the other, but not conversely. This explanation is wholly consistent with
sentimentalism, but it makes essential appeal to self-directed emotions.
Let’s turn to the final objection. We observed earlier that emotional intensity
does not always correlate with intensity of moral values. One can read about a
great atrocity and feel only moderate outrage, while insisting that the atrocity was
far more evil than some minor act of inconsideration in your personal life, which
made you fly into a rage. If emotional intensity does not account for moral intensity,
sentimentalism may be in trouble.
Here again, sentiments hold the key. In determining how intense a value is, the
answer is determined not by the emotions that the value happens to engender on
some specific occasion, but rather the emotions that the value disposes one to expe-
rience. To determine which of two values is stronger, we need to ask which disposes
us to stronger emotions. That question can be answered only if we control for vari-
ables such as personal proximity to the transgression. We cannot compare an atrocity
on the other side of the planet to an inconsiderate act perpetrated by a spouse at
home. Instead we should compare how one react to transgressions that are equally
proximate (e.g., being the victim of an atrocity vs. being the victim of an incon-
siderate spouse). Controlling for proximity in this way allows us to make a more
accurate ranking of which values are more intense. It is likely that an atrocity (e.g.,
being tortured) would cause far more intense and enduring outrage than some minor
inconsideration at home.
So far, this response to the intensity objection makes no appeal to self-directed
emotions, but such emotions may play an important role in assessing intensity.
There is a crucial difference between self- and other-directed emotions. You can feel
an other-directed emotion about a transgression that you had no personal involve-
ment in. We feel rage about distant times and places. But self-directed emotions
are always personal. We feel guilt and shame about actions that we ourselves have
done. Thus, guilt and shame are not subject to the same proximity effects. They
do not vary as a function of whether one is personally connected to a transgres-
sion, because they always arise when there is a personal connection. Thus, guilt and
shame may give us a more reliable measure of moral intensity than anger. To decide
how intense a value is, we can ask, How guilty or ashamed would you be if you
violated that value?
Indeed, without asking this question, we could not have an accurate measure of
intensity. Imagine the moral hypocrite, who gets outraged at the misdeeds of oth-
ers, but feels only mild remorse about her own misbehavior. A complete account
of this person’s moral attitudes would require that we take both responses into con-
sideration. We can say, “You got outraged when I broke my promise, but felt little
guilt when you broke yours, so clearly you don’t think promise breaking is a cardi-
nal sin in general.” The overall intensity of a value is a function of how intensely
emotions are felt for all proximate violations of that value including violations com-
mitted by one’s self. If we use both self- and other-directed emotions as a measure,
we can expose hypocrites and discover cases where values are more or less intense
than they should be. Having proportionate self- and other-directed emotions can
serve as a regulative ideal when it comes to intensity. This balance is more crucial
than the balance between the emotions felt about proximate and distant transgres-
sions, because it may not be very useful to get overly angry about events that are
152 J. Prinz
too far away to control. But having self-directed emotions that are proportionate
to other-directed emotions can help ensure that each of us will practice what we
preach.
8.4 Conclusion
This discussion began with an assessment of the empirical case for sentimentalism.
I argued that recent evidence strongly favors the view that emotions are compo-
nents of moral judgment. This empirical defense, however, is threatened by the
fact that sentimentalism faces some very damaging objections. I discussed four of
these, and argued that sentimentalism can be saved if we define moral judgments in
terms of moral values and define moral values in terms of sentiments. Sentiments, in
turn, essentially involve self-directed emotions. Without these, moral judgments and
values would be impossible. Self-directed emotions help us distinguish non-moral
emotions from moral emotions, and they explain asymmetries between conflicting
values, and help provide an adequate measure of evaluative intensity. Self-directed
emotions have received less attention than other-directed emotions in the recent
empirical literature, but this is clearly an oversight. I tried to make some progress
by sketching a non-cognitive theory of guilt and shame. More work is needed to test
these proposals and to show that these two emotions play the role in morality that
I have speculatively suggested. If I am right, guilt and shame are key components
in a workable sentimentalist theory, so more empirical and theoretical investment in
these self-evaluative feelings would be very worthwhile.
Acknowledgements I am extremely grateful to Anita Konzelmann Ziv and other members of the
Basel workshop on self-evaluation. My deepest debt is to Axel Seemann, whose detailed comments
served as my main guide while revising.
References
Blair, R. 1995. “A Cognitive Developmental Approach to Morality: Investigating the Psychopath.”
Cognition 57:1–29.
Blumenthal, J. A. 2005. “Does Mood Influence Moral Judgment? An Empirical Test with Legal
and Policy Implications.” Law & Psychology Review 29:1–28.
Brink, D. 1986. “Externalist Moral Realism.” Southern Journal of Philosophy 24:23–41.
Campagna, A., and S. Harter. 1975. “Moral Judgment in Sociopathic and Normal Children.”
Journal of Personality and Social Psychology 31:199–205.
Casimir, M. J., and M. Schnegg. 2002. “Shame Across Cultures: The Evolution, Ontogeny and
Function of a “Moral Emotion”.” In Between Biology and Culture: Perspectives on Ontogenetic
Development, edited by H. Keller, Y. H. Poortinga, and A. Scholmerich, 270–300. Cambridge:
Gibbard, A. 1990. Wise Choices, Apt Feelings. Cambridge, MA: Harvard University Press.
Greene, J. D., R. B. Sommerville, L. E. Nystrom, J. M. Darley, and J. D. Cohen. 2001. “An fMRI
Investigation of Emotional Engagement in Moral Judgment.” Science 293:2105–08.
Greenspan, P. 1995. Practical Guilt: Moral Dilemmas, Emotions, and Social Norms. Oxford:
Haidt, J. 2001. “The Emotional Dog and its Rational Tail: A Social Intuitionist Approach to Moral
Judgment.” Psychological Review 108:814–34.
Harenski, C. N., and S. Hamann. 2006. “Neural Correlates of Regulating Negative Emotions
Related to Moral Violations.” Neuroimage 30:313–24.
Heekeren, H., I. Wartenburger, H. Schmidt, H. Schwintowski, and A. Villringer. 2003. “An fMRI
Study of Simple Ethical Decision-Making.” Neuroreport 14:1215–219.
Hume, D. 1739/1964. A Treatise of Human Nature. Oxford: Clarendon Press.
Hutcheson, F. 1738/1994. “An Inquiry into the Original of Our Ideas of Beauty and Virtue.” In
Francis Hutcheson: Philosophical Writings, edited by R. S. Downie, 7–44. London: J. M. Dent.
Koenigs, M., L. Young, R. Adolphs, D. Tranel, F. Cushman, M. Hauser, and A. Damasio.
2007. “Damage to the Prefrontal Cortex Increases Utilitarian Moral Judgements.” Nature 446:
908–11.
Lerner, J., J. Goldberg, and P. Tetlock. 1998. “Sober Second Thought: The Effects of
Accountability, Anger, and Authoritarianism on Attributions of Responsibility.” Personality
and Social Psychology Bulletin 24:563–74.
Lewis, H. B. 1971. Shame and Guilt in Neurosis. New York: International Universities Press.
Moll, J., R. De Oliveirra-Souza, and P. J. Eslinger. 2003. “Morals and the Human Brain: A Working
Model.” Neuroreport 14:299–305.
Nichols, S. 2004. Sentimental Rules: On the Natural Foundations of Moral Judgment. Oxford;
New York: Oxford University Press.
Prinz, J. J. 2004. Gut Reactions. Oxford; New York: Oxford University Press.
Prinz, J. J. 2007. The Emotional Construction of Morals. Oxford; New York: Oxford University
Press.
Rozin, P., L. Lowry, S. Imada, and J. Haidt. 1999. “The CAD Triad Hypothesis.” Journal of
Personality and Social Psychology 76:574–86.
Schmader, T., and B. Lickel. 2006. “The Approach and Avoidance Function of Guilt and Shame
Emotions: Comparing Reactions to Self-Caused and Other-Caused Wrongdoing.” Motivation
and Emotion 30:42–55.
Schnall, S., J. Benton, and S. Harvey. 2008a. “With a Clean Conscience: Cleanliness Reduces the
Severity of Moral Judgments.” Psychological Science 19:1219–222.
Schnall, S., J. Haidt, and G. Clore. 2008b. “Disgust as Embodied Moral Judgment.” Personality
and Social Psychology Bulletin 34:1096–109.
Smith, A. 1759/2002. The Theory of Moral Sentiments. Cambridge: Cambridge University Press.
Tangney, J. P., and R. L. Dearing. 2002. Shame and Guilt. New York: Guilford Press.
Valdesolo, P., and D. De Steno. 2006. “Manipulations of Emotional Context Shape Moral
Judgment.” Psychological Science 17:476–77.
Wheatley, T., and J. Haidt. 2005. “Hypnotically Induced Disgust Makes Moral Judgments More
Severe.” Psychological Science 16:780–84.
Chapter 9
Psychopathic Resentment
John Deigh
Hitchcock, in Shadow of a Doubt, introduced a new kind of villain into his films.
His previous villains, while conventionally ruthless and sinister, were largely plot
devices. His most successful earlier films were about espionage, reflecting the grow-
ing political tensions in Europe at the time, and the villains in those films were spies
and assassins whose intrigues created the suspense for which Hitchcock had become
famous. In Shadow of a Doubt, by contrast, the suspense comes from the villain
himself: one Charles Oakley, masterfully played by Joseph Cotton. Oakley, to his
sister and her family, is the worldly Uncle Charlie, a man who exudes boyish charm,
displays impeccable manners, and is attentive to all in his company, especially the
women. And they in turn adore him. He is also, as we see in the opening scenes, in
flight from the law, and as we later learn, a serial killer who preys on widows and
who has sought refuge in the small California town of Santa Rosa, where his sister
and her family live. As the film progresses and Oakley comes under the suspicion of
his niece, he becomes increasingly secretive, manipulative, and menacing towards
her. Slowly we come to realize that this charming, debonair fellow connects with no
one and has no conscience. He is Hitchcock’s first major attempt at exploring the
chilling personality of a psychopath.
Hitchcock continued this exploration in Rope, which is loosely based on the infa-
mous Leopold and Loeb case,1 and then in Strangers on a Train, which is based on
Patricia Highsmith’s eponymous novel. The villain in the latter film, Bruno Anthony,
is, if anything, an even more gripping example of a psychopathic killer. Robert
Walker, who gives a brilliant performance as Bruno, endows the character with a
naive egocentricity, like that of a child, and in doing so bids us to see his lack of
conscience and inability to connect with others as products of a severely arrested
moral development. Bruno, who has murdered the estranged wife of the film’s pro-
tagonist, Guy Haines, in the belief that he and Guy have entered into a compact in
which in return for his killing Guy’s wife, Guy would kill Bruno’s father, becomes
1 The film is an adaptation of Patrick Hamilton’s play Rope.

J. Deigh (B)
Department of Philosophy, University of Texas, Austin, TX, USA
e-mail: jdeigh@law.utexas.edu
156 J. Deigh
enraged when it becomes evident to him that Guy has no intention of killing his
father. Feeling double crossed, Bruno seeks revenge. The last part of the film is then
taken up with his repeatedly frustrated attempt to return to the scene of the murder
so as to plant evidence against Guy and Guy’s attempt to stop him. Bruno’s dis-
plays of resentment and anger in this part of the film seem entirely in keeping with
his personality. They match, in this respect, the resentment and bitterness Oakley
exhibits in a scene in which he suddenly and vehemently denounces the world as a
“foul sty” and belittles his niece as a know-nothing. It is, I believe, no accident that
Hitchcock made the capacity for these volatile, retributive emotions part of both
villains’ personality profile.
While the anger and resentment these villains exhibit seems entirely in place
in Hitchcock’s portrayal of them, their having a capacity for these emotions may
nonetheless seem incongruous to those who are familiar with what has become
the standard view in Anglo-American philosophy about their nature. Resentment,
in particular, is now commonly understood as an emotion a person cannot expe-
rience unless he or she is capable of moral judgment, and indeed rather complex
moral judgment. Stephen Darwall’s recent treatment of the emotion in his book
The Second Person Standpoint is representative.2 Darwall writes, “Resentment . . .
is felt in response to apparent injustice, as if from the victim’s point of view. We
resent what we take to be violations against ourselves or those with whom we iden-
tify. If you resent someone’s treading on your foot or, even more, his rejecting your
request or demand that he stop doing so, you feel as if he has violated a valid claim
or demand and as if some claim-exacting or responsibility-seeking response by you,
or on your behalf, is justified.”3 What makes such accounts of resentment seem ill-
suited to the resentment displayed by the villains in Hitchcock’s films is not so much
the complexity of the judgment that the emotion is represented as containing as the
moral orientation that they presuppose. It is not an orientation one would think was
available to someone like Oakley or Bruno. It is not, one would think, the orientation
of a psychopath. Yet it is essential to these accounts.
On Darwall’s account, for instance, resentment is one of a set of emotions that
he calls, following P. F. Strawson, reactive attitudes.4 The set also includes grati-
tude, moral indignation, guilt, and forgiveness, and to have the capacity for these
emotions one must, according to Darwall, see oneself and others as members of
a moral community in which every member has authority to address any other, to
make obligation-creating claims on him or her, and to hold that person responsible
for failures to meet those obligations.5 Accordingly, the injustice to which or to the
appearance of which Darwall characterizes resentment as a felt response consists in
a failure to meet some such obligation. And similarly for the two other emotions,
moral indignation and guilt, that Darwall takes to be felt responses to injustice. What
2 Darwall (2006). See also G. Watson (1988), and R. J. Wallace (1996).

3 Ibid., 68.
4 P. F. Strawson (1962). All references to the reprint.
5 Darwall (2006, 67).
9 Psychopathic Resentment 157
differentiates each of these emotions from the other two, then, is the viewpoint from
which the injustice is seen. Thus, when one feels resentment toward another, one
takes oneself to have been treated unjustly by him or her; when one feels moral
indignation toward another, one takes the person to have acted unjustly toward a
third party on whose behalf one is taking this retributive attitude; and when one
feels guilt, one takes oneself to have acted unjustly toward another. Consequently,
on Darwall’s view, to be liable to any of these emotions is to be capable of mak-
ing moral judgments that reflect a view of oneself as joined together with others
in mutual relations defined by reciprocal rights and obligations.6 And this means,
given the capacity of psychopathic killers like Hitchcock’s villains for resentment,
that they too are capable of making such moral judgments. Yet the lack of a con-
science and inability to connect with others that are principal marks of this type of
personality imply that they are not.7
Explaining how such killers could have a capacity for resentment consistently
with Darwall’s account of the emotion may not, however, seem all that problematic.
For it requires explaining how they can regard themselves as victims of injustice
though they lack a conscience and are unable to connect with others, and one might
give such an explanation by appealing to their egocentricity. After all, being ego-
centric does not obviously imply that one is incapable of judging that another has
treated one unjustly, even if there is no question that it implies that one lacks a con-
science and cannot connect with others. Accordingly, since the primary evidence
for taking the lack of a conscience as a principal mark of psychopathy is that psy-
chopaths exhibit no guilt or remorse for their crimes and since the absence of these
emotions is also evidence of their inability to connect with others, it may suffice for
understanding Darwall’s account as consistent with acknowledging the capacity of
psychopaths for resentment to point out that a capacity for resentment requires only
that a person be capable of making moral judgments from his viewpoint. Hence,
that psychopaths are incapable of making moral judgments from others’ viewpoints
need not be a problem on Darwall’s account. Or so one might argue.
Of course, if one makes this argument, one must abandon the common character-
ization of psychopaths as incapable of moral judgment. But this departure from
settled opinion may not seem that worrisome. As long as it is understood that
psychopaths are incapable of taking the viewpoints of others and making moral
judgments from those viewpoints, one might think that it is possible to attribute to
them, consistently with accounts of resentment like Darwall’s, a limited capacity for
moral judgment. Thinking this, however, would be a mistake. On such accounts, the
capacity for moral judgment cannot be limited in this way.
The reason is that the capacity for moral judgment that these accounts presuppose
is a capacity for judgments that reflect an understanding of oneself as belonging to
a moral community in which one’s relations to the other members are relations of
reciprocal obligation. Because of this fact, one cannot, according to these accounts,
6 Ibid., 66–68.
7 See R. D. Hare (2003).
158 J. Deigh
judge, when another fails to meet an obligation he owes one, that he has acted
unjustly toward one unless one is also able to judge, when one has failed to meet
a similar obligation that one owes him, that one has acted unjustly toward him. Or
to put the point in the language Darwall favors,8 the obligations that any member
of a moral community has are reasons to do that which he is obligated to do, and
therefore, if one sees oneself as a member of a moral community, whatever reasons
one thinks these obligations give others to act are reasons one also recognizes one-
self as having. Consequently, if one expects others to meet their obligations to one
in view of these reasons and likewise demands of them that they make amends if
they don’t, one must then see oneself, in view of the same reasons, as required to
meet the obligations one has to others and be disposed to demand of oneself that
one make amends if one doesn’t. Resentment, on Darwall’s account, that one feels
toward someone who has not met his obligations to one implies both that one expects
him to meet those obligations and that one demands of him that he make amends
since he didn’t. And it does so because the moral judgment it contains reflects an
understanding of oneself as a member of a moral community. So, on this account,
resentment implies that one expects the same of oneself and is disposed to make
similar demands on oneself if one fails to meet one’s obligations. Plainly, then, it is
not possible, on this account, to be liable to resentment if one is incapable of taking
the viewpoints of others. The resentment of which psychopaths are capable remains,
therefore, inexplicable.9
No doubt, this problem will seem to many to be merely an issue of labeling.
Surely, they will think, we can restrict the term “resentment” to that species of anger
that is moral in character and felt in response to one’s being treated unjustly or at
least to what one takes to be unjust treatment. To say that it is moral in character is
to say that the judgment about being treated unjustly it implies is a moral judgment.
And it follows, then, that the anger toward others of which psychopaths are capable,
while it may be similar to resentment, should not be called “resentment” since psy-
chopaths are incapable of making genuine moral judgments. After all, the standard
view in Anglo-American philosophy about the nature of resentment emerged in the
process of our regimenting our terms for moral emotions so as to have a precise
way of discussing the emotions distinctive of moral agency. Such regimentation is
necessary because language is used loosely in everyday speech. The standard view,
then, once one understands that it has come about through stipulation necessary for
making discussions of the psychology of moral agents precise, is not threatened by
the anger displayed by psychopaths toward those whom they see as having betrayed
them.
The problem, however, with this way of understanding the standard view is
that it makes any representation of the liability to resentment as a mark of moral
agency vacuous. This result is especially problematic for accounts of resentment
8 Darwall,74–79.
9 Forfurther discussion of what it means to say that psychopaths lack the capacity for moral
judgment see Deigh (1995, 743–63).
like Darwall’s, for the apparent force of these accounts comes from their repre-
senting the liability as such a mark.10 Typically, their proponents put forward the
liability to resentment and other emotions like guilt and indignation as an essential
feature of being a moral agent so that the accounts of these emotions they propose
are seen as illuminating the nature of moral agency. Such illumination, however,
would be illusory if the emotions were defined in advance as the emotions of a
moral agent. Hence, to be genuinely illuminating, an account of resentment, like
accounts of other natural phenomena, must presuppose that one can identify the
emotion prior to giving the account. For to give an illuminating account of some
natural phenomenon one must first be directly acquainted with the phenomenon,
which is to say, one must first know how to identify it by name. Having so identified
the phenomenon, one then brings its nature to light by developing an account that fits
it. One cannot, therefore, understand the liability to resentment as a mark of moral
agency unless the attribution of resentment to psychopaths is a misidentification.
Yet nothing supports taking such attributions to be misidentifications apart from
the thesis in question, that the liability to resentment implies a capacity for moral
judgment. While the source of the anger Oakley expresses when he denounces the
world as a “foul sty” is obscure,11 it is clear that Bruno’s anger at Guy is due to his
belief that Guy has double-crossed him. He is not, therefore, merely furious with
Guy for thwarting his plan to end his father’s life. His anger, that is, is more than
the fury of someone whose efforts are frustrated by another’s refusal to do what he
wants him to do. Such fury is no different from anger at some machine or beast for
refusing to work as it should. The farmer who beats his mule mercilessly because
it refuses to pull the plow may thereby show anger amounting to rage that is due
to the mule’s obstinacy and the farmer’s frustration at being unable to get the mule
to move, but because the farmer need not see the mule as having betrayed him, the
farmer’s anger need not be vengeful or retributive. Bruno’s, by contrast, is, which is
why we readily identify it as resentment.
Bruno’s sense of being betrayed by Guy comes from an attachment he has formed
to Guy. His fondness for Guy develops during the conversation they have on the
train that sets the story in motion and continues afterwards. He thinks of Guy as his
friend and expects that this new found friend will follow through on the agreement
he imagines they have made. His attachment to Guy is childlike to be sure, but even
small children regard their playmates as friends. A psychopath is not necessarily a
loner. Thus Bruno’s anger, when he realizes that Guy does not plan to follow through
on the agreement he imagines they have made and, indeed, is not his friend, is anger
at what he feels to be a betrayal. It is resentment.
The standard view in Anglo-American philosophy about the nature of resentment
is therefore mistaken. A capacity for moral judgment is not necessary to be capable
of the emotion. Someone whose personality is entirely egocentric may be capable
10 Darwall, 17.
11 Hitchcock plants in the film’s dialogue the suggestion that the cause of Oakley’s anti-social
personality is a childhood injury to the brain.
160 J. Deigh
of it as well. Consequently, its place in human psychology is different from that of

the other emotions that are commonly explained in conjunction with it as emotions
that imply judgments about injustice and that differ from each other according to the
viewpoint with respect to the injustice from which the emotion is experienced. Thus,
think again of Darwall’s account of these emotions according to which resentment
is a response to injustice experienced from the viewpoint of the victim, indignation
a response to it experienced from the viewpoint of a third party, and guilt a response
to it experienced from the viewpoint of the perpetrator. While the symmetry among
these emotions that structures this account neatly fits the underlying thesis that each
emotion is distinctive of moral agency, it is belied by the psychopath’s liability to
resentment. Only indignation and guilt imply moral judgments. The question then
is how resentment, which does not imply such judgment, is related to them.
Ironically, the work that exponents of the standard view about the nature of
resentment usually cite as its source, Strawson’s “Freedom and Resentment”, offers
an answer. It is embedded in Strawson’s explanation of the role of these emotions in
the dynamics of interpersonal relations. The emotions belong to a set whose mem-
bers, as we’ve already noted, Strawson calls reactive attitudes, and the insight on
which Strawson proceeds in this essay is that the dispute over whether determinism
threatens the soundness of our practice of attributing moral responsibility to peo-
ple for their actions can be resolved through a study of these emotions. Strawson’s
explanation of their role in the dynamics of interpersonal relations consists mostly
of his description of the conditions of their arousal and expiration, and to see the
answer to our question that Strawson’s essay offers, the question, that is, about
the relation of resentment to the other reactive attitudes, requires elaboration of his
explanation. The explanation begins with some plain, natural facts – what Strawson
calls commonplaces – about men and women as social beings. The facts he adduces
are these.12
A central concern of people who live together in communities is what their
intentions and attitudes are toward each other. They place great importance on
these intentions and attitudes, and they interact with each other with an expecta-
tion that generally their fellows behave with goodwill toward them. The goodwill
each expects of others consists in the others’ keeping, as best they can, from harming
him in their conduct toward or interactions with him and in their being responsive to
harm they may nonetheless cause him by being solicitous of his injury and feelings.
Such goodwill may be basic to human personality generally (though not univer-
sally), or it may derive from more basic emotions and primitive attachments. Either
is compatible with the general fact that Strawson means to emphasize: that people
are highly sensitive to and greatly concerned with whether others are well- or ill-
disposed toward them. This sensitivity and concern gives rise in a person to certain
attitudes and feelings when others – particularly certain others – behave in ways that
12 The following is taken from part I of my essay “Reactive Attitudes Revisited” in Morality and
the Emotions, edited by Carla Bagnoli. Oxford: Oxford University Press, 2011.
manifest good or ill will toward him. Thus, while one naturally takes pleasure in ben-
efits that come one’s way on account of another’s behavior, one responds, not just
with pleasure, but with gratitude too, when one sees that the other intended to bene-
fit one through his actions and because of his affection, esteem, or goodwill toward
one. Likewise, while the harm inflicted on one by another’s actions is disagreeable,
one will feel, not just displeasure, but also resentment if one recognizes in his actions
malice or ill will toward one. Gratitude and resentment are Strawson’s initial exam-
ples of reactive attitudes that result from one’s seeing in another’s actions evident
good or ill will toward one. Other examples are love, forgiveness, and hurt feelings.
He calls these personal reactive attitudes.
One cannot overstate the importance to Strawson’s explanation of the distinc-
tion between being benefited or harmed inadvertently or accidentally and being
benefited or harmed intentionally out of evident good or ill will toward one. It is
essential to how he conceives of the personal reactive attitudes. They are, on his
conception of them, emotional responses in which one discerns (or thinks one dis-
cerns) in the beneficial or harmful action of another the person’s good or ill will,
regard or indifference towards one. Hence, they differ from one’s merely being
pleased or delighted with the benefit, pained or aggravated by the harm. They dif-
fer in that the cognition implicit in the response in either case, the cognition by
virtue of which the response is an intentional state, is an awareness of the attitude
toward one that is expressed in the action to which one is responding. The ability
to discern in actions that benefit or harm one the actor’s dispositions and attitudes
towards one is, therefore, necessary to one’s being liable to the personal reactive
attitudes.
Indeed, the ability, by virtue of the feelings and attitudes it makes possible, is
basic to human sociability. Strawson makes this point implicitly when he invites us
to “think of the many different kinds of relationship which we can have with other
people – as sharers of common interest; as members of the same family; as col-
leagues; as friends; as lovers; as chance parties to enormous range of transactions
and encounters” and to think, “in turn . . . of the kind of importance we attach to the
attitudes and intentions toward us of those who stand in these relationships to us,
and of the kinds of reactive attitudes to which we ourselves are prone.”13 A decade
later, John Rawls, in A Theory of Justice, makes it explicitly. Concerning the three
psychological laws that, on his theory of moral education, explain the transitions
between the stages of moral growth in a child’s development of a sense of justice,
Rawls writes, “[T]he three laws are not merely principles of association or of rein-
forcement. While they have a certain resemblance to these learning principles, they
assert that the active sentiments of love and friendship, and even the sense of jus-
tice, arise from the manifest intention of other persons to act for our good. Because
we recognize that they wish us well, we care for their well-being in return. Thus
we acquire attachments to persons and institutions according to how we perceive
our good to be affected by them. The basic idea is one of reciprocity, a tendency to
13 Strawson, 76.
162 J. Deigh
answer in kind. . .. [T]his tendency is a deep psychological fact. Without it our nature
would be very different and fruitful social cooperation fragile if not impossible.”14
The depth of this fact (and so the ability to discern the dispositions and attitudes
of others toward one) was observed two centuries earlier by Rousseau, whom Rawls
acknowledges as his predecessor.15 For Rousseau, the fact has to do with the tran-
sition in the very young child from the solipsism characteristic of infancy and the
corresponding regard for others as mere instruments of or threats to the satisfaction
of its needs to its uniting with or setting itself in opposition to others and the cor-
responding regard for them as fellows or foes. “Every child,” Rousseau writes, “is
attached to his nurse. . . . At first this attachment is purely mechanical. What fos-
ters the well-being of an individual attracts him; what harms him repels him. This is
merely a blind instinct. What transforms this instinct into sentiment, attachment into
love, aversion into hate, is the intention manifested to harm us or be useful to us.
One is never passionate about insensible beings which merely follow the impulsion
given to them. But those from whom one expects good or ill by their inner dispo-
sition, by their will – those we see acting freely for us or against us – inspire in us
sentiments similar to those they manifest toward us. We seek what serves us, but
we love what wants to serve us. We flee what harms us, but we hate what wants to
harm us.”16
While Rousseau takes account of both relationships of mutual affection and of
mutual enmity as products of our awareness of and responsiveness to the attitudes of
others toward us, Strawson concentrates on the former. In general, Strawson thinks
of these relationships as based on the parties to them expecting and demanding
“some degree of goodwill or regard” from each other.17 This is another of the plain,
natural facts about us as social beings he adduces. The personal reactive attitudes
occur, then, against the background of these expectations and demands. They and the
responses they elicit in turn are evidence of them. Resentment is Strawson’s prime
example. Thus, when one party in a relationship of mutual regard injures another and
the injured party feels resentment in response, the offender may try to remove these
feelings with expressions of regret or apologies and such explanations as “It was an
accident”, “I didn’t mean to”, “I didn’t see you coming” or similarly “I couldn’t help
it”, “I slipped”, or “I had no choice.” Each of these explanations is typically offered
to assuage resentment by denying that the offending action manifested ill-will. They
are meant, that is, to restore the relationship to the prior state of mutual goodwill or
regard by giving the injured reason to abandon his belief that the offender acted with
indifference or malice toward him. The offender in making such pleas is in effect
saying that he did not, despite his having committed the offense, lack the goodwill
toward the person he injured expected of him. He is saying, in other words, that
14 Rawls (1971, 494–95).

15 Rawls (1963, 281–305).
16 Rousseau (1979, 213).
17 Strawson, 76.
while he is an appropriate object of reactive attitudes particularly at the time he

committed the offense, the resentment the injured feels is misplaced in this case.
Strawson compares such pleas with pleas that are offered to assuage resentment
by denying that one was even, at the time one committed the offense, an appropri-
ate object of reactive attitudes. “I lost my head,” “I was only a child”, “It happened
during a psychotic breakdown,” “I was in a complete trance” etc. Each of these
explanations is meant to exempt the offender, at least at the time of the offense,
from the class of people of whom one expects some degree of goodwill in their
dealings with one. It is in effect a plea to look on the offender, not as a fellow
participant in the kind of friendly or cooperative relations that presuppose mutual
goodwill and regard, but rather objectively as someone to be managed, controlled,
or handled. In taking this outlook toward someone, one ceases to see his actions as
manifesting good or ill will towards one and consequently one loses one’s disposi-
tion to feel resentment, gratitude, or any of the other reactive attitudes toward him.
One may of course still be liable to some feelings toward him. That one looks on
the offender objectively does not mean that one becomes incapable of feeling sym-
pathy, fear, or irritation towards him. What it means, though, is that one no longer
engages with him with an attitude of goodwill or regard and an expectation of its
being reciprocated. While his actions can, then, provoke in one emotional responses,
these responses do not belong to the class Strawson defines as that of the reactive
attitudes. Hence, the objective view of someone that pleas of this second type invite
is sharply opposed to the view one takes of another with whom one participates in
an interpersonal relationship.
Strawson makes this comparison between the two types of plea to bring out the
difference between our viewing someone as a participant in an interpersonal rela-
tionship with us and our viewing him as an object whose behavior is the product of
the external forces and conditions to which he is subject and who is to be managed
or controlled accordingly. The latter is consistent with the view we would take of
people and their actions if we were to accept the thesis of determinism and regard
them in its light. We are disposed to take such a view of someone when we see him
as immature or dysfunctional in a way that incapacitates him for normal interper-
sonal relationships. In that case, we abandon the view of him as a participant with
us in such relationships and regard him, instead, with detachment of a sort that an
objective view of him entails. Nor is this the only occasion on which we would
view someone objectively. We also, typically, take such a view of people gener-
ally when we are making social policy or designing a program to provide certain
services to a given population. And we sometimes take such a view when, because
of the strain of our relations with someone, we need to put distance between him
and us of the sort that a detached attitude toward him creates. In each of these cases,
we maintain an objective view only temporarily or only toward certain individuals.
We could not, Strawson argues, permanently hold such a view of all people, for
doing so would require our doing something we cannot in fact do, namely, cease
to have interpersonal relationships. Nor, to continue Strawson’s argument, would
it be wise permanently to adopt the objective view even if we could, for to do so
would mean giving up human life as we know it, and abandoning our humanity is
164 J. Deigh
an unreasonable sacrifice. Determinism, Strawson concludes, is therefore no threat

to the personal reactive attitudes. We would not cease to be liable to them even if
we did accept its truth, nor would we have good reason to suppress them.
Whether Strawson’s argument, at this point, succeeds in vanquishing the bogey
of determinism is not our concern, however. What we are after is how Strawson
explains the relation of resentment to the moral emotions of indignation and guilt. To
make this explanation he adds to the class of reactive attitudes vicarious analogues
of the personal reactive attitudes. These analogues are attitudes one takes toward the
goodwill or ill will of another, not in reaction to the manifestation of that will in the
person’s actions toward one, but in reaction to its manifestation in his actions toward
others. Specifically, Strawson calls the vicarious analogue of resentment “indigna-
tion” or “disapproval”. Plainly, his use of “indignation” and “disapproval” is, as he
notes, somewhat stipulative. In common speech, for instance, we are as comfortable
with saying that one feels indignation at one’s own mistreatment as that one feels
it at the mistreatment of another. But since his point is to have distinctive names
for vicarious resentment and not to give an independent account of indignation or
disapproval, the stipulative use of these terms is unproblematic.
Strawson also remarks that the vicarious character of indignation, that one can
conceive of it as “resentment [felt] on behalf of another, where one’s own interest
and dignity are not involved,” makes the adjective “moral” an apt modifier of the
attitude.18 Strawson, in other words, takes indignation to be a moral attitude in virtue
of its being impartial or disinterested. He does not, then, in conceiving of it as a third
party analogue of resentment, suppose that to feel it in response to someone’s injur-
ing another in a way that shows a lack of due regard for the victim one must have a
special attachment to or specially identify with the victim. Rather he supposes that
people’s expectations of and demands for a reasonable amount of goodwill toward
each other that form the backdrop of interpersonal relationships generalize as they,
as individuals, come to recognize that everyone reasonably expects and demands
such goodwill and regard on the part of others toward himself or herself in his or
her transactions with them. The recognition of this general fact, Strawson further
supposes, naturally disposes a person to respond with indignation or disapproval
when, say, the person witnesses an injury done to someone to whom he has no
special attachment by another, who manifests attitudes towards the injured that fall
short of the goodwill and regard for others that is generally expected of people in
their transactions with others. Strawson, apart from a quick reference to human con-
nection, does not say why it is natural for people’s expectations of and demands for
a reasonable amount of goodwill from others toward themselves to generalize in this
way. Nonetheless, one could, to explain this development, appeal in the way Hume
did to something like our general sympathy with humankind.
Strawson, in fact, seems to be tacitly following Hume in this part of his essay.
Hume too distinguished attitudes that reflect a disinterested view from those reflect-
ing an interested one and took the former to be moral because they reflected this
18 Ibid., 84.
view and occurred in response to a perception of the quality of the will manifested
in the action that is their object. Indeed, Hume took the distinction between the
personal or interested view and the general or disinterested view to be crucial to
the explanation of our praising some people for their moral virtue and censuring
others for their moral depravity. Such praise and censure, Hume held, express the
sentiments of moral approbation and blame that we experience in response to our
perceiving behind the actions of those whom we praise and censure such motives
as kindness and compassion, on the one hand, envy and malice, on the other, and
for our perception of such motives to give rise to these sentiments we must perceive
them from a general or disinterested view. We could not otherwise, Hume argues,
explain our praising and censuring the heroes and villains of the ancient world, like
the Roman senators who defended the republic at the time of its crumbling and
the Roman emperors who came later and whose cruelty and injustice are infamous,
since their actions can hardly affect our interests.19 What Strawson then adds to
Hume’s account is the idea of certain moral feelings and attitudes being analogues
of feelings and attitudes that reflect a personal or interested view.
In addition to the vicarious reactive attitudes, Strawson introduces reflexive ones
into the class of reactive attitudes. He calls these “self-reactive attitudes”. They
naturally arise, he suggests, as a result of one’s generalizing the expectations of
and demands for goodwill on the part of others toward one and thereby coming to
expect and demand such goodwill of others toward each other in their transactions
with themselves. For once one has generalized these expectations and demands,
one realizes that they apply to oneself as well, that one is no different from oth-
ers in being someone of whom such goodwill in interpersonal relations is expected
and demanded. Accordingly, one becomes liable to feeling obliged or bound to act
as the goodwill toward others that is expected and demanded of one requires, to
feeling compunction, and to feeling guilt or remorse if one fails to meet this require-
ment. These are what Strawson refers to as the self-reactive attitudes. He mentions
them, he says, for the sake of completing the picture.20 But he also notes that an
agent who was not liable to the self-reactive attitudes would be abnormal in a way
that supported, were he to injure another, a plea like the pleas of immaturity and
derangement that invite taking an objective view of him and thus suppressing the
resentment or indignation toward him that otherwise would be forthcoming.21
Strawson, having completed his picture of the reactive attitudes, returns to the
dispute over whether determinism poses a threat to our practice of attributing moral
responsibility, and to show that it doesn’t, he reprises the argument he gave to show
that accepting the thesis of determinism would not affect our being liable to the
personal reactive attitudes. The argument, he believes, applies mutatis mutandis to
the vicarious reactive attitudes, since they are analogues of the former. At the same
19 Hume (1975, 215–18).

20 Strawson, 84.
21 Ibid., 86.
166 J. Deigh
time, he suggests that we can directly see in the argument’s application to this sub-
class of the reactive attitudes that our accepting the thesis of determinism is no
threat to our practice of attributing moral responsibility. The argument, as applied
to the vicarious reactive attitudes, affords this observation, he argues, because our
susceptibility to these attitudes, unlike our susceptibility to the personal reactive atti-
tudes, is the same as our expecting and demanding of everyone in our community
some degree of goodwill or regard toward others generally in his or her dealings
with them, and this generalized, impersonal expectation and demand amounts to
an understanding of everyone in the community as lying under moral obligations
to act with some degree of goodwill toward others generally in his or her dealings
with them and as therefore someone to whom moral responsibility for failures to
meet these obligations and for harms that result from some such failures is, in the
absence of an exculpatory plea, properly attributed. Attributions of moral respon-
sibility, according to Strawson, require an understanding of people as moral agents
and as members of a moral community, and the interpersonal relationships in which
one participates are seen as falling among the relationships that a moral commu-
nity comprises only when one regards people in one’s community from a general or
disinterested standpoint.
Strawson initially makes this point by supposing someone whose view of himself
and others is entirely interested or personal.22 Such a person is liable to the personal
reactive attitudes but not to the vicarious ones. He is someone, then, who expects and
demands from others a degree of goodwill toward himself in their dealings with him
but does not expect of or demand from others such goodwill or regard in their deal-
ings with each other. Such a person, Strawson contends, would be a moral solipsist.
He would not, that is, see the interpersonal relationships in which he participates
as embedded within a moral community. He would not see the demands he makes
on others that they act with a degree of goodwill or regard toward him as general
requirements to act with that degree of goodwill toward others to which everyone is
subject and that as such represent moral obligations under which everyone lies. The
ideas of belonging to a moral community, of having in common with others recip-
rocal moral obligations to act with a degree of goodwill toward one another, and
of being, as everyone else is, morally responsible for actions that affect others do
not arise, then, until one takes up the standpoint of others in one’s community and
their expectations and demands on those with whom they interact.23 These ideas
22 Ibid., 85.
23 Strawson then reaffirms the point in considering how pleading, in response to injuring some-
one, that one was at the time incapacitated for normal participation in interpersonal relationships
(“I was suffering hallucinations, undergoing a psychotic breakdown, in a trance, etc.”) affects our
susceptibility to moral indignation. Thus, Strawson, after introducing the case by writing of seeing
the agent in a different light, “as one whose picture of the world is an inane delusion; or as one
whose behaviour . . . is unintelligible to us” etc., goes on, “[A]bstracting now from direct personal
interest, we may express the facts with a new emphasis. We may say: to the extent to which the
agent is seen in this light, he is not seen as one on whom demands and expectations lie in that
particular way in which we think of them as lying when we speak of moral obligation; he is not, to
do not arise, in other words, until one regards oneself and others from a general or
disinterested standpoint.
Here too Strawson may be meaning to follow Hume in supposing that perceptions
and judgments of moral relations and moral conduct must come from a view we
share with others. The intelligibility and practical force of the language of morals,
Hume argued, presuppose a common point of view, and therefore we must vacate
our respective interested or personal viewpoints, which are peculiar to each of us,
and take up a disinterested or impersonal one (or suppose that we have), which is
common to all, to take in the moral character of people’s relationships and actions
within them and to speak properly of it. Nonetheless, though Strawson may be
meaning to follow Hume here, he departs, I think, from Hume in supposing that
we get our ideas of moral obligation and moral responsibility directly from gen-
eralizing the expectations of and demands for goodwill toward ourselves that we
make on others, which is to say, by vacating our personal view and taking up an
impersonal or disinterested one. For Hume does not think we come to have these
ideas until we apply the generalized expectations and demands that we acquire in
taking up the impersonal view to ourselves in those circumstances in which we find
ourselves lacking the natural motives by which we typically manifest goodwill or
due regard in our dealings with others.
Specifically, Hume thinks we acquire these ideas as a result of a kind of indig-
nation with ourselves for lacking the motives of goodwill toward others that we
come to demand of everyone including ourselves when we take up the impersonal
or disinterested view.24 That is, seeing that we lack the motive of benevolence, say,
or fellow feeling that in our dealings with another would show us to bear goodwill
toward him and at the same time expecting and demanding on his behalf that we so
act, we reproach ourselves for our lack of benevolence and fellow feeling and are
thereby moved by a sense of duty to do the action that we would have done had
we been naturally moved by either to do it. It is this experience of self-reproach
for lacking the proper natural motive of goodwill and the consequent sense of duty
to do what we would naturally do if we had that motive that gives us the idea of
moral obligation, the idea that acts displaying goodwill or due regard for others in
our dealings with them are obligatory even when we lack in ourselves the motives
of benevolence or fellow feeling that would naturally produce them. And presum-
ably, though Hume does not expressly say this, the same experience that gives us
the idea of moral obligation gives us as well the idea of being morally responsible
for doing or failing to do what we now understand as obligatory. Be this as it may,
the important point is that Strawson, like Hume, holds that the acquisition of these
ideas occurs only after one becomes liable to the vicarious reactive attitudes. Being
liable to the personal reactive attitudes is by itself insufficient for having them.
that extent, seen as a morally responsible agent, as a term of moral relationships, as a member of
the moral community.” Ibid., 86.
24 See Hume (1978, 479).
168 J. Deigh
Let us distinguish, then, between Strawson’s original account of the reactive atti-
tudes and that account modified to reflect Hume’s view of how we come to see
ourselves as having moral obligations. I will refer to the latter as the modified
Strawsonian account. Clearly, either account shows how someone might be liable
to resentment though not to indignation. Such a person, while capable of experi-
encing personal reactive attitudes, would be incapable of experiencing the vicarious
ones. To be capable of the former he must be able to form attachments to others, to
regard them as friends, mates, pals, or the like, and thus to have some affective tie
to them. Such attachments are possible, though, even if one is incapable of experi-
encing the vicarious reactive attitudes. For one can form an attachment to another
without being able to understand the other’s interests sympathetically, and on either
account, to be capable of the vicarious reactive attitudes means that one can take
an impartial view of others, a view in which one regards their interests sympathet-
ically and thereby feels personal reactive attitudes on their behalf. A person who is
incapable of taking such a viewpoint is thus incapable of experiencing the vicari-
ous reactive attitudes. On the modified Strawsonian account, such a person is also
incapable of experiencing the self-reactive attitudes, for on this account, a person
can experience such attitudes only if he is capable of experiencing the vicarious
reactive attitudes. Being capable of experiencing the latter attitudes, in other words,
is a necessary condition for being capable of experiencing the former. Hence, at
least on the modified Strawsonian account, a thoroughly egocentric person, some-
one lacking a conscience and incapable of empathy, could nonetheless be liable to
resentment. This account, in short, supports an understanding of psychopaths as
capable of resentment.25
If psychopaths, by virtue of their lacking a conscience, are the paradigm of
human agency that is incapable of moral judgment but otherwise rational, then the
modified Strawsonian account yields the better explanation of this incapacity and
thus yields a better understanding of the nature of moral agency. The two accounts
differ in identifying the capacity for moral judgment with the acquisition of differ-
ent liabilities to reactive attitudes. The modified Strawsonian account identifies it
with the acquisition of a liability to the self-reactive attitudes. Strawson’s original
account, by contrast, identifies it with the acquisition of a liability to the vicarious
reactive attitudes. On his account, acquiring the capacity for these attitudes is inde-
pendent of one’s acquisition of a liability to the self-reactive attitudes. Yet if the
hallmark of the capacity for moral judgment is the ability to see oneself as others
see one and to react accordingly, then Strawson’s original account, in identifying
the capacity for moral judgment with the liability to experience the personal reac-
tive attitudes vicariously, falls short. A general sympathy with others that gives rise
25 Strawson’s characterization of the moral solipsist is that of someone who is capable of experi-
encing the personal reactive attitudes but not the vicarious ones, and Strawson allows that such a
person might also be capable of the self-reactive attitudes. Consequently, the moral solipsist, as
Strawson characterizes him, does not fit the profile of a psychopath. At the same time, Strawson
raises the possibility of a moral idiot, who does. See Strawson, 85.
to one’s being liable to the vicarious reactive attitudes is, in other words, insuffi-
cient for understanding oneself as joined with others in relations of reciprocal moral
obligations and for understanding, in consequence, the moral responsibility that can
attach to failures to meet those obligations. Hence, only with the development of the
self-understanding that comes from being able to see oneself as others see one and
of the liability to the self-reactive attitudes that is consequent to it does one acquire
a capacity for moral judgment. Such development marks a significant advance over
the development of a capacity for experiencing vicarious attitudes.26 The modified
Strawsonian account of the reactive attitudes captures this advance. By doing so it
explains, as Strawson’s original account does not, the moral deficit that is so mem-
orably displayed by Hitchcock’s villains in Shadow of a Doubt and Strangers on a
Train.
References
Darwall, S. 2006. The Second Person Standpoint: Morality, Respect, and Accountability.
Cambridge, MA: Harvard University Press.
Deigh, J. 1995. “Empathy and Universalizability.” Ethics 105:743–63.
Deigh, J. 2011. “Reactive Attitudes Revisited.” In Morality and the Emotions, edited by C. Bagnoli.
Oxford; New York: Oxford University Press.
Hare, R. D. 2003. Manual for the Revised Psychopathy Checklist, 2nd ed. Toronto, ON: Multi-
Health Systems, Inc.
Hoffman, M. 2000. Empathy and Moral Development: Implications for Caring and Justice.
Cambridge: Cambridge University Press.
Hume, D. 1975. Enquiries Concerning Human Understanding and Concerning the Principles of
Morals, edited by P. H. Nidditch, 3rd ed. Oxford: Clarendon Press.
Hume, D. 1978. A Treatise of Human Nature, edited by P. H. Nidditch, 2nd ed. Oxford: Clarendon
Press.
Rawls, J. 1963. “The Sense of Justice.” Philosophical Review 72:281–305.
Rawls, J. 1971. A Theory of Justice. Cambridge, MA: Belknap Press.
Rousseau, J. J. 1979. Emile or On Education. Translated by A. Bloom. New York: Basic Books.
Strawson, P. F. 1962. “Freedom and Resentment.” Proceedings of the British Academy, 48, 1–25.
Reprinted in G. Watson (ed.), Free Will, 2nd ed. Oxford: Oxford University Press, 2003, 72–93.
Wallace, R. J. 1996. Responsibility and the Moral Sentiments. Cambridge, MA: Harvard University
Press.
Watson, G. 1988. “Responsibility and the Limits of Evil: Variations on a Strawsonian Theme.”
In Responsibility, Character, and the Emotions, edited by F. Schoeman, 256–86. Cambridge:
26 See Hoffman (2000, 93–112).

Chapter 10
Self-Knowledge, Knowledge of Others,
and “the thing called love”
Edward Harcourt
10.1
Under the influence of psychoanalysis and psychoanalytically inspired research
in child development, philosophers have recently begun to acknowledge the role
played by loving nurture in the development of the capacity to think about one-
self or, as I shall put it – though the thoughts in question obviously sometimes fall
short of knowledge – the capacity for self-knowledge. As Neera Badhwar has said,
drawing on the work of D. W. Winnicott and others:
The look of love does more than see the loved individual veridically: it also shows the
loved individual what it sees. . . . The loving mother reflect[s] the baby’s facial expressions
and mental states on her own face, thereby giving the baby a concrete image of its own
psychological states. The mother’s look of love is, then, the first avenue to self-awareness
and self-understanding.1
In this paper I approach love’s role in the development of self-knowledge via

Wittgenstein’s account of knowledge of other minds. I identify a gap in the account
of knowledge of others often credited to Wittgenstein, and try to repair it using
some relatively neglected Wittgensteinian materials. The conclusion of that part of
the discussion is that a given subject’s capacity to self-ascribe psychological states,
and that same subject’s capacity to other-ascribe them, depend on knowledge of that
subject by another. I then turn to child development, where some recent work points
tentatively towards the conclusion that the capacity to think about oneself in mental
state terms, as indeed about others, depends on quality of nurture – very roughly,
then, that receiving the right kind of loving nurture promotes one’s capacity to be a
self-knower. But arguably loving another involves, among many other things, see-
ing the other as they really are. So philosophy gives us reasons of its own to accept
the tentative conclusions of empirical research: if self-knowledge depends on oth-
ers’ knowledge of us and others’ knowledge of us is implied by their love of us, the
1 “Love”, in N. Badhwar (2003).

E. Harcourt (B)
Faculty of Philosophy, Oxford University, and Keble College, Oxford, UK
e-mail: edward.harcourt@philosophy.ox.ac.uk
172 E. Harcourt
conclusion that others’ love of us works in favour of self-knowledge is more or less

what we should expect.
The shape of the paper is as follows. First I outline two problems about other
minds and Wittgenstein’s general approach to them (Section 10.2). I then amplify
the solution to the more fundamental of these problems that’s usually credited to
Wittgenstein, and identify some difficulties with it (Section 10.3). Next, and drawing
on further materials from Wittgenstein, I sketch a way of repairing these difficul-
ties which makes use of the connections between self-knowledge and knowledge
of others, both our own knowledge of others and others’ knowledge of ourselves
(Section 10.4). In the final section I argue that a focus on child development makes
vivid the priority over self-knowledge of the knowledge others have of us by
showing how love is implicated in both (Section 10.5).
10.2
Philosophers distinguish at least two problems of other minds, the epistemic prob-
lem and the conceptual problem. The epistemic problem is the problem of how we
can know that others have minds, or what particular mental state they’re in. But in
order so much as to pose these questions, we must already have a concept of mental-
ity, or of different particular mental states, such that it’s possible at least to entertain
the thought that persons other than ourselves have minds, and are in those states. We
deploy concepts of mentality in others in framing the epistemic problem, and indeed
in the well-known sceptical answer to it, which combines the thought that we can’t
ever know whether others really have minds with the thought that they do appear to.
So the conceptual problem is prior: the very fact that the epistemic problem can be
posed shows that the conceptual problem – the problem, that is, of how we can so
much as conceive of mentality in others, or possess concepts of mental states that
both I and others can in principle satisfy – has a solution, though of course we may
not yet be able to state what that is.
So much, I assume, is agreed. But Wittgenstein takes the priority of the concep-
tual over the epistemic problem one step further. According to Wittgenstein, if we
can solve the conceptual problem – if we can show how it’s possible to possess
concepts of mental states such that both I and others may intelligibly satisfy them –
the epistemic problem evaporates. Let me expand a little. The threat which the epis-
temic problem engages with is other minds scepticism – mightn’t I be the only one?
But the other minds sceptic assumes two things. The first is that I can have direct,
supposedly sceptic-proof, knowledge of a mental state only through being the sub-
ject of that state. The second is that I can acquire mental state concepts from my own
case alone: just being in a mental state is enough to give me a concept of that state
that’s applicable both to myself and to others. For convenience let’s call concepts
of mental states that are applicable to myself and to others person-neutral concepts.
The sceptic must allow that his mental state concepts are person-neutral, otherwise
the question whether or not others really instantiate them can’t be genuinely open,
as the sceptic needs it to be. But thanks to the sceptic’s first assumption, the question
is then readily settled in the sceptic’s favour, since all I can know directly of others –
their behaviour – is (so it’s said) compatible in every case with the absence of the
state.
To this, Wittgenstein replies as follows:
If one has to imagine someone else’s pain on the model of one’s own, this is none too easy
a thing to do: for I have to imagine pain which I do not feel on the model of the pain which
I do feel. That is, what I have to do is not simply to make a transition in imagination from
one place of pain to another. . . . For I am not to imagine that I feel pain in some region of
[another’s] body. (Which would also be possible.)2
To summarize: I can’t (according to Wittgenstein) acquire a person-neutral mental

concept just from my own case. At the same time it’s beyond question that I have
mental concepts, and that I can apply them to myself. Now either these concepts
are person-neutral or they are not. If they are not, then solipsism is true: it’s no
longer doubtful whether there can be minds other than my own, but a conceptual
truth that there can’t be, because my mental concepts are such that I’m the only one
who can conceivably satisfy them. Thus the second horn of the dilemma puts the
other minds sceptic, and so the epistemic problem, out of business. If my concepts
are person-neutral, on the other hand, the sceptic’s first assumption – that the only
mental states I can know directly are my own – must be false, so once again the
epistemic problem is out of business, because it relies on that assumption to get
going. On Wittgenstein’s view, then, the epistemic problem of other minds occupies
an unstable middle ground between the possibility – fallible, of course – of knowl-
edge of other minds, and solipsism. The conceptual problem of other minds is the
only game in town.3
To note that this is Wittgenstein’s view of the relation between the epistemic and
the conceptual problems is not, of course, to argue for it. What’s more, even if it’s
true, the Wittgensteinian view I have sketched is disjunctive – either solipsism or the
possibility of shared knowledge – and I haven’t said anything to favour one of these
alternatives over the other. The burden of eliminating the first alternative – that one
can acquire non-person-neutral mental concepts from one’s own case – is carried, in
Wittgenstein’s own work at least, by his considerations about the (im)possibility of
a private language.4 My topic here, however, is the second alternative. If the mental
2 L. Wittgenstein (1953) [henceforth PI], §302; cf. also especially §351: “Yet we go on wanting to
say: ‘Pain is pain – whether he has it, or I have it; and however I come to know whether he has a
pain or not.’ – I might agree. – And when you ask me ‘Don’t you know, then, what I mean when I
say that the stove is in pain?’ – I can reply: These words may lead me to have all sorts of images;
but their usefulness goes no further.”
3 I am indebted here to Hacker’s (1986) discussion: see the comparison of the “weak-kneed sceptic”
and the “tough-minded solipsist”, pp. 262–64. Note, however, that to claim philosophical priority
for the conceptual problem is not to claim priority in the order of acquisition for concepts of
others’ mental states over knowledge of such states. On the contrary, I take it that the concepts and
the knowledge are developmentally coeval (and that the development of both is very gradual).
4 See e.g. PI §293.
174 E. Harcourt
concepts I have are indeed person-neutral, then there must be a story to tell about
how I have come to have them (and indeed a story to tell about how I can have
knowledge, directly, of the mental states of others and they of mine). Assuming
this second alternative is roughly the right way to go, I’m going to sketch some
shortcomings of a story often credited to Wittgenstein, and then try to say something
about how to put them right.
10.3
Wittgenstein is often credited with the view that, far from knowing our own mental
states directly, we cannot properly be said to know that we have them at all:
It can’t be said of me at all (except perhaps as a joke) that I know I am in pain. What is it
supposed to mean – except perhaps that I am in pain?
Other people cannot be said to learn of my sensations only from my behaviour, – for I
cannot be said to learn of them. I have them.5
In place of what we can call the first-person-only direct access model, he proposes
the following account of primitive self-ascriptions of mental states:
How does a human being learn the meaning of the names of sensations? – of the word
‘pain’, for example. Here is one possibility: words are connected with the primitive, the
natural, expressions of the sensation and used in their place. A child has hurt himself and
he cries; and then adults talk to him and teach him exclamations and, later, sentences. They
teach the child new pain-behaviour. . . . [T]he verbal expression of pain replaces crying and
does not describe it.6
As Hacker puts it, “the utterance, like the groan, is an expression or manifestation . . .
of pain. It is a learnt form of pain-behaviour”.7 Let’s call this the expressive account
of primitive psychological self-ascription.
The expressive account of primitive psychological self-ascription has been the
target of various objections. First of all the account cannot, it’s said, apply to all
mental concepts, because not every mental state is like pain in having a natural,
language-independent expression: in Anscombe’s view, intention does not8 ; nor
does belief. Secondly, it can’t be a complete account, just as it stands, even of
those mental concepts to which it does apply. For though some psychological self-
ascriptions are “wrung from us”9 by the states they express – “that hurts”, perhaps,
5 PI §246.
6 PI §244, my italics. See also e.g. PI II, p. 174, “[T]hink of the sensations produced by physically
shuddering: the words ‘It makes me shiver’ are themselves a shuddering reaction”; Wittgenstein
(1992, 14), “[An] exclamation is [an expression] . . . in a different sense from [a] report. It is wrung
from us. It is related to the experience as a cry is to pain.”
7 Hacker (1986, 293–94).
8 Anscombe (1957, §2).
9 PI §546: “And words can be wrung from us – like a cry”. Cp. Wittgenstein (1992, 14): “[An]
exclamation is [an expression] . . . in a different sense from [a] report. It is wrung from us. It is
related to the experience as a cry is to pain.”
as uttered when prodded by an exploring doctor – by no means all of them are,

and this is not yet to mention compound uses (and therefore uses which obviously
aren’t doing any expressing) such as “I’m not in pain” and “If I’m in pain I want the
strongest painkillers I’m allowed”. Nonetheless the concept “pain” that features in
these uses is surely the very same one that features in the uses most congenial to the
expressive account.
To these objections Wittgenstein could, I think, reply that the expressive account
of primitive psychological self-ascription isn’t meant to be a complete account of
how we get into the way of ascribing mental states to ourselves: it’s only meant to
be an account of the primitive beginnings of this capacity, and the story of how we
get from the primitive beginnings to the full-blown capacity will of course need to
be elaborated step by step. The person-neutrality issue, however, gives rise to a more
serious pair of objections, more serious because they threaten the adequacy of the
expressive account of primitive psychological self-ascription even as an account of
mere beginnings, and for this reason I’m going to focus exclusively on this pair of
objections and the solution to them, leaving the step by step elaboration of the full
story to another day. The first-person-only direct access model of self-knowledge
creates difficulties, as Wittgenstein says, for the idea that anyone other than me
could even possess the concept I deploy when I say I am in pain – let alone know
that I satisfy it. But does this problem get any easier on the expressive account? On
the face of it, no. Crying out in pain doesn’t seem to be the deployment of a concept
at all, so the more closely psychological self-ascription shadows the natural expres-
sions of the states so ascribed, the less we have an account of concept-acquisition.
Moreover, the only mental states I can express are my own, so if the story about my
acquisition of the concept I deploy when I say that I’m in pain appeals only to the
idea that “I’m in pain” replaces an unlearnt pain-expression, then this concept too
looks as if it will be one which only I can possess.
This second worry can be developed a bit further. A standing difficulty with men-
tal concepts if the sceptic’s assumptions are granted is that the concepts fragment,
so instead of one concept that’s exercised both in the first and third person we have
two: one that’s usable only in the first person (because it gets its meaning from
something only the subject of the state can be exposed to, namely the state itself),
the other that’s usable only in the second or third person (because it gets its mean-
ing from what we have to make do with when talking about other minds, namely
observable behaviour). So even if you and I both mean the same when we each say
“I am in pain” – that is, forgetting for the moment about the threat of solipsism –
I don’t mean the same when I say “I am in pain” as you do when you say, of me,
“he is in pain”. But this makes it, absurdly, appear as an accident that “I’m in pain”
as said by me is true if and only if “he is in pain” is true as said of me by another.
Something like this problem seems to resurface on the Wittgensteinian view, at least
unless more is said. On the Wittgensteinian view no less than on its rivals, we ascribe
psychological states to others on the basis of behaviour. But, on the expressive just
as on the first-person-only direct access model, self-ascriptions of mental states are
not evidence-based – that is, they aren’t answerable to evidence of any kind. So on
the face of it the old worry about person-neutrality hasn’t gone away: if two different
176 E. Harcourt
types of evidential basis led, on the old picture, to the fragmentation of (apparently)
person-neutral mental concepts into distinct first- and third-person concepts, why is
this problem any less urgent when, instead of two different types of basis we have a
behavioural basis (in the third person) and no basis (in the first)?10
To raise these problems is not to say that the expressive account of primitive
psychological self-ascription, even when clear-sightedly presented as an account
only of that, is wrong, but rather that it is incomplete: the approach needs to tell a
story about how first- and third-person uses of the same psychological words (“pain”
and so on) can constitute exercises of the same psychological concepts, and we
haven’t told that story if all we do is replace the first-person-only direct access model
with the expressive model. More needs to be done to show how the first- and third-
person uses of these words stitch together. In the next section I want to make some
(I hope) new efforts in that direction.
10.4
The thought I want to explore here, by way of filling in some gaps in the
Wittgensteinian story, is that self-knowledge depends, at least in part, on the knowl-
edge others have of us. But though I do want to assert the priority “ontogenetically”
over self-knowledge of the knowledge others have of us – that is, to assert that my
self-knowledge (as indeed my knowledge of others) typically depends for its expan-
sion on the presence in my environment of others who know me – one place I would
like to end up is a point at which the capacity for self-knowledge and the capac-
ity for knowledge of others become, in the case of a single individual, two aspects
of the same complex capacity. That, at any rate, is the ambition. Secondly, I don’t
want to abandon the expressive account of primitive psychological self-ascription:
what’s wrong with the Wittgensteinian account I’ve set out so far is not that it places
self-expression in some sense at the beginning of the story of our acquisition of
person-neutral psychological concepts – I think it’s quite right to do this – but that it
doesn’t say enough about what else belongs with it. Thirdly, I suggest that the ambi-
tion of an appropriately filled-out Wittgensteinian account should be not to offer a
transcendental deduction of what we know to be the case – that we possess person-
neutral psychological concepts – but rather to tell a story about how we acquire
them that actually takes place. So there will be more by way of empirical material
than there is in Wittgenstein – though I think this is wholly consistent with his own
approach. Finally, though perhaps it doesn’t matter how it happens as long as we
can point to some way in which it happens, I’m going to discuss just one family of
ways in which it happens – but of course there may be others.
10 This problem was emphasized by P. F. Strawson, though it has been overlooked often enough
since: see Strawson (1959, 99–100).
Commentaries on Philosophical Investigations 244 tend to make a great deal of

just one idea in it, the idea that the child’s own sensation-language is a verbal substi-
tute for expressive behaviour – what I’ve called the expressive account of primitive
psychological self-ascription. This account has the apparent merit of solving one
problem: if ground-level sensation-language is nothing but a verbal substitute for
expressive behaviour – that is, if it is expressive behaviour – then we don’t have to
worry about how the child finds out that he’s in the state he says he’s in when, e.g.,
he says he’s in pain. As Wittgenstein himself says,
It is not as if he had only indirect, while I have internal direct evidence for my mental state.
Rather, he has evidence for it, (but) I do not.11
There isn’t, at this level, an epistemology of sensations (in the sense of a way we
know which ones we’re having) any more than there’s an epistemology of crying
or shivering. This helps along the thought that the knowledge others have of our
mental states when we thus express them is direct, because there is no more direct
knowledge of them which the child himself has and with which others’ knowledge
can be unfavourably contrasted.
But in embracing the expressive account of primitive psychological self-
ascription, commentators tend also to miss something. The child – obviously
enough – doesn’t get from natural expression to verbal substitute unaided: as
Wittgenstein says in a sentence I didn’t emphasize the first time I quoted
Investigations 244, “adults talk to him and teach him exclamations and, later,
sentences”.12 But a great deal is presupposed by this “teaching of exclamations”
(etc.) which needs to be brought to the fore. In particular, teaching exclama-
tions (etc.) presupposes that the adults in question know or can recognize what
psychological state the child is in.
It’s at this point that the question how we can know others’ – in particular pre-
verbal children’s – psychological states is of special importance. A crucial part of
Wittgenstein’s answer to this is summed up in a – compared to the many sources for
the expressive account of self-ascription – rare but invaluable passage:
It is a help here to remember that it is a primitive reaction to tend, to treat, the part that hurts
when someone else is in pain; and not merely when oneself is – and so to pay attention to
other people’s pain-behaviour, as one does not pay attention to one’s own pain-behaviour.13
The first thing to say about this is that it seems to be true: we experience a range
of sympathetic reactions to others which are as involuntary as are, sometimes, the
expressions of our own psychological states and which, even if it might be mislead-
ing to describe them as unlearned – they might have an interesting developmental
11 Wittgenstein (1992, 67).

12 Cp. “We teach the child to use the words ‘I have toothache’ to replace its moans, and this was
how I too was taught the expression”, Wittgenstein (1968, 295).
13 Wittgenstein (1967, §§540–41), first italics mine. Compare Wittgenstein (1980, §27), “When I
see someone else in a terrible situation, even when I myself have nothing to fear, I can shudder,
shudder out of sympathy. . . . We are afraid along with the other person, even when we have nothing
to fear”; PI §287, “Pity, one might say, is a form of conviction that someone else is in pain”.
178 E. Harcourt
history which would give the lie to that phrase – it is natural and typical for human
beings to display. A child cuts its finger and sheds blood – you wince; a trusted
person panics in a situation that seemed to be under control, and you panic; a loved
one – or perhaps not a loved one, given the right context – bursts into tears, and
you burst into tears too. The moral Wittgenstein wants to draw from this is that
other-ascriptions of psychological states are rooted in prelinguistic behaviour – our
reactions to others – no less than the corresponding self-ascriptions are. Of course
if the disposition to react to others in these ways were found, unpredictably, in
some people only, or in all people but only some of the time, these dispositions
couldn’t ground anything which has the regularity characteristic of language-use.
But this is not the way things are. The very high degree of regularity in the types
of states we ascribe on the basis of others’ expressive behaviour is explained by the
fact that we converge in our dispositions to react to others’ reactions (and to other
related aspects of their behaviour and of its context). The fact that we converge in
our unlearned dispositions to react to others’ expressive behaviour means that these
reactive dispositions can provide a foundation for our other-ascriptions of psycho-
logical states which predates identification, i.e. the subsumption of some observed
phenomenon under a concept. Here’s how the Zettel passage goes on:
But what is the word ‘primitive’ meant to say here? Presumably that this sort of behaviour
[sc., reacting to others’ reactions] is pre-linguistic: that a language-game is based on it, that
it is the prototype of a way of thinking and not the result of thought.
And again a few remarks further on,

Being sure that someone is in pain, doubting whether he is, and so on, are so many natural,
instinctive kinds of behaviour towards other human beings, and our language is merely an
auxiliary to, and further extension of, this relation. Our language-game
– and here I take it he doesn’t mean, or doesn’t only mean, our language game of
psychological self-ascription –
is an extension of primitive behaviour. (For our language-game is behaviour.)14
In other words it’s no less true of other-ascriptions of psychological states than

it is of self-ascriptions that they are, at the ground level, linguistic replacements
of prelinguistic behaviour. Let’s call this the expressive account of primitive
psychological other-ascription.15
There’s an objection to the overall Wittgensteinian picture of knowledge of mind
that the expressive account of other-ascription helps with. I’ve presented the expres-
sive account of self-ascription so far as a way of countering the thought that others’
access to our mental states is indirect. But it buys this advantage at the cost – or so
it would seem – of maintaining that talk of knowledge of our own mental states (at
least where pains etc. are concerned) is out of place altogether (“it can’t be said of
14 Wittgenstein (1967, §§541, 545).

15 For something very like the expressive account of other-ascription (though I do not mean to
burden her with the elaboration of it I give here), see Bar-On (2004, 288). This book came to my
attention too late for me to be able to do it justice in this paper.
me at all . . . that I know I am in pain”). But why say that, because there’s no way in
which we find out that we have pains (no evidential basis for the self-ascriptions),
we don’t know that we have them? And anyway, don’t we know we have them? We
are aware of our pains (that is to say, we have them), and it might be said that the
gap between “awareness” and “knowledge” seems too small to be significant. These
are tricky issues which I cannot hope to resolve here, but the expressive account of
other-ascription allows us to sidestep them. We can have direct knowledge of others
not because they don’t have knowledge of themselves, but because our natural dis-
positions to react are geared as much to others’ expressive behaviour as they are to,
for example, injuries to our own bodies. Maybe what the verbal substitutes for one’s
own pain-reactions express is knowledge of oneself, but never mind – it’s something
others can have of me too, and for not dissimilar reasons.
Now the ability to other-ascribe psychological states comes into the explanation
of the child’s ability both to self-ascribe them (“teaching exclamations (etc.)”) and to
other-ascribe them, insofar as adults must possess that ability in order to teach chil-
dren which words to substitute for their pre-verbal reactions. But at the adult level,
other-ascription isn’t wholly to be explained by the expressive account of it any
more than adult self-ascription is to be wholly explained by the expressive account
of that. But there is another, if anything more important moral of the Zettel passage
which concerns the child’s own ability to other-ascribe psychological states. That
knowledge of one and the same fact is expressed by my primitive verbal substitutes
for my pain-reactions and by a child’s primitive verbal substitutes for its reactions
to my pain-reactions point to the co-ordinated pattern of “primitive behaviour” –
my expressive behaviour and a child’s expressive behaviour that’s attuned to it –
on which our language-game of psychological ascription is based. And for some-
one plagued by the worry that our psychological vocabulary expresses unrelated
concepts in its first- and second- or third-person uses, this co-ordination of primitive
behaviour holds the promise of deliverance. Let’s begin with the fact that any child’s
reactions to my pain-reaction – to stick with that example, though my joy-reaction
would do as well – not only have the same degree of regularity as nature displays
in giving shape to my expressive behaviour itself: they are also sometimes the same
reaction. If I can teach a child to say, of itself, “my finger hurts” thanks to its expres-
sively wincing and sucking its finger, I can also teach it to say, of me, “your finger
hurts” on the basis of that same child’s wincing when I hurt my finger. Thus it looks
as if there is a neutrality between the first and second or third persons already at the
pre-verbal level: just the sort of thing one needs if one is seeking, in Wittgensteinian
fashion, to account for the possibility of a concept by finding its “prototype”, that
is, by displaying its use as the elaboration of some more primitive phenomenon.
But how often is the reaction really the same?16 In some cases, indeed, not only
is the reaction the same for both persons but so is the psychological state expressed,
as clearly in the panic case above. Sometimes, however, the reactions are the same
but the psychological states are distinct or merely overlapping, as in the case of my
16 This question is pressed in Hobson (2009).

180 E. Harcourt
cutting myself, where perhaps shock or distress are shared, but pain in the finger
is not, though at a certain level of description the stimulus is the same: someone’s
cutting their finger. However, I am not sure that we should look upon the cases
where there is sameness of reactions (and a fortiori sameness of states) as basic,
and seek to work outwards from these, attractively simple as it might seem to claim
that it is sameness of reactions that, in the basic case, underwrites the sameness of
the psychological concept expressed in first- and second- or third-person uses of the
same substituting psychological words. After all, it is not as if the word “pain” is
offered, on the expressive account of self-ascription, as a first-person substitute for
just one particular reaction (it can just as well replace crying as wincing or rubbing
the injured spot). So even if what underwrites the person-neutrality of the concept
expressed by the word “pain” is that your and my twinned reactions to my injury
are both characteristic pain-reactions, it seems unnecessary that they both be the
same one. But I do not think the concept’s person-neutrality is underwritten by the
fact that our twinned (co-ordinated) reactions to my injury are both characteristic
pain-reactions, if indeed they are: what it’s underwritten by is the fact that they are
co-ordinated, in a sense I’ll explain. What appears crucial is that the other-ascription
I offer you (“you’re in pain”, to be said by you to me) be offered as a substitute for
a reaction of yours which is a reaction to my reaction, and which – obviously, if I
am in a position to do the offering – I understand as such.
However, this too isn’t quite enough. The phrase “a reaction to my reaction”
covers at least two distinct phenomena. On the one hand, it fits the case where I
panic and you “catch” my panic and panic as well. In this case, you are reacting
to my reaction (indeed, you are reacting to the state my reaction expresses), but the
relation my state bears to yours is merely that of cause to effect. By this I do not mean
that it is just as if you, about to be inoculated, had panicked at the sight of a needle,
since even our primitive “catching” reactivity to one another (as opposed to our
reactivity to stimuli of other sorts) is doubtless an important part of the scaffolding
on which the distinct form of reactivity I wish to focus on is built. Nonetheless, in the
“catching” case, the fact that the cause was a state of mine has, as it’s tempting to put
it, no echo in the nature of the state your reaction expresses. The reactions to others’
reactions I have in mind, by contrast, show a more intelligent reactivity.17 These
differ from “catching” cases for a start because they are behaviourally different, in
that the reactions in question form a much looser family: I can only catch panic
from panic, but though I can intelligently wince at your wincing, I can also react
with a look of distress, or a sharp intake of breath, or by reaching out my hand
towards you. Moreover in the intelligent cases, it may be that there is no way to
characterize your reaction otherwise than to say that it’s a reaction to my reaction,
and indeed to the state my reaction expresses. Contrast panic: it is a reaction to
17 “Recent studies of children in the second year of life indicate that they have the . . . [capacity]
to display integrated patterns of concern for others in distress. During this period of development,
children increasingly experience emotional concern ‘on behalf of the victim’”, Zahn-Waxler et al.
(1992). In another study, in which parents simulated hitting themselves with a toy hammer and
crying out in pain, 100% of normal children looked at their parent’s face and 68% showed “facial
concern”, Sigman et al. (1992).
my reaction, but an intrinsic characterization is also available, viz. that it simply

expresses your panic, and that panic is what it is no matter what caused it. In the
panic case all the adult will think of the child is that his own panic has “set the
child off”; in the intelligent case, by contrast, the adult will think, of the child, “he
is upset that I am upset”, or “she is shocked that I am in pain”. Now because, in the
intelligent case, I understand your reaction (whatever typical form it takes) to be an
acknowledgment of the very state expressed by my own reaction, there’s a point to
my teaching you to substitute for it the same word, in the second person, which I
use, in the first person, to express the state I see you are reacting to (just as when I
teach you to self-ascribe pain, I teach you the first-person form of the word which, in
the second person, I learnt to substitute, years back, for my own intelligent reaction
to the state you express with it). It’s the co-ordination of your intelligent reactivity
with my expressive behaviour, and of my intelligent reactivity with your expressive
behaviour, that gives point to the introduction of one and the same word to express
both your pain and (with the relevant shift of person) your reaction to mine, and thus
that underwrites the person-neutrality of the psychological concept it expresses. In
this way by supplementing the expressive account of primitive psychological self-
ascription with the expressive account of primitive psychological other-ascription
we are able to remedy a troublesome defect in the account of our knowledge of
others frequently credited to Wittgenstein.18
But is the story we have told really one about the acquisition of a concept? The
worry is that the closer we get to describing what’s distinctive about intelligent reac-
tivity as the possession of an attitude with an object (such as the child’s distress at
the adult’s pain), the closer we get to conceding that the pre-linguistic child already
18 Anita Avramides does well to emphasize the importance both of untutored reactions to others and
of the person-neutrality issue: see Avramides (2001). However, Avramides seems to have especially
in mind the fact that, just as each person is alike in their reaction to their own pain, each is also
alike in their reaction to another’s pain. (I take it that this is what she means when she says (p. 196)
that “[Wittgenstein] does not intend simply to draw our attention to our reaction to, say, pain; he
also intends to call attention to the way we react to each other when we are in pain. We share
these reactions just as we do all others”.) But notwithstanding her claims to the contrary (“the
description we have given of the asymmetry [between first- and second- or third-person uses] is
such as to make it clear how it is that the word ‘pain’ has a univocal meaning”, p. 201), it’s not
clear how the person-neutrality problem is solved, if the two sets of common reactive propensities
Avramides describes, together with the asymmetrical linguistic practice they are said to ground, is
all there is to our practice with sensation-words. Let’s agree that the two sets of common reactive
propensities explain how the first-person use of “pain” and its second- and third-person uses both
constitute the deployment of a concept. For if these propensities were not common to almost all
of us, there would be no pre-linguistic regularity on the basis of which to introduce these words
in such a way that they express concepts: “if rule became exception and exception rule; or if both
became phenomena of roughly equal frequency – this would make our normal language-games
lose their point”, PI §142, quoted by Avramides, p. 196. What these common propensities do not
explain, however, is why the first-person use of “pain” and its second- and third-person uses do
not simply constitute the deployment of two distinct concepts, one to go with each set. To explain
why they do not requires not only that each person be alike in their reaction to their own pain and
alike in their reaction to another’s pain, but, as I have emphasized here, that each person’s reaction
to another’s pain be such that its nature cannot be specified without reference to the state it’s a
reaction to.
182 E. Harcourt
has the person-neutral concept whose possession we’re trying to explain, and so
the closer also to conceding that our story is simply about fitting words to concepts
the child already has. The objection tries to get us to choose between two alterna-
tives: either the child’s learning to say “you are in pain” or “that hurts” (pointing to
another) marks a leap from no concepts to concept-possession, or it adds nothing
to the child’s conceptual capacities. If we choose the second, we must give up on
explaining the person-neutrality of the concept expressed by the word “pain” (or
whatever it might be) by appeal to something more primitive; and we cannot choose
the first if we are not to deny ourselves the means to distinguish the intelligent
from the merely “catching” reaction, since such an important addition to the child’s
behavioural repertoire surely marks an expansion of its conceptual capacities too.
Perhaps we could dig in our heels here and say that the expansion in repertoire is
merely behavioural, but nonetheless it is a critical expansion because it puts in place
the scaffolding for the introduction of a new psychological concept – that of some-
one else’s pain – that wasn’t there before. But suppose we do not dig in our heels.
Why accept that the two stated alternatives are the only ones? Let’s suppose first of
all that language-learning and concept-acquisition are not coeval, a view suggested
by the child who can react intelligently to others’ reactions but who is as yet inca-
pable, even if not of articulate speech altogether, at least of other-ascription. This
supposition is consistent with rejecting the first alternative, as the objector wants
us to do. But it’s also consistent with claiming that, whatever conceptual capacities
the child already has in virtue of his intelligent reactivity, the transition to articulate
speech adds something to them, so it’s consistent with rejecting the second alter-
native as well. For our story to be a story of concept-acquisition, it is not required
that the child make a sudden leap from a lack of any concepts at all to possession
of the person-neutral concept: all that’s required is that its conceptual repertoire be
gradually expanded. A similar reply can be made if we suppose, alternatively, that
language-learning and concept-acquisition are coeval, a view we may find attractive
once we remember that language-acquisition is itself very gradual. (For example,
children learn the rhythm and cadence of sentences, and to look at the person who
is speaking to them, before they learn any words, but even at this stage they already
have a foothold in the kind of structured interactivity characteristic of language-
use.) Given this, we can credit the child who hasn’t yet learned the words “you are
in pain” with some conception of the other’s pain, thanks to his intelligent reactivity.
But does this mean we have to credit him with the very concept he would possess
were he able to say (as he surely won’t be for a while yet) “I am sorry that you are
in pain”? No: on this picture, concepts come as gradually as language does. Thus
whether we suppose that language-learning and concept-acquisition are coeval or
not, we can explain the possession of a person-neutral concept by appeal to some-
thing more primitive, without claiming that that more primitive something implies
the total absence of concepts of any kind.19
19 Nonetheless there is no way of identifying the concepts learners possess at their various stages
of development otherwise than as the developmental forerunners of the relevant adult ones.
Let’s summarize where we have got to so far. Children express their

psychological states in unlearned but nonetheless highly uniform reactions. In order
for them to advance to the linguistic substitutes for these, they must be offered the
substitutes by adults, so adults must be able to match (reliably) words to children’s
reactions. Thus, at this basic level, the child’s capacity to self-ascribe psycholog-
ical states rests on others’ capacity to ascribe these same psychological states to
it. But something similar is also true of the child’s capacity to ascribe psycho-
logical states to others: this rests on the child’s disposition to react intelligently
to others’ reactions, but in order to advance from that disposition to its linguis-
tic elaboration, the environment must contain others who recognize these reactions
for what they are. Moreover, because the reactions for which other-ascriptions of
psychological states are the second- or third-person verbal substitutes are (and are
understood to be) reactions to states expressible, by the subject of ascription, in the
first-person form of the very same words, putting the expressive account of psy-
chological other-ascription together with the more familiar expressive account of
psychological self-ascription yields the beginnings of an account of psychological
concept-possession that’s person-neutral.
10.5
It’s time now that the third element of my title made an appearance, so in this final
section I want to leave the person-neutrality issue on one side and focus in more
detail on the priority of others’ knowledge of us over our knowledge of ourselves.
Iris Murdoch is well known for a view of love which is captured in the following
remarks of hers:
[R]eal things [need to be] looked at and loved without being seized or used, without being
appropriated into the greedy organism of the self. . . . [T]he ability . . . to direct attention
[away from the self] is love.20
Again,
It is in the capacity to love, that is to see, that the liberation of the soul from fantasy
consists.21
We don’t have to agree that this is what love is to agree that one of the characteristic
marks of love at its best is the capacity to disentangle one’s own needs, interests,
thoughts, desires from those of the loved one and to see them disinterestedly, as a
person in the round – that is, to see them (insofar as this is possible) truly; conversely
one of the things that commonly happens when love goes wrong is that the loved
one is seen merely as an extension or echo of oneself. So, we might think, those
who are best placed to know us best are those who love us best. In fact, however, we
can distinguish a strong and a weak thesis here. The strong thesis is that knowledge
20 Murdoch (1970, 65–66).

21 Murdoch (1970, 66).
184 E. Harcourt
of another is found only where we find love of that other: that is, knowledge implies
love. (I suspect this is what Murdoch thought.) Now if self-knowledge depends on
other’s knowledge of us and other’s knowledge of us implies their love for us, it
would seem to follow that self-knowledge requires other’s love for us. That would
be a very strong connection between self-knowledge and love. The weaker thesis is
that, of those who love us at all, those who love us best are those who know us –
which leaves it open that we may be known by others who don’t love us. So on
the weaker thesis, knowledge of us is implied by others’ love of us. But here too
there’s a connection. For if self-knowledge depends on other’s knowledge of us and
other’s knowledge of us is implied by their love for us, it would seem to follow
that self-knowledge is fostered by other’s love for us: other’s love for us might be a
standard route to self-knowledge, even if self-knowledge doesn’t require it. This is
the connection between self-knowledge and “the thing called love”22 that I shall be
exploring here.
Now the capacity for self-knowledge is unevenly distributed so, if the connection
between self-knowledge and other-knowledge is as I say, we should expect to find
the capacity for self-knowledge at its most fully developed where there is a context
of other-knowledge – that is, knowledge by others of us – to favour it. Moreover for
this very reason, if those who are best placed to know us best are those who love us
best, we should expect the uneven distribution of the capacity for self-knowledge to
track the obviously uneven distribution of the availability to potential self-knowers
of love. Though they don’t quite put it like this, I think this conclusion is supported
by some recent work in child development, which I now proceed both to summarize
and to interpret.
Before I go on, however, I’d like to try to forestall an objection. If there’s
truth in the thought that love fosters self-knowledge – so the objection runs – the
thought can hold good at most for self-knowledge in the sense in which the Delphic
inscription adjured us to accumulate it, that is, for a form of wisdom. For it’s
only self-knowledge in this sense that is unevenly distributed. The Wittgensteinian
considerations rehearsed up to now, on the other hand, hold good not for this self-
knowledge but for its mere homonym, self-knowledge in the minimal sense of the
capacity for (true, more or less reliable) self-ascription of things like pains, and
since everyone has that, it can’t depend on quality of interactions with others.23 A
modest reply to this objection would be that though mundane self-knowledge is so
mundane that no one can lack it, this doesn’t show that it’s wholly independent of
the presence in the environment of knowing others – it’s just that what’s required
for it from others is so basic that they cannot fail to provide it. But perhaps one
can overdo the distinction between the mundane and the “Delphic”. Though no one
22 I have chosen this phrase of D. W. Winnicott’s (1964, 17) for the same reasons I assume he chose
it himself: there is a range of phenomena including attunement to another, concern, emotionally
toned responsiveness and so on for which “love” is a perfectly good word, but if you object to the
word, drop it and think of another one – at this point at least, it is the phenomena and not the word
that matter.
23 For a statement of the mere homonym view, see for example Jäger (1999, 1).
counts as wise just because they know they are currently in pain, it is a mistake to
insist that there are two different senses of “self-knowledge” in play here rather than
a spectrum of different (and differently significant) kinds of things about oneself
which one can know. If the Wittgensteinian story I have told so far has any truth
in it, even self-knowledge of the mundane kind depends on social interaction, and
if the interaction were impaired, we would expect the capacity even for mundane
self-knowledge to be impaired too – which is not to deny that, the further we ascend
towards the Delphic, the more is demanded of quality of social interaction for the
capacity for self-knowledge to be in place, and the more chancy it is whether or not
those demands are met.
Now to a point about terminology. In the child development literature, the capac-
ity to think of oneself and others as minded is often referred to as the capacity to
“mentalize”. The term covers the capacity to deploy psychological words in relation
to oneself and to others, though it needn’t be tied – because the capacity to think of
oneself and others as minded needn’t be – to the use of a clearly demarcated vocab-
ulary.24 The capacity to mentalize, which I shall focus on in this section, therefore
clearly covers what I have been calling the capacity for self-knowledge (though of
course it also covers more – the capacity for knowledge of others too25 ). By con-
trast developmentalists don’t tend to talk about love or have any single word that
covers the same ground. This is presumably because the definition of love is itself
controversial, and even if it were not, because love is manifestable in a great variety
of ways and it’s easier to track the causes and effects of these singly or in small
clusters than to track those of love itself. I hope to show, however, that this doesn’t
matter, so I won’t mention love in introducing the empirical material but try instead
to bring it back in at the end.
There is now a body of evidence that seems to show that the development of
mentalizing ability is favoured by social, interactional factors as well as by physio-
logical endowment. If physiological endowment were all that mattered, one would
expect physiologically indistinguishable children to be developmentally the same.
But deaf children of deaf parents are apparently swifter to develop mentalizing abil-
ities than deaf children of hearing parents, though they are physiologically alike.
Indeed pre-school age deaf children of hearing parents show roughly the same
delays and deficiencies in development of mentalizing ability as autistic children
(and deafness is a common misdiagnosis of autism). Why? The thought is that deaf
24 Carpendale and Lewis do well to make this point: “In studying talk about the psychological
world it is important to remember that researchers should not just be concerned with mental state
terms but more broadly with talk about human activity”, Carpendale and Lewis (2004, 88). Some
of the studies (e.g. by Meins) cited below might have got on more easily if they had adopted a more
liberal view of what they were actually after.
25 Since in the previous section I have been arguing for the interdependence, in the case of a
single individual, of knowledge of oneself and knowledge of others, if there is a connection
between self-knowledge and love it would be a surprise to discover there wasn’t also a connec-
tion between love and knowledge of others – indeed the connection might be expected to confirm
the interdependence. But exploration of those further connections must await another occasion.
186 E. Harcourt
parents are much more likely to communicate with one another in ways accessi-
ble to a deaf child – e.g. in sign language – than hearing parents are. So the deaf
child of deaf parents grows up in a family context in which it can “overhear”, and
share in, conversations, including conversations (if they’re going on) in which men-
tal state concepts are deployed. The deaf child in a hearing family, by contrast, will
be relatively socially isolated.26
But which interactional factors? Some seem to be especially important.
According to Carpendale and Lewis,
mental state understanding [as measured e.g. by false belief tests at 4 or 5 years] is signifi-
cantly correlated with factors in the child’s social environment, like attachments, parenting
styles and parent-child communication,27
that is, with the quality of early child-caregiver interactions. Or as Peter Hobson
puts it, “appropriate forms of interpersonal engagement” favour children’s devel-
opment of concepts, and indeed of concepts of mind.28 Since intuitively one thing
that makes early child-caregiver relations good is love – or at least love from the
caregiver and the counterpart proto-attitude on the part of the infant or child – these
interactions would seem to be the place to look for the kind of evidence we want.
That said, the connections between early child-caregiver interaction and mentaliza-
tion are very much work in progress, so I’m not only going to be highly selective
but also necessarily take a good deal on trust.
One line of thought connecting the child’s mentalizing ability and the caregiving
environment runs the connection through maternal “mindmindedness” – mothers’ or
other caregivers’ propensity to “treat their infants as individuals with minds, rather
than merely as entities with needs that must be met”.29 Thus it is argued that mater-
nal use of mindminded language at five months predicts better performance on false
belief tests at five years.30
There’s also another more complex connection. Security of attachment in infants
as measured by the Strange Situation Test (SSn)31 predicts good mentalizing capac-
ities later.32 Why? One hypothesis notes a strong correlation between the mother’s
26 Wellman and Lagattuta (2001).

27 Carpendale and Lewis (2004, 79).
28 Hobson (2004, 109). There are also other interactional factors beyond parent–child interactions
e.g. the “sibling effect” (see Ruffman et al. (1998), but this doesn’t seem to depend simply on
number of siblings or even of older siblings (Carpendale and Lewis 2004, 79), so perhaps the
family environment for which the parents are responsible is what really does the work even here.
29 Carpendale and Lewis (2004, 92), citing Meins and Ferneyhough (1999, 332).
30 Meins et al. (2002, 1715–16).
31 See e.g. Ainsworth et al. (1978).
32 See Fonagy and Target (1997, 687): children’s SSn ratings at 12 months (with mother) and 18
months (with father) were compared with their performance on three “theory of mind” tests at five
years. Of those rated secure at 12 months 82% passed one of the tests (“the belief-desire reasoning
task”) as compared with 54% rated as insecure (77% secure at 18 months passed as compared
with 55% insecure). See also Meins, Ferneyhough et al. (1998, 1–24): 83% of securely attached
children passed a false belief task at four years compared with 33% insecure, and the ratio was
attachment classification during pregnancy (on the Adult Attachment Interview

(AAI)) and that same infant’s attachment classification to the mother (on the SSn)
at around 12 months.33 A possible explanation of the correlation exploits the fact
that scoring on the AAI, in which adults are asked about their childhood, is based on
narrative style – fluency, richness of vocabulary, gaps, hesitations, coherence, dis-
missiveness etc. – not on content, so those who are rated secure on that test will tend
to be good at talking about themselves and making sense of their experience without
either being overwhelmed by it or dismissing it – that is, they will be mindminded,
at least about themselves. Since maternal mindmindedness seems to be what’s mea-
sured by tests of maternal attachment security, it’s not surprising (it’s said) that
infant attachment security predicts the child’s mentalizing ability, since maternal
mindmindedness predicts it too – whatever in fact explains the link between the
mother’s attachment security and that of her infant.
On the other hand infant attachment security may independently favour the devel-
opment of mentalizing ability, for more than one possible reason.34 Mentalization
depends “on optimal prefrontal cortex functioning” which in turn depends on
optimal arousal (roughly, over-arousal drowns out the capacity for “flexible reflec-
tive” responses to others35 ), and secure infant attachment facilitates optimal levels
of arousal.36 Security of attachment also favours other interactive patterns (such
as mother-child conversations) which themselves favour the development of the
capacity to mentalize.37
Another connection between quality of infant-caregiver interaction and men-
talization exploits the idea of maternal attunement to the infant: poor attunement
impacts negatively on mentalizing ability later on. Thus depressed mothers are less
well-attuned to their infants, more likely to be “hostile and intrusive, withdrawn,
85:50 on a test at five years. Also cited by Carpendale and Lewis (2004, 92), who note that security
of attachment is also correlated with proto-declarative pointing which is an early indicator of social
understanding (Bretherton et al. 1979).
33 Fonagy, Steele, Steele (1991, 891–905).
34 Fonagy (2004, 106).
35 Fonagy (2004), citing e.g. Mayes (2000, 267–79).
36 Field (1985).
37 Fonagy and Target (1997, 688), citing e.g. Dunn (1996). The relationship between attachment
security and mentalizing capacity would seem to be complex, however, since it looks possible –
for example for children in care – to develop their mentalizing capacity thanks to the fact that
they have multiple caregivers with whom they interact positively, but to none of whom they are
attached (securely or otherwise). (Thanks to Peter Hobson for this point.) If that’s so, security of
attachment is a context that favours mentalization, but isn’t necessary for it. The putatively loose
link between attachment security and interactivity of the right kind does not on its own, however,
challenge the link between mentalization and love of the right kind, since that is itself a looser
notion than secure attachment: I’d be happy to say, with David Velleman (Velleman (2005)), that
teacher–pupil relations for example can exemplify Aristotelian philia.
188 E. Harcourt
or showing negative feelings”38 ; similarly mothers with borderline personality dis-

order (BPD), “were . . . less sensitive and more intrusive towards their infants” in
videotaped two-minute interactions.39 With regard to the later effects of maternal
depression, evidence is somewhat unclear: at 18 months children in the study did
comparatively badly on “searching for hidden objects” and (boys only) on more gen-
eral intelligence tests at five years.40 On BPD, Hobson and his colleagues did not do
a follow-up study at an age at which false-belief tests would have been appropriate,
but noted that in a follow-up at 12 months, children of BPD and non-BPD mothers
performed similarly on “non-social” tasks but children of BPD mothers tended to
do less well on social tasks, i.e. tasks which involved “engaging with another per-
son’s engagement with the world”.41 The capacity so to engage may in its turn be a
predictor of the development of mentalization later on.
Finally, mentalization is said to be fostered by a certain kind of caregiver’s “mir-
roring” response. This is consistent with the last point since depressed caregivers
are likely either not to react to their babies’ reactions at all, or to react in inappro-
priate and inconsistent ways – so not even the beginnings of “mirroring” here. But
there is more than one way of being a “mirror”. It might be said to be enough to
count as reflecting or mirroring a child’s mental state that the caregiver’s response
matches the infant’s expression of mental state and is appropriately prompt. But
some researchers in this area are also interested in another property of caregivers’
responses which they call “markedness”. This is explained as
the caregiver’s capacity to incorporate into her expression a clear indication that she is not
expressing her own feelings, but those of the baby.42
Unmarked responses simply duplicate the baby’s state, for example catching the
baby’s panic and panicking oneself. Marked responses, on the other hand, don’t
simply express a duplicate of the infant’s mental state: while they have something
in common with what a response would be which did simply duplicate the infant’s
state, they also somehow “indicat[e] that [the caregiver’s] display is not for real”.43
The caregiver “combines a ‘mirror’ with a display incompatible with the child’s
affect”, e.g. “smiling, questioning, mocking display”.44 It has been argued that
marked responsiveness favours the child’s capacity to mentalize. This would seem
to be so indirectly because eight-month old infants have been shown to calm down
38 Murray et al. (1996), cited by Hobson (2002, 136n). Incidentally BPD mothers typically score
insecure-disorganized on the AAI (Hobson 2002, 133), and mothers of securely attached infants
are more sensitive to their infants’ needs (Fonagy and Target 1997, 689, citing various authors), so
this is consistent with the claims about security and mentalization.
39 Hobson (2002, 133–34).
40 Hobson (2002, 136), citing Murray et al. (1996).
41 Hobson (2002, 135).
42 Bateman and Fonagy (2003, 193).
43 Fonagy et al. (2004, 9).
44 Fonagy and Target (1997, 684). This exaggerated not-for-real response to the infant’s distress
could be seen as an elaboration of “motherese” – see Fonagy et al. (2004, 177, n. 7).
more rapidly after an injection when their mothers’ responses are “marked”,45 link-
ing marked responsiveness to affect-regulation, which in turn fosters the capacity to
mentalize (see above). But it may also affect mentalization directly.46
There is thus some evidence that maternal mindmindedness, good maternal
attunement, infant security of attachment, and “marked” responsiveness all favour
the development of the capacity to mentalize. I leave security of attachment out of
the story for now because its relation to mentalization raises complex explanatory
issues of its own. What’s of interest is why the difference between mindminded-
ness and treating infants “merely as entities with needs that must be met”, between
attunement and failure of attunement, and between marked and merely “catching”
responsiveness are all philosophically significant differences. Here again is Hobson
on the BPD mother’s difficulty in attuning to her infant:
[The mother] maintained a monologue rather than a dialogue with her [two-month-old]
baby. She repeated her own thoughts about her infant’s state in a rather insistent way, often
cutting across what appeared to be the latter’s attempts to vocalize and play some more
active part in the interchange. What the mother actually said also seemed to reflect some of
her own preoccupations, rather than elaborate her baby’s current feelings and actions.47
The vocabulary here is strikingly close to Iris Murdoch on the way “fantasy” clouds
the ability to “see”. Again, Bateman and Fonagy explain the significance of marked
responsiveness because the capacity to mentalize
develops through a process of having experienced oneself in the mind of another during
childhood within an attachment context,48
which marked responsiveness favours because only the marked response makes the
infant’s state of mind available to it truly, and in a form it can digest. (Otherwise
the infant does not see its own state in the caregiver’s expression: what it sees is
simply that the caregiver herself is also in that state.) Compare Badhwar, whom I
quoted right at the start: “The look of love does more than see the loved individ-
ual veridically: it also shows the loved individual what it sees”.49 The differences
between attunement and its absence (and so on) don’t amount to differences in
quality of care just because the first apparently makes more easily available than
the second something that’s of independent value, namely self-knowledge. Though
they do not say as much, what Fonagy, Hobson and others seem to be talking about
is love. If it’s a piece of wisdom that love at its best involves the ability to disen-
tangle one’s own needs/desires/fantasies and to see the other truly – to allow one’s
thoughts about, and behaviour towards, the other to take their shape from how they
are – then the relatively low-level phenomena of mind-mindedness, attunement and
45 Fonagy and Target (1997, 684), citing Fonagy et al. (1995).

46 See Fonagy et al. (2004, esp. chaps. 4, 7); also Gergely and Watson (1999).
47 Hobson (2002, 128).
48 Bateman and Fonagy (2003, 191).
49 Of course there is another dimension of interactivity here too, namely the learner’s capacity to
understand the other’s knowledge of them as such, and which might be absent while the other’s
knowledge is present – but I cannot do justice to this extra loop here.
190 E. Harcourt
marked responsiveness all embody, at a micro-level, the difference between a virtue

or quasi-virtue of loving relations and one of its corresponding (quasi-)vices.
Let me put this conclusion together with the more abstract considerations about
self-knowledge I advanced earlier. One conclusion of those considerations was that
self-knowledge depends on other-knowledge, specifically on the knowledge others
have of us. Moreover if love at its best is centrally manifested in the capacity to
see another disinterestedly and therefore truly, then those best placed to know us
will be those who love us best. If self-knowledge depends on others’ knowledge
of us and others’ knowledge of us is implied by their love for us then, empirical
considerations aside, it should follow that we will find self-knowledge where we
find others’ love of us. The child development material I have been reviewing points
towards the conclusion that the capacity for self-knowledge varies with the qual-
ity of early interactions – perhaps especially early child-caregiver interactions. This
independent connection should strengthen our conviction that the philosophical con-
siderations I’ve advanced are correct, while these in their turn should encourage us –
for all the gaps that need to be filled in – to interpret the empirical material boldly,
to the effect that self-knowledge owes a debt to “the thing called love”.
Acknowledgements Thanks for valuable comments on this paper and on surrounding issues
to Anita Avramides, Dorit Bar-On, Charlie Lewis, Peter Hobson, Angelika Krebs and Roger
Teichmann, and to audiences at the Oxford Department of Continuing Education, the University
of Hertfordshire, the Universidade Nova de Lisboa, the University of Southampton, and at the
Self-Evaluation: Individual and Collective workshop, University of Basel.
References
Ainsworth, M. et al. 1978. Patterns of Attachment. Hillsdale, NJ; New York; London: Erlbaum.
Anscombe, G. E. M. 1957. Intention. Oxford: Blackwell.
Avramides, A. 2001. Other Minds. London; New York: Routledge.
Badhwar, N. 2003. “Love.” In The Oxford Handbook of Practical Ethics, edited by H. La Follette,
42–69. Oxford: Oxford University Press.
Bar-On, D. 2004. Speaking My Mind. Oxford: Oxford University Press.
Bateman, A., and P. Fonagy. 2003. “The Development of an Attachment-Based Treatment for
Borderline Personality Disorder.” Bulletin of the Menninger Clinic 67 (3):193.
Bretherton, I. et al. 1979. “Relations Between Cognition, Communication, and Quality of
Attachment.” In The Emergence of Symbols, edited by E. Bates, 223–69. New York; London:
Academic Press.
Carpendale, J., and C. Lewis. 2004. “Constructing an Understanding of Mind: The Development
of Children’s Social Understanding Within Social Interaction.” Behavioral and Brain Sciences
27:79–151.
Dunn, J. 1996. “Children’s Relationships.” Journal of Child Psychology and Psychiatry 37:507–18.
Field, T. 1985. “Attachment as Psychobiological Attunement.” In The Psychobiology of Attachment
and Separation, edited by M. Reite and T. Field, Orlando, FL: Academic Press.
Fonagy, P. 2004. “The Roots of Social Understanding in the Attachment Relationship.” Behavioral
and Brain Sciences 27 (1):105–6.
Fonagy, P., and M. Target. 1997. “Attachment and Reflective Function.” Development and
Psychopathology 9:679–700.
Fonagy, P. et al. 1995. “Attachment, the Reflective Self, and Borderline States.” In Attachment
Theory, edited by S. Goldberg, R. Muir, and J. Kerr. New York: Analytic Press.
Fonagy, P., G. Gergely et al. 2004. Affect-Regulation, Mentalization, and the Development of the
Self. London: Karnac.
Fonagy, P., H. Steele, and M. Steele. 1991. “Maternal Representations of Attachment During
Pregnancy Predict the Organization of Infant-Mother Attachment at one Year of Age.” Child
Development 62:891–905.
Gergely, G., and J. Watson. 1999. “Early Social-Emotional Development. Contingency Perception
and the Social-Biofeedback Model.” In Early Social Cognition, edited by P. Rochat, 101–36.
Hillsdale NJ: Erlbaum.
Hacker, P. M. S. 1986. Insight and Illusion. 2nd ed., 293–94, Oxford: Oxford University Press.
Hobson, P. 2002. The Cradle of Thought. Basingstoke: Macmillan.
Hobson, P. 2004. “Understanding Self and Other.” Behavioral and Brain Sciences 27 (1):109–10.
Hobson, P. 2009. “Wittgenstein and the Developmental Psychopathology of Autism.” New Ideas
in Psychology 27:243–57.
Jäger, Ch. 1999. Selbstreferenz und Selbstbewusstsein. Paderborn: Mentis.
Mayes, L. C. 2000. “A Developmental Perspective on the Regulation of Arousal States.” Seminars
in Perinatology 24:267–79.
Meins, E. et al. 2002. “Maternal Mind-Mindedness and Attachment Security as Predictors of
Theory of Mind Understanding.” Child Development 73:1715–16.
Meins, E., C. Ferneyhough et al. 1998. “Security of Attachment as a Predictor of Symbolic and
Mentalizing Abilities.” Social Development 7:1–24.
Meins, E., and C. Ferneyhough. 1999. “Linguistic Acquisitional Style and Mentalising
Development: the Role of Maternal Mind-Mindedness.” Cognitive Development 14 (3):363–80.
Murdoch, I. 1970. The Sovereignty of Good. London: Routledge & Kegan Paul.
Murray, L. et al. 1996. “The Impact of Postnatal Depression and Associated Adversity on Early
Mother-Infant Interactions and Later Infant Outcome.” Child Development 67 (5):2512–26.
Ruffman, T. et al. 1998. “Older (but not Younger) Siblings Facilitate False Belief Understanding.”
Developmental Psychology 34 (1):161–74.
Sigman, M. et al. 1992. “Responses to the Negative Emotions of Others by Autistic, Mentally
Retarded, and Normal Children.” Child Development 63 (4):796–807.
Strawson, P. F. 1959. Individuals. London: Methuen.
Tappolet, C. 2000. Emotions et Valeurs. Paris: PUF.
Velleman, J. D. 2005. “Love as a Moral Emotion.” In Self to Self, edited by J. D. Velleman.
Wellman, H., and K. Lagattuta. 2001. “Developing Understandings of Mind.” In Understanding
Other Minds, edited by S. Baron-Cohen, H. Tager-Flusberg, and D. Cohen, 2nd ed. Oxford:
Winnicott, D. W. 1964. “A Man Looks at Motherhood.” In The Child, the Family, and the Outside
World, edited by D. W. Winnicott. Harmondsworth: Penguin.
Wittgenstein, L. 1953. Philosophical Investigations. Translated by G. E. M. Anscombe. Oxford:
Blackwell.
Wittgenstein, L. 1967. Zettel, edited by G. E. M. Anscombe and G. H. von Wright, Translated by
G. E. M. Anscombe. Oxford: Blackwell.
Wittgenstein, L. 1968. “Notes for lectures on “Private Experience” and “Sense Data”.”
Philosophical Review 77 (3):275–320.
Wittgenstein, L. 1980. Remarks on the Philosophy of Psychology, vol. 2, edited by G. H. von
Wright and H. Nyman, Translated by C. Luckhardt and M.A.E. Aue. Oxford: Blackwell.
Wittgenstein, L. 1992. Last Writings on the Philosophy of Psychology, vol. 2, edited by G. H. von
Wright and H. Nyman, Translated by C. Luckhardt and M.A.E. Aue. Oxford: Blackwell.
Zahn-Waxler, C. et al. 1992. “The Development of Empathy in Twins.” Developmental Psychology
28 (6):1038–47.
Chapter 11
Is Shame a Social Emotion?
Julien Deonna and Fabrice Teroni
Shame is a psychological state of special relevance for those interested in the nature
of the self-evaluative attitudes. It is particularly so in the context of this volume:
while shame is experienced by individual people, it is only as members of col-
lectives, many have argued, that individuals may be susceptible to such negative
self-evaluations. Indeed, classical and recent philosophical and psychological treat-
ments of shame regard it as an essentially social emotion. This is how advocates of
the social nature of shame think that it differs from other negative emotions directed
at the self such as, for example, guilt or self-disappointment. This claim is also the
foundation for important conclusions about shame and its role. First, the influen-
tial distinction between “shame cultures” and “guilt cultures” (e.g. Creighton 1990)
is based on it and, second, most discussions of the links of shame to morality are
directly concerned with it. For some, the social character of shame disqualifies it
from playing a significant ethical role; for others, this same feature is what is ethi-
cally important about it.1 For these reasons, one of the central aims of an account of
shame is to assess the claim that it is a social emotion. In this article, we will argue,
first, that this claim is potentially confusing since different understandings of “being
a social emotion” are possible and, second, that shame is not more “social” than
many other emotions. Although therefore modest, the aims of the coming discus-
sion are we believe crucial for establishing the foundations of a full-fledged theory
of shame.
1 Good representatives of the first approach are Lamb (1983) and Tangney (1991). Good represen-
tatives of the second are Calhoun (2004), Williams (1993, chap. 4) and Wollheim (1999). Although
we shall not address these issues at all in this paper, relevant discussions of the relations between
shame and morality are taken up in Deonna and Teroni (2008) and Bruun and Teroni (2011).
J. Deonna (B)
Department of Philosophy, University of Geneva, Geneva, Switzerland
e-mail: Julien.Deonna@unige.ch
F. Teroni
Department of Philosophy, University of Berne, Berne, Switzerland; Center for Affective
Sciences, Geneva, Switzerland
e-mail: Fabrice.Teroni@philo.unibe.ch
194 J. Deonna and F. Teroni
11.1
What are the different possible understandings of the claim that an emotion is social?
This question is best addressed by noting that emotions have several aspects, reflect-
ing their complexity and intimate links with other mental states. Emphasis can be
put on the particular objects of the emotions, on their specific genesis, on the types
and properties of their cognitive antecedents, as well as on the typical contexts or
situations that elicit them. If we now address our main question in the light of these
different aspects (see Table 11.1), distinct issues emerge.
Table 11.1 Questions regarding the social nature of shame
Object Is the particular object of shame always social?

Genesis Does society play a fundamental role in the acquisition of the values relevant for
shame?
Cognitive Do the reasons for which we feel shame have fundamental links with society?
antecedent
Context Does shame always take place in a social context?
Let us first clarify our use of the terms “object” and “cognitive antecedents”. In
shame, it is often claimed, a self is evaluated as degraded, unworthy or lacking in
value. Now, this self-reflexive evaluation should be distinguished from the reasons
why a subject makes it: she thinks she failed with respect to something, or that others
believe she has so failed. That is, in shame, the negative evaluation of the particular
object (the self) is based on such cognitive antecedents or reasons. In what follows,
“object” ranges only over selves and not, as it is sometimes used, over the reasons
for which one is ashamed. If John is ashamed of being mean-spirited, his awareness
of his mean-spiritedness (cognitive antecedent) grounds an apprehension of himself
(object) as unworthy.2
Having clarified this distinction, we can now note the following. If we were to
give a positive answer to any of the above questions, we would conceive shame
as a social emotion. So, let us now look at some of these answers and start by
discarding those that are not good candidates for supporting the conclusion that
shame is essentially social.
Could we make sense of a social thesis focussing on the object dimension of
shame? Well, not really, as it would consist in claiming that we are always ashamed
for someone else. Although the phenomenon indeed exists – Sam is ashamed for a
candidate in a TV show making a fool of himself – it is obviously not what anyone
2 Taking our lead from the theories we discuss, the coming discussion will refer to “judgements” in
connection with the evaluations subjects make in shame. This should not be read as implying that
a satisfactory account of shame should ultimately appeal to judgements rather than to conceptually
less demanding states such as is now customary in emotion theory. See for instance Goldie (2000,
chap. 3), Roberts (2003) and Tappolet (2000).
11 Is Shame a Social Emotion? 195
has in mind who defends the claim that shame is essentially social: shame is more
often than not directed toward ourselves.
Consider now what a social claim focussing on the genetic dimension of shame
would amount to. The idea that springs to mind is that we learn to feel shame in
virtue of our belonging to specific communities and that the other members of these
communities play an essential role in our acquiring the values explaining why we
feel shame. Now, while true, this genetic claim cannot be at the heart of what is
meant when it is said that shame is distinctively social. For this claim of course
holds true of many emotions nobody regards as particularly social. What angers me
or makes me proud might very well be anchored in values I have acquired at the
contact of others, but this has never been offered as a reason to conclude that anger
or pride are essentially social.3
More interesting and contentious claims arise from positive answers to the
remaining two questions. As regards the cognitive antecedents of shame, it is neces-
sary to distinguish three distinct social theses. The first thesis has it that the reasons
that lead us to apprehend ourselves as degraded consist in taking a perspective “from
the outside” on what we do or who we are. According to the second thesis, these
reasons are judgements regarding failures that the subject feeling shame does not
think he has, or does not conceive as failings. Shame is then social because it is felt
independently of our agreeing with the negative judgement that is passed upon us –
something often referred to as the heteronomy of shame. The third and last thesis is
that shame essentially relates to our “social selves”, i.e. those properties related to
the standing we have in the eyes of others.
Two substantial social claims also arise from positive answers to the question
about the contexts in which shame is elicited. Shame may be thought to occur only in
the presence of real-life audiences, or, perhaps more plausibly, also in the presence
of merely imagined audiences.
In this paper, we will argue that while all these theses are true of some impor-
tant families of shame episodes, none of them should be used as a springboard for
defending a conception of shame as an essentially social emotion. We will consider
all these claims in turn, explaining in the process their connections with one another
as well as the constraints on a theory of shame they help uncover. Finally, we show
how a non-social picture of shame is not only capable of meeting these constraints,
but has the further virtue of shedding light on those situations in which others seem
to play no role at all in why we feel shame. Table 11.2 presents the five social theses
about shame on which we shall focus.
3 This is not to say that an investigation into the genesis of shame is philosophically irrelevant, but
only that we cannot derive social theses about it from genetic considerations. A psychoanalytically-
informed philosophical treatment of the genesis of shame can be found in Nussbaum (2004,
chap. 4). For a developmental perspective, see e.g. Ferguson, Stegge, and Damhuis (1991) and
Olthof et al. (2004).
Table 11.2 Possible social theses about shame
Social theses about shame’s Thesis (1): shame requires taking an outside perspective on what
cognitive antecedents we do or who we are
Thesis (2): whether we agree or not with the negative judgement
that is passed on us is irrelevant to the self-evaluation present in
shame
Thesis (3): shame is concerned with our standing vis-à-vis others
Social theses about shame’s Thesis (4): shame requires a real-life audience
context Thesis (5): shame requires at least an imagined audience
11.2
Let us start with the least plausible claim, thesis (4), according to which shame
requires the context of a real audience. Its main support comes perhaps from the fact
that, when we start thinking about shame, situations in which we are in front of an
audience immediately spring to mind. For example, one is seen naked in front of
one’s partner’s parents, or more generally in an awkward situation by people one
neither expected, nor wanted, to be seen by. Other potential evidence for this thesis
may come from the study of children’s or foreign cultures’ shame. It is nevertheless
all too easy to find episodes of shame in which no one is actually present. Indeed,
surprisingly for classical conceptions of the differences between shame and guilt,
one important empirical study shows that the presence of an audience features less
often in people’s descriptions of their shame experiences than in their descriptions
of their guilt experiences.4 So, thesis (4) might hold true perhaps of some central
cases, but cannot be generalized to all instances of shame.
A slight modification of this thesis leads to a much more plausible claim accord-
ing to which shame always occurs in the context of a real or imagined audience
(thesis (5)). In her classical study, for instance, Ruth Benedict conceives Japan as
a “shame culture” for the reason that it assigns the uttermost importance to exter-
nal sanctions, shame being “a reaction to other people’s criticisms” (Benedict 1946,
222ff.). But she adds that imagining being criticized or ridiculed by an audience
is enough. Imagining someone, say our beloved or our superior, disapproving us is
required in shame. This straightforward understanding of thesis (5), however, proves
problematic and invites the following considerations.5
While shame episodes often involve imagining others criticizing us, this does not
appear to be required. Defending such a requirement entails defending an implausi-
ble phenomenological claim about shame, given that the act of imagining something
is subject to the will and open to introspection. In shame, we should be expected thus
4 According to Tangney et al.’s studies (1996), subjects claim that more of their shame than of their
guilt episodes are experienced privately. The figures are 18.2% of shame experiences and 10.4%
of guilt experiences (pp. 1259–260).
5 These problems and considerations also bear for example on the closely related accounts of
Williams (1993) and Elster (1999).
not only to be aware of an imagined audience, but to be also aware of who this audi-
ence is. And this means being able to answer the question “Who is criticizing me?”.
But we are often unaware of an audience and unable of answering this question,
which suggests that there is something amiss with a straightforward reading of the-
sis (5). Often-heard claims about a phenomenology of “the look”, “the eyes” or “the
gaze” just appear to be an overstatement of what sometimes happens in shame, or
the result of an exclusive focus on some specific cases.
In the light of these observations, it proves more appealing to allocate a differ-
ent role to such imagined audiences. Some may pursue the idea that shame often
involves narratives in which imagined audiences play a critical role without being
part and parcel of all shame episodes.6 Alternatively, and this in our opinion more
adequately locates the role of imagination in shame, we may say that we engage
in imaginative projects featuring an audience in order to explore and assess which
emotional reactions fit which situations and/or would be deemed adequate by a given
audience. One can for instance imagine oneself doing in front of one’s mother what
one presently does in her absence and, as a result, feel shame. Or one can, after
feeling shame, evaluate it as inadequate as a result of imagining one’s lover being
amused by the apparent misdeed. But these imaginative projects are not necessary:
the matter at hand may be too evident for such heuristics to be needed.7
If this were to prove insufficiently convincing as an objection against thesis (5),
note that it also faces the following difficulty. We have just seen that real or imag-
ined audiences are not necessary for shame, but we have also good reasons to think
that they are not sufficient for it. Why, we may indeed ask, think that being aware
that a real or imagined audience judges us as unworthy or contemptible would result
in shame? If we agree with these judgements, shame may indeed ensue. But what
of all these cases where we disagree with these judgements? As Wollheim has con-
vincingly shown (1999, 167ff.), such a disagreement would typically predict other
affective reactions. Indeed, if someone looks down on you for an alleged failing
you do not think you have, anger or indignation are much more likely responses
than shame. Imagining an audience cannot explain why these more likely responses
would be supplanted by shame.
The upshot of our discussion of the social claim about the context of shame is
then the following. Audiences, real or imagined, are neither necessary nor sufficient
for the occurrence of shame. This is true unless our objections were based on a
too superficial conception of what imagined audiences in fact consist in. There may
be something in shame that requires that we understand these imagined audiences
along different lines. Only in this way, some have argued, can we account for the
heteronomy of shame as formulated in thesis (2). This subtle interaction between
different social theses about shame will be the topic of the next section. Before we
6 On the role of imagined narratives in emotions, see Goldie (2000).

7 Arguing for any stronger requirement concerning audiences in shame may also come from a
conflation between shame and shaming. While it is true that we often feel shame as a result of
shaming, this is surely not always the case (see Smith et al. 2002 and Deonna, Rodogno, and
Teroni, 2011).
turn to that, let us stress that the present discussion still motivates the following
constraint on a theory of shame:
Constraint (i): Although insufficient to explain shame, audiences are often relevant deter-
minants of shame.
11.3
Richard Wollheim is probably the staunchest defender of the thesis according to
which shame involves an imagined criticizing agent, but not one conceived along
the lines of the straightforward reading of thesis (5) we have presented and criti-
cized above. This, Wollheim argues, follows from the fact that shame essentially
involves the subject’s submission to pressures exerted upon him. This strategy con-
sists in stressing the fact that agreement with the negative judgement passed on us
is irrelevant to explain the occurrence of shame (thesis (2)).
Wollheim starts his discussion by observing that shame involves the experience
of an external agent assaulting oneself, an assault felt as an “alien force”. This looks
like a phenomenological version of thesis (2). More often than not, we disagree
with the negative assessment formulated against us and experience it as unjust crit-
icism.8 Accounting for this phenomenological fact is the task that Wollheim sets
himself. But as we have seen, he concurs with our previous conclusion that real or
imagined criticizing audiences are neither necessary nor sufficient for shame. These
audiences, Wollheim correctly diagnoses (1999, 167–70), may be seen at best as
triggers of or aids to the onset of the shame experience that might have occurred
unaided. For this reason, they cannot account for the sense of an alien force assault-
ing the self which Wollheim claims to be essential to shame. Who is then assaulting
us in shame? Which understanding of audiences is apt to circumvent the difficulties
we raised regarding the supposedly social context of shame?
Wollheim’s strategy is to appeal to phantasised figures. These phantasised oth-
ers need neither stand in for real-life agents, nor should they be viewed as the
subject taking a distinct perspective on himself. They should not be so external
as to become, like real-life or merely imagined audiences, accessory or superflu-
ous, but at the same time they should not of course merely mirror the subject’s
own self-reflexive attitudes. They should have enough authority to induce shame
and thus account for the submission characteristic of this emotion. This is how, in
Wollheim’s psychoanalytic framework, the criticising other is not merely imagined,
but is a figure the subject has gradually internalised through a complex multi-stage
mechanism. The radical otherness which is, for Wollheim, at the centre of shame is
explained by the fact that its origin is traceable to “whole internalised figures” phan-
tasised through mechanisms the subject has no control over. Hence, these figures
8 For Wollheim, all primitive instances of shame can be so characterized, even if, as we grow older,
the force of these assaults may wane.
“continue to address the person who now harbours them as an alien force” (1999,
179). Their special authority derives from the history of their formation, i.e. the ways
in which various properties are ascribed to them within the phantasies in which they
play a central role. Now, while the finer points of this process do not have to detain
us here, it becomes understandable how, so characterized, the disapproval of these
internal figures may well not induce reactions of anger or indignation. For, Wollheim
argues, these figures have, so to speak, no face. They are “pure insistence” and no
indication as regards the criteria according to which their judgement is made is ever
available to the subject. As a result, the negative emotion felt cannot be directed
towards these elusive figures and is bound to be directed at the self.
Wollheim, then, defends a specific version of thesis (5) according to which shame
requires the context of an audience. Defending this social thesis is, according to him,
the only way to secure the truth of a social thesis about the cognitive antecedents
of shame, i.e. the thesis that shame is the emotion of submission or, as he describes
it, the emotion of radical heteronomy (thesis (2)). The thought is that phantasized
others are required to explain why we submit in shame. Thus, everything now rests
on the question of whether shame is indeed the heteronomous emotion it is claimed
to be. Where does this common and plausible idea that shame is heteronomous come
from?
Well, it is indeed a remarkable fact that some instances of shame occur even
though we disagree with the criticisms triggering it. One may for instance feel shame
at one’s accent being ridiculed although one does not believe there is anything wrong
with it and might even be inclined to cultivate it. But is there any reason to use these
specific cases as our model for understanding all shame episodes? In particular, is
it really the case – as Wollheim seems to think – that when we happen to agree
with the relevant criticism, this agreement plays no role in explaining why we feel
shame? Why say, for example, that it is only the charge of cowardice I am subjected
to, and not my conviction that I behaved cowardly, which explains at least in part my
shame? As far as we are aware, Wollheim only offers two reasons for this strong and
counterintuitive conclusion. The first consists in his appeal to the phenomenology of
shame as assault, and the second in his claim that adult shame should be understood
on the model of early shame episodes that are radically heteronomous (1999, 151–
52). These arguments might be appealing within the psychoanalytical framework
in which they are couched, but one might question whether this phenomenological
intuition remains intact when the subject agrees with the criticism that is directed at
him. It is all the more questionable when we realize that alternative and less theo-
retically laden explanations of shame without agreement are available (see Section
11.6). In conclusion, the second constraint we want to hold on to is:
Constraint (ii): a theory of shame should accommodate the fact that we often disagree with
the judgement that triggers shame.9
9 In a similar vein, Deigh claims that “we must admit cases of shame felt in response to another’s
criticism and ridicule in which the subjects do not accept the other person’s judgement of them and
so do not make the same judgement of themselves.” (1983, 233)
11.4
Disagreement with the relevant criticism is, we just argued, not likely to be a social
aspect common to all instances of shame. An alternative to the explanation of the
essentially social nature of shame might thus be thought to lie not in the attitude we
take with regard to the criticism that is directed upon us, but in the specific subject
matter of the judgements that elicit this emotion.
One interesting idea here consists in claiming that the reasons motivating shame
concern our social selves: we are always ashamed because of a perceived threat
to our standing vis-à-vis others (thesis (3)). Sam, say, is a wealthy individual liv-
ing in an underprivileged neighbourhood. Were he to feel shame about his wealth,
it would be because he perceives it as a threat to the way he wants to be appre-
hended by others. John Deigh may have such examples in mind when he criticizes
Rawls’ conception of shame on the ground that “a satisfactory account [of shame]
must include in a central role one’s concern for the opinions of others”, such a role
being “internally related to shame.” (1983, 238) Shame, then, is “a reaction to a
threat, specifically, the threat of demeaning treatment one would invite in giving the
appearance of someone of lesser worth.” (1983, 242) Note how this claim differs
from those we have already discussed. It neither implies real or imagined audi-
ences, nor trades on the fact that the subject submits to other people’s opinions. On
the contrary, shame occurs because the subject agrees that the way he is or behaves
constitutes a threat. But how should we understand the idea that shame is social
because it is a reaction to such a threat?
The first possible interpretation of Deigh’s claims appeals to the idea that shame
is elicited when, and only when, our social standing is threatened. On this interpreta-
tion, shame is social in the sense that we perceive the image we project onto others
as under threat. This idea points towards an important family of shame episodes.
Indeed, we often feel shame when made aware that the way we are or behave is
such that a negative image of ourselves is likely to be projected.
While these cases are indeed common and of great interest, there is once again
no reason to think that we should model our theory of shame on them. For even
though one’s concern for others’ opinions is no doubt essential to a proper analysis
of these cases, a concern of this nature is not always the relevant determinant of
shame. Is it really impossible to envisage cases where shame is felt as a result of
a failure to uphold a privately held ideal? Aren’t such cases even common at least
for some of us? Can’t I be ashamed for not having seen my children in the past two
weeks independently of what I believe they, my wife or anybody else for that matter
will think of me? Or if we are not convinced by this example, we may consider the
following case. A writer may happen to live in a society that despises intellectual
achievements. He may well still desire to write a good novel and, when perceiving
his inability, be ashamed of what he has written. Such an episode of shame has
nothing to do with the way he defines himself socially, for he can be well aware of
his actual social context.
This supports the following two conclusions. Firstly, although threats to our
social standing are often the reasons for which we feel shame, this fact cannot jus-
tify the claim that shame always connects with our social standing. Secondly, while
shame may sometimes have to do with the appearance we project onto others, it is
also sometimes felt in direct connection with our realizing that what we are or do
presently violates our ideals.
The discussion of an alternative interpretation of Deigh’s claims will confirm
the conclusion we just reached. It consists in emphasizing the social source of the
demeaning treatment as opposed to the social nature of the reasons for which one
would be treated that way. Shame is social because the demeaning treatment is
represented as coming from others: in shame, one perceives others as potentially
menacing in virtue of who one is or what one does.
Can we substantiate the thesis according to which a perceived threat from others
always features among the cognitive antecedents of shame? As a first step towards
an answer, the following distinction will prove helpful. There are situations we
deem shameful for the reason that they are seen, and situations we deem shameful
independently of their being seen (see e.g. Wollheim 1999, 158–59). Thus, being
naked in one’s living room can cause shame if one is unexpectedly seen, whereas
morally repugnant behaviour often causes shame independently of anyone witness-
ing it. Moreover, it seems to us that this distinction is often connected to a distinction
between what we can call “superficial” and “deep” shame, since the impact of shame
about what is shameful independently of its being witnessed often happens to be
more dramatic.
Now, the idea that shame is a reaction of fear of demeaning treatment by others
is particularly attractive to account for what we have just called superficial shame.
But for this very same reason, the idea, when applied across the board, flouts the
distinction between superficial and deep shame. First, as we have already argued,
shame has not always to do with the management of our social image, since many
cases of shame, and certainly what we have just called deep shame, prove difficult
to account for in terms of fear of social mistreatment. When one thinks that one
has done something that is shameful independently of anyone witnessing it, shame
is not or at least does not have to be a function of the perceived degree of social
threat one invites. Any situation, like that of the above writer, where someone feels
shame as a result of values not shared by members of his community provides an
example where it is not a function of the perceived threat of social mistreatment,
even though the thought of such a treatment might heighten one’s shame. The crucial
point here is that what motivates our considering an instance of shame as deep is
precisely our belief that shame would be felt independently of any thought of others
witnessing it, thinking ill of us or treating us badly as a result. That is, shame’s
depth (as well as the depth of the attachment to the values it discloses) is a function
of the following counterfactual: it reveals deeply held values if it would have been
felt independently of one’s perceiving it as a reason for mistreatment by others. For
example, to say of someone that he is shameless is never to think of him as simply
insensitive to social threats. Any claim to the contrary implies, quite implausibly,
that shame is wholly instrumental, only serving the function of alerting us to those
threats.
Taken as a general claim about shame, our discussion shows that thesis (3) is
on the whole unsupported. The grain of truth it does contain is that shame is often
linked with threats to our social image or standing.10 Thus, the third constraint is
the following:
Constraint (iii): a theory of shame should adequately deal with the fact that what others do
and think is often but not always of paramount importance in shame.
11.5
According to the final social thesis we shall discuss, the route that leads us to appre-
hend ourselves in negative terms in shame involves a change of perspective that
is often glossed in spatial terms (thesis (1)). In shame, we allegedly consider our-
selves or our activities “from the outside”. What does this spatial metaphor exactly
mean? We start by introducing the relevant phenomenon in terms of a change of per-
spective. We next criticize two social interpretations of this phenomenon and also
explain why it is not essential to shame. We conclude by giving reasons to think
that the change of perspective that is typical of shame should not be confused with
a characteristic “shift of focus” that is essential to this emotion.
The phenomenon that typically motivates thesis (1) is best conveyed by way of
an example.
Sam has been writing for three hours now and it has gone quite smoothly. Transitions come
naturally, examples come easily to mind, and paragraphs pile up on one another. The clock
rings four o’clock, time for Sam’s afternoon walk; “well deserved” he thinks. An hour later,
he is back at his desk and starts to read what he has been writing that day. The beginning
is fine, if a bit quick. As he reads along, he spots a few weak links, one premise not really
argued for. Nothing he cannot remedy though. And suddenly it dawns on him: the conclu-
sion now seems very much to say the same thing as the premise that was not argued for. He
thinks a bit more, re-reads some paragraphs, and now the entire argument appears circular.
He reads some more, all the examples look contrived and badly written, the assertive tone
unwarranted, the boldness pure defensiveness. Sam feels so ashamed.
In examples of this kind, shame is provoked by a “change of perspective”.

Suddenly – or more gradually but in some sort of prise de conscience – some
aspect of the situation in regard to ourselves appears in a new and unfavourable
light. What was intent activity, idle play, or not reflected upon is now apprehended
in a dark and menacing colour. Philosophers who, on the surface at least, do not
subscribe to any of the social claims we have discussed so far have argued that a
10 While David Velleman also speaks of threats to one’s image, his account differs markedly from
Deigh’s in that Velleman (2001) explicitly allows for the possibility of private shame. However, we
have argued elsewhere that his theory suffers from difficulties related to the failure to distinguish
between deep and superficial shame of the kind discussed in this section. See Deonna and Teroni
(2009).
social reading of these examples is needed to capture this aspect of shame. Assessing
whether this phenomenon should be conceived in social terms is thus important. We
will first discuss Taylor’s other-perspective reading of thesis (1) and next O’Hear’s
objective-perspective reading.
Gabriele Taylor puts forward a social reading of thesis (1). She relies on a general
distinction between “explanatory judgements”, which correspond to what we call
cognitive antecedents, and “identificatory judgements”, our object cognitions (see
Section 11.1). In shame, Taylor believes that the latter is a negative judgement made
by the subject about herself, whereas the former is a judgement about herself that
might be neutral, hostile or even friendly. So, a subject who undergoes shame typi-
cally (i) apprehends herself in a negative light as a result of perceiving a discrepancy
between (ii) a certain hoped for conception of herself and (iii) a given judgement.
Moreover, Taylor is clear about her disagreeing with social theses about the context
in which shame is elicited: according to her, this emotion neither requires a real, nor
an imagined audience (1985, 66).
But Taylor also thinks that something is correct in the venerable metaphor of an
audience. It is by way of locating this grain of truth that she writes that “feeling
shame is connected with the thought that eyes are upon one” (1985, 53), and that
shame “centrally relies on the concept of another, for the thought of being seen
as one might be seen by another is the catalyst [i.e., the explanatory judgement]
for the emotion.” (1985, 67) These claims are meant to explain the phenomenon
conveyed by the above example, which she glosses by underlining the change from
the state the subject is in just before feeling shame, and the “revelation” that what he
is doing “may be seen under some description” (1985, 66). According to Taylor, for
this change of perspective to occur, one of the cognitive antecedents of shame must
be a judgement with a content centrally relying on the concept of another.11 We
judge that our behaviour or trait would look in a given way to an external observer.
This is surely too articulate a judgement, but for Taylor the onset of shame always
traces back to something of the sort. Clearly, her thesis is not that a real or imagined
audience is required; what is required is only a judgement about how one would
look to an audience. Is this correct?
Taylor’s conclusion relies on the supposition that expressions like “observer-
description”, “object of observation” or “detached description” can be clarified
through reference to the concept of another. But we should not grant that much,
since, taken at face value, the former expressions do not imply the use of this con-
cept. For the above example and those Taylor uses to convey the spirit of her thesis
imply no more than a change from a “doer-perspective” to a “reflective-perspective”.
11 She writes that “this [identificatory] judgement is brought about by the realization of how her
position is or may be seen from an observer’s point of view. But there is no reference to such a point
of view in her final self-directed judgement. It is because the agent thinks of herself in a certain
relation to the audience that she now thinks herself degraded, but she does not think of this degra-
dation as depending on an audience.” (1985, 68) Taylor is thus sensitive to the distinction between
what is shameful in itself and what is shameful when observed we emphasised in Section 11.4.
Let us come back to Sam the writer to clarify what we mean. The expression “doer-
perspective” simply refers to perspectives like that occupied by Sam when writing.
He is immersed in the task of writing. He takes a “reflective-perspective” when, back
at his desk, he makes judgements about what he wrote. He is not writing anymore,
is not immersed in this task, but disengages himself from what he wrote and takes
an evaluative stance. Such examples convey two important lessons. First, that such
a change of perspective often occurs in shame, since to be ashamed of what one is
wholly immersed in seems impossible. Second, that this change is, on the face of
it, merely a change in the perspective one happens to take upon what one does or
did. In these cases, one moves from the role of unreflective doer to that of evalua-
tor. So, the claim that this change of perspective essentially relies on the use of the
concept of another is a further and questionable step. This claim rests on the thesis
according to which taking a reflective stance upon one’s traits or one’s behaviour
consists in vicariously adopting the vantage point of someone else (although, it is
true, of no one in particular). One might well prefer one’s analysis of shame not to
be committed to such a strong thesis.12
So, although an analysis of shame must respect the specific phenomenology aris-
ing from such changes of perspective (and Taylor refers to nothing more with her
use of expressions such as “sudden realization”, “revelations” or “(re)discoveries”),
there is no need to translate this as meaning that the subject takes another’s perspec-
tive. For, while this “sudden realization” may be due to her taking this perspective,
it may also be due to her simply adopting a reflexive stance. And, in this last case,
no thought about a third party need be implied. Taylor’s reading of thesis (1) is thus
too strong.
These may well be the reasons which lead Anthony O’Hear (1977) to offer a
different reading of this thesis. For him, shame starts and finishes when the subject
takes an unfavourable view of himself on a dimension that matters to him. The role
of the audience is therefore explicitly framed as ancillary (1977, 77). Moreover,
against the thesis that shame involves one’s submission to the opinion of others
(thesis (2)), he argues that “being publicly shamed produces feelings of shame, as
opposed to feelings of embarrassment, resentment, rebellion or anger, only when
one feels that the public disapproval is in some way in harmony with one’s own
feelings of how one would like to see oneself.” (1977, 77, our emphasis) This is why
O’Hear puts emphasis on examples of shame triggered by the subject’s own neg-
ative judgements about himself. This does not prevent him, though, from insisting
on a social aspect of shame he believes to be often overlooked. The idea here is
that the route that leads one to a negative assessment of oneself relies on objective
judgements about oneself. “If a judgement is objective” he writes, “it is in principle
acceptable to anyone. As one’s own judge, one is, if one is reasonable about it, in
principle judging oneself as any reasonable person would.” (1977, 78–79).
12 Thisis of course compatible with the thesis that reflexivity depends on the possession of the
concept of another. What we claim is that this concept, however acquired, need not be deployed in
shame.
How should we understand this reference to objectivity? Perhaps O’Hear’s point

is a purely logical one: the relevant judgement can in principle be made by anybody.
Alternatively, the idea is rather that shame requires the subject to apprehend the
negative judgement as one that could be made by anyone. At first sight, it may seem
that only the second idea offers a social reading of thesis (1), but this is clearly not
what O’Hear has in mind. For, as his own examples convincingly show, adverse
judgements we pass on ourselves can trigger shame independently of any thought
that others might also make them. This direct route to a social interpretation of
thesis (1) is thus blocked. This then means that we are back to the purely logical
point that shame involves the making of an objective judgement upon ourselves,
which, because it is objective, could be made by others. Why this qualifies as a
social reading of thesis (1) is not only unclear but also gives rise to the following
dilemma.
The objectivity requirement is (i) either satisfied by all judgements, or (ii) appeals
to a feature only some judgements possess. Now, if all judgements are objective, the
objectivity requirement only means that a judgement about oneself that can be made
by anyone is required in shame. This might well be true, but this fact adds nothing
to the explanation a simple reference to the occurrence of this judgement would pro-
vide, and a fortiori nothing of a social nature. Moving on to the second horn of the
dilemma, then, means uncovering a further feature of some judgements that endows
them with objectivity. The distinctive feature of the relevant judgements O’Hear
seems at times to allude to is their rational character (1977, 78). The idea is that the
judgement can be made by anyone in virtue of its being reasonable in the light of the
available evidence together with, or so we might think, the prevailing norms about
what is shameful. But, in addition to facing the above difficulties, taking this route
deprives us of an account apt to cover rational as well as irrational shame.
The upshot of our discussion of O’Hear’s as well as Taylor’s suggestions is then
that perspectival changes occurring in shame should not be elucidated in social
terms. We will now argue that these changes are not a necessary ingredient of shame
and contrast them with what we shall call the “shift of focus” distinctive of this emo-
tion. These changes are not a necessary ingredient of shame because, even though
shame is often accompanied by a change from an attitude uninformed by one’s val-
ues to a discovery that what one is doing stands against them, there are reasons to
think that this temporal sequence is only typical of this emotion. Indeed, at least
two common types of cases are difficult to reconcile with this model. First, one
can engage in an activity while aware that it conflicts with one’s values: this is so
when one is about to knowingly do something shameful. Second, the same is true
of what could be called dispositional shame, i.e. recurrent shame triggered by some
enduring trait of ours.
Now, Max Scheler’s seminal discussion of shame proves helpful in order to pin
down the grain of truth in the idea that shame involves taking a perspective from the
outside (thesis (1)). Scheler stresses the occurrence of a specific shift of attention
in shame. “Shame”, he claims, “is a feeling which belongs to feelings of ourselves.
For in all shame there is an act of ‘turning to ourselves’.” This shift is most notice-
able, we are told, when “shame sets in all of a sudden after an intensive interest
of ours in external affairs had prevented our being conscious and having a feeling
of our own self.” (1913/1987, 15) Whether gradual or sudden, this picture and its
phenomenological underpinning are, in our opinion, essentially correct.13 It is remi-
niscent of Taylor’s perspicuous description of the phenomenon minus its social side.
Note moreover that this is compatible with the claim that the change of perspective
does not always occur in shame. For Scheler’s claim is merely that shame essentially
involves a shift of focus on the self, an idea we shall elaborate in the next section
and contrast with the change of perspective Taylor and O’Hear try to elucidate. For
now, our discussion leaves us with this fourth and final constraint:
Constraint (iv): a theory of shame should satisfactorily account for the change of perspective
typical of this emotion.
11.6
How should we account for the four constraints we have uncovered within a non-
social picture of shame? We shall start by giving a non-social explanation of the
above example of Sam the writer and show how the same type of explanation can
handle more complex cases, in particular those lending initial plausibility to social
theses about shame.
Back to Sam, then. What triggers his shame is of course what he wrote. “Well,
it’s terrible”, he thinks. His thinking that it is bad and that good writing is important
to him (cognitive antecedents) motivates an apprehension of himself as unworthy
(object cognition). So, the cognitive antecedents capture the subject’s awareness
that what he does goes against his values, i.e. they manifest the subject’s awareness
of a value conflict or discrepancy of the kind we have already seen at work in both
Taylor’s and O’Hear’s accounts of shame. This awareness of a conflict between
privately held ideals or values on the one hand and what we are or what we do
on the other hand constitutes the barebones of the type of explanation we want to
appeal to in order to account for all shame episodes in non-social terms. Let us now
give some flesh to it by elaborating on the shift of focus alluded to at the end of the
previous section.
The main claim is that shame episodes are essentially characterized by the onset
of a specific negative evaluative focus upon oneself in the light of what one is
(was/will be) or is (was/will be) doing. The cognitive antecedents of shame exem-
plify a personal stance since they consist in an awareness of a conflict with one’s
13 Unfortunately, Scheler goes on to gloss this in ways that are too specific to fit all shame episodes.
Roughly, he argues that the relevant shift in the subject’s own perspective is to be understood either
as that between a judgement about himself that construes him as an individual and another judge-
ment that construes him merely as an instance of some general category, or vice versa. Although
some well-known examples of Scheler’s (the model and the artist) lend themselves nicely to such
a treatment, it does not extend easily to all cases of shame, not even some of the central ones we
mention.
personal values. One is aware of a conflict between the (dis)value exemplified by

what one is ashamed of and a value one is committed to. This conflict motivates
an apprehension of oneself as unworthy or degraded, a central aspect of shame that
qualifies as a shift in evaluative focus. But can we really develop this idea while
accounting for the intuitions behind the social theses we saw failed to apply across
the board, i.e. meeting the constraints brought to light in our discussion (see Table
11.3)?
Table 11.3 Constraints on a theory of shame
Constraint (i) Although insufficient to explain it, Revealed by theses (4) and (5)
audiences are often relevant
determinants of shame
Constraint (ii) One often does not agree with the Revealed by thesis (2)
judgement that triggers shame
Constraint (iii) Shame is often but not always connected Revealed by thesis (3)
with the social self
Constraint (iv) Shame often involves a change of Revealed by thesis (1)
perspective
Let us start with constraint (iv) according to which changes of perspective typ-
ically occur in shame. The explanation should proceed along the following lines.
The negative focus on ourselves is often due to our moving from an attitude unin-
formed by our values to our discovering that what we are doing stands against them.
Sam becomes aware that his writing is silly and that intelligent writing is impor-
tant to him; he moves from an engagement with writing to an evaluative stance by
means of which he returns to what matters to him. What is crucial is the taking of
a personal stance towards what triggers shame so as to motivate a negative stance
towards oneself. This is sometimes a matter of discovery, sometimes of rediscov-
ery, but sometimes we knew it all along. Although reminiscent of how both Taylor
and O’Hear would treat this type of cases, our treatment differs in crucial respects
from theirs. As opposed to Taylor, the specific point of view or perspective rele-
vant for shame is not social since it is a perspective an agent has towards himself: a
personal evaluative perspective motivates a negative evaluative stance towards our-
selves. A thought about how someone else would look at us is only one of the
possible occasions for occupying this personal stance. As opposed to O’Hear, occu-
pying a perspective informed by one’s values is not equivalent to the making of an
objective judgement. What is required is not that we make a judgement about our-
selves that anyone could make, but that we occupy this personal stance so as to be
aware of a conflict which accounts for the onset of a negative evaluative focus upon
ourselves (and note in passing that we can in this way account for irrational shame).
The example of Sam the writer is of course the most congenial to the picture
of shame we put forward in this article, since it was set up in a way that made
a social reading of it at best contrived. How, starting with such an example and
the explanation we just offered, are we going to meet the remaining constraints?
Consider constraint (i), according to which a theory of shame should satisfactorily
explain why, although insufficient, audiences are typical in shame. Suppose that,
returning to his office, Sam meets his dear friend and colleague John in the corridor
and invites him in. He gives John his writing of the day to read. While John is
reading, Sam’s attention is focused on his face and, remembering his main argument,
it suddenly dawns on him how foolish it is. Here, John, it is fair to say, is merely a
trigger for Sam’s emotion. His presence plays no other role than that of making Sam
apprehend his work in a bad light – after all, Sam might just as well have merely
imagined his colleague reading his work, or imagined nothing at all. And note that
for this very reason, the mere contextual presence of an audience is insufficient to
explain the episode of shame. For it has no explanatory power; it is only when the
subject himself comes to be aware of a conflict, aided or not aided by an audience,
that he feels shame. When playing such an ancillary role, audiences are dispensable.
Now, the force of the intuition that audiences play an important role in shame can
be elucidated by means of the already mentioned contrast between what is “shame-
ful because it is seen” and what is “shameful independently of being seen” (see
Section 11.4). Suppose it is summer and Sam cannot stand the insufferable heat
in his office. He decides to lock his door, removes his shirt and trousers, and falls
asleep. Some time later, he is woken up. He opens his eyes to realise he is face
to face with the secretary of the department who stares at him in disbelief. Sam is
deeply ashamed. These cases indeed depend on the presence of an audience but only
because shame is triggered by something that is “shameful because seen”.
There are two options with respect to such examples. The first is to agree that
some instances of shame indeed depend on there being an audience, or at least a
belief that there is one. But if so, we should say that the presence of an audience
is not sufficient, since shame must be mediated by a concern not to be seen by
such kinds of people in such circumstances. We can then use the aforementioned
distinction to say that the dependence claim is true only when the relevant value
refers somehow to one’s being seen by an audience. And this, of course, is a local
truth about some instances of shame, one that fails to extend to shame episodes
motivated by a perceived failure to uphold those values of ours for which reference
to an audience is irrelevant. Alternatively, one may want to defend a stronger claim.
This would consist in saying that there are no such values as “shameful because
seen” such that they could elicit shame. This amounts to denying the existence of
“superficial” shame in favour of a distinction between embarrassment and shame.
Let us illustrate both options. Sam has a negative emotional reaction because he
cares about what being seen in this position entails for his standing in some relevant
circle. According to the first option, such a concern is likely to reveal a serious
conflict between Sam’s values and the way he stands with respect to them. Enough
anyway to produce shame. If so, the presence of the secretary is needed for him to
feel shame. Alternatively, one may say that Sam’s concern about being seen half
naked is not a value of enough weight for his reaction to be shame. He feels slightly
embarrassed of course, or even very much so, but not ashamed. We prefer the first
option, since it does not prejudge the kinds of things people can be ashamed about.
Be that as it may, on both options, our account meets constraint (i).
Let us now address constraint (ii), which is concerned with the fact that one is
often not committed to the truth of the negative judgement that triggers shame, by
modifying once again our example. Sam walks back into his office. A colleague
of his for which he has little respect, Mark, sits at his desk engrossed in what can
only be what he wrote earlier that day. Mark is obviously enjoying it wholeheart-
edly. Sam realises immediately why his piece of writing would appeal to Mark. He
feels shame. Here, shame is triggered by the positive reaction of an audience. The
cognitive antecedent is the awareness of a conflict between what such a positive
evaluation reveals to Sam with respect to what he wrote, and what his own evalu-
ation of it would demand: for a philosophical piece to appeal to Mark is for it to
exemplify values Sam despises. This is one way in which one might feel shame
without being committed to the judgement made by the audience. For although he
agrees with Mark that his piece exemplifies some specific evaluative properties, he
disagrees with him that they make his piece a good one. It is by coming to see such
a positive judgement upon his work as depending on a preference for values he does
not share that he comes to feel shame.
Now, change the example slightly. Mark is in the same position, but obviously
not liking what he reads. What would it take for Sam to feel shame, as opposed to
anger at or fear of Mark (see Section 11.2)? For shame to occur, the explanation
we favour must entail that Mark’s judgement connects in one way or another with
values Sam is attached to. The natural suggestion in this case is that it connects
with Sam’s concern for his reputation or standing as a colleague or as a philosopher.
Suppose for instance that Mark is an influential colleague whose opinion is very
much regarded within the department. If concern for his reputation or standing is
part of Sam’s values, then he will be likely to feel shame independently of his agree-
ing with the judgement Mark passes on his work. The conflict Sam is here aware
of is that between a concern for his reputation and the adverse publicity Mark’s
judgement is likely to produce. This shows what is wrong with Wollheim’s starting
point (Section 11.3), which consists in developing a contrived context in which one
submits to the negative judgement of an audience to account for the cases where
one appears to disagree with the judgement that is passed on us. For, according
to the simple explanation we favour, shame occurring when we disagree with the
judgement triggering it only makes sense in the light of its impact on the subject’s
own values. Constraint (ii) is met without understanding shame as the emotion of
submission.
These observations directly speak to constraint (iii), according to which one’s
social self or projected image often plays an important role in shame. Consider once
again the scenario involving Sam half naked in his office. We can now understand
the idea underlying talk about a “social self” as referring to the many ways one is
concerned about how one appears in various social contexts. In the example just
discussed, Sam values being regarded inside his department as a good philosopher
and becomes aware that this is incompatible with being negatively judged by Mark.
Similarly, when Sam is ashamed at being caught half-naked in his office, his emotion
depends on his valuing the opinions of others. The importance of those values we
hold that are concerned with our image or reputation should then not lead us to think
that these values are the relevant determinants in all cases of shame, but rather that
these are among the many values we personally hold which, when undermined, may
motivate shame.
We have now reviewed all the examples that motivate social theses about shame,
explaining how we can account for the four constraints they give rise to in non-
social terms, i.e. in terms of the personally held values these examples trade on.
And we did not do that at the expense of the intuition that we are social crea-
tures for whom the regard of others is often of paramount importance and figures
among the determinants of many shame episodes. Firstly, threats to our honour or
reputation – which are potent triggers of shame – do indeed often come from how
others look or would look at us, either literally or metaphorically. A concern with
such socially defined values is of paramount importance for many of us and no doubt
more important in some cultures than in others. But we can explain why honour
and reputation are central triggers of shame without endorsing a picture of shame
that makes it essentially social: these values are only some among the many values
which when threatened can motivate this emotion. We saw no reason to conclude
that when we are ashamed of our dishonesty, or lack of compassion, what matters
to us is always our reputation or social image. Secondly, there is another sense we
did not discuss in which shame is interestingly social: we often learn through shame
in virtue of the impact our social interactions have on the values we come to be
attached to.14 Some of the examples we have discussed illustrate paradigmatic cases
of such learning. Because of the respect we have for some people, the unfavourable
view they take of us is indeed likely to elicit shame. And this may well then lead
one to reflect on one’s values and to adopt new ones. This much is secured within
the picture of shame sketched above: a fundamental ethical role can be assigned to
others without conceiving shame as essentially social. Indeed, in this picture, the
way others look at us, often depending on who they are for us, plays more than the
ancillary role we have seen it playing in some of the theories we have discussed.
11.7
We have defended the idea that shame only makes sense in the context of our attach-
ments to personal values and our failures with respect to them. This is the case, we
argued, even when the values concerned are contingently or essentially related to
how others look at us. Only within this picture that refuses to see shame as essen-
tially social can we account for what we have called deep shame: the shame we feel
at personal failures which is only heightened when, in addition, they become known
by relevant others.
14 As O’Hear notes, others have the power to reveal to us things about ourselves that we should be
ashamed of. Others, in particular those we respect, might help us identify some aspect of ourselves
that is not viewed as problematic but becomes so through their negative assessment. See O’Hear
(1977, 79) and Williams (1993, chap. 4).
Of course, rejecting the well-entrenched idea that shame is essentially social has
potentially costly consequences for the theory of shame. For note that not all per-
ceived discrepancies between our personally held values and what we presently
are or do lead to shame, i.e. to an evaluation of ourselves as unworthy. Although
they face serious problems, social claims about shame at least hold the promise of
explaining why only some perceived discrepancies, i.e. those essentially trading on
our standing vis-à-vis others, motivate shame. If this route is, for the reasons we
adduced here, blocked, another explanation altogether has to be found in order to
individuate shame from other negative self-reflexive emotions. Only when such an
alternative explanation is brought to light will a full-blow theory of shame emerge.15
Acknowledgements We are grateful to Otto Bruun, Tjeert Olthof, Kevin Mulligan, Martha
Nussbaum and Raffaele Rodogno for their helpful comments on previous versions of this paper.
This paper was written with the support of the Swiss National Centre for Competence in Research
(NCCR) in Affective Sciences.
References
Benedict, R. 1946. The Chrysanthemum and the Sword. Reprint New York: New American Library,
1974.
Bruun, O., and F. Teroni. 2011. “Shame, Guilt, and Morality.” The Journal of Moral Philosophy
8:221–243.
Calhoun, C. 2004. “An Apology for Moral Shame.” The Journal of Political Philosophy 12
(2):127–46.
Creighton, M. R. 1990. “Revisiting Shame and Guilt Cultures: A forty-year Pilgrimage.” Ethos 18
(3):279–307.
Deigh, J. 1983. “Shame and Self-Esteem: A Critique.” Ethics 93:225–45.
Deonna, J., and F. Teroni. 2008. “Differentiating Shame from Guilt.” Consciousness and Cognition
17:725–40.
Deonna, J., and F. Teroni. 2009. “The Self of Shame.” In Emotions, Ethics, and Authenthicity,
edited by V. Maier and M. Salmela, 33–50. Amsterdam: John Benjamins.
Deonna, J., R. Rodogno, and F. Teroni. 2011. In Defence of Shame. Oxford: Oxford University
Press.
Elster, J. 1999. Alchemies of the Mind. Cambridge: Cambridge University Press.
Ferguson, T. J., H. Stegge, and I. Damhuis. 1991. “Children’s Understanding of Guilt and Shame.”
Child Development 62:827–39.
Goldie, P. 2000. The Emotions: A Philosophical Exploration. Oxford: Oxford University Press.
Lamb, R. E. 1983. “Guilt, Shame, and Morality.” Philosophy and Phenomenological Research 43
(3):329–46.
Nussbaum, M. 2004. Hiding from Humanity. Disgust, Shame, and the Law. Princeton: Princeton
University Press.
O’Hear, A. 1977. “Guilt and Shame as Moral Concepts.” Proceedings of the Aristotelian Society
77:73–86.
15 We have developed a non-social explanation apt, we think, to individuate shame from other
negative self-reflexive emotions elsewhere. See Deonna and Teroni (2008) and, especially, Deonna
and Teroni (2009) and Deonna, Rodogno and Teroni (2011).
Olthof, T., T. J. Ferguson, E. Bloemers, and M. Deij. 2004. “Morality- and Identity-Related
Antecedents of Children’s Guilt and Shame Attributions in Events Involving Physical Illness.”
Cognition & Emotion 18:383–404.
Roberts, R. 2003. Emotions: An Essay in Aid of Moral Psychology. Cambridge: Cambridge
University Press.
Scheler, M. 1913/1987. Person and Self-Value. Three Essays. Dordrecht: Martinus Nijhof.
Smith, R. H., J. M. Webster, W. G. Parrott, and L. Eyre. 2002. “The Role of Public Exposure
in Moral and Nonmoral Shame and Guilt.” Journal of Personality and Social Psychology 83:
138–59.
Tangney, J. P. 1991. “Moral Affect: The Good, the Bad, and the Ugly.” Journal of Personality and
Social Psychology 61:598–607.
Tangney, J. P., R. S. Miller, L. Flicker, and D. Hill Barlow. 1996. “Are Shame, Guilt,
and Embarrassment Distinct Emotions?” Journal of Personality and Social Psychology 70
(6):1256–69.
Tappolet, C. 2000. Emotions et Valeurs. Paris: PUF.
Taylor, G. 1985. Pride, Shame and Guilt. Oxford: Oxford University Press.
Velleman, D. J. 2001. “The Genesis of Shame.” Philosophy and Public Affairs 30 (1):27–52.
Williams, B. 1993. Shame and Necessity. Berkeley, CA: University of California Press.
Wollheim, R. 1999. On the Emotions. New Haven: Yale University Press.
Part IV
Evaluating the Social Self
Chapter 12
Feeling Up to It – The Sense of Ability
in the Phenomenology of Action
Trust in one’s abilities is an important condition of success. If you believe the words
of a recent presidential candidate, lack of confidence is the main reasons why we
are living such compromised lives in a world stricken by poverty, war, illness and
pollution. What’s stopping us from doing better is our skepticism about what we –
individually as well as collectively – could achieve. It is this negative attitude which
lead us to accept us the deal we’ve gotten so far as good enough for us. In actual
fact, the future president suggested, we could get much more out of life, if only we
accepted that this is in fact possible. It is time, he argued, to break the bonds of
our negative self-image and to choose to pursue the things we have always secretly
wanted and accepted as goods worthwhile having, but never really thought to be
within our reach. The candidate claimed that there has never been anything wrong
about having high ambitions: yes, we can!
Yet there is another successful politician’s voice, coming from the distant past.
As the leader of the March of the Ten Thousand, Xenophon is an expert in achiev-
ing unlikely successes by means of motivational speech. But in his memorabilia
(book 4, chap. 2, 26), he is far from recommending unlimited confidence. It is of the
utmost importance, Xenophon argues, to be realistic about one’s abilities – individ-
ually as well as collectively – and that includes knowing and accepting one’s limits.
Exaggerated confidence in one’s powers is a clear sign of hubris. A bloated sense of
one’s abilities will lead agents to engage in futile endeavors. Today’s personal train-
ers tend to argue against fear of failure, as failures may teach important lessons. But
Xenophon reminds us that where the stakes are high, failure may mean death and
destruction to individuals as well as to communities. Thus in Xenophon’s view, a
keen sense of the limits of our abilities and a skeptical attitude towards increasing
ambitions is far from being an obstacle in the way of getting a better deal; rather, it
protects us from harm.
H.B. Schmid (B)

e-mail: Hans-Bernhard.Schmid@unibas.ch
216 H.B. Schmid
This paper will not decide this dispute; rather, it focuses on the attitude around
which this controversy revolves. Agents have some sort of awareness, conscious-
ness, or sense of what they can and cannot do – individually or together with others,
as a team. And while this sense may not be freely changed at will, it is clear that it
is not merely a given, but an attitude that can be cultivated and developed. Having
a healthy sense of our abilities will help us to do the best we can while avoiding
hubris and self-destruction.
Cognitive psychology has long discovered this topic. Here, the sense of ability
is the focus of a whole branch of research, especially in Self-Efficacy Theory (e.g.,
Bandura 1997). Self-Efficacy theorists argue that the main role of the sense of ability
in people’s agency is that it determines the range of considered alternatives, and thus
settles the question of what kind of tasks people will set themselves and take on.
What people do depends on how much they want it as well as on how likely they take
themselves to succeed. By assigning expected probabilities of success to available
alternatives, the sense of ability determines the choices people make, and the amount
of effort they are likely to make in pursuit of the goals they choose. People tend to
make bigger efforts in the pursuit of tasks they sense to be more difficult for them to
achieve, if they decide to act accordingly. In many cases, however, a weak sense of
ability will discourage people from engaging in an action; this is true for collective
actions (Bandura 2000) as well as for individual ones.
Thus psychological research shows just how fundamental the sense of ability is.
In a more general perspective, it determines the degree of involvement people take
themselves to have in their lives; it determines the degree to which they rely on
themselves rather than on other agents, or on external forces, for the fulfillment of
their needs and desires; it determines the degree to which they see themselves as
responsible for their lives rather than dependent on other agents or external forces.
In short, people’s sense of ability determines the degree to which they see their lives
as being up to themselves, instead of a matter of fate or destiny. Thus the sense of
ability is indeed central to people’s view of themselves, as well as to their outlook
on the world; it is an important and fundamental feature of agency.
It is rather surprising that there is relatively little to be found on this topic in the
received philosophical literature. In analytical action theory, there is a discussion
about whether or not agents may intend to do things they believe to be impossible
to do (e.g., Mele 1989; Ludwig 1992, 1995). While this is certainly pertinent to
the topic, it is a rather specific issue. (It will be argued below that this particular
question can be dealt with successfully only on the base of an adequate account of
the sense of ability.) As the consciousness or awareness involved in action are the
proper domain of phenomenology, and as phenomenology of action has received
increasing attention over the last couple of years (e.g., Pacherie 2008; Roessler/Eilan
[eds.] 2003), one would expect there to be a phenomenology of the sense of ability.
In fact, there are analyses on the senses of ownership (Marcel 2003; Ehrsson et al.
2004), purpose (Falvey 2000), and control (Nahmias 2005) involved in action. While
the sense of ability plays a part in all of these features, these analyses are limited
to the phenomenology of actual engagement in action, while the sense of ability
extends to potential action, and also concerns the “planning stage”.
12 Feeling Up to It – The Sense of Ability in the Phenomenology of Action 217
Among the first philosophical questions to be asked about the sense of ability is
what kind of attitude it is. Self-Efficacy theorist do not waste much thought on this
question, and they tend to conceive of the sense of ability in doxastic or percep-
tual terms, talking of the agent’s beliefs about – or perceptions of – their abilities.
Similarly, current analytical action theory focuses on the sorts of beliefs agents may
have concerning the feasibility their goals, thus focusing on the cognitive dimen-
sion of the sense of ability. The main aim of this paper is to argue for an affective
account: the sense of ability is an (extistential) feeling. This is not to say that it does
not involve beliefs, or perceptions. Rather, the point is that it is important to con-
ceive of the cognitive or perceptual component of the sense of ability as features of
a feeling.
This paper has three sections. The first section opens with some preliminary con-
siderations about the concept of ability, and presents some conflicting intuitions
concerning the structure of the sense thereof. The second section argues that an
affective account of the sense of ability is likely to be more successful in dealing
with these issues than the purely doxastic account implicit in most of the received
literature. The concluding section turns to the special case of joint action, where it
is argued that the sense of collective ability is the feeling of mutual trust.
12.1 Some Conflicting Intuitions
What is it that our sense of ability is of or about? What is that state, quality, power,
or disposition called ability?
It is typical of philosophical problems that they tend to proliferate rather than
disappear in the course of philosophical research, and this seems to be no different in
the case at issue here. There are a wide variety of analyses of the concept of ability in
the received literature.1 The matter is further complicated by the fact that the topic is
approached from very different angles, ranging from the question of compatibilism
(cf., e.g., Oddie/Tichy 1982, 1983) to virtue ethics and virtue epistemology (e.g.,
Sosa 1993) to the evaluative basis of normative economics (cf. Cohen 1989).
Still, there are some basic features of the concept of ability, which seem to be
recognized in all or most of these analyses. In an early contribution, the German phe-
nomenologist Karl Konstantin Löwenstein (1911) distinguishes two basic aspects:
ability combines possibility (possibilitas) and potency (potestas) – a distinction that
is closely related to the distinction between opportunity and capability. Possibility
is the objective side of ability and describes the mode of being of what an agent
can achieve. For somebody to be able to ϕ, it has to be objectively possible that she
or he ϕs. More precisely, the possibility-aspect is that by virtue of which our abili-
ties depend on our (internal or external) circumstances. (The external circumstances
1 Examples from the past decades include, among many others, Kenny (1976, 1989, chap. 5);
Millikan (2000, chap. 4); Morriss (2002); Hacker (2007, chap. 4) (for a critical discussion of John
L. Austin’s thoughts on ability cf. Graham (1977, 250ff.)).
218 H.B. Schmid
describe the agent’s situation, the internal circumstances those inner constraints or
those “inner realities” which are not subject to the agent’s will, and which he might
therefore experience as alien to himself, such as compulsions.) Possibility is mea-
sured in degrees of probability, and this is reflected in the fact that our sense of
ability can be weaker or stronger, depending on how likely we think we are to
succeed. (Needless to say that the possibility aspect of ability is somewhat more
complicated in cases in which collective acceptance is part of the success conditions
of an action.)
Ability, however, is not exhausted by possibility. This is revealed by the fact
that there are many things that are possible to do but which we still can’t do for
reasons that have to do with the kind of person we take ourselves to be (our personal
or moral integrity), rather than with external constraints. And conversely, there are
things we would be perfectly able to do if only they were in fact possible under
the given circumstances.2 The potency-aspect of ability is that by virtue of which
it is, in some sense or another, up to the agent whether or not he in fact realizes a
certain possibility. Löwenstein calls potency a “dynamic principle” which cannot be
further analyzed. But it seems that various factors play a role here. In order to be
“potent”, an agent has to be goal-oriented, she has to have some motor control, and
some representation or concepts of what it is she is capable of doing, even though
these are necessary rather than sufficient conditions (it is important not to conceive
of representations in terms of the use of conceptual capacities, as this would lead
into a conception according to which only beings who have a concept of using a
concept would have any ability at all).
I already emphasized that possibility and potency do not coincide; at the same
time, it seems that they are more closely interrelated than simply in terms of overlap-
ping extensions. Everyday descriptions of the surrounding world are deeply imbued
with action opportunities, and descriptions of competences are always made against
the background of normal circumstances. The fact of the matter is that it is hard to
think of the everyday world in other ways than in terms of action opportunities, and
it seems impossible to think of competences in other ways than in terms of what can
actually be done in the world as we know it, i.e. in terms of typical situations. So it
seems that possibilities are always individuated against the background of potencies,
and vice versa.
Some more distinctions are in order. Ability can either refer to action types, or
to token actions. It is usually thought that under normal circumstances at least, the
former is implied in the latter, whereas the converse is not the case. An agent may
be an able piano player, but he may be unable to play that piano right now for some
reason having to do with the situation or other factors. The distinction at issue here
is between two meanings of ability: ability in terms of “scire” or “savoir”, or in the
sense of “posse” or “pouvoir”, where the former refers to action types, and the latter
to token actions.
2 Throwing one’s rich spinster aunt from the train would be an example for the former case, opening
yet another bottle after one has emptied the last remaining one an example for the latter.
Another distinction that is relevant to the topic is the one between complex and
simple actions, as ability might be something quite different depending on the kind
of action in question. Concerning the phenomenology of simple actions, I find the
distinction between causally basic actions and intentionally basic actions especially
important (Hornsby 1980). Moving one’s hand is causally more basic than turning
on the light, because the movement of one’s hand causes the light to go on. But turn-
ing on the light is intentionally basic for most agents because we are able to do it
“just like that”, that is, without having to do something else intentionally. Intentional
basicness and thus the bottom level of our awareness of our agency, is determined by
our abilities. Needless to say, intentional basicness is relative; what is intentionally
basic for one person might not be basic for another. Under normal circumstances,
however, bodily movements are intentionally basic and are thus our fundamental
abilities. As to complex actions, it seems that whoever is able to perform that action
must be able to perform some possible combination of simple actions, which real-
izes the complex action; there might be many, however, and the agent need not be
committed to any single one of them.
What is the relation between ability and the agent’s sense thereof? In his paper
on the concept of ability, Löwenstein strongly urges his readers not to confuse the
two. Our sense of ability might not tell us much about our actual abilities; as in
many other kinds of self-evaluative attitudes, our sense of ability is far from being
infallible. Having a sense of ability isn’t a necessary, let alone a sufficient condition
of ability. There might be abilities of which we have no sense whatsoever, and con-
versely, people routinely take themselves to be able to do things of which they turn
out to be utterly unable. Still, the fact of the matter seems to be that agents do have a
sense, i.e. some form of awareness, of whatever they take to be their abilities; there
is some phenomenal quality involved in this attitude, something “it is like” for the
agents to have that sense.
With these preliminary remarks in mind, let us now turn to some conflicting
intuitions, which seem to make the topic interesting from a philosophical point of
view. It seems plausible to assume that our sense of ability is subject to rational
constraints (one’s sense of ability can be under- or over-developed to the degree
of irrationality), that it is informed by experience (especially by precedent), and
that it limits the class of possible objects of intention (a sense of complete inability
prevents us from forming a corresponding intention: there is some plausibility in
assuming that one cannot seriously intend to do what one takes to be impossible to
do). Yet there are plausible objections to all of these claims. It seems that a strong
case can easily be made for the view that first, the sense of ability is, to a large
degree, self-confirming rather than subject to rational constraints, and that second,
it is a matter of convention rather than a matter of experience, and third, that it
does not limit the class of possible objects of intention, as one may easily think of
cases of intending the subjectively impossible. The following remarks touch on the
first two conflicts only briefly and go into some more depth with regard to the third
issue.
Here is why it might seem plausible to assume that our sense of ability is subject
to tight rational constraints. If agent A has some pro-attitude towards ϕ-ing, but A
220 H.B. Schmid
does not intend to ϕ only because A takes herself to be unable to ϕ, and where
there is no evidence for A to take herself to be unable to ϕ, or where there is even
evidence that A could ϕ, we would probably call A irrational. In that case, A’s sense
of ability is irrationally under-developed (this is highlighted by the fact that people
routinely object to such attitudes in trying to convince the respective person of his
or her abilities: “You can do it!”). The converse case is where A’s sense of ability
seems to be over-developed to the degree of irrationality. This is the case where A
does intend to ϕ in spite of the fact that there is reason to take him- or herself to
be utterly unable to ϕ. So there are limiting cases, in which an under-developed or
over-developed sense of ability is irrational.
But the rational constraints to which the sense of ability is subject go even further.
Not just the question of whether or not one should take oneself to have any chance
of success seems to be a question of rationality, but also the degree to which one
should be confident of one’s abilities. There is a rational ideal in which one’s actual
abilities and one’s sense thereof coincide, and the task of becoming a fully rational
agent does not only seem a matter of having a sense of what is within or beyond
one’s scope, but having a sharpened sense of just how much one can rely on one’s
abilities under the given circumstances.
So much for the first intuition, which seems to be implicit in much of received
action theory, as well as in folk psychology. Now, here is a reason for doubt. I
assume that anyone who has spent some time in different milieus and cultures will
agree that there are cultural standards concerning the question of just how confident
people are expected to be of their own abilities, and that these standards vary a great
deal. In some milieus, people are expected to show a greater amount of confidence
than in others. Some milieus foster or enforce what might appear to others as an
overly high degree of confidence (“Believe in yourself! You can do it, if you really
want!”), whereas other milieus seem to instill a level of confidence that might appear
to outsiders as self-debilitatingly low.
According to the predominant stereotypes, these differences are particularly
blatant between rural and urban milieus, between professional groups such as
philosophers and business people, and perhaps between Europe and the United
States. If we limit ourselves to the intercultural case, this observation puts us in
a rather uncomfortable position. If we hold on to the rationality claim, and if we
agree that the difference is not one of people’s actual abilities, but rather of peo-
ple’s sense thereof, it seems that we have to rank cultures and milieus in accordance
with the degree of rationality they seek to instill in people’s self-image. This, how-
ever, collides with our intuition that attempts to judge the rationality of cultural
standards usually simply amounts to claiming rationality for one’s own cultural stan-
dard, which amounts to chauvinism rather than epistemic rationality. If we accept
this intuition, we seem to be caught in a conceptual dilemma. The two intuitions,
both of which are plausible enough for themselves, turn out to be incompatible. The
problem with which we are left therefore seems to be to reconcile the intuition that
the sense of ability is subject to rational standards with the intuition that the sense
of ability is subject to conventions, especially to norms of propriety, character, or
culture.
It might seem, though, that this is only an apparent conflict, and that we shouldn’t
be too worried about cross-cultural claims to rationality. After all, it seems that there
is an easy way to find out whether or not one’s sense of ability is on target: one
simply has to put the matter to the test, and to undergo a reality check. This seems
to be how intentional under- or over-confidence is conclusively revealed. This leads
to the second pair of conflicting intuitions (which is really just a version of the first
one). It seems plausible to assume that the acquisition and development of one’s
sense of ability is a matter of making good use of available evidence, especially
precedent, which is a matter of epistemic rationality. Developing a sense of ability,
one might think, is a matter of learning from experience, from success and failure
providing a standard to which our sense of ability can and should conform.
Plausible as this intuition might seem, it is rather obvious that in actual fact, peo-
ple do not usually let experience interfere with their sense of ability, but rather tend
to interpret available evidence in such a way that their sense of ability is confirmed.
Thus people with a rather strong sense of ability will be quick to ascribe their fail-
ures to their own not having tried hard enough – “I would have succeeded if only I
had put my back into it”. And conversely, people with a rather weak sense of ability
will be quick in ascribing unexpected successes, where they occur, to anything but
their own abilities: “I guess I was just lucky”. So neither of these will let experience
interfere with their sense of ability.
This is to say that there are ways in which an agent’s sense of ability might
be immune to experience, and largely self-confirming. It is an open question if we
can dismiss such cases simply as irrational, or pathological: how should we tell?
Doubting the irrationality of self-immunization is not to deny that experience can
and should play some role in the development of an agent’s sense of ability. But
the fact of the matter is: the question of whether or not any given case of success
(or failure) speaks for (or against) an agent’s abilities – rather than being due to
the agent’s sheer luck (or lack of willpower) – is, to some degree at least, left to
the agent’s (or some other observer’s) discretion. You simply never know. It is true,
of course, that complete immunity against experience, as it occurs in cases such
as those just mentioned, might become increasingly difficult to uphold in repeated
cases; but the fact remains that our sense of ability inevitably has to draw from
other sources than experience. Part of our sense of ability is a priori. This becomes
apparent if we construe a purely empiricist counter-example to the abovementioned
self-immunizers. Think of an agent who takes himself to be able to ϕ if and only
if his last attempt to ϕ was successful. It seems obvious that we would call such an
agent’s sense of ability inadequate, because in many cases, success and failure are
due to the circumstances or to lack of effort rather than to ability or inability. Thus
a further qualification has to be added: the circumstances need to be normal and
the agent’s effort sufficient. A reasonable empiricist about ability would take herself
to be able to ϕ if under normal circumstances and with adequate effort, she has
successfully ϕ-ed before. This, however, leads into a circle: in order to determine
which circumstances are “normal”, and which level of effort is “adequate”, we have
to appeal to the agent’s (or some generalized agent’s) abilities; the explanandum
reappears in the explanans. If this line of thought is sound, we end up with another
222 H.B. Schmid
pair of conflicting intuitions. In this respect, the problem of an adequate concept of

our sense of ability is to reconcile our intuition that our sense of ability is informed
by experience with the equally plausible intuition that it is basically a priori, immune
against experience, and, to some degree at least, self-confirming.
The third issue is whether or not the agent’s sense of ability limits the class of
possible objects of intention. As mentioned above, this question has attracted some
attention in received action theory (the following discussion is focused on prior
intentions3 ).
Why should one think that there are any limitations to one’s objects of intentions
at all? The point of departure is that there seem to be things, which – in one’s right
mind at least – one simply cannot intend to do. I can intend to run seven miles, but
I cannot intend to run one hundred (as I know I would collapse trying). I can intend
to improve the paper I’m writing, but I cannot, in my right mind, intend to discuss
it with Wilfrid Sellars (as I don’t think I can talk to the dead). I can intend to take
my sunglasses with me, but I cannot intend to make the sun shine (as I assume that
this is beyond my powers). This is not to say that such intentions are meaningless
(such as the intention to pink). All of the above are actions, and therefore, to some
degree, suited for intention on a purely conceptual level. But they are not suited for
my intentions, or the intentions of average beings. I can well imagine to be doing all
of these things; the range of imagination (and perhaps of desire) might be unlimited,
but the range of intention is not. So one might wonder: How precisely is the limit
and scope of actions one can make the objects of one’s intention determined?
Considering the above examples it seems plausible to assume that one’s sense of
ability plays a crucial, perhaps even the decisive role here. I will defend this view
by discussing a couple of alternative ways in which it has been formulated in the
received literature, and by trying to accommodate a plausible objection.
A first thought is the following. Philosophers such as Paul Grice (1971) and
David Velleman (1989) seem to have argued that for action A to be the object of
agent P’s intention, P has to believe (or at least to accept) that she or he will A.
The claim appears to be that one cannot intend to do what one does not believe one
3 Most philosophers of action use a distinction of the sort of the one between prior intention and
intention-in-action (Searle 1983). Prior intentions are the “plans” of (or “projects” for) an action
that do not, as such, involve any practical engagement; one might have a prior intention without
currently doing anything about it, such as in the case of the intention to spend next year’s summer
in Spain. Intentions-in-action, by contrast, are the feature by virtue of which a given complex of
behavior is an action. Along similar lines, Michael Bratman (1987) distinguishes between future-
directed and present-directed intention; Alfred Mele (1992) has the distinction between distal and
proximate intention. This distinction is usually taken to be a matter of the temporal structure – the
prior intention comes first and terminates in the intention-in-action –, but one can also interpret this
difference as a difference in terms of the kind of practical reasoning at work in these cases, and of
the kind of representation of the content. Prior intentions are a matter of deliberation, intentions-in-
action a matter of implementation and practical skills (with different sets of abilities involved). It
seems obvious that our sense of ability extends to all levels of intentionality, ranging from a sense
of the reach of our limbs (or, in the case of internal actions, the scope of mental resources available
to us) up to the kinds of plans to which we can commit ourselves.
will do. And obviously, one cannot believe that one will do what one does not take
oneself to be able to do.
It has been objected – by Bratman (1999), among others – that this is too restric-
tive, at least if belief (or acceptance) is taken to be all-or-nothing. People intend to
do all sorts of things they do not really believe (or accept) they will actually achieve.
In his Change in View (1986), Gilbert Harman provides an example. He considers
the case of a sniper aiming at an ambassador; the sniper takes himself to be a ter-
ribly bad marksman, and therefore he does not really believe (or accept) that he is
able to shoot the ambassador. Rather, he thinks he will probably miss his target.
Nevertheless, he has a go at it, he aims and pulls the trigger. In this case, Harman
argues, the object of the agent’s intention differs from what he thinks he will do.
What the sniper intends to do is to shoot the ambassador. But he does not believe
that he will in fact shoot the ambassador. Rather, he thinks that he will miss his tar-
get. So Grice and Velleman’s statement seems too strong. For ϕ to be the object of
A’s intention, A does not have to believe (or accept) that she or he will ϕ.
But there are at least two weaker versions of the limitation claim. The first is
the following: for ϕ to be an object of A’s intention, A has to believe (or assume,
or accept) that there is a chance that she will ϕ; she might take her chance to be
minimal, but in her view, it has to be greater than zero. A needs some confidence con-
cerning her prospects of success, however minimal it may be (e.g., Wallace 2001).
All that is said is that there has to be some sense of ability, however weak, and
however minimal.
Another version (endorsed by Bratman 1987 and Mele 1989) is that A cannot
take herself to be utterly unable to ϕ if she intends to ϕ (cf. Baier 1970). This is even
more liberal a limitation claim: Nothing is said about the background conditions of
intention except the following: the fact that ϕ is the object of A’s intention is incom-
patible with A’s sense that his or her chances at success are zero. Intention does
not involve any certainty of success, and is perfectly compatible with any degree
of self-doubt, as far as it is self-doubt, and not certainty of inability. It is clear that
one can intend to do things one is all but certain to be unable to carry out – remem-
ber Harman’s sniper. Ambitious intentionality is not in conflict with the limitation
claim. Life would be boring indeed if we limited our intention to objects which we
are confident to be able to carry out (even though we might have good reasons to
take Xenophon’s advice to heart and try to avoid constant failures). All that is said
is that we cannot intend to do what we are convinced we cannot possibly achieve.
Before proceeding any further and discussing the problematic consequences of
the limitation claim, a possible alternative has to be addressed. In current action
theory, the limitation claim is often bypassed rather than rejected. It is assumed that
it can be avoided by choosing an alternative approach to intentionality. We normally
think of intentions in terms of what we express with the words “A intends to ϕ”.
Put this way, intentions are action referential, as the object of the intention is an
action. Moreover, this limits the class of possible objects of intention to the agent’s
own actions, since it seems plausible that one can directly intend only what one does
oneself. In other words, if intention is action-referential, it is action self-referential.
If I express my intention in terms of “I intend to close the door”, it is clear that I
224 H.B. Schmid
intend the closing of the door as my own action. If you pre-empt my plan and close
the door before I get around to doing it, my intention does not reach its conditions
of satisfaction. It is true that I wanted the door to be closed, so that my desire is
now satisfied, but as I intended to close the door myself, my intention isn’t. I cannot
intend your closing the door directly; all I can intend is to make you closing the
door, which is action self-referential as the making is something I have to do myself
(e.g., by means of asking you to close the door).
Now what about intentions of the form “I intend that the door be closed”? Let us
call this way of putting intention propositional, since the objects of these inten-
tions are propositions (or propositional contents) rather than actions. If I intend
that the door be closed, it does not seem to matter how this is achieved: the job
might be done by myself, anybody else, or perhaps by the wind. The subject of
the action, if there is any, need not be identical with the subject of the intention.
I might intend that I close the door, or that you close the door, or that we do it
together. This widens the scope of possible objects of intention beyond the limi-
tations mentioned above (Vermazen 1993). I can now, it seems, intend all sorts of
things which are not my own actions: for example, I can intend that we go for a
walk together, which is clearly something I do not take myself to be able to do,
since it takes at least two to be done: we can do it, I can’t. So the question is:
shouldn’t intention be expressed in propositional rather than in action referential
terms?
Wilfrid Sellars (1992, 183f.) has argued rather forcefully (and convincingly) that
propositional intentions are expressions of practical commitments only by virtue of
their conceptual tie to action referential intentions. Sellars says that the intention
that X, when made explicit, spells as the intention to do whatever is necessary to
make it the case that X, which is action self-referential, because again, one has to
do the making oneself. I still think he is right. But it is completely sufficient for the
purpose of this paper to accept action referential intention as one important kind of
intention, and that the action-referential mode of expressing intention should not be
abandoned completely, even though it might not be the only way of thinking about
intention. So let us limit the limitation claim to action referential intentions. The
following, then, seems acceptable: A cannot intend to ϕ if A takes himself/herself
to be utterly unable to ϕ. So it seems that our sense of ability does limit the class of
possible objects of intention.
This claim, however, is not uncontested, either. Here is why. It seems plausible
to say that if A takes herself to be unable to ϕ, she could at least try to ϕ. (After
all, this is one possible way in which learning gets started; a person who cannot
bring herself to plunge into the water because she takes herself to be unable to swim
might be encouraged to try to swim.) This is not incompatible with the limitation
claim, insofar as “trying” functions as a proper action term, that is, if in order to try
to ϕ one does not have to intend to ϕ (for an early defense of this thesis cf. Mele
1989). If “trying” is a proper action term, “trying to ϕ” and “ϕ-ing” have entirely
different success conditions: for one’s intention to try to ϕ to reach its conditions of
satisfaction, one may have to have made one’s best effort; but one does not have to
have ϕ-ed. Thus in the case where “trying” is a proper action term “trying to ϕ” and
“ϕ-ing” are really different actions. If this is the case, one does not have to intend to
ϕ in order to try to ϕ; it is enough to intend to try to ϕ.
So the question is: are cases in which people who take themselves be unable to ϕ
may at least try to ϕ limited to cases in which “trying” functions as a proper action
term? Are cases of aiming at the subjectively impossible limited to proper-action-
tryings? If so, the limitation claim is unproblematic. If not, we’re in trouble.
Even though I think that there are many cases in which “trying” does function as a
proper action term, I agree with Kirk Ludwig (1992, 1995) that there are some cases
of aiming at the subjectively impossible in which trying is not a proper action term.
Clear-cut examples can be found in Ludwig’s papers. Let me instead give an exam-
ple that could be construed either way, in order to come to a clearer understanding
of the conceptual issues at stake here.
Consider a group of people joined around a breakfast table. A jar of jam with a
tightly closed lid is passed among the members, each taking turns in trying to open
the lid. After all the strong women and men in the group have had their go at it
in vain, the jar is passed to little Benjamin, last in the group, sitting at the far end
of the table. Benjamin may be small, but he is sensible. Assume that as such, he
is perfectly aware of the situation: he knows very well how his forces compare to
those of the other group members, he is aware of the fact that tightly closed lids do
not loosen gradually. Making good use of the available evidence, Benjamin believes
he doesn’t stand the slightest chance at opening the jar. But as all the others in the
group have had their go, he thinks that he should have his, too.
Obviously, Benjamin intends to try to open the jar. But does “trying”, in this
description, work as a proper action term or not? If the case is to be compatible with
the limitation claim, we should think it does so, and in fact, this seems perfectly
plausible given the circumstances: Benjamin does not intend to open the jar; rather,
he intends to make an effort, and his intention will be satisfied by his making an
effort, not by his opening the jar. But the example is designed so as to allow for
another interpretation. Consider two versions of Benjamin, assuming two different
frames of mind: BenjaminR , the realist, has just finished his reading of Xenophon’s
Memorabilia, and he is in a “know thy limits”-frame of mind. BenjaminO , the opti-
mist, is just back from one of Obama’s stomp speeches. He is in a can-do mood, but
he is not as stupid as to ignore the facts. BenjaminR tries to open the jar in terms of a
proper action term; avoiding any intentional hubris, he is perfectly aware of the fact
that he does not stand any chance of success. But happens to think that in terms of
a group ritual, he should have his go at the jar, too, if all the others have had theirs.
His intention is not to open the jar, but rather to exert as strong a force as he can
muster on the closed lid. This is all he wants to do, and expects to be able to carry
it out. His intention will be satisfied by his having had a go at the matter, i.e. his
having pushed as hard as he can at the closed lid.
Now assume that in spite of all appearances, and much to the surprise of all
participants in the event, the jar suddenly opens in BenjaminR ’s hands (perhaps it
was all a set-up, and the lid was loosened by remote control). What will BenjaminR ’s
reaction be? One might assume that he will be pleased by the result; he might even
start to boast about his hidden forces, mistaking the result for a proof that he is
226 H.B. Schmid
stronger than he appears, after all. While all this might well be the case, however,
there will be an element of slight embarrassment in his reaction to the event. Even if
he takes the result to speak for his greater hidden forces, which he likes, he will have
to perceive the result as somewhat undermining his status as a competent agent.
After all, competent agents know what they can do. And there is a sense in which
this is not the case here: there is a lacuna between BenjaminR ’s intention, and his
action: instead of merely trying, he ended up opening the jar. He did something he
didn’t strictly want to do, even though the result may be far from unwelcome.
Now consider BenjaminR ’s optimistic twin, BenjaminO . BenjaminO takes a
wholly different approach to the matter. The difference, however, is not one of epis-
temic rationality. BenjaminO may be optimistic, but he is not ignorant. He is as
certain as BenjaminR that he does not stand any chance at opening the jar. Thus
when BenjaminO has his go at it, resulting in the unexpected opening of the jar, he
will be just as surprised by the lid’s coming loose as BenjaminR , and he, just as
BenjaminR , welcomes this unexpected result. But by contrast to his realist coun-
terpart, his joy is not impaired by any embarrassment. In spite of his not having
taken himself to stand any chance beforehand, he perceives the opening of the jar as
something he did intentionally. Opening the jar was what he was trying to do. So by
contrast To BenjaminR , he doesn’t feel his competence undermined by the events
the least bit; rather, he feels his agential competence to be unexpectedly expanded
by his turning out to be able to do what nobody – himself included – had taken him
to be able to carry out.4 So in a sense, his reaction is contrary to his realist twin’s;
the events satisfy his intention, if rather unexpectedly, rather than crossing his plans.
But how is this possible? How could Benjamin intend to open the jar while think-
ing this is impossible? If cases such as BenjaminO ’s are possible, we are left with
a further dilemma. The problem now is to reconcile our intuition that we cannot
intend to do what we take ourselves to be unable to perform with the intuition that
in some cases of trying the impossible, “trying” is not a proper action term, that is,
that sometimes, people do intend to do what they take themselves to be unable to
carry out. We have to find a way to reconcile the intuition that the objects of inten-
tion are limited by what one takes oneself to be able to do with the intuition that
there are cases of intending the subjectively impossible.
4 In his “Impossible Doings” (1992), as well as in some later papers on the topic, Kirk Ludwig
contested that claim. Ludwig discusses the following example. P assumes (with certainty) that the
battery of his car is dead. Upon another person’s request, he turns the ignition key. Contrary to what
he expects, the engine starts. Ludwig claims that it would be wrong to say that P started the engine
unintentionally. I agree with Ludwig that there are some cases of trying the subjectively impossible
where “trying” does not function as a proper action term. I argue, however, that in such cases, the
agent must take himself to have a chance at success, however minimal, which might be in conflict
with his conscious assessment of the situation. Sometimes the agent’s intentional self-confidence
is not in tune with their beliefs concerning their ability. If this is true, Ludwig’s point does not
prove that it is not the case that people cannot intend to do what they take themselves to be unable
to carry out. The question is how to understand the “taking”: insofar as it is mere belief, Ludwig is
right; insofar as it is confidence, he is not.
The authors in the received literature are divided between the two horns of the
dilemma. It is either argued that BenjaminO ’s case does not exist, i.e. that one cannot
intend to do what one believes to be impossible (e.g., Adams 1995). Or it is argued
that agents may intend impossible doings, and that the intuition that agents cannot
intend what they take to be impossible to carry out is wrong (Ludwig 1995).
In BenjaminO ’s case, this amounts either to saying that he does not really intend
to open the jar after all (but rather to try to open the jar, in terms of an action that
is different from opening the jar), or that his optimism leads to ignorance so that he
does not really believe that he is unable to open the jar. The first alternative could
be worked out by allowing for cases in which people can intentionally do things,
which they did not intend to do (BenjaminO could then rightly credit the result of
his effort to his having intended this effect, without having to act on an intention
which is inconsistent with his beliefs). But this distinction between intending to
ϕ and ϕ-ing intentionally seems rather ad hoc. Introducing this ambiguity into the
concept of intention needs strong arguments; until such arguments are provided, I
see no reason to depart from the straightforward view that A ϕ-s intentionally if and
only if A intends to ϕ. The second alternative seems unattractive because it forces
us to assume complete cognitive irrationality: why should BenjaminO ignore such
pertinent factors such as his beliefs about his strength? The question therefore is:
is there a third way which allows us to accommodate such cases without resorting
either to ad hoc conceptual “ambiguations”, or to ascriptions of complete epistemic
irrationality?
In the following, I suggest an alternative which accommodates both intuitions:
people do sometimes intend the subjectively impossible; but I argue that in these
cases, there is a sense in which it is true that they do not take themselves to be
unable to attain their goals, even though they continue to have the belief that they
cannot attain their goal.
12.2 The Sense of Ability as an Affective Attitude

The discussion of the limitation dilemma has left us with the puzzle that in some
cases, it seems that people intend to do the subjectively impossible; at the same time,
it seems plausible to assume that one cannot intend to do what one takes oneself to
be utterly incapable of doing. In the received literature, an incompatibilist line is
taken; either the first or the second intuition is dropped in favor of the other. I do
not think that this is viable, and I suggest that we look for an alternative: a way to
accommodate both intuitions. I shall argue that this is possible within an affective
account of the sense of ability, which will also make the two other dilemmas appear
much less problematic.
Let us start with a closer look at the problem. If we hold on to both intuitions, it
seem that agents such as BenjaminO do and do not take themselves to be unable to
achieve their goals at the same time. It seems plausible to say that BenjaminO does
intend to open the jar, and in order to intend to open the jar, he cannot take himself
228 H.B. Schmid
to be utterly unable to open the jar. At the same time he is convinced that he cannot
open the jar, and thus he does take himself to be unable to open the jar. The position
underlying most of the received literature seems to be that such cases of practical or
cognitive incoherence simply cannot occur.
But let us have a closer look. Why should it be a problem that agents believe that
p and do not believe that p at the same time? Why shouldn’t we say that this is just
another case of “cognitive dissonance”, or inconsistent attitudes, as they often occur
in real world psychologies? The problem with this liberal view is the following. For
an attitude to be of the kind of a belief, there need to be some degree of consistency.
It is true that people often hold inconsistent beliefs. But these do not usually have
the form of occurrent beliefs in the same domain of an agent’s doxastic system,
and they certainly cannot remain beliefs once awareness their conflict is achieved.
Perhaps one can entertain the thought that something is p, and that the same thing
is not p at the same time, but this way of entertaining a thought is not the same
as having beliefs. If an agent has incompatible beliefs, as it often happens, at least
one of the conflicting beliefs cannot be occurrent and be conscious as conflicting,
such as in the case of incompatible conclusions following from accepted premises,
which the agent fails to draw, or some such. In such cases the agent “could have
known” one his beliefs was false, but he failed to draw the necessary conclusions,
and thus may be called holding on to incompatible beliefs. Since in BenjaminO ’s
case, both attitudes – his belief and his intention – are clearly occurrent, and in the
same domain of his doxastic system, it does indeed seem problematic to say that
BenjaminO does and does not take himself to be unable to open the jar. The epis-
temic restrictions on belief make it plausible to choose either horn of the dilemma.
How could there be a third alternative?
Recent theory of practical reason has tended to become somewhat less restrictive
with regard to the “Myth of Practical Consistency” (Kolodny 2008). And indeed
there is a special area in philosophy of mind and action in which case of agents
subscribing to incompatible cognitive contents is quite familiar. Recent philoso-
phy of the emotions has devoted a considerable amount of attention to the fact that
emotions do involve cognitive components (it is an open issue not to be discussed
here whether the cognitions implied in emotions should be conceived of in terms
of perceptions, judgments, or simply in terms of beliefs). Moreover, the fact that
the cognitions or judgments that are implicit in affective states can be in conflict
with the agent’s beliefs has become a well-studied phenomenon under labels such
as “recalcitrant emotions”. The example that is usually chosen in the current debate
is that of a person being afraid of flying in spite of her belief that this poses little or
no danger. Earlier examples include David Hume’s discussion of the fear a person
feels at the sight of a precipice under her feet, even when she is protected by an
iron cage and therefore knows that there is no imminent danger (Hume [1739/1740]
2000, book 1, part 3, section 13, §10). In his Expression of Emotion, Charles Darwin
(1872, 38) discusses a similar case when he reports being afraid of a snake even if
he knows that it poses no danger, because it is behind a glass wall.
In cases of recalcitrant emotions, the agent is subject to some degree of irrational-
ity; there is some inconsistency in his attitude. What makes emotional recalcitrance
interesting for our topic, however, is that it allows for conflicting cognitions to be
occurrent in agents. People who experience recalcitrant emotions are often perfectly
aware of the conflict between how they feel about the object of their emotion, and
how they think about it – just remember Darwin’s experimental setting. Recalcitrant
emotions allow for occurrent cognitive inconsistency without undermining the sta-
tus of the agent as a cognizer, or holder of beliefs. And this is exactly what seems
to be needed in order to explain the case where agents have an intention to ϕ (and
cannot, therefore, take themselves to be utterly unable to ϕ), and believe that they
are unable to ϕ. We have argued above that it is not possible to have the occurrent
belief that p and that non-p at the same time in the same domain, because for an
attitude to be of the kind of a belief, there need be some consistency. The case of
emotional recalcitrance, however, shows us how cognitive inconsistency is possible:
it is well possible, and even quite frequent in our lives, that we believe that p and
yet feel that non-p. This inconsistency does not undermine the status of the belief; a
belief that p is a belief that p, no matter of how one feels about it.
The Platonic answer to the question of how such inconsistency is possible is that
there are two separate systems of cognitive appraisal or evaluation at work here: the
emotions do a quick and dirty job at identifying the most relevant features of a given
situation; our reason, by contrast, works slowly and deliberatively. In most cases,
the two cognitive systems work smoothly hand in hand, and one learns from the
other. My anger informs me of the depravation I have experienced with the loss of
my bicycle, and it is directed at John as the suspected thief. Here, the emotions serve
as watchdogs for reason. As soon as I am presented with evidence from which I can
infer that John isn’t the culprit after all, my anger at him vanishes. In some cases –
particularly in cases of recalcitrant emotions – however, the conflict persists. These
cases undermine, to some degree, our overall rationality, as our perspective on the
matter at hand is split. But such cases do not undermine our status as bearers of
beliefs. My belief that the snake poses no danger is still a belief, and no less firm
a belief, in the face of my feeling threatened by the snake. I’m subject to some
degree of irrationality, of cognitive and perhaps practical inconsistency; but I’m no
less having a belief and experiencing an emotion. Such conflicts are conceptually
possible.
My suggestion is that this is not only a parallel, but that cases in which agents
intend to do what they believe they are utterly incapable of doing are in fact cases
of emotional recalcitrance, because the sense of ability is an affective attitude, i.e.
an attitude of the kind of an emotion. The sense of ability is the feeling of being up
to a task, the feeling of self-trust, or self-confidence. Part of the reason why this is
not listed among the affective attitudes in the received literature is that it belongs to
a special class of affective attitudes (I will come back to this topic at the end of this
section).
As a case of emotional recalcitrance, the feeling of being up to a task, which one
perfectly knows one cannot fulfill, is abnormal and irrational. But such cases are
not conceptually impossible. I submit that this is exactly what happens in the case
of conflict between what we intend and what we take ourselves to be able to do.
These are cases of conflict between what we know in our heads and what we feel
230 H.B. Schmid
in our guts. Sometimes we’re in an overly self-confident gung-ho mood and feel
up to things we know in our head we won’t be able to do. And perhaps even more
frequent is the converse case. People might want to ϕ, they might perfectly know
that they could easily ϕ – and they still fail to form a corresponding intention for the
sole reason they just don’t feel up to it, that is for sheer lack of self-confidence, or
self-trust.
This problem is usually misunderstood as one of the will or one of reason in the
philosophical literature, but it is a matter of the third part of the platonic soul, a
problem of the thymos, that is, of our affective attitudes. In order to act, one has to
desire, one has to have some beliefs, but one also has to feel up to the task, and in
some cases at least, this turns out to be an entirely different matter. In these cases,
our sense of ability – our feeling up to the task or its opposite – turns out to be
recalcitrant with regard to our beliefs.
The fact that our sense of ability is basically an affective attitude is further
substantiated if we turn back to the other two conceptual dilemmas that we have
encountered above. The first problem was to reconcile our intuition that our sense
of ability is subject to rational constraints (and a rational ideal) with the intuition
that it is largely a matter of convention (i.e. of character and culture). This double
nature is typical of many affective states – consider the cases of joy or anger. It is
within the conventional standards of a culture that we judge a feeling appropriate or
inappropriate; at the same time, affective states are subject to rational constraints,
especially with regard to their cognitive components.
The second problem was to reconcile our intuition that our sense of ability is
informed by experience with the intuition that it is immune to experience and to
a large degree self-confirming. Again, I can only point out that we know this phe-
nomenon from the philosophy of the emotions. It is a well-known feature of affective
states that they seek out evidence for themselves: just as to the loving eye, every-
thing tends to look beautiful, it seems that to the confident eye, everything tends to
look feasible.
Thus I suggest that our sense of ability is basically an affective attitude, i.e. a
matter of the question of whether or not, and to what degree, an agent “feels up to
a task”. This suggestion might seem rather unusual, as affective attitudes are nor-
mally divided into moods (such as boredom or nervousness, where there is no clear
intentional object) and emotions (such as anger, joy, surprise or fear), neither list
of which includes anything coming close to the sense of ability. Yet recent philo-
sophical thinking has taught us to expand the class of affective attitudes beyond
the classical moods and emotions. In particular, Matthew Ratcliffes’ (2005, 2008)
important work on existential feelings includes such feelings as the feeling of home-
liness and alienation, belonging and separation, power and control. Ratcliffe argues,
in a somewhat Heideggerian vein, that these feelings are not “emotional” in the nar-
row sense of the term, but that they are affective in nature, and that they are the basic
way in which we are “in the world”. I suggest that the sense of ability is the most
fundamental existential feeling, and that many other existential feelings – such as
the sense of belonging or separation, and even the “sense of reality” – are ultimately
based on our sense of ability.
12.3 The Sense of Ability in Joint Action

The previous chapters focused on self-directed forms of the sense of ability, i.e. on
the sense of one’s own abilities. Yet our sense of ability need not be self-directed; it
may also be other-directed, general, or group directed. In this concluding section of
the paper, I shall concentrate on the latter form, and sketch out the bearings of the
result of the previous sections for the theory of joint action.
As any parent will easily admit, there are situations in which our sense of another
person’s abilities is equally direct and immediate, and perhaps even sharper and
more alert as the sense of our own. In such situations, we have a keen sense, i.e. a
gut feeling for what is within and what is beyond another’s powers. This, however,
is not the only way in which our sense of ability might not be simply self-directed.
There are two mixed cases. Competent members of a community will usually have
a sense of general abilities, that is, a gut feeling for what “one”, i.e. the anonymous
average individual person (the agent himself included), is capable of doing. Thus
the general sense of ability provides a standard for our notion of an agent. Last but
not least, people have a group-directed sense of ability. Just as people’s individual
conduct is in part guided by a sense of what they can do individually, there is a
sense of what they, as a group, can do together. This sense of collective ability
guides people’s joint actions. And it is perhaps not surprising at all that this sense is
frequently made appeal to in political rhetoric (“Yes, we can!”).
Joint actions imply individual actions, but these individual actions are not inde-
pendent of each other, but closely intertwined, and an important task in the analysis
of joint agency is to find out more about the precise nature of the interrelations at
stake here. Some characteristics seem rather obvious: if an individual intends his or
her contribution as his or her contribution to a joint venture, she or he cannot be
completely indifferent as to what the other contributors do, and how successful they
are likely to be, because the success of her contribution, and even the question of
what it really is she is doing, depends on the relevant others’ actions. Participatory
intentionality presupposes certain attitudes towards other people’s intentions and
actions. I cannot intend to score the match-winning goal if I know nobody is play-
ing in a way similar to the way in which I cannot intend to run one hundred miles
if I know I couldn’t possibly make it past thirty. I might imagine me scoring the
decisive goal while playing around with a ball on an empty field. But this would
be doing something while pretending it to be something else, just as in the case in
which one imagines to have ninety of the hundred miles completed and intends to
run the remaining miles.
In the case of joint action, the sense of ability is therefore, in part, an interpersonal
relation. Now let us have a quick look at one of the most prominent analyses of
what it means to intend to do one’s part in a collective venture, in order to learn
more about how the interpersonal attitudes at work in the background of collective
intentions are usually conceived of, and how this fits to our above consideration
concerning the role and structure of the sense of ability.
Here is Raimo Tuomela and Miller’s (1988) and Tuomela’s 2007) analysis of
we-intention:
232 H.B. Schmid
(WI) A member A of a collective g we-intends to do X if and only if

(i) A intends to do his part of X (as his part of X);
(ii) A has a belief to the effect that the joint action opportunities for an intentional
performance of X will obtain (or at least probably will obtain), especially that
a right number of the full-fledged and adequately informed members of g, as
required for the performance of X, will (or at least probably will) perform
their parts of X, which under normal conditions will result in an intentional
joint performance of X by the participants;
(iii) A believes that there is (or will be) a mutual belief among the participating
members of g (or at least among those participants who perform their parts of
X intentionally as their parts of X there is or will be a mutual belief) to the
effect that the joint action opportunities for an intentional performance of X
will obtain (or at least probably will obtain);
(iv) (i) in part because of (ii) and (iii).
As is obvious from the quote, Tuomela agrees that we-intention requires some sense
of collective ability. But he conceives of this attitude purely in terms of belief, and
the belief requirement is rather strong: as is evident from clauses ii and iii, where
Tuomela mentions beliefs concerning the action opportunities and the contributions
of the other group members, A has to believe that they, together, will actually do the
joint action to which he or she intends to do his or her part.
Remember the reason mentioned above for rejecting Grice’s and Velleman’s
claim that in order to intend to ϕ individually one has to believe that one will ϕ:
it is that intention is compatible with any degree of self-doubt, and that only the
weakest version of the limitation claim seems plausible. The weakest version of the
limitation claim states that in order to intend to ϕ, an agent cannot believe that it is
impossible for him to ϕ. Now it seems plausible to assume that the same argument
can be made in the collective case. In order to do one’s part in a collective venture,
one does not have to assume that anything will come off it; any degree of (collec-
tive) self-doubt is compatible with participatory intention, as long as it is self-doubt,
and not certainty of inability.
The question, however, is: are the cases strictly parallel? Do beliefs concerning
the limits of one’s collective abilities exert a similar rational pressure on the range
of one’s participatory intentions in the same way as they do on the formation on
one’s purely individual action? Consider agent A in the following two situations.
First, A intends to ϕ while believing that she is unable to ϕ (remember the examples
discussed above in Section 12.2). Second, A, as a member of group G, intends to
do her part in G’s ϕ-ing while believing that G cannot ϕ. An example for this latter
case might be construed of Tuomela’s (1995, 235ff.) case of a group of people push-
ing a broke-down bus. Assume that the group’s shared intention is to move the bus
out of the lane, that the bus is quite large, and that the group is very small, perhaps
consisting of just two persons. In this case, there is evidence for the participating
individuals to believe that the bus cannot be moved out of the lane, even if they
act jointly. At first sight, it might seem that there are some differences between the
individual aiming at a subjectively impossible individual goal, and the individual
intending to do her part in what she takes to be a collective venture. The collective
case, it might seem, isn’t subject to the same degree of irrationality as the individ-
ual is. After all, A might just intend to do his part, i.e. to push as hard as he can
(perhaps with the aim of not letting the other down, who isn’t convinced of their
collective inability, and still believes that they together can do it). Thus it seems
that the agent’s beliefs concerning collective inability do not (rationally) limit their
participatory intention. But upon closer consideration, it becomes apparent that this
line of reasoning is wrong. If A’s intention is simply to push as hard as he can,
he does not thereby intend to move the bus out of the lane together with the other
person, but simply intends to do what would be her contribution to their collective
action, if that action were to come into being. In other words, he intends her part de
re rather than de dicto. In his analysis of we-intention, however, Tuomela has made
it clear that for an individual to we-intend, she has to intend to do her part as her
part (cf. clause i. in the above analysis); for her to intend her contribution as her
part, however, she cannot believe that there is no whole to which her part is a part.
In other words: if A intends to push, without assuming that she, together with the
other group member, will thereby move the bus out of the lane, she is in the exact
same situation as BenjaminR , who makes an effort at the closed lid, but does not
intend to open the jar.
Thus we have to distinguish cases in which people simply throw in their efforts
independently of what they expect to come off it from cases in which people gen-
uinely share an intention and make their contribution. The first case is parallel to
individual tryings in terms of proper actions. The question to be addressed here
is: is it possible to share an intention (as opposed to simply throwing in one’s
forces) while believing that the jointly intended action cannot possibly be carried
out?
I assume that anyone who has ever been involved in university administration
will be quite familiar with that kind of experience. One does not have to believe
that we will ϕ in order to intend to do one’s part of ϕ as one’s part, i.e. to share an
intention. Indeed, the experience of taking part in a collective venture which one is
convinced is doomed to failure is even more frequent than the experience of intend-
ing to so something one knows in one’s head one will not be able to carry out. Just
as their individual parallels, such cases are, to some degree at least, irrational cases.
They involve a degree of cognitive inconsistency, or of practical incoherence. But
as they are not conceptually impossible, an adequate analysis of collective inten-
tionality should be able to accommodate such cases. This is obviously not the case
in Tuomela’s analysis, because Tuomela demands of the participants in joint action
to believe that the collective action opportunities hold, and that they, together, will
perform the action in question (cf. ii. and iii. in the analysis above).
If it is possible for agents to share an intention, which they believe to be impos-
sible to carry out, the basic mode of how participants in joint action are interrelated
cannot simply be belief. They have to be interrelated in a way that leads them to take
themselves to be able to ϕ even against their beliefs concerning what they are able
to do. So what, then, is the basic mode of interpersonal relation at work in collective
intention, if not belief?
234 H.B. Schmid
Let me just mention two features to be found in competing accounts of collective

intentionality, which sound promising with regard to the issue at stake here (I will
not address the weaknesses of these accounts). Michael Bratman has emphasized
that in joint action, the intentionality of the participants is intertwined in that each
somehow “aims at the efficacy of the intention of the other” (1999, 124). “Aiming at
the efficacy” of another’s intention is obviously a different attitude from simply hav-
ing a belief that the other will carry out his intention; it involves some commitment
to support, some disposition to make the other’s aim one’s own. Margaret Gilbert
(1996) proposes a striking alternative to Tuomela’s account in claiming that the
basic attitudes towards the other participant’s actions are normative rather than cog-
nitive. In her analysis, she claims that for two agents jointly to intend to ϕ it has to
be true that they are obliged to each other to perform their respective parts, and that
they are entitled to rebuke the other if he fails to deliver. It has been pointed out (by
Sugden [2000], among others), that this requirement is too strong. There are many
cases where agents act jointly, without there being proper obligations involved in
the case. But Gilbert seems to be right in pointing out that the inter-individual atti-
tudes involved in joint action cannot be purely cognitive. So the question is: how
are we to understand an attitude that is not purely cognitive, even though it involves
cognitive elements, and not purely normative, even though there is some element of
normativity involved?
Again, affectivity seems to offer a convincing alternative. I propose to conceive
of the basic mode of how agents relate to the other participant’s abilities in the same
way as they relate to their own in the individual case. I have argued above that the
sense of one’s own abilities is basically affective; it is a matter of self-confidence,
conceived of in affective terms so as to allow for the possibility of recalcitrance.
Affective confidence is the attitude that determines the class of possible objects of
intentions, and serves as a basic guiding line for agency, the sense of reality, and the
sense of self. Concerning the case of joint agency, I propose to see the basic inter-
individual attitude involved as matter of trust, conceived of in affective terms. If trust
is basic to joint action, and joint action is basic to sociality, this proposal assigns
trust a fundamental role in the social world. Of course, this proposal goes beyond
the views to be found in most of the received literature on trust, where trust is often
seen as a remedy against collective action problems rather than the stuff of which
social relations, as such, are made. But my proposal is not entirely idiosyncratic,
either. It conjoins two received views in the literature on trust, combining Martin
Hollis’ (1998) insight that trust is the stuff of which even the most rudimentary
forms of interactions are made,5 with Karen Jones’ (1996) insight that trust should
be seen as an affective attitude.
To sum up: I have argued that some conceptual puzzles about the sense of abil-
ity at work in the background of intention can be resolved if we conceive of it in
affective terms. Action requires not only volition and cognition, it also requires an
affective element. In order to form an intention, one has to feel up to its object. I
5 This is a point taken up by Robert Sugden in his analysis of team trust (Sugden 2000).
have argued that the affective attitude at play here is basically a matter of self-trust
and confidence, and that in collective action, interpersonal trust or confidence plays
a similar role.
Acknowledgements I presented earlier versions of this paper on conferences and workshops at
the universities of Basel, Berkeley, and Bielefeld. I would like to thank the audiences there, and
especially Kevin Mulligan and Kirk Ludwig.
References
Adams, F. 1995. “Trying: You’ve got to believe.” Journal of Philosophical Research XX:549–61.
Baier, A. C. 1970. “Act and Intent.” Journal of Philosophy 67:648–58.
Bandura, A. 1997. Self-Efficacy. The Exercise of Control. New York: WH Freeman.
Bandura, A. 2000. “Exercise of Human Agency through Collective Efficacy.” Current Directions
in Psychological Science 9 (3):75–78.
Bratman, M. E. 1987. Intentions, Plans, and Practical Reasons. Cambridge: Cambridge University
Press.
Bratman, M. E. 1999. Faces of Intention: Selected Essays on Intention and Agency. Cambridge,
MA: Harvard University Press.
Cohen, G. A. 1989. “On the Currency of Egalitarian Justice.” Ethics 99 (4):906–44.
Darwin, Ch. R. 1872. The Expression of the Emotions in Man and Animals. London: John Murray.
Ehrsson, H. H. et al. 2004. “That’s My Hand! Activity in Premotor Cortex Reflects Feeling of
Ownership of a Limb.” Science 305:875–877.
Falvey, K. 2000. “Knowledge in Intention.” Philosophical Studies 99:21–44.
Gilbert, M. 1996. Living Together: Rationality, Sociality, and Obligation. Lanham, MD: Rowman
and Littlefield.
Graham, K. 1977. J.L. Austin – A Critique of Ordinary Language Philosophy. Stanford Terrace:
Spiers.
Grice, P. 1971. Intention and Uncertainty. British Academy Lecture. Oxford: Oxford University
Press.
Hacker, P. M. S. 2007. Human Nature: The Categorical Framework. Oxford: Blackwell.
Harman, G. 1986. Change in View. Cambridge, MA: MIT Press.
Hollis, M. 1998. Trust Within Reason. Cambridge: Cambridge University Press.
Hornsby, J. 1980. Actions. London: Routledge.
Hume, D. [1739/1740] 2000. In A Treatise of Human Nature, edited by D. F. Norton and M. J.
Norton. Oxford: Oxford University Press.
Jones, K. 1996. “Trust as an Affective Attitude.” Ethics 107 (1):4–25.
Kenny, A. 1976. “Human Abilities and Dynamic Modalities.” In Essays on Explanation and
Understanding. Studies in the Foundations of Humanities and Social Sciences, edited by J.
Manninen and R. Tuomela, 209–32. Dordrecht: Reidel.
Kenny, A. 1989. The Metaphysics of the Mind. Oxford: Oxford University Press.
Kolodny, N. 2008. “The Myth of Practical Consistency.” European Journal of Philosophy 16
(3):366–402.
Löwenstein, K. K. 1911. “Über den Akt des „Könnens“.” Zeitschrift für wissenschaftliche
pädagogische Psychologie und experimentelle Pädagogik XII:403–20.
Ludwig, K. 1992. “Impossible Doings.” Philosophical Studies 65:257–81.
Ludwig, K. 1995. “Trying the Impossible. Reply to Adams.” Journal of Philosophical Research
XX:563ff.
Marcel, A. 2003. “The Sense of Agency. Awareness and Ownership of Action.” In Agency and Self-
Awareness, edited by J. Roessler and N. Eilan, 48–93. Oxford; New York: Oxford University
Press.
236 H.B. Schmid
Mele, A. R. 1989. “She Intends to Try.” Philosophical Studies 55 (1):101–06.

Mele, A. R. 1992. Springs of Action. Oxford: Oxford University Press.
Millikan, R. G. 2000. On Clear and Confused Ideas. An Essay about Substance Concepts.
Morriss, P. 2002. Power. A philosophical Analysis. Manchester: Manchester University Press.
Nahmias, E. 2005. “Agency, Authorship and Illusion.” Consciousness and Cognition 14:771–85.
Oddie, G., and P. Tichy. 1982. “The Logic of Ability, Freedom, and Responsibility.” Studia Logica
41:227–48.
Oddie, G., and P. Tichy. 1983. “Ability and freedom.” American Philosophical Quarterly 20:
135–47.
Pacherie, E. 2008. “The Phenomenology of Action. A Conceptual Framework.” Cognition
107:179–217.
Ratcliffe, M. 2005. “The Feeling of Being.” Journal of Consciousness Studies 12 (8–10):43–60.
Ratcliffe, M. 2008. Feelings of Being. Phenomenology, Psychiatry, and the Sense of Reality.
Oxford; New York: Oxford University Press.
Searle, J. 1983. Intentionality. Cambridge: Cambridge University Press.
Sellars, W. 1992. Science and Metaphysics. Variations on Kantian Themes. Atascadero: Ridgeview.
Sosa, E. 1993. “Proper Functionalism and Virtue Epistemology.” Nous 27 (1):51–65.
Sugden, R. 2000. “Team Preferences.” Economics and Philosophy 16:175–204.
Tuomela, R. 1995. The Importance of Us. A Philosophical Study of Basic Social Notions. Stanford:
Stanford University Press.
Tuomela, R. 2007. Philosophy of Sociality. The Shared Point of View. Oxford; New York: Oxford
University Press.
Tuomela, R., and K. Miller. 1988. “We-Intention.” Philosophical Studies 53:367–89.
Velleman, D. J. 1989. Practical Reflection. Princeton: Princeton University Press.
Vermazen, B. 1993. “Objects of Intention.” Philosophical Studies 71:223–65.
Wallace, J. R. 2001. “Normativity, Commitment, and Instrumental Reason.” Philosophers’
Imprint 1.
Chapter 13
Self-Evaluation in Intention: Individual
and Shared
Lilian O’Brien
13.1 Introduction
Intention has long been at the forefront of discussions in the philosophical
psychology of intentional action. In spite of its prominence, aspects of it have been
neglected. For example, its role in the phenomenology of ability, as discussed by
Hans Bernhard Schmidt, Chapter 12, in this volume, is rarely treated. I focus on
a different but similarly neglected aspect in this paper. This is the self-evaluative
aspect of intention. I argue that in intention the agent takes an evaluative attitude
towards herself. This attitude of self-evaluation is best understood in terms of what
John Broome has called normative requirements.
With these preliminaries in place, the main focus of my discussion shifts from
individual to shared intention. My aim is to draw on what I say about self-evaluation
in individual intention to shed light on shared intention. In discussions of shared
intention, there has been considerable controversy over whether shared intention
alone generates obligations for the participants to one another. Michael Bratman
and Margaret Gilbert have taken opposing positions on this issue. They also dis-
agree about whether or not shared intention can be understood in terms of personal
intentions – intentions that can be formed and rescinded unilaterally – and whether
we must accept what Gilbert calls “plural subjects” in our ontology if we are to char-
acterize what it is to share an intention. In this paper, I will focus on the first dispute.
Drawing on self-evaluation and its role in normative requirements, I will argue for
a middle way between their views on the relationship between shared intention and
obligation.
In Section 13.2, I characterize some of the first-personal aspects of intention in
an attempt to motivate what I call the Self-Evaluation Condition. In Section 13.3,
I turn from individual to shared intention, and offer a brief account of the debate
about whether shared intention involves the imposition of obligations on the partners
to one another. I offer a criticism of Bratman’s view of shared intention, and also
L. O’Brien (B)
Department of Philosophy, University College Cork, Cork, Ireland
e-mail: l.obrien@ucc.ie
238 L. O’Brien
discuss aspects of Gilbert’s opposing view. Finally, I draw on the Self-Evaluation

Condition of Section 13.2 to present my middle way between the views of Bratman
and Gilbert.
13.2 Self-Evaluation and Normative Requirements

Intention, like promising, appears to be a species of commitment – specifically, it
is a commitment by an agent to undertake some action.1 As Michael Bratman has
persuasively argued, intentions settle matters for the agent and are stable.2 They
settle matters in that they resolve for the agent the problem of what she is to do. They
are stable insofar as although they are not immune to revision, they are resistant
to it. Moreover, once formed, intention imposes constraints on what it is rational
for an agent to deliberate about, intend, and do. If you form the intention to move
to the USA, then this places a rational demand on you to foreclose deliberation
about whether to move to Mexico. Intentions, Bratman argues, play an important
role in planning agency. These features reflect the idea that intention is a species of
commitment. Commitment has a normative force that a mere desire does not have,
but that something like a promise does have – once in existence, it tends towards
shaping deliberation and action rather then being shaped by them.
It is worth thinking about intention not only in terms of its functional role in the
agent’s planning life, but also as it is from her first-person perspective. These are two
sides of a coin, and exploration of both allows us to enrich our understanding. As
intention involves the agent in a commitment to act, it stands to reason that she takes
some sort of positive attitude to the action that she is committed to performing or
the state of affairs that she intends to bring about. However, it also stands to reason
that she takes some sort of attitude towards herself. After all, she commits to acting
in a particular way, and thus, plausibly holds herself to the standards involved in
tokening the relevant act-type. She would seem, then, to regard herself as subject to
evaluation depending on whether or not she performs the relevant action. In forming
the intention to drive to town, for example, the agent regards herself as evaluable as
a success or a failure with regard to whether or not she drives to town. This attitude
of self-evaluation runs as follows:
The Self-Evaluation Condition: in forming an intention3 to do X, the agent

thereby regards herself as evaluable as a success or a failure depending on
whether or not she does X.
1 What I say about intention is not uncontentious and departs from what, for example, Gilbert
(2006) says about it. However, my view is consistent with that of others e.g. Broome (2001), and
as I do not propose to develop a theory of intention here, I will not offer a full discussion the ways
in which my view differs from the views of others.
2 See Bratman (1999a).
3 This applies to either intention for the future or intention in action.
13 Self-Evaluation in Intention: Individual and Shared 239
It is not that the agent must actually go through the process of evaluating herself if
she is to have an intention (Although this will be helpful in certain circumstances –
for example, in keeping her action on course and in recognizing whether or not she
has completed it.) What is essential is that, in forming the intention, she regards
herself as subject to evaluation depending on whether she performs the action that
is intended.
This applies no less to intention in action. Consider an agent who, in raising
her hand, has the intention of attracting the attention of a passerby. Her intention
in action involves her in assigning a task to herself in her raising her hand, and
thus, holds herself to the standards involved therein. In taking this attitude towards
herself, she regards herself as amenable to failure or success depending on whether
or not she attracts the attention of the passerby.
Let us try to say a little more about the content of the evaluative attitude involved
in intention. There are two ways in which one can succeed or fail, but only one,
the first, is essential to intention. The first way has to do with the content of the
I-thought that is involved in intention, the second has to do with the reasons that the
agent thinks that she has for performing the intentional action in question. I will call
the first component “I-thought” and the second “Reasons”.
Intentions are de se thoughts, such as, “I will drive to town” or “I will eat dinner
in an hour”. With respect to I-thought, success as an agent involves complying with
the contents of the I-thought involved in intention: (i) tokening the act-type that is
referred to in the intention in the conditions specified (e.g. driving to town or eating
one’s dinner at the relevant time), where (ii) the tokening of the act-type is something
that is produced by the agent herself and for which she can be held responsible (she
who is referred to by the de se thought).
Failure as an agent with respect to (i) is failure to token the relevant act-type or
do so in the conditions you intend. Suppose that you intend to go to the cinema
tomorrow afternoon. Failure could involve simply lying in bed all day without hav-
ing revised one’s intention to go to the cinema on rational grounds. Alternatively, it
might involve walking to the cinema but finding it shut.
Failure with respect to (ii) has become something of a preoccupation of philoso-
phers of action and pertains to the metaphysical conditions that must obtain if an
intentional action is to be performed. It is possible to fail to have the tokening of
an act-type be appropriately up to you, as the many deviance cases in the literature
show. One can succeed or fail, relative to I-thought if, for example, nervousness
intervenes between your intention and your intended movement, such that it is not
true of you that you are responsible for your movement. This is another way in
which one can fail relative to I-thought, but these cases will not concern us here.
The second way in which you can succeed or fail, Reasons, pertains to the reason
or reasons for which the agent acts. If the agent has reasons for her intentional action,
she regards herself as amenable to success or failure in light of those reasons.4 For
4 One may be able to intend or act without having reasons for intending or acting. For example, it
may be that I can intend to eat a saucer of mud without thinking that anything favours this, and
240 L. O’Brien
example, the reason that you go to the cinema is to be a good friend to someone
who needs company. In going to the cinema or failing to go, you regard yourself as
amenable to success or failure as a friend. It is important to note that our primary
concern is with the first component, for this first component alone is essential to
intention. This is because it is possible to form an intention to act without having
reasons for the intention or the action.
One troubling thing about the Self-Evaluation Condition is that we sometimes
feel indifferent to our intentional actions, and indeed, we may even revile them. Is
it coherent to say that even in those cases the agent regards herself as evaluable as
a success? I think that this is possible if success has to do with the first of the two
components mentioned, I-thought. In a case of akratic or addicted action where the
agent thinks of her intentional action as the wrong thing to do, she may still regard
her intentional tokening of an act-type as her successful completion of a task that she
has set herself. For example, if she forms the intention to get a drink, she succeeds if
she herself tokens the relevant act-type. “Success” in this context does not, however,
imply any overall positive evaluation by the agent of her intentional action or of
herself. Consequently, it is possible to regard oneself as positively evaluable for X-
ing – a success in that regard – while not overall approving of one’s action of X-ing
or oneself for X-ing.
This raises the question of what kind of evaluation an agent makes of herself
when she judges herself a success or failure with respect to I-thought. I believe
that this should be understood in terms of what John Broome calls “normative
requirements”.5 He says,
. . . a large part of rationality consists in conforming to normative requirements, and is not
concerned with reasons at all. For instance, one part of rationality is doing what you believe
you ought to do, and this does not necessarily mean acting for reasons.6
An agent may not have reason to believe or intend to do something, and yet, she may
be subject to a normative requirement to believe or do that thing because of her prior
epistemic and practical commitments. For example, suppose p implies q, and that S
believes both p, and that p implies q. Intuitively, there would be something defective
about S as a rational agent if she believes these things, but doesn’t believe q. This
is so even if there is no reason for S to believe q. In the case of intention, we may
judge of an agent that she has no reason to do something, X, but we nevertheless
think that there is something defective about her as rational practical agent if she
does not do X or attempt to X, once she has formed the intention to do so.
then act on my intention. In this case, one judges oneself in light of the first element, I-thought,
but not the second, Reason. Such intentions are puzzling, not only to the bystander, but to the
intender herself. Insofar as I do not have a reason for what I’m doing, I don’t know why I’m
doing it, and so, am to some extent rationally incomprehensible to myself (and others). Plausibly,
the rational practical agent is characterized by the fact that her intentional actions are rationally
comprehensible to her. Acting without rationally comprehending why one acts is, necessarily, an
anomalous activity of rational practical agents. See Kim (1998).
5 See Broome (1999, 2001).
6 Broome (1999, 410).
This evaluation of agents from the third-person perspective of the onlooker is

mirrored in the first-person perspective of the agent on her action. In fact, there is a
strong case to be made that the third-person evaluation depends on the first-person
evaluation, but I set that aside here. From the first-person perspective, although the
agent is likely to think that she has reason to do what she does, she also regards
herself as subject to evaluation depending on whether or not she carries out what
she has committed to carrying out. Being committed in the form of an intention
involves, from the first-person perspective, a judgment that one is evaluable as a
failure if one does not fulfill one’s intention. That is what it is to have an intention, a
commitment, to doing something. I intend the Self-Evaluation Condition to pertain
to the agent’s regarding herself as subject to evaluation in terms of what she is nor-
matively required to do, given the content of the I-thought involved in her intention
to act.
An important advantage of what I have said about the Self-Evaluation condition
is that it is consistent with the features of intention that Michael Bratman describes
in his theory of intention and planning agency. When one arrives at the judgment
that one would be a failure were one not to do something, one has settled the ques-
tion of whether one should do it in quite definitive terms. Similarly, the relationship
between self-evaluation and stability is not difficult to discern: intentions are sta-
ble, that is, resistant to revision, because revision requires rationally challenging the
judgment that one would be a failure were one not to comply with one’s intention.
Moreover, because arbitrary abandonment of intentions without revision norma-
tively requires that the agent judge that she is a failure as an agent, rational revision
is preferable, but potentially arduous. The fact that intention involves the possibil-
ity of failure plausibly contributes to its role in shaping deliberation: its normative
status makes it eligible for being accorded weight in deliberation.
Having looked at individual intention, focussing on how it is from the first-person
perspective of the agent, I will turn in the next section to a discussion of shared
intention. As aforementioned, my focus will be on the debate about whether shared
intention alone generates obligations on the participants to one another to partici-
pate in the actions that they intend to carry out. I will focus on the opposing views
of Bratman and Gilbert, offering a criticism of both. In the course of this critical
discussion, I will formulate a necessary condition on shared intention, which I call
the Evaluative Dependence Condition. This condition draws on the Self-Evaluation
Condition. The Evaluative Dependence Condition maintains that in shared intention,
the participants take the interests of their co-intenders into account in the conditions
in which the participants judge themselves a success or failure. It is not just a plau-
sible necessary condition on shared intention, it allows us, I argue, to avoid the
excessive permissiveness of Bratman’s view on shared intention. It also allows us
to explain some of the phenomena that are associated with shared intention that are
focused on by Gilbert. It does this, however, without committing us to some of what
Gilbert has said about the relationship between obligations and shared intention. In
short, I maintain that it offers us a middle way between their opposing positions on
the relationship between shared intention and obligation.
242 L. O’Brien
13.3 Shared Intention

In this section, I will be exclusively concerned with shared intention in cases of what
Bratman calls “modest sociality”:
Cases of modest sociality are cases of small scale shared intentional agency in the absence
of asymmetric authority relations. Examples include: our singing a duet together, but not the
activities of an orchestra with a conductor; our going to NYC together, or (to use Margaret
Gilbert’s example7 ) our walking together, but not a school trip led by a teacher . . . 8
Like others in the literature, I am assuming that individual and shared intentions will
have the same core features: they will have the features of stability that Bratman has
discussed, and they will foreclose possibilities and shape deliberation. I am also
assuming that they will satisfy the Self-Evaluation Condition.
Michael Bratman’s theory of shared intention relies on the idea that shared inten-
tion is a matter of the appropriate interlocking of individually-held intentions. These
individually-held intentions have relevantly similar content, and stand in complex
relationships to one another, thereby constituting a shared intention:
We intend to J if and only if
(1) (a) I intend that we J and (b) you intend that we J.

(2) I intend that we J in accordance with and because of (1)(a), (1)(b), and mesh-
ing subplans of (1)(a) and (1)(b); you intend that we J in accordance with and
because of (1)(a), (1)(b), and meshing subplans of (1)(a) and (1)(b).
(3) (1) and (2) are common knowledge between us.9
In Bratman’s account, there is no reason to think that shared intention generates

nonconditional obligations of the co-intenders to one another, that is, obligations
that derive from the shared intention alone. There is a burden on Bratman to explain
how it is that shared intention and shared activity is often accompanied by mutual
obligation. He goes on to explain how this is possible by appeal to the creation of
expectations in the partners to shared intention:
My conjecture, then, is that the typical etiology of yours and my shared intention involves
the intentional, purposive creation by me of expectations on your part about what I will do,
and vice versa. When you and I have a shared intention to sing the duet, it is likely that I
will have purposively led you to expect that I will participate if you do; that you will have
purposively led me to an analogous expectation; and that each of us wants such grounds for
such expectations.10
In her work on shared intention, Margaret Gilbert has argued that shared inten-
tion requires the existence of a “plural subject”, and that shared intention generates
7 See Gilbert (1990).

8 Bratman (2009). See also Gilbert’s clarification of shared intention in (2009).
9 Bratman (1993). Bratman offers further developments of this view in later work. See, for example,
chap. 8 (1999b), chap. 13 (2007, 2009).
10 Bratman (1997a).
nonconditional obligations on the partners to one another.11 On her view, if one

abandons a shared intention, one does not simply disappoint reasonable expecta-
tions of one’s partner, one violates a (non-moral) obligation that one has to the other.
She formulates the following “obligation criterion” for shared intention:
An adequate account of shared intention will entail that each party to a shared intention is
obligated to each to act as appropriate to the shared intention in conjunction with the rest.12
Both sides of the dispute about whether or not shared intention generates obligation
have appeal. If we think that it is possible to share an intention where a certain
amount of coercion is involved, then we may be wary of the idea that the mere
sharing of an intention with another could generate an obligation to her. Similarly, if
the shared intention were to perform morally questionable acts, we may be reluctant
to think of this as a situation in which there are obligations on the partners to one
another.13 Finally, if we believe that obligations provide reasons to act, then the view
that shared intention generates obligations is committed to the claim that intentions
taken alone give reasons to act, but this seems to involve bootstrapping, and so, is
counter-intuitive.14
On the other hand, Gilbert’s claims resonate with the fact that in shared intention,
we place our agential success in the hands of another. Surely, we may think, a defec-
tor from a shared intention does not merely disappoint expectations generated by her
behaviour, rather, she violates some norm that is grounded in the shared intention
alone. Also, if there is a sense of obligation in which it is clear that boot-strapping is
avoided, but the genuine bonds of sociality are nevertheless captured, this, it might
be thought, is the correct way to characterize shared intention.
First, I will critically discuss a putative counterexample to Gilbert’s thesis of
non-conditional obligation that is offered by Bratman. I will argue that it is not a
case of shared intention (Section 13.3.1). Second, I will use this discussion to argue
for a necessary condition on shared intention that, at least, goes beyond his view
in one respect (Section 13.3.2). This makes use of the Self-Evaluation Condition of
Section 13.2, and entails that would-be defectors from shared intention are norma-
tively required to rationally assess the cost to their partner of their defection from
shared intention. This is the Evaluative Dependence Condition mentioned at the
end of Section 13.2. Third, I argue that this necessary condition fits with some
of the appeal of Gilbert’s side of the dispute, and helps to explain some of the
11 See Gilbert (1993, 2006, 2009).

12 See Gilbert (2009, 175).
13 See Bratman (1999a).
14 Gilbert (1993, 686) says that obligations are reasons to act. In her (2009) she says “I take it
that given either a standing, unrescinded decision or a current intention, the person in question
has sufficient reason to act in a particular way even without this having been the case prior to
the formation of the intention or decision. In saying that one has sufficient reason to act in some
way I mean that, if all else is equal, one ought so to act. The ‘ought’ here is a matter of what
might be referred to as a rational requirement . . . I do not mean to imply that either decisions or
intentions are ‘reasons’ . . . considerations for or against a particular action that would appropriately
be weighed prior to making a decision whether or not to perform it” (179–80).
244 L. O’Brien
phenomena that she cites to motivate her view about the relationship between shared
intention and obligation (Section 13.3.3). It does not, however, support her view. In
effect, drawing on the Self-Evaluation Condition and spelling this out in terms of
John Broome’s concept of normative requirement, I argue for a middle way between
their views that captures some of the intuitive appeal of both.
13.3.1 A Counterexample to the Thesis of Non-conditional

Obligation
Bratman objects to Gilbert’s thesis that shared intention alone generates obliga-
tion by presenting cases of shared intention and shared intentional action that do
not involve the partners in having obligations to one another. In one example, I
threaten to blow your house up unless you sing a duet with me, and we sing the duet
together. In another case, we each have the intention to sing a duet together but we
also announce our intention to stop the shared activity at any time. In such cases,
Bratman maintains, we have genuine shared intentions but they are only contin-
gently related to mutual obligations of the partners to follow through on that shared
intention and intentional action. They can abandon the intention without failing to
fulfil an obligation to their partner. I will take a closer look at the first case, arguing
that, contrary to what Bratman says, it falls short of shared intention.
In the first case, one agent coerces another to sing a duet with her by threatening
to blow up her house if she doesn’t. We can suppose that, ultimately, the conditions
in which the coerced agent, let’s call her Corina, regards herself as a success as an
agent are conditions in which she prevents her house from being blown up. In the
medium term, with no satisfactory alternatives in sight, success for Corina is singing
a duet with the would-be house-bomber, let’s call her Betty. We can imagine Corina
forming the intention: “I intend that we sing” where this intention is implicitly con-
ditional: as long as a better means to saving her house does not emerge, she regards
herself as evaluable as a failure if she doesn’t sing the duet. We can safely sup-
pose that Betty’s intention (“I intend that we sing”) is not conditional in this way.
We can also suppose that Betty and Corina satisfy the other conditions on shared
intention that Bratman offers. Corina formulates sub-plans with Betty about how to
go about the duet-singing and co-ordinate their activity. These sub-plans are mutual
knowledge between them. Nevertheless, if Corina finds a better means to her goal
of saving her house and defects in such a scenario, it seems that she abandons a
shared intention. Moreover, as Bratman maintains, it seems that she doesn’t thereby
violate an obligation that she has to Betty. Consequently, it seems that shared inten-
tion doesn’t entail obligations on the co-intenders to one another to participate in
the actions that they had intended.
Let’s take a closer look at the case. We can suppose that Corina discovers an alter-
native means to protecting her house midway through the duet, and she stops singing
the duet and abandons the shared intention. It does not seem that Corina’s abandon-
ment normatively requires her to rationally revise her individually-held intention.
It also does not seem that she is rationally criticizable for terminating the singing.
Taking what we said about intention in Section 13.2 into account, this indicates that
she is not abandoning an intention. And in fact, this makes sense: her abandonment
appears to be rationally licensed by her conditional individually-held intention. She
would be violating the normative requirements imposed by this intention if she did
not abandon the activity just as soon as she found alternative means to her end.
One worry is that if she is not abandoning an intention, then this case does not
provide an appropriate test of the hypothesis of nonconditional obligation. To test
this, we need genuine abandonment of a shared intention that does not involve the
violation of an obligation – there must be violation of the shared intention without
violation of an obligation.
Of course, Corina ceases to satisfy the conditions on shared intention that are
presented in Bratman’s account. The dependence of her intention and activity on
the intentions and activity of Betty ceases. This, it might be pointed out, is just
what abandoning a shared intention involves, and it is question-begging to sup-
pose otherwise. But is it not counter-intuitive to suppose that Corina can abandon a
shared intention without being at risk of violating any normative requirement that is
imposed on her by the shared nature of the intention? There is no normative require-
ment on Corina to avoid her defection because of the mutual dependence between
her and Betty. Nor is there any rational pressure on her to consider her defection
in light of Betty’s interests or their mutual dependence. In short, the fact of mutual
dependence appears to impose no normative requirement on her when it comes to
her defection. Given this, although it seems correct to say that Corina violates no
normative requirement in defecting, this strikes me as a reason to deny that Betty
and Corina really share an intention.
Supporting this intuition is the fact that if shared intention is a case of intention,
then it should have the features that individual intention has, such as, stability.15 The
stability at issue here is the stability that derives from resistance to re-consideration.
The stability of individual intention allows it to play, as we have seen, a signif-
icant role in planning agency: intention tends to shape the deliberative landscape
rather than be shaped by it. If Corina and Betty’s individually-held intentions con-
tinue to interlock in the kind of relationships described by Bratman – for example,
they have meshing sub-plans that concern the duet-singing and mutual knowledge
of those sub-plans – it is only because circumstances allow that. Specifically, it is
only because these circumstances do not affect the conditions on the individually-
held intentions that the agents have (i.e. “I intend that we sing a duet”). Corina, for
example, may not find an alternative means to her end of self-protection, and so,
the conditions in which her intention to sing a duet no longer apply to her do not
eventuate. Corina’s individually-held intention may be stable, but the shared nature
of the intention is not obviously stable. Perhaps it is tempting to infer that because
the parts of the shared intention have stability, the whole has stability. However,
15 Gilbert
also thinks that Bratman’s view fails to provide stability. See her (2009). See also Gold
and Sugden (2007).
246 L. O’Brien
such an inference is invalid. The individual parts of a wall may all be cubes (e.g.
the bricks) but the wall may not be a cube. In fact, in the case of shared intention, at
least as Bratman characterizes it, it seems that there is good reason to think that the
stability of the individual intentions constituting what is, putatively, a shared inten-
tion, will not accrue to the shared intention. This is because the agents who are party
to the putatively shared intention may have starkly diverging overall goals. This is
so even if they both have intentions that are superficially similar (i.e. “I intend that
we J”) that fit Bratman’s conditions and if they have meshing subplans of which
there is mutual knowledge. Because of the possibility of diverging overall goals, the
normative requirements to which each is individually subject may run contrary to
the “stability” of the interlocking structure of which shared intention is putatively
constituted. I think that we should conclude that if we accept Bratman’s theory of
shared intention, it does not necessarily have stability.
However, the stability of intention is arguably a defining feature of it, so without
stability, it’s not clear that we should accept Bratman’s conditions as a theory of
shared intention. Indeed, it seems more appropriate to regard the co-ordinated and
mutually dependent intentions of Corina and Betty as a case of shared activity that
does not amount to a shared intention.
Actual world cases of modest sociality are diverse, and there are cases that are
so different from that of Corina and Betty that they seem to merit being placed in a
different category. For example, there can be duet-singers who have a sense of duty
of some sort to their partner. In these cases, there is shared activity where the mutual
dependence of the co-intenders on one another is taken to have some normative force
for participants, and which, in turn, generates stability. The symmetry between these
cases and cases of individual intention makes them well suited to being classed
together, where cases like Betty and Corina are not well suited to being classed in
that way. Because Bratman’s view generates an asymmetry between individual and
shared intention, where shared intention does not have what is plausibly understood
as a defining feature of intention, I think that we have good reason to suspect that the
view does not quite carve the social world at the joints. It is plausible to suppose that
we can regard Betty and Corina as a case of co-ordinated and mutually dependent
intentions that do not amount to a shared intention – they share activity without
sharing an intention.16
Bolstering the case for this claim requires saying in positive terms what shared
intention would involve that would distinguish it from mere shared activity. Let’s
turn to that now.
16 Itis not that Bratman is opposed to sub-categories within modest sociality. He distinguishes
between shared intentional activity and shared cooperative activity. “Shared intentional activity . . .
is activity suitably explainable by a shared intention and associate forms of mutual responsive-
ness. Shared cooperative activity requires, further, the absence of certain kinds of coercion, and
commitments to mutual support in the pursuit of the joint activity.” Bratman (1997b, 142). My
view entails that shared intentional action will involve a certain amount of what might be called
co-operation, but elaborating on the relationship between shared intention and co-operation goes
beyond my scope here.
13.3.2 A Necessary Condition on Shared Intention

Among the features that suggest that the case of Corina and Betty is not a shared
intention are that there is no normative requirement on Corina (i) to take her
partner’s interests and expectations into account when she initially forms her
individually-held intention, (ii) nor is there any normative requirement on her to
revise her individually held intention in light of her partner’s interests and expec-
tations when she considers defecting. If the intention is genuinely shared, then its
formation would put rational pressure on the partners to consider the permissibil-
ity of the conditional nature of their individually-held intentions in light of their
partners’ diverging interests. In addition, abandonment of the intention would ratio-
nally require the agent to revise her intention in light of whether or not it is rationally
defensible for her to abandon her partner. These features would enhance the stability
of the intention, bringing shared intention into line with individual intention.
How would this work in practice? Shared intention would be one in which an
individual participant (let’s call her P1) would regard herself as a failure, not just
if she (P1) didn’t participate in the shared activity, but also if her partner did not
participate as a result of her (P1’s) defection. This is a case that illustrates what I
have in mind.
Suppose that Peter and Paul both have an intention to sing a duet together, and both have
the thought “I intend that Paul/Peter and I sing a duet”. Let’s suppose that the content of the
intention is such that each individual thinks he is a failure in circumstances in which they
both do not sing the duet together.
Their intentions are not just similar in content (“I intend that we sing a duet”), they
make reference to the other participant, implying that the intender himself is a failure
if his partner does not engage in the shared activity as the result of his defection.
Let’s suppose that Peter suddenly feels bored with singing and wants to defect. Becoming
bored is a circumstance in which his intention still applies, as his individually-held intention
is not conditional on this eventuality. Subject to the Self-Evaluation Condition, he considers
whether or not he can rationally defeat the judgement that he would be a failure as an agent
if he doesn’t sing. Given that his intention involves him judging himself a failure if he and
Paul don’t sing the duet together, it also requires that he show that Paul’s not singing the
duet together with him must be considered in rationally evaluating whether or not he is a
failure for not singing.
In effect, in this case of shared intention, rational revision of shared intention nor-
matively requires of a would-be defector that he show that the disvalue created by
defection both for him and his partner is rationally defensible. This contributes to
the stability of the shared intention: participants must take their partner’s interests
into account when rationally revising it. As with the case of individual intention,
this stability derives from the nature of shared intention itself, independent of any
other features of the situation.
We can formulate this idea more precisely in the following condition, which
I will call the Evaluative Dependence Condition to indicate that Self-Evaluation
in the shared case depends on evaluation of oneself in light of the interests and
expectations of one’s partner:
248 L. O’Brien
Evaluative Dependence Condition: If individuals share an intention, then each

individual considers herself evaluable as a failure if her partner does not
participate in the shared activity due to her defection.
My proposal is that we add the Evaluative Dependence Condition to whatever theory

of shared intention we endorse.17
This proposal has the advantage of ruling out cases like Corina and Betty. It
is intuitively problematic that the mutual dependence between them imposes no
normative requirement on Corina when she considers defection. Far from suggest-
ing the permissive nature of shared intention, it suggests that this is not a case of
shared intention at all. Second, the intuition about their case is supported by the
consideration that intentions are characteristically stable. The proposed condition
helps to bring shared intention into alignment with individual intention: the char-
acteristic stability of intentions, whether shared or individual, has its source in the
normative requirements that derive from the intention itself independent of extrin-
sic and contingent features of it. Third, the proposed condition and the restricted
understanding of shared intention that it supports helps to make sense of signif-
icant differences among actual world cases of modest sociality. If the Evaluative
Dependence Condition is necessary for shared intention, we can rely on it in trying
to draw up a taxonomy that fruitfully distinguishes shared activity that is based on
shared intention from shared activity that is not. In the latter case, various relation-
ships of interdependence may occur between participants, but without the relevant
interdependence that is captured in the Evaluative Dependence Condition, this is
not shared intention. Finally, I will argue below (Section 13.3.3) that the proposed
condition accommodates the intuitive force of some of Gilbert’s views without com-
mitting us to the claim that the fact that an intention is shared gives rise to an
obligation on the part of the co-intenders to participate in the shared activity.
Of course, shared intention is not immune to revision and abandonment, but it
does imply that there is a rational burden on the co-intender that makes revision
more difficult. Moreover, we should note that there is more than one way in which
a would-be defector from a shared intention can take her partner’s interests into
account in pursuing the shared activity. Let’s consider this in the case of Peter,
who is bored. Peter may take into account only the reasons that he believes Paul
has for singing the duet together when he is weighing the disvalue to Paul of their
not pursuing the duet-singing against the value of his defection. Alternatively, he
may take into account the reasons that Paul takes himself to have for duet-singing,
and consider them authoritative for him. He may also take both perspectives into
account, and try to adjudicate between them. Whatever the case, it is possible for
the would-be defector to be more or less deferential to their partner, or more or less
in agreement with them about what counts as a good reason to abandon or con-
tinue with the intended shared activity. I do not take a stand on whether one way
or another is normatively required by shared intention. My aim here is to establish
17 Amore comprehensive treatment of the relevant issues would involve a discussion of Raimo
Tuomela’s discussion of the “we-mode” e.g. (1991, 2003).
that there is a normative requirement on the would-be defector to take the disvalue
of abandonment to their partner into account in their revision.
13.3.3 Obligation and Self-Evaluation

Because I have been concerned only with normative requirements on agents, none
of what I have said shows us what reasons agents have to continue to intend or act,
and hence, nothing that I have said pertains to whether or not would-be defectors
have an obligation to the abandoned party to continue to have the intention or to
participate in shared activity. If Gilbert is to be understood as taking shared intention
to generate obligations to act, and if these are understood as reasons to act, the
necessary condition that I have argued for is weaker than what shared intention
requires on her view.
However, it seems that Gilbert does not think of obligations as reasons to act,
and so, she appears to avoid the boot-strapping problem mentioned above.18 Even
so, her vision of shared intention is that it generates some species of obligation. She
maintains that the species of obligation involved in sharing an intention is directed
obligation, and that it results in the partners owing one another the actions that their
shared intention pertains to – indeed, she maintains that co-intenders own these
future actions of their partners. She also argues that this is a result of the structure
and not the content of shared intention. And so, irrespective of the moral or other
standing of the actions at stake, these features of shared intention obtain.
Although there is much to be said to assess the relevant sense of obligation that
Gilbert works with, here is not the place to do it. It is sufficient to note that Gilbert
believes that the agents owe one another the pertinent actions just as a matter of
the structure of the shared intention. But this strikes me as a strong claim. It seems
that whether one really owes someone an action depends precisely on the nature
of the action that is putatively owed: the moral standing of the intended action, for
example, should play a role in determining whether or not it is really owed, as should
other features of the situation. Of course, from the agent’s first-person perspectives
on their shared action, they may think that they owe those actions, but it does not
follow that they do, in fact, owe those actions.
A conclusive assessment of Gilbert’s complex views on shared intentions and
also on promises and directed obligations would take us too far afield here. Having
raised a doubt about her approach that is also voiced by Bratman, I would like to
proceed by presenting one of the strong motivations that she has for defending the
relationship between shared intention and obligation. Ultimately, I will argue that it
can be explained by the less contentious Evaluative Dependence Condition.
18 Inher (2006) Gilbert says that intentions give sufficient reason to act. She compares what she
says about sufficient reason to Broome’s normative requirements: “Broome (2001) argues that
intentions (which he conceives of much as I conceive of decisions) are not reasons but ‘normatively
require’ conforming action. This may allow that they give their possessor sufficient reason, in my
sense, to conform” footnote 8, p. 128. See also footnote 15 of this paper.
250 L. O’Brien
Gilbert’s view is motivated by a careful consideration of everyday cases, and

specifically those aspects of them that strongly suggest the obligation of the partners
to one another to participate in the intended action. Consider the following scenario:
Queenie and Rom share an intention to do some shopping in a nearby town. In order to get
there in time they must walk some miles along a dusty road at a certain pace.19
Gilbert imagines Queenie falling behind, and then unilaterally deciding to quit.
Rom’s response is imagined as follows:
Rom is likely to be taken aback. Whatever he says, his thoughts may well run along these
lines: “You can’t just decide to stop here, not just like that! . . . “We’re on our way to the
shops!”20
Gilbert has the intuition that Rom’s response sounds natural, and that it indicates
that shared intention involves the concurrence of all parties to the arrangement, and
the obligation of each to continue in it. Like Gilbert, Rom’s response sounds nat-
ural to me. It does not sound, as it does to Bratman, like “bluff”.21 If Rom was
merely bluffing, then he would be guilty of an exaggerated sense of the importance
of their commitments to one another, overstating his entitlements. However, it is not
at all clear that Rom’s unstated and mildly formulated sense of grievance fits these
descriptions.
Gilbert draws on such phenomena to motivate her thesis about obligation. In
short, she maintains that these are ubiquitous features of shared intention and that
any robust account of shared intention must explain them. Where her view does take
these features into account, Bratman’s does not.
However, I think that we can equally well explain the naturalness of Rom’s
response in terms of the Evaluative Dependence Condition. On the assumption
that their intention is shared, Rom feels entitled to be aggrieved about the fact that
Queenie does not appear to take his interests into account in defecting. If Queenie
really shares the intention with Rom, and is not merely pretending to, then she is
normatively required to consider his interests before defecting, and is criticizable
if she does not. Insofar as her unilateral withdrawal does not indicate that she has
taken his interests into account, she appears to violate the Evaluative Dependence
Condition.
The advantage of relying on the Evaluative Dependence Condition to explain
this and similar cases is that it appears to do the explanatory work without involving
contentious claims about obligations and owing. In fact, there is reason to conclude
that there is a middle way between Bratman and Gilbert, in which we do not think
of Rom as bluffing, but in which we do not think that Queenie really owes her action
or that Rom really owns her future action.
If this is correct, it is in the nature of a genuinely shared intention that
agents take their partner’s success or failure into account when forming their
19 Gilbert
(2009, 173).
20 Gilbert
(2009, 173).
21 Bratman (2009, 152).
individually-held intention, and when considering abandonment of the shared

intention. Consequently, a would-be defector who does not do this is at least
deficient as a rational practical agent. Like an agent who wantonly abandons an
individual intention, she is similarly insensitive to the judgment that she made of
herself when she shared the intention, namely, that she is a failure if her partner
doesn’t participate in the intended shared activity because of her defection. Given
this, the abandoned party is entitled to have the harm to her of abandonment taken
into account by the would-be defector, and if the defector does not do this, she has,
plausibly, an entitlement to be aggrieved. In circumstances where the defector can-
not be excused on other grounds, and where there are no other bars to rebuke, this
entitlement to be aggrieved grounds an entitlement on the part of the abandoned
co-intender to rebuke the defecting co-intender.
This entitlement to be aggrieved or to rebuke is not dependent on the co-intenders
having any particular desires, attitudes, needs, or proclivities. It is also not dependent
on their having reasonable expectations on the basis of their co-intender’s behaviour,
nor on the shared intention’s relationship to their well-being. Similarly, it is not
dependent on any other features of the agents’ circumstances, or the moral status
of the intended activity. Even co-intenders in heinous activity are subject to the
normative requirements imposed by intentions, individual and shared.
I have not attempted to fully adjudicate the dispute between Gilbert and Bratman:
the problem of whether or not obligation is a feature of shared intention invites
further discussion that goes beyond my scope here. I have tried to establish, how-
ever, that Bratman’s view omits what is a plausible necessary condition on shared
intention and that Gilbert’s claims, while sensitive to important features of shared
intention, can be explained in a way that invites less contentious claims about the
relationship between shared intention and obligation and the owing and owning of
actions. Drawing on the Self-Evaluation Condition and spelling this out in terms
of Broome’s idea of a normative requirement provides, I believe, a middle way
between Bratman’s and Gilbert’s views on the matter of whether or not shared inten-
tion, taken alone, generates nonconditional obligations on the co-intenders to one
another.
13.4 Concluding Remarks

Arbitrary abandonment of an individual intention displays disregard for what one
takes to be one’s own success or failure as an agent, and indeed, for the author-
ity of one’s own evaluative judgments of oneself. In the shared case, arbitrary
abandonment involves disregard for another’s success as well. In the case of the
abandoned party who can rationally expect her interests to be taken into account
in cases of abandonment, this may generate an entitlement to be aggrieved. In both
the individual and the shared cases of intention the intender is bound by a normative
requirement to rationally revise her intention before abandoning it, and in the shared
case, rational revision requires taking the partner’s interests into account.
252 L. O’Brien
Acknowledgements My thanks to the participants at the Workshop on Self-Evaluation at the

University of Basel for helpful comments. Thanks to Antti Kauppinen, Federico Lauria and Alain
Pé-Curto for very helpful comments on earlier drafts of this paper.
References
Bransen, J. and Cuypers, S. E. (eds.). 1998. Human Action, Deliberation and Causation.
Netherlands: Kluwer.
Bratman, M. 1993. “Shared Intention.” Ethics 104:97–113.
Bratman, M. 1997a. Shared Intention and Mutual Obligation. Originally published as “Intention
Partagée et Obligation Mutuelle.” In Les limites de la rationalité, vol. 1, edited by J. Dupuy
and P. Livet, 246–66. Paris: Editions La Découverte. Reprinted in Bratman (1999b).
Bratman, M. 1997b. I intend that we J. In Holmstrom-Hintikka and Tuomela (eds.) 1997. Reprinted
in Bratman (1999b).
Bratman, M. 1999a. Intention, Plans, and Practical Reasoning. Harvard University Press, 1987;
re-issued by CSLI, Stanford, 1999.
Bratman, M. 1999b. Faces of Intention. Cambridge: Cambridge University Press.
Bratman, M. 2007. Structures of Agency. Oxford: Oxford University Press.
Bratman, M. 2009. “Modest Sociality and the Distinctiveness of Intention.” Philosophical Studies
144:149–65.
Broome, J. 1999. “Normative Requirements.” Ratio 12:398–419.
Broome, J. 2001. “Are Intentions Reasons? And How Should We Cope With Incommensurable
Values?” In Practical rationality and preference: Essays for David Gauthier, edited by
C. Morris and A. Ripstein, 98–120. Cambridge: Cambridge University Press.
Gilbert, M. 1990. “Walking Together: A Paradigmatic Social Phenomenon.” Midwest Studies 15:
1–14.
Gilbert, M. 1993. “Agreements, Coercion, and Obligation.” Ethics 103:679–706.
Gilbert, M. 2006. A Theory of Political Obligation: Membership, Commitment, and the Bonds of
Society. Oxford: Oxford University Press.
Gilbert, M. 2009. “Shared Intention and Personal Intentions.” Philosophical Studies 144:167–87.
Gold, N., and R. Sugden 2007. “Collective Intentions and Team Agency.” The Journal of
Philosophy 104 (3):109–37.
Holmstrom-Hintikka, G., and R. Tuomela 1997. Contemporary Action Theory, vol. II. Dordrecht:
D. Reidel.
Kim, J. 1998. “Reasons and the First Person.” In Human Action, Deliberation, and Causation,
edited by J. Bransen and St. Cuypers, 67–87. Dordrecht: Kluwer Academic Publishers.
Morris, Ch., and A. Ripstein. 2001. Practical Rationality and Preference: Essays for David
Gauthier. Cambridge: Cambridge University Press.
Schmitt, F. 2003. Socializing Metaphysics: The Nature of Social Reality. Lanham, MD: Rowman
and Littlefield.
Tuomela, R. 1991. “We Will Do It: An Analysis of Group-Intentions.” Philosophy and
Phenomenological Research 51:249–77.
Tuomela, R. 2003. “The We-Mode and the I-Mode.” In Socializing Metaphysics — The Nature of
Social Reality, edited by F. Schmitt, 93–112. Lanham, MD: Rowman & Littlefield.
Chapter 14
Where Individuals Meet Society: The Collective
Dimensions of Self-Evaluation and
Self-Knowledge
Ulla Schmid
14.1 Introduction
Self-evaluation is not a private affair. Inasmuch as it consists in obtaining a value-
oriented perspective towards oneself and involves value judgements with regard to
one’s own conduct, self-evaluation is built upon the normative system in which
an agent is embedded. And inasmuch as the agent is immediately concerned by
its conclusions, self-evaluation has a remarkable impact on the agent’s further
behaviour.
Broadly understood, self-evaluation is the self-referential feedback process by
which we constantly survey the course and success of our actions, by which we
are able to intervene and adjust our behaviour. Self-evaluation in this sense ranges
from basic physiological mechanisms, which are not accessible to us consciously,
serving the adjustment of our movements and bodily position, via self-evaluative
attitudes such as shame or pride to theoretical judgements about one’s agency –
most salient, of course, our reflection upon our last ends and fundamental principles
of acting. Self-evaluation therefore plays an essential part in our lives on every level
of behaviour. I will concentrate on those kinds of self-evaluation involving self-
reflexive attitudes (emotions, thoughts etc.), and thereby conceive of self-evaluation
as a self-referential (mental) activity, which consequently requires the evaluator to
have a sense of her own purposiveness and hereby a minimum understanding of
herself as an agent. This includes that the subject can regard herself as the cause
of certain bodily movements and “mental events” occurring in her surroundings.1
I will argue that to be capable of evaluating oneself involves a certain form of
self-knowledge, which does not originate from observation, but rather consists in
intentional activity itself, as well as certain standards, which determine under which
circumstances an action counts as good or successful. This account will dispense
1 Cf. Elisabeth Pacherie (2007), Axel Seemann (2008), John Campbell, Past, Space, and Self,
Cambridge, MA: MIT 1995, esp. chap. 3.
U. Schmid (B)
e-mail: ulla.schmid@unibas.ch
254 U. Schmid
with the tendency to take self-evaluation and a fortiori self-knowledge as primarily

epistemic relation an agent stands in towards herself. Instead, I will argue that the
primary form we gain knowledge of our minds is by making them up, by taking a
stance towards the world – towards what we are concerned with in our respective
situation.
Understanding self-knowledge as an intentional activity sheds light on the special
aspects of the first-personal relation an agent stands in to herself giving rise to the
authority we hold with regard to our own minds in contrast to other people’s minds.
That first-personal knowledge can occur in two forms, as observational empirical
and as non-observational agential knowledge, provides the explanatory grounds for
the difference between self-referential and other-referential evaluative attitudes, i.e.
why we react with attitudes such as shame or guilt to our own misconduct, but
not to the wrong-doing of others, and why we nevertheless can blame or be angry
with, approve or praise ourselves for normatively relevant behaviour just as we do
to others.2 I will focus on those evaluative attitudes which are directed to the agent’s
own behaviour and thereby require the agent to stand in a first-personal relation to
what she evaluates. The phenomenon of feeling shame, guilt or pride as reaction to
the behaviour of others entails that the agent establishes a first-personal relation to
it – be it by ascribing herself partial responsibility for the other’s action, be it that
she puts herself in the other’s shoes, be it that the other’s action belonged to a joint
project they pursue together.
Finally, inasmuch as self-evaluation consists in reflexive intentional activity, it
is grounded in the individual agents’ collective context in two respects: On the one
hand, insofar as I can be said to know my mind, I genuinely do so intersubjec-
tively; I am able to communicate my intentional attitudes using intersubjectively
accessible verbal and non-verbal means of expressing myself. On the other hand,
self-evaluation per se entails a normative component insofar as it comprises value
judgements upon one’s own behaviour with regard to standards of appropriateness
the agent obtains from her social context.
14.2 The Relevant Kind of Self-Knowledge

I will first distinguish agential self-knowledge from empirical self-knowledge and
outline two corresponding kinds of self-evaluation. Agential self-knowledge, on
which self-evaluation resulting in self-evaluative attitudes is based, displays the
distinctive features of epistemic authority and immediacy with respect to one’s
own mind. These characteristics are internal to the first-personal relation a sub-
ject stands in with her own mind qua being its author.3 Correspondingly, agential
2 Cf.J. Prinz (2007), and his contribution to this volume (Chapter 8).
3I will elaborate later, why self-knowledge qua being the author of my mind does not entail that
my mind’s constitution simply implements my would-be image of it. Let me allow only three
brief remarks here: First, imagining my mind is itself a mental activity already presupposing the
14 Where Individuals Meet Society: The Collective Dimensions of Self-Evaluation . . . 255
self-evaluative attitudes are grounded in the first-personal character of the evaluated

attitudes inasmuch as the subject of self-evaluation acknowledges her authorship
of and thereby her responsibility for the constitution of her mind. Empirical, or
theoretical self-knowledge in contrast does not differ from other-knowledge, it is
rather comparable with a collection of reportable facts about the agent’s mind. Self-
evaluation based on theoretical self-knowledge accordingly results in theoretical
judgements being the conclusions of a comparison between what one actually did
(or felt, believed etc.) with how one ought to have done (felt, believed etc.) in the
respective situation. Theoretical self-evaluation is analogous to the evaluation of
other people’s behaviour apart from its object. This is indicated by a change of
meaning that becomes visible in the comparison of utterances about mental phe-
nomena across grammatical persons and tenses. For example, we use the sentence
“I felt sad” (at time t1 ) in the same way as the sentence “Bernard felt sad”. In this
case, we deal with an ordinary predication of Bernard: S(bt1 ) or me: S(ut1 ). But in
describing someone’s present intentionality, the analogy between assertions about
myself and Bernard breaks in one decisive respect. Though I can also report my
present state of mind from a descriptive standpoint (S(unow )) there is a sense in that
“I feel sad” does not express a propositional content, but the mode of my perspective
on my current situation. The verb feel sad such as all other psychological predicates,
plays a twofold role which differs between third-personal predication and the first-
personal present tense case4 partly with the consequence that in the latter there is no
space for uncertainty about it – I will know that I feel sad just by feeling sad.5
The difference between knowing that I felt sad and knowing that I feel sad
roughly parallels the difference between knowledge I generate by observing my
own behaviour and knowledge I generate just by entertaining a certain affective
state without observation. In this sense, my feeling sad – and my intentionality in
general – are familiar to me without requiring that I have already formed explicit
beliefs upon them. Or, in Moran’s formulation, whenever I actually endorse an inten-
tional attitude with respect to a particular state of affairs, this is transparent to me
in the sense that it contributes to determining the position from which I further
constitution of the mind. Second, the realisation of a mentally represented object is not restricted
to cases of “making up one’s mind” and therefore cannot account for the specificity of the first-
personal relation to my intentional attitudes. Third, constituting one’s mind by self-description/self-
interpretation resembles creating a set of third-personal propositions and hence does not provide
sufficient grounds for the responsibility that comes along with being the subject of one’s mind. Cf.
Moran (2001, esp. chaps. 2.2–2.5).
4 Peter Strawson has famously argued for the distinctiveness of psychological predicates compared
with physical predicates in this respect. It is crucial for this view that although psychological
predicates comprise different modes in the first and third person case, none of these modes is
conceptually privileged over the other. This is partly because psychological predicates are taken
not to designate states of mind as kept in an inner realm, but to refer to primarily embodied inten-
tional activities like smiling, going for a walk etc. As a consequence, the problem of other minds is
disguised as a pseudo-problem insofar as intentional attitudes are basically perceivable by others
in their bodily component. Cf. Strawson (1990 [1959]).
5 Cf. Wittgenstein (1958, §247), spelled out further by Richard Moran (2001) and (1997).
256 U. Schmid
proceed in my behaviour rather than being itself explicitly present in the contents of
my deliberation. If in contrast, I were in need to verify that I can justifiably attribute
the affection feel sad to myself by seeking evidence e.g. via introspection, say, or
observing my behaviour, I will paradoxically already have lost the grounds for that
very feeling. Attempts to uncover my intentionality by exclusively empirical means
are likely to fail (such as Hume’s endeavour to find the self where he could only
find its material contents) due to this categorical mismatch between the way I try to
discover it and its success condition: I am only able to recognise my state of mind
as sadness if I actually feel sad. But if I feel sad, I will know about it and have no
reasons to gather further evidence. How this is possible and in how far I actually
gain knowledge of my mind by entertaining intentional attitudes, will be explored
in the next sections.
14.3 Reading One’s Own Mind: Theoretical Self-Knowledge

To a large extent, what we know and come to know about ourselves does not differ
from the knowledge others (can possibly) have of us, neither in the way it is gained,
nor in the facts known. Sometimes, others seem to know us even better than we do
so ourselves – they can predict quite accurately our further actions, tell how we feel
momentarily or can remember our past behaviour. Theoretical self-knowledge pri-
marily consists in observationally or inferentially based statements of facts about the
subject who knows about them in an ordinarily epistemic manner that does not differ
from other-knowledge or knowledge about the world. The acquisition of theoretical
knowledge of my psychological phenomena functions analogously to the formation
of beliefs, judgements about the world external to my mind (including the minds of
other people) does. For me theoretically knowing that I feel sad makes no structural
difference to knowing that Bernard feels sad; the predication simply changes its
subject from K(u, S(u)) to K(u, S(b)). However, due to this analogy with theoretical
knowledge about others theoretical self-knowledge lacks the two features which I
earlier indicated as distinctive for agential self-knowledge, epistemic authority and
immediacy – it therefore exhibits the structure as if we knew ourselves as others,
the identification of “u” and “I” is missing. I can see myself in a mirror, thereby
acquiring knowledge about my appearance whilst believing that it was somebody
else or, even worse, knowing that it must be me, but not believing it.6 Theoretical
self-knowledge can be subsumed under observational knowledge and thereby cate-
gorically differs from first-personal self-knowledge insofar as theoretically known
propositions have no normative impact upon the agent’s intentionality, it does not
practically concern her in her behaviour. Situations in which we act against better
knowledge or as notorious acratics fail to do things we know we should do are very
familiar (think of starting to write up a paper in due time to meet the publisher’s
deadline).
6 InGeorge Orwell’s Nineteen Eighty-Four, Winston looks into a mirror after having been heavily
tortured for a long time, without recognising that it is himself he looks at. Cf. also Wright (1998).
Traditionally, the idiosyncratic features of the first-personal relation to one’s

mind have been thought to be accounted for by reference to introspection as epis-
temic entrance to the mental realm that was reserved for the respective subject.
However, the Cartesian picture has equally traditionally been rejected because – as
the term introspection already suggests – the subject’s epistemic privilege of her
own mental life is modelled upon the perceptual structure of approaching ordi-
nary objects. That the subject should possess exclusive epistemic authority about
a special realm of things due to its location “within” her, then appears as arbi-
trary stipulation (which is often supposed to become obsolete once the empirical
neuropsychological sciences have made sufficient progress). And in fact, any intro-
spective account gets into trouble once it comes to explaining why the truth of
certain propositions concerning my intentionality could depend on my mental activ-
ity on the one hand, and could have certain normative implications for my further
behaviour on the other hand. Protagonists of as well as opponents to an introspec-
tive and, correspondingly, sceptical, empirical or reductive model of self-knowledge
both give “[i]nadequate attention [. . .] to the person as epistemic agent, and
hence to the mutual interaction between mental life and the first-person awareness
of it.”7
14.4 Making Up One’s Mind: Agential Self-Knowledge
In addition to theoretical self-knowledge – both in its open, empirical and its

disguised, introspective conception – I want to outline a model of agential self-
knowledge that accounts for the subject’s intentionality, authority and her immediate
access to herself and thus can provide the first-personal perspective from which self-
evaluative attitudes such as shame, pride or guilt proceed.8 Agential self-knowledge
on this view comes along as a by-product of intentional attitudes, as it were, which
are formed by the subject when she takes a stance with respect to a certain issue.
Thereby, agential self-knowledge indicates that an intentional attitude is related in a
special way to the rest of the subject’s psychological economy that differs from the
epistemic relation obtaining in theoretical self-knowledge. Knowing agentially that
I feel sad rather amounts to an adverbial specification of my feeling sad (I knowingly
feel sad) and does not per se imply that I hold a corresponding belief that I feel sad –
its structure reads SK (u) rather than Kagential (u, S(u)).9 That is to say that entertaining
an intentional attitude simpliciter already suffices for being in the position to become
aware of it. In order to comprehend the agency-based model of self-knowledge I am
suggesting here, it will be helpful to clarify the following questions: (1) What are
intentional attitudes? (2) Is there a distinction between intentional attitudes I know
agentially and intentional attitudes I hold without being aware of them? (3) In which
7 Moran (2001, 27).

8 Cf. Moran (2001, esp. chap. 2); McGeer (1996, here p. 506ff.); Pamela Hieronymi (2009).
9 Cf. Moran (2001, 31ff.).
258 U. Schmid
respect are intentional attitudes sensitive to being known agentially? (4) If agential
knowledge of my mind does without me holding any corresponding beliefs, how
can we sensibly speak of knowledge in this case at all?
We should, however, not regard agential self-knowledge as being somewhat
supervenient on an underlying unknown or, in psychoanalytical terms, unconscious
state of mind which has the potential of becoming known by its subject if she
engages in a suitable mental process.10 The reason is that this model would bring
us back to opening a gap between the subject and her attitudes as it is present in the
theoretical model of self-knowledge. Instead, taking the relation between intentional
attitudes and agential self-knowledge to be internal as I suggested above implies that
we know of our intentional attitudes immediately by entertaining them as long as
they are transparent to us.
14.5 Investigating Intentional Attitudes

Employing intentional attitudes – be it cognitive, affective or conative – means
responding to what concerns us in our current situation, having settled a ques-
tion or a number of related questions and thereby taking a stance to the world.11
My belief that it is raining involves that I have positioned myself with regard to
the question whether it is raining or not; my intention to have coffee for break-
fast includes that I find drinking coffee the appropriate thing to do and my feeling
sad is a reaction to a corresponding state of affairs. Constituting my mind primar-
ily does not result from explicit deliberation what attitude would be best to obtain,
all things considered (if anyone at all would be prone to claim this). Rather, mak-
ing up one’s mind consists in considering a certain state of affairs; questions about
my belief, say, concerning p are accordingly answered by reference to p and not
to my belief whether p or not-p.12 Moran focuses on beliefs, intentions and to
some extent on desires when discussing his transparency- and commitment-based
account of self-knowledge and first-personal authority suggesting that in this case,
the Transparency Condition is most obviously met, i.e. that we arrive at knowing our
beliefs and desires just by having them and acting accordingly. The stronger claim
is that for all kinds of intentional attitudes, not only for belief and intention, enter-
taining them entails positioning oneself knowingly towards a particular issue. If all
intentional attitudes are the results of a subject’s mental activity, of her considering
a certain issue that is important to her, the subject can be expected to know any of
them agentially due to the transparency condition for deliberative self-knowledge.
Moreover, that corresponding self-ascriptions set up certain normative constraints
on the subject’s further behaviour even if the expressed intentional attitudes are not
yet actualised will hold for non-cognitive attitudes and cognitive attitudes equally.
10 Cf. ibid., 13.

11 Cf. Hieronymi (2009, 1ff.).
12 Cf. Moran (2001, chap. 3).
Once one has introduced them into the realm of public reason by making them
explicit, one’s behaviour underlies the rationality-based cognitive expectations of
others.13
The stronger view maintains that every intentional attitude, insofar as it is actively
generated, is transparent, but articulable and hence, belongs to the deliberative
stance the agent takes in her reasoning. Or, reversely, passively acquired parts of the
agent’s mind do not belong to her deliberation, are not transparent to her, and so are
not knowable agentially. However, the stronger claim also holds that all intentional
attitudes are possibly be known agentially. Consequently, proponents of the stronger
claim must either limit the intentionality of emotions, dispositions etc. to the same
extent that they are knowable active products of the agent’s will and declare the
phenomenally passive “rest” to merely third-personally accessible, and otherwise
unalterable, opaque facta of one’s mind. Alternatively, one can adopt a cognitivist
view towards the mind – including emotions, feelings, dispositions etc. – that fun-
damentally grounds the mental in cognitive operations. Whereas the first view holds
that all intentional states entail propositionally structured elements that are agen-
tially knowable, on the second view, an agent’s intentional relatedness to the world
consists in a set of judgments about it.
It is crucial to intentional attitudes that they are no object-like entities stored in
their subject’s mind, but that they make up the various relations to the world in
which the subject actively situates herself. The role conceded to the subject with
regard to her attitudes is better characterised as that of an author than that of a pas-
sive observer of what emotions occur to her, what beliefs are forced upon her or
what desires she has to satisfy. Conceiving of intentional attitudes as mental activity
proceeding from (and contributing to the constitution of) the subject’s perspective
allows for its first-personal character as well as for the possibility of regarding them
from a distanced third-personal viewpoint. With regard to her own mental activity,
the subject can depart from both perspectives14 : Inasmuch as her attitudes are trans-
parent to her and thereby immediately responsive to the subject’s positioning with
regard to the issue at question, the subject adopts what Moran calls a deliberative
13 The normativity involved is rational, not moral. That is, if one does not behave in accord with
the contents of one’s avowals one will be considered unsteady, irrational or unaccountable. The
intersubjective consequences of irrational behaviour do not so much consist in specific punishment,
but rather in the loss of recognition as an equivalent member of one’s social context. Cf. McGeer
(1996, 508).
14 The formulation can be misunderstood as suggesting that the subject’s obtaining one of the two
positions is a deliberate act based on the subject’s decision to either identify with her mind or
distance herself from it. But voluntarism with regard to the position we take up towards ourselves
begs the question, because if taking a position was based on a deliberative process, this process
itself wouldn’t do without a standpoint from which it departed. Moreover, this view if conceivable
at all would suggest that the subject was primarily standing at distance to herself, that she was
in a position to choose from either accepting her authorship of her mental activity or maintaining
an objective relation to it. However, this amounts to returning to the picture of mental activity
occurring in some special realm the subject has privileged access to, from which we started, rather
than illuminating the special features of the subject’s first-personal perspective.
260 U. Schmid
stance towards her mind.15 In contrast, if the subject has to engage in reflection in
order to know the attitudes she currently employs viz. the position she has taken
towards a certain issue, her relation to her own intentionality will not categorically
differ from the relation she holds to the intentionality of others. One can be alienated
from one’s intentional attitudes if one stands at a certain distance to the reasons one
originally had for forming them. Alienated attitudes lose their efficacy as elements
in the subject’s deliberation process although they still can have causal effects on
the subject’s behaviour. Likewise, the agent cannot avail herself of her dispositions,
passions and inarticulable parts of her mental life from her deliberative stance, but
has to approach them from a bystander’s theoretical stance.
The distinction between deliberative and theoretical stance finds – according to
Moran and others – an analogy in a context-depending distinction between two
categorically different modes first-personal statements that refer to the speaker’s
intentional attitudes can come in. However, even though there might be cases in
which the line between deliberative and theoretical character of a first-personal
statement is easily to draw, in usual circumstances, I suggest, this is not the case.
When I violate my resolution not to eat chocolate anymore, this seems to indicate
that I find myself at a distance to it and so leaves the possibility of being over-
ruled by my politeness not to reject the offered piece, together with my craving for
its euphoriant effect or simply its delicate taste. However, inasmuch as I will have
a bad conscience thereafter because of having acted against the reasons favouring
my original resolution they still matter to me. This indicates that I still endorse the
resolution in the sense that is further at work, so to speak, having effects on my feel-
ings, my self-image or my further behaviour (e.g. I might take up means to avoid
tempting situations – I might, of course, reconsider the reasons I had for not eat-
ing chocolate and dispense with my resolution after all). In contrast, the panic of
my childhood in face of climbing anything higher than 7 ft might be worth only a
footnote in conversations concerning hiking in the mountains because it is not only
in temporal, but also in personal terms at far distance from myself. This also gives
an additional emphasis to the Strawsonian understanding of mental predicates as
equivocally comprising both, third and first-person-perspective.
The character of first-personal utterances thus oscillates between expressing and
reporting, between avowing and listing one’s psychological states, reflecting the
distance from which the speaker regards herself. It is notable that the categorical dif-
ference is often not mirrored by the verbal composition of the statement in question,
the avowal “I feel sad” does not part with one iota from the report “I feel sad”. For
this reason, to determine whether a particular first-personal statement tends to be an
avowal or rather a report one must examine how it is embedded in the rest of the sub-
ject’s psychological economy and to what extent it corresponds with the subject’s
behaviour. The statement “I feel sad” uttered in a mournful voice by someone, who
shows signs of heavy-heartedness (such as walking slowly, being absent-minded,
making a melancholic face) or of whom we know that he has reason to feel so,
15 Cf. Moran (2001, chap. 3) and Hieronymi (2009, 3).

will be an expression of his current state of mind rather than a report intended to
communicate some fact about himself. As a first criterion to differentiate between
avowals and reports we can refer to the different way language functions in both
cases, as manifestation of the speaker’s otherwise embodied and enacted attitudes
on the one hand, as a means to transport information on the other hand.16
A second feature that distinguishes avowals from reports draws attention to the
relation the uttered intentional states stand in with the rest of the subject’s mental
states, especially with respect to her future behaviour. In contrast to descriptive self-
reports, self-ascriptions of intentional attitudes by avowing them are commissive,
i.e. they have normative impact on the range of future behaviour one can and will
perform in the light of one’s current state of mind. Taking intentional attitudes as
determining in this sense partly accounts for the subject’s responsibility for employ-
ing them and acknowledges that the commitments entailed in intentional attitudes
maintain a guidance function for the subject’s acting even if they are not consciously
represented at all times. In contrast to reports, which are restricted to describe a
present state of mind, avowals are future-directed with respect to their correctness-
conditions, i.e. their truth depends on whether the subject’s further behaviour is in
accordance with what she attributes to herself. On this view, the commissive ele-
ment of avowals partly constitutes the subject’s special authority about predicting
her future actions. One caveat: The predictive authority one has over one’s own atti-
tudes is gradual and holds ceteris paribus – i.e. if one’s commitments do not change.
It does not necessarily come down to the predictive authority one can acquire over
others given a certain background knowledge which fosters according cognitive
expectations, only if one’s future self is considered as a different person. Rather,
“genuine” first-personal predictions are undertaken from the deliberative standpoint
and as such approximate the explication of one’s intentions.17
However, the thesis that intentional attitudes carry an internal commitment, if
commitment is understood similarly to a “promise” to oneself as McGeer first con-
siders18 or if in the notion of commitment the subject’s awareness of its normativity
is included, is justifiably objectionable.19 Therefore, I propose to use a somewhat
thin notion of commitment that enables us to account for the distinctiveness of
self-evaluative attitudes in comparison to other-evaluation that is at stake, whilst
16 This point is made and discussed in extenso by and in succession to Wittgenstein’s argument on
the possibility of a private language. Cf. Wittgenstein (1958, §304) and Wright (1998).
17 Cf. Anscombe (1950), McGeer (1996, 508ff.), Moran (1997, 154). The remoter the future is at
which one’s commitments are directed, the more converge the relation one obtains to oneself in the
future and the relation to one’s future self. This is convincingly argued especially by Carol Rovane
in Rovane (1998).
18 Cf. McGeer (1996, ibid.).
19 Gertler finds the latter conception entailed in Moran’s deliberative account of self-knowledge,
which actually departs from knowing one’s beliefs as the default case of deliberative self-
knowledge. This is indeed highly demanding – too demanding, as O’Brien remarks – regarding
the subject’s cognitive capacities and hence should be taken with caution. Cf. Gertler (2008), and
O’Brien (2003).
262 U. Schmid
minimising the cognitive requirements for a subject to know her own mind. In that,
I want to draw on Moran’s Transparency Condition for the endorsement of inten-
tional attitudes and on Hieronymi’s suggestion to interpret intentional attitudes as
a response to a certain issue. My aim is to make plausible that the commitment in
question comes along with the acquisition of a position to a state of affairs (be it
one’s own “internal” state of mind or an “external” event “in the world” so to speak)
and that due to its internal relatedness to an intentional attitude the commitment
itself is transparent to the subject as long as she entertains the attitude in question.
I will have to deny the second question I departed from in this paragraph, whether
there is a distinction between intentional attitudes I know agentially and intentional
attitudes I hold without being aware of them. Rather, I want to reserve the notion
“intentional attitudes” for such psychological states that differ from mere thoughts
and sensations we have en passant in that they present us with a practical con-
clusion with respect to a certain issue that concerns the subject, without claiming
though that every such “conclusion” has to be preceded by a process of deliberation
or amounts to practical knowledge.
According to Moran, beliefs must meet what he calls the “transparency condi-
tion” in order to qualify as specifically first-personal, proceeding from the subject’s
deliberative stance.20 If someone is asked a question about her beliefs concerning p,
she will consider whether it is the case that p by considering what speaks in favour
or against p. She will not, however, consider an inventory list of her mind, so to
speak, in order to see whether she believes that p. Knowing one’s beliefs according
to the transparency model therefore requires the subject to direct her attention “out-
ward” rather than “inward”, to look at the issue her belief is concerned with rather
than to introspect her mind. Although not reducible to the question whether p is the
20 Moran (2001, chap. 2.6). As I already mentioned, he focuses on beliefs (chap. 3) as paradig-
matic case where the transparency condition holds. He leaves open the possibility of extending the
notion to non-cognitive intentionality as well. I think that this is actually the right thing to do in
the light of the notion of intentional attitudes I am spelling out here. On such an account, inten-
tionality amounts to the mode of “aboutness” by which one propositionally relates to an object
in the world and is transparent insofar as this relating is actively undertaken by the intentional
subject. The main challenge I see for Moran’s account consists in the need to reconcile content-
externalism (transparency) with the distinctive subjectivity of first-personal knowledge (authority,
immediacy). Referring to the agent’s agential knowledge about his intentional attitudes goes as
far as these actually are determinable by the agent – insofar as they are actions, not passions. Not
only does this exclude the passive elements of conative and affective intentionality, but it also
presupposes the agent’s conceptual knowledge of his attitudes, and a fortiori the mastery of cor-
responding intentional concepts. Introducing a meta-cognitive level into first-personal knowledge
of non-cognitive intentionality, however, alienates those parts of experience that are not knowable
agentially. This applies to non-cognitive intentional attitudes insofar as they are not deliberately
caused by the agent and hence represented as external to his deliberative stance (i.e. external to
the (rational) agent himself). How then can other intentional attitudes than beliefs and intentions
be first-personally known? I think that on Moran’s account, there is no space for this. Fortunately,
there are alternatives – that I cannot elaborate here – that draw on non-conceptual action-awareness
as basis for agential self-knowledge rather than on rational self-consciousness. Cf. O’Brien and
Soteriou (2009), esp. the contributions by Christopher Peacocke and Joëlle Proust.
case, the question whether I believe that p is answered by the same considerations.
Whenever I hold a belief concerning the question whether p is the case the belief
itself is transparent to me in that my belief constitutes the position I have taken
with regard to p, i.e. my belief specifies the mode of my relation to p, it metaphori-
cally speaking functions as the glass I look through to p. In this sense, knowing my
beliefs concerning a certain state of affairs is internal to having made up my mind
with regard to the state of affairs in question, i.e. internal to the response I would
give to the question whether p is the case.
However, the transparency model seems to contribute to explaining the formation
of new intentional attitudes and the subject’s awareness of doing so rather than to
account for the detection of those intentional attitudes the subject already endorses
and so seems to fail as a model of acquiring knowledge of prior existing facts. The
scope of intentional attitudes we can detect by the outward-looking method pre-
scribed by the transparency model moreover seems to be restricted to those cognitive
conclusions achieved by our current deliberation and thus seems to be ignorant of
intentional attitudes other than beliefs.21
It seems to me that the first objection begs the question. If we do not consider
those attitudes of ours we come to know of theoretically but those which the trans-
parency condition is supposed to apply to, we have already focused on attitudes
we hold as parts of our current deliberative process.22 Moreover, in favour of the
transparency model it can be argued that currently endorsing an intentional attitude
entails that the subject responds to certain situations in the light of the attitude she
employs. On this understanding, the acquisition of intentional attitudes does not nec-
essarily derive from a preceding process of deliberation what belief or other attitude
would be best to endorse given the relevant evidence we have at hand concern-
ing our situation. Rather, endorsing an attitude implies that the subject has settled
a question concerning a state of affairs for herself and thus can answer questions
about this state of affairs without reference to her attitude. Along these lines, the
transparency condition is not restricted to the formation process of new attitudes
alone, but also holds for the reformulation of already endorsed attitudes bringing
them to light by considering issues they are concerned with. Still, one should note
that endorsing an attitude is a temporally extended condition of someone’s mind
affecting the subject’s way of behaving. The transparency of intentional attitudes
now derives exactly from their being integrated in the structure of the perspective
from which the subject approaches the world. As such, they are equally known to
the subject in responding to related issues as how she acquires new attitudes when
she addresses new questions. In this sense, I want to deny that there is a categorical
difference between conscious and non-conscious attitudes – the difference breaks
21 Cf. Gertler (2008). There are further objections related to the possible mismatch between com-
ing to a theoretical conclusion and abolishing pre-existing contrary beliefs. One can in Gertler’s
example conclude that black cats are not responsible for one’s bad luck and still hold the belief that
it is black cats that caused one’s own misfortune. However, this case is an instantiation of Moore’s
paradox and as such considered by Moran (2001, chap. 3).
22 Cf. Moran (2001, chap. 2.5 et passim).
264 U. Schmid
down to whether one currently entertains a certain intentional attitude or not and
this again amounts to the question what the relations between the subject and the
world are.
As to the second objection, I agree that an account of self-knowledge focus-
ing primarily on knowing one’s beliefs is too restrictive, not only regarding the
kinds of intentional phenomena it considers, but also regarding the mental capaci-
ties demanded of the subject on this picture. Apart from presupposing conceptual
capacities in general, self-knowledge on this model also requires the subject to
have meta-cognitive capacities, i.e. to be able to perform cognitive operations
self-consciously. That rules out the possibility that beings who lack the according
capacities can engage in processes that contain self-reflexive knowledge of one’s
intentionality as it is present in self-evaluation, the reflexive relation we departed
from. Thus, it excludes children, animals, comatous people etc. from having first-
personal knowledge as well as non-cognitive and non-conscious ways of self-access
and, a fortiori, self-regulation in full-fledged concept-users. The strength of Moran’s
transparency-based account however consists in that its applicability to intentional
attitudes of all kinds insofar as they involve conceptual representations that are
actively formed and can, for this reason, be articulated.23 And still despite the overly
demanding picture of self-relatedness, I want to maintain the basic insight that we
have immediate self-awareness of our intentional attitudes (a sense of agency, so to
speak) because we are actively involved in their generation and that this is the actual
basis for all higher-order forms of self-knowledge.24
14.6 Commitment and Knowledge Entailed in Intentional

Attitudes
If I have succeeded in making the transparency of intentional attitudes to their sub-
jects sufficiently plausible, how does their commissive element come into play? I
suggest the answer is best approached by clarifying what kind of knowledge we are
concerned with in agential self-knowledge and thereby also answering the sceptical
question in how far the notion knowledge is justified in this case given that know-
ing one’s attitudes agentially need not entail holding a belief about them that would
be based on observable evidence.25 Three claims about agential self-knowledge as
23 Moran does indeed include desires in his account of self-knowledge. Cf. Moran (2001, chap. 4).
Moreover, Hieronymi suggests treating all kinds of intentional attitudes alike as their subject’s
response to the issues she is concerned with.
24 Christopher Peacocke has spelled out this idea and an according functional model on various
occasions, most recently in Peacocke (2009).
25 I actually want to put forward a stronger claim. It does indeed seem to me that agential self-
knowledge does not even require the subject to be conscious of the intentional attitudes she
endorses. For if agential self-knowledge is internally related to and hence comes into play together
which a particular intentional attitude, and if further an intentional attitude is a transparent con-
stituent of the subject’s perspective from which she acts and deliberates, it is neither necessary for
I have characterised it so far seem to provide most reason for opposing the use of
the term knowledge: that self-knowledge should be immediate, non-observational,
and not necessarily accompanied by an epistemic or other cognitive attitude – and
still be knowledge of contingent “worldly” facts, namely the subject’s intentional
attitudes.
I have already tied the immediacy of agential self-knowledge to there being an
internal relation between entertaining an intentional attitude and knowing of it, i.e.
knowing oneself agentially is intrinsic, though not reducible, to having an inten-
tional attitude. The internal relation holding here is characteristic of all sorts of
non-observational knowledge, i.e. knowledge of facts about oneself that is not based
on either behavioural inference or (empirically accessible) perceptual evidence. In
her account of practical knowledge, Anscombe for instance considers someone’s
knowledge of the position of his limbs, bodily movements and intentions as paradig-
matic cases for non-observational self-knowledge. Similarly, I know that I feel sad,
that my pulse is running or that I know that Paris is the capital of France with-
out looking into the mirror, at my ECG or at a virtual list of propositions currently
known by me. Common to all these cases is that they involve direct awareness of
one’s own mental and bodily states that enables the subject to make according judg-
ments without considering further evidence. This is not to say that the subject could
not follow an evidential route to gain knowledge about facts concerning herself, but
rather that particular facts about her are genuinely known by the subject and can be
ascribed to herself once she directs her attention to them. Agential self-knowledge
is similar to Anscombian practical knowledge inasmuch as it is essentially first-
personal and generally associated with world-directed, in this sense intentional,
activities the subject exercises as soon as she is capable of a minimal sense of being
the origin of a variety of mental and bodily events. Certainly, in order to be able to
ascribing intentional attitudes to oneself, certain cognitive and conceptual resources
are necessary; in particular, being conscious of one’s intentional attitudes and even
more expressing them verbally require the subject to have an according conceptual
repertory at her disposal. Employing certain intentional attitudes itself is not suffi-
cient for the subject to actually conceive of them as such, although it suffices for
the subject’s agential awareness of them (qua their internal relatedness) and is a
necessary condition for an adequate conceptualisation of the attitudes in question.26
If someone actually has this sense of himself as an intentional subject, he will
thereby be taking a stance towards the world, i.e. he will be actively respond-
ing to what is going on around him, rather than just reacting to it. The difference
between responding and reacting lies in the role conceded to the subject’s delibera-
tion, reactions can occur as automatic reflexes to sensory stimuli independent from
the subject’s will, whereas responding, or adopting a position to a situation requires
the endorsement of an intentional attitude unlike for having sensations – nor for agentially knowing
it that the subject is aware that it is in place. Cf. Moran (2001, 41 et passim). This, however, is true
of theoretical knowledge as well, I know for example that Paris is the capital of France even if I
am not constantly aware of this fact.
26 Cf. Moran (2001, chap. 2.1).
266 U. Schmid
the subject’s intentional effort. Taking a stance viewed that way involves that the
subject sees herself as a subject, that she is capable of taking (causal) responsibil-
ity for certain events and that she makes up and therefore is able to express her
mind. And expressing one’s mind simply consists in uttering sentences that contain
the ascription of particular intentional attitudes to oneself. Similarly to express-
ing one’s sensations or colour impressions, uttering propositions concerning one’s
intentional attitudes gives immediate insight to the constitution of the speaker in her
relatedness – be it responsive or reactive – to the world. Sentences of both types
are best characterised as avowals because they do not only make claims about the
subject’s current state of mind, but also give the subject reason to act accordingly,
and if she fails to do so, reason to make them fit by changing, modifying or adding
either her attitudes or her behaviour. For we should bear in mind that the verbal
elaborations of her intentional attitudes are actions, too, the subject is responsible
for as a rational being.27
That the correctness-conditions of avowals of this kind in this way partly
can change retrospectively, depending on the subject’s further behaviour, cred-
its for the authority of agential self-knowledge. That the correctness-conditions
likewise depend on the rules governing the correct application of sensation- and
intentionality-language grants for the normativity of agentially knowing oneself
and thereby for the first of two collective elements entailed in self-evaluation.
Inasmuch as endorsing intentional attitudes means having answered a question to
a certain issue for oneself, they can be explicated. In the self-ascribing explica-
tion of her intentional attitudes a subject follows the rules governing the use of
intentionality-language if she avails herself of language in a meaningful manner
rather than uttering meaningless sounds. It is debatable whether content-externalism
with regard to self-knowledge is reconcilable with first-personal authority at all, but
failure to give accurate self-ascriptions in that a subject’s avowals and behaviour
diverge must be accounted for when expressive utterances are to count as articula-
tions of self-knowledge at all. Self- and other-deception aside, which both are (more
or less) voluntary activities initiated by the subject herself, the failure entailed in giv-
ing wrong statements about one’s state of mind lies either in the subject’s inability to
use the right concepts (the mistake lies in the articulation of one’s self-awareness)
or in a lack of coherence throughout her psychological system. The failure here
arises on the basis of entertaining conflicting attitudes, whereby the attitudes artic-
ulated by the subject are not efficacious for her acting. Due to the intrinsic relation
holding between one’s intentional attitudes and one’s agential self-knowledge it is
logically impossible to be mistaken about one’s intentionality although the con-
tents of one’s knowledge is determined by the intentional attitudes one actually
endorses and although one’s articulate self-ascriptions do not necessarily corre-
spond with one’s attitudes.28 Agential self-knowledge bears a normative quality in
27 Cf.McGeer (1996, 509); Searle (1983).

28 Insome cases, our action-awareness can be mistaken, though, due to external influences – e.g.
paralysed people can have action-awareness of a movement they cannot perform: That we are
the sense that, although a subject’s failure to avow her intentional states does not
compromise her first-personal epistemic authority, she nevertheless can be wrong in
her self-ascriptions and thereby undermine the credibility of her avowals. That the
content of first-personal utterances underlies subject-independent correctness condi-
tions should prevent us from interpreting the constitution of one’s mind as allowing
for self-interpretation to be self-fulfilling. The constitution of a person’s mind is not
entirely up to her, but hallmarked by her engagement with her situational context
and the commitments deriving from doing so.29
Until now, I have characterised agential self-knowledge in contrast to theoretical
self-knowledge as genuinely first-personal in quality, internally related to the inten-
tional attitudes a subject obtains in making up her mind regarding the states of affairs
she is concerned with. I have argued that holding intentional attitudes amounts to
having settled a question for oneself, bringing about the subject’s ability to express
her intentional attitudes verbally as avowals entailing the rational commitment to act
in the light of her self-ascriptions. This account of agential self-knowledge therefore
provides for the asymmetry between the relation a subject holds to her own mind
and those relations she stands in with the intentionality of others, characterised by a
special authority of constituting her mind on the one hand and a normative require-
ment to bring her – tacit or public – self-ascriptions in agreement with her behaviour.
Failure to do so can result in questioning the subject in her first-personal authority
and gives the subject an incline for disapproving self-evaluative attitudes, such as
shame or guilt, as I will argue in the following.30
14.7 Self-Regulation by Evaluating Oneself

From the discussion of agential self-knowledge three results have direct impact
on the understanding of self-evaluation that I will elaborate in the following: its
first-personal character manifested in its immediacy and the subject’s authority, the
subject’s active role in making up her mind as response to the world and the nor-
mativity of holding intentional attitudes. If we conceive of intentional attitudes as
constituents of their subject’s perspective, from which she gets involved with the
world, that themselves entail the subject’s response to a certain state of affairs, we
are able to understand the regulative function of self-evaluation for the situational
adjustment and refinement of this class of mental states. I have closed the last section
by characterising the commitment entailed in holding intentional attitudes as a com-
mitment to act in the light of one’s self-ascriptions which are embodied in holding
nonetheless entitled to self-knowledge is argued for by Christopher Peacocke (1996) and Tyler
Burge (1996).
29 For a comprehensive discussion of self-interpretative views as held by Charles Taylor, Paul
Ricœur and others (most of them subsequent to Sartre’s views as elaborated in Being and
Nothingness) and a clear-cut dissociation from them, cf. Taylor (1985) and Moran (2001, chaps. 2
and 4).
30 Cf. McGeer (1996, 508ff.).
268 U. Schmid
intentional attitudes. Here, self-evaluation comes into play as that mental activity
by which one monitors one’s own bodily behaviour, one’s dispositions and mental
activity bringing to light discrepancies between one’s intentional attitudes and one’s
behaviour or within one’s rational commitments. In self-evaluation, one provides
oneself with reasons for adjusting them actively or taking alternative measures to
avoid incoherence.31 Exerting control over the constitution of one’s mind can, in
principle, take three forms. One can acquire new or modify one’s existing dispo-
sitions in order to be subsequently caused to act in the right way – this, however,
does not have impact on one’s deliberatively generated attitudes. One’s delibera-
tive stance is open to direct control by reconsidering the grounds of the attitudes
one obtains and thereby re-constituting or altering one’s commitments. One can,
however, exert indirect rational influence over one’s own attitudes – for example,
by bringing oneself into a situation in which one is likely to respond in the desired
way.32
The proper kind of adjustment depends on whether the inconsistencies among the
subject’s attitudes obtain between her rational commitments and her dispositions or
among her intentional attitudes viz. the corresponding rational commitments. If the
subject’s intentional attitudes conflict with dispositions she holds, she will have rea-
son to improve herself in the light of her commitments – she will have reason to take
measures to alter her dispositions in order to meet up with the rational requirements
originating from her commitments. This adjustment can be undertaken without actu-
ally changing one’s mind for one can succeed just by manipulating oneself on a
sub-rational level. If on the other hand the subject has discovered inconsistencies
among her intentional attitudes, she is provided with a reason to actively revise
the constitution of her own mind. She will do so by reconsidering the questions
answered by the intentional attitudes at stake and if necessary by additionally exert-
ing manipulative influence upon them in order to change them according to her
purpose.33 This is for example the case when instrumental actions do not succeed
in realising the goal they are directed at so that the subject has to modify either, her
goal-directed behaviour or the goal itself, or if she has committed herself to several
incompatible goals.
The requirements to engage in self-evaluation on a minimal level include know-
ing one’s mind in terms of agential self-knowledge and additionally taking a reflex-
ive stance towards oneself, i.e. a stance from which one directs one’s intentional
31 Alternatives can include self-deception, ignorance towards or repression of the mismatch. Which
method is applied by the subject presumably depends to some extent on the relevance of the
incoherence in question, to some other on the method she is prone to pursue given her mental
constitution. At any rate, since holding intentional attitudes implies that the subject knows that she
employs them, she is accountable for the way she handles situations of incoherence.
32 The two ways of controlling the constitution of one’s set of intentional attitudes are presented in
detail by Hieronymi (2009). Exercising rational influence over oneself parallels rational influence
over other people to a certain extent (cf. Rovane 1998), its success still hinging on an exertion of
direct rational control, the endorsement of the response in question.
33 Cf. Hieronymi (2009, 1).
activity to one’s own deliberative stance without entirely distancing oneself from
it. Although self-evaluation can be performed analogously to the evaluation of oth-
ers and in this case basically consists in a theoretical process of comparing one’s
self-ascriptions with one’s behaviour or other standards, the conclusions of evalu-
ative self-reflection do not per se give the subject sufficient reason to change her
behaviour.34 It is conceivable that the outcome of self-reflection consists in a prac-
tical Moore-situation, i.e. a situation in which I believe that I ought to do, feel or
think p, but still do, feel or think the contrary. An appeal to rationality alone will not
help to make the subject act according to the conclusion she arrived at. For this to
be the case, the conclusion has to be commissive for the evaluating subject – which
is the case if she endorses it as integral part of her deliberative stance.
Rather than being a reflective process that remains in the domain of theoretical
reasoning, I propose to conceive of self-evaluation as reflexive form of control the
subject intentionally, but not necessarily consciously exercises over her intentional
attitudes. Understanding self-evaluation as a kind of intentional activity however
does not imply that the subject engages in some form of practical reasoning whilst
forming, monitoring and revising her mental agency, if practical reasoning is under-
stood Kantianwise to contain the subject’s representation of its object as cause of
its conclusion. In contrast, self-evaluation as well as making up one’s mind by
positioning oneself to certain questions seem to belong to neither practical nor
theoretical reasoning inasmuch as their results, i.e. intentional or self-evaluative
attitudes, immediately have normative impact on the subject’s behaviour, but are
not represented or otherwise considered by the subject in advance.35
Self-evaluation itself presents us with a complex exercise of mental agency
involving two different aspects under which the subject engages with the inten-
tional attitudes constituting her mind: Whilst self-evaluation contributes to forming
the subject’s intentional attitudes and in this respect evidently proceeds from the
subject’s deliberative stance, the question she is settling here is concerned with the
appropriateness of her own intentional attitudes. Self-evaluation now appears as a
process bringing the subject into some distance to her deliberative stance without
separating from it by lapsing into theoretical reasoning about herself. To under-
stand this properly, I would like to recall that the attitudes one is concerned with
in evaluating oneself are known by the subject agentially, that is they primarily
are transparent features of the stance from which she approaches the world, and
internally bear a commitment to accord with them. The most basic version of eval-
uating oneself consists in complying with the variety of commitments entailed in
the subject’s intentional attitudes by responding to changes in the grounds of the
attitudes with changing the subject’s mind and thereby adjusting her commitments.
Contributing to making up the subject’s mind, the entire process of evaluating her-
self in its most basic form remains transparent to the subject insofar as her attention
in doing so is directed “outward” to the question at issue.
34 Thanks to Anita Konzelmann Ziv for helping me to clarify this point.

35 Cf. Hieronymi (2009, 12 (footnotes)) and Moran (2001, chap. 3.3).
270 U. Schmid
14.8 Self-Evaluation as Response to One’s Responsive Stance

However, this thin notion of evaluating oneself does not provide sufficient explana-
tory grounds either for the self-reflexive processes giving rise to self-evaluative
attitudes, or for their explicit normative dimension. To account for these, we must
develop a fuller notion of self-evaluation referring to the subject’s responsibility for
her attitudes entailed in the notion of commitment I elaborated above. The commis-
sive character of intentional attitudes includes the subject’s responsibility not only
for having generated them, but also for taking them into consideration in my further
behaviour. Whenever I engage in settling a question, the attitudes I employ provide
the framework from which I depart in my deliberation. However, in the light of my
considerations of the situation I am concerned with the attitudes I previously held
might turn out to be incomplete, inappropriate or simply obsolete with regard to
what is appropriate to do, think or feel in the respective situation. Typically, the role
of self-evaluation in a rich sense is that of a negative feedback mechanism: it serves
the purpose of finding out whether there actually is a mismatch between my inten-
tional attitudes or my behaviour and the requirements set up by my situational or
social context and thereby provides me with reasons for refining, adjusting or giv-
ing up parts of my intentional attitudes. Metaphorically speaking, in self-evaluation
I will “bring [my attitudes] out of the frame and install them within the scope of the
problem itself, on the negotiating table and there my relation to them is unlike any-
one else’s. [. . .] they cannot enter into my thinking as the fixed beliefs and desires
of some person or other, who happens to be me.”36 The downside of the author-
ity over my own intentional attitudes, which I am conceptually endowed with by
standing in a first-person relation with my beliefs, desires and emotions, is that I
am accountable for taking measures such that my actions make sense in the light of
the assertions I previously made about my mental states.37 Misrepresenting my own
intentionality in a significant number of cases affects my scope of acting as well as
my status as rational and authoritative agent, in extreme cases with the result that I
cannot cope any longer with the requirements of my situation or that I have to suffer
from social forms of disregard. The effect presents the subject’s conversion from
an agent into a patient having lost authority and control over her own mind whilst
having lost the capacity of engaging in self-evaluation or at least, taking its results
seriously enough.38
I have mentioned above that the rational and conceptual requirements implied
by knowing oneself agentially accompanied by self-evaluative processes account
for one of two collective roots of self-evaluation. The second one eventually shifts
the perspective of the investigation to considering the morally relevant dimen-
sion of self-evaluation focusing on self-evaluative attitudes. Self-evaluative attitudes
36 Moran (2001, 164).

37 Remember that it is not necessary for my argument that these claims be made in public. It is
sufficient to have taken a stance and thereby be able to avow one’s intentionality.
38 Cf. McGeer (1996, 509).
resulting from self-evaluative deliberation, such as shame or guilt, are special in two
respects: first, employing them amounts to taking a stance with regard to my own
behaviour whilst maintaining the commitments in question as I explained in the
last section; and second, they open the way into a social dimension by grounding
in the respective systems of norms, rules and values I observe as participant in the
according collectives.
That we respond to our own wrong-doing with feelings of shame or guilt rather
than with anger or disapprobation, which are typical emotions caused by the mis-
behaviour of others, corresponds to the first-personal relation we stand in with
our own behaviour. Shame or guilt will be due results from a self-evaluating pro-
cess in the course of which one has found a mismatch between the attitudes one
holds or the actions one performed and what one finds one ought to have done
in the according situation. Although one here takes up a somewhat distanced self-
reflexive perspective similar to a spectator’s stance, one’s self-approach towards
one’s inappropriate conduct remains personal, it is not entirely detached. In contrast
to theoretical evaluative judgments, one’s self-response is affective and so closer
to anger or disapproval that one equally feels towards one’s own and the moral fail-
ures of others. Self-referential evaluative attitudes engage with the subject’s conduct
from her first-personal, evaluative stance, based on the subject’s agential knowledge
of the misplaced behaviour and her knowledge about the rational relations hold-
ing among her commitments. The affective nature of these responses as well as
the adjustments of one’s mental and bodily activity subsequent to self-evaluation
make plausible that these are best characterised analogously to the kinds of influ-
ence and manipulation we can exercise over others. In self-evaluation, we take a
second-person stance towards ourselves that is characterised by our own perspecti-
val stance – including our endorsement of both, the standards by which we evaluate
ourselves and the evaluated behaviour.
I have taken the notion “situational appropriateness” for granted as reliable stan-
dard by which a subject’s behaviour can be judged right and wrong. However, what
counts as adequate way of feeling, desiring and also of believing in a certain situa-
tion is determined by the conventions, norms and values forming the background
conditions from which the subject approaches the world together with the sub-
ject’s bodily and mental dispositions. In the course of evaluating my behaviour, I
primarily take into account what would be adequate behaviour according to the nor-
mative principles I have learned to comply with when becoming a member of a
certain community before I develop my own system of behaviour-relevant norms
in critical reflection on the common sense.39 Inasmuch as this entails that I take
39 That we first adopt the norms of our immediate social environment without questioning their
objectivity is captured in Heidegger’s notion of everydayness. The self-concept developed in social
interaction first and foremost sets the usual way of life as normative standard. It is only after the
individual has recognised her own possibilities and talents that she is able to detach her self-view
from a non-individual oneself. Still, she cannot entirely leave the conventions and language of her
social context behind because they constitute the means by which she performs changes in her
normative system. Cf. Heidegger (2008 [1927]).
272 U. Schmid
responsibility for my behaviour and position myself towards it when evaluating

myself, self-evaluation understood as responsive activity constitutes a practice
regarding the ways I settle the questions posed by the outer world as well as by my
own behaviour. For the correctness-conditions of my self-ascriptions as well as the
standards I observe when judging the adequacy of my actions, beliefs and emotions
are of genuinely social origin. My capacity to be ashamed for my misbehaviour and
to feel guilty when I wronged somebody else is acquired by internalising rules and
values just as it is influenced by my emotional sensitivity. The phenomenon of gen-
erating self-evaluative attitudes is so to speak the place where an individual meets
the social.
14.9 Conclusion
Whenever I engage with a problem I apply techniques of problem-solving.
Whenever I engage in settling a question I apply techniques of responding by form-
ing and employing intentional attitudes. The way I make up my mind observes
certain standards of correctness in a conceptual as well as in a moral dimen-
sion, both being constituted by the values, norms and rules obtaining in my social
environment. I have argued that the process generating self-referential evaluative
attitudes comprises both collective elements and likewise rests on the subject’s
first-personal relation to her mind bringing about the special authority and responsi-
bility she holds over them. The activity of self-evaluation thus conceived usually is
performed without the subject being conscious of it; however, she knows of it agen-
tially by endorsing the evaluative attitudes constituting her response to herself. The
account of self-evaluation I propose thus regards it as vitally influenced by both, the
evaluator and her collective environment.40
References
Anscombe, G. E. M. 1950. Intention. Oxford: Blackwell.
Burge, T. 1996. “Our Entitlement to Self-Knowledge.” Proceedings of the Aristotelian Society
96:91–116.
Gertler, B. 2008. Self-Knowledge. The Stanford Encyclopedia of Philosophy (Winter 2008
edition), edited by E. N. Zalta, URL = <http://plato.stanford.edu/archives/win2008/entries/
self-knowledge.
Heidegger, M. 1927. Being and Time. Oxford: Blackwell. 2008. J. Macquarrie (ed.).
Hieronymi, P. 2009. “Two Kinds of Agency.” In Mental Actions, edited by L. O’Brien and M.
Soteriou. Oxford: Oxford University Press. Page numbers refer to the draft version, available
from http://www.humnet.ucla.edu/humnet/phil/faculty/Phieronymi (download 22/06/08).
40 Iam indebted to the audience of the Workshop Self-Evaluation: Individual and Collective at
Basel in January 2009, in particular to Anita Konzelmann Ziv and Carol Rovane by whom I
received more than helpful and detailed comments.
McGeer, V. 1996. “Is “Self-Knowledge” an Empirical Problem? Renegotiating the Space of

Philosophical Explanation.” The Journal of Philosophy 93 (10):483–515.
Moran, R. 1997. “Self-Knowledge: Discovery, Resolution and Undoing.” European Journal of
Philosophy 5 (2):141–61.
Moran, R. 2001. Authority and Estrangement. An Essay in Self-Knowledge. Princeton, NJ:
O’Brien, L. 2003. “Moran on Agency and Self-Knowledge.” European Journal of Philosophy
11:375–90.
O’Brien, L., and M. Soteriou 2009. Mental Actions. Oxford: Oxford University Press.
Pacherie, E. 2007. “The Sense of Agency and the Sense of Control.” Psyche 13 (1):1–30.
Peacocke, C. 1996. “Our Entitlement to Self-Knowledge.” Proceedings of the Aristotelian Society
96:117–58.
Peacocke, C. 2009. “Mental Action and Self-Awareness (II): Epistemology.” In Mental Actions
and Agency, edited by L. O’Brien and M. Soteriou, 192–214. Oxford: Oxford University Press
Prinz, J. 2007. The Emotional Construction of Morals. Oxford: Oxford University Press.
Rovane, C. 1998. The Bounds of Agency. Princeton: Princeton University Press.
Searle, J. R. 1983. Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge
University Press.
Seemann, A. 2008. “Person Perception.” Philosophical Explorations 11 (3):245–62.
Strawson, P. F. 1990 [1959]. Individuals. An Essay in Descriptive Metaphysics. New York:
Routledge.
Taylor, C. 1985. Philosophical Papers I. Cambridge: Cambridge University Press.
Wittgenstein, L. 1958. Philosophical Investigations, Translated by G. E. M. Anscombe. Oxford:
Blackwell.
Wright, C. 1998. “Self-Knowledge: The Wittgensteinian Legacy.” In Knowing Our Own Minds,
edited by Wright C. et al., 13–46. Oxford: Oxford University Press.
About the Authors
Lynne Rudder Baker is distinguished professor of Philosophy at the University of

Massachusetts, Amherst.
John Deigh is professor of Philosophy at the University of Texas, Austin, TX.
Julien A. Deonna is SNF-professor of Philosophy at the University of Geneva,
Switzerland.
Juliette Gloor is Ph.D. candidate at the Philosophy Department of the University
of Basel, Switzerland.
Edward Harcourt is lecturer in Philosophy at Keble College, Oxford, GB.
Anita Konzelmann Ziv is assistant professor at the Philosophy Department of the
University of Geneva, Switzerland.
Federico Lauria is Ph.D. candidate and SNF assistant at the Philosophy
Departments of the Universities of Geneva and Fribourg, Switzerland.
Keith Lehrer is the Regent’s professor emeritus of Philosophy at the University of
Arizona and the University of Miami, Florida.
Lilian O’Brien is lecturer at the Philosophy Department at University College Cork,
Ireland.
Alain Pé-Curto is Ph.D. candidate and SNF assistant at the Philosophy
Departments of the Universities of Geneva and Fribourg, Switzerland.
Jesse Prinz is distinguished professor of Philosophy at the Graduate Center of the
City University of New York and research professor at UNC/Chapel Hill.
Carol Rovane is departmental chair professor of Philosophy at Columbia
University, NY.
Hans Bernhard Schmid is SNF-professor of Philosophy at the University of Basel,
Switzerland.
DOI 10.1007/978-94-007-1266-9, C Springer Science+Business Media B.V. 2011
276 About the Authors
Ulla Schmid is Ph.D. candidate at the Philosophy Department of the University of

Basel, Switzerland.
Axel Seemann is associate professor of Philosophy at Bentley University, Waltham,
Massachusetts.
Fabrice Teroni is lecturer at the Philosophy Department of the University of Berne
and researcher at the Center for Affective Sciences in Geneva, Switzerland.
Index
A Chisholm, R., 4, 45–46, 52, 54, 56

Ability, 4, 9, 19, 34, 36–37, 49, 51, 53, 55, Christine, K., 6, 84, 98
101–102, 143, 161–162, 168, 179, 183, Commitment, 5–6, 12, 21, 24, 27, 84, 87,
185–187, 189, 215–235, 237, 267 89–90, 92–94, 110, 115–116, 224, 234,
Achievement (term), 5, 12–16, 23, 84, 107, 238, 240–241, 246, 250, 258, 261–262,
114, 120–121, 125, 129, 144, 200, 264–271
215–217, 223–224, 227–228, 263 Comparative evaluation, 81–82, 95
Action, 97–117, 215–235 Consensus, 121–124, 131
Affective attitude, 3, 99, 106, 227–230, Convergence, 122–123
234–235
Agency, 2, 5–6, 10, 45–46, 82–83, 90–91,
D
94–95, 97–98, 103–104, 106–114,
Darwall, S., 156–160
158–160, 168, 216, 219, 231, 234, 238,
Deigh, J., 7–8, 155–169, 199–202
241–242, 245, 253, 257, 264, 269
Aggregation, 42, 121–122, 124–128, 131 Deliberation, 5–6, 15, 81, 83–89, 91–92, 111,
Akeel, B., 84 136, 222, 238, 241–242, 256, 258–260,
All things considered judgments, 83–84, 88–92 262–263, 265, 270–271
Anger, 7, 10, 47–50, 53, 56, 136–137, Deonna, J., 8, 193–211
139–144, 146–151, 155–156, 158–159, Desires, 1, 6, 12, 24–25, 55, 68–71, 73, 86, 91,
195, 197, 199, 204, 209, 229–230, 271 97–110, 113, 116, 183, 189, 216, 251,
Attachment theory, 187 258–259, 264, 270
Avowal, 4, 27, 45–57, 259–261, 266–267 Diggory, J., 19, 26
Axel, S., 4, 45–57, 253 Direct apprehension, 16
Discursive, 15
B Disgust, 7, 34–35, 68, 137, 139–143, 146–147,
Baker, L. R., 3–4, 33–43 149–150
Behaviour, 59, 62–64, 66–68, 73, 75–77, 166, Dissociative identity disorder, 90
173–175, 177–179, 181, 189, 201, Diversity, 6–7, 18, 125–129, 131–132
203–204, 243, 251, 253–261, 266–272
Benedict, R., 196 E
Ben Ze’ev, A., 25 Edward, H., 8, 171–190
Emotional recalcitrance, 228–229
C Emotions, 7–9, 70–71, 135–152, 155–169,
Carol, R., 40, 81–95, 261 171–190, 193–211
Causation, 4, 45, 101, 103, 108 Ends in themselves, 94–95
Character, 2, 11, 17–23, 25, 46, 50, 56, 59–77, Evaluation, 6–7, 11–15, 17–18, 20–27, 74–75,
83, 128, 131, 140, 155, 158, 164, 167, 81, 94, 98–99, 105–106, 111, 113–114,
193, 205, 220, 230, 255, 259–260, 267, 117, 119, 121–122, 128, 136, 194, 209,
270 211, 229, 238–241, 247
277
278 Index
Evaluative judgment, 14–15, 17–18, 21, 24–25, Intentional attitudes, 254–270, 272
110, 251, 271 Interpretation, 4, 10, 45–57, 98, 102, 130, 132,
Exemplarization, 6, 120–121, 128–131 138, 200–202, 205, 225, 255, 267
Expertise, 14–17, 125–128 Intersubjectivity, 4, 53, 121–122
Expression, 1, 12, 14, 52–53, 98–101, 105,
144, 162, 171, 174–177, 188–189, J
203–204, 224, 228, 261 Janus-faced, 130
Externalism, 23, 26, 33, 39, 43, 84, 129–131, Joint action (action, joint), 9, 115, 217,
262, 266 231–235
F K
Fabrice, T., 1, 8, 102, 114, 193–211 Kant, I., 113
Kantianism, 81–82, 86–87, 94, 109–110, 113,
Felt evaluation, 24
269
First-personal perspective, 257, 259
Knowledge
First-person perspective, 3–4, 33–34, 36–38, discursive, 15
41–43, 47, 51, 238, 241, 249, 260
implicit, 13, 17, 25, 45, 56, 85, 91, 161,
First person point of view, 84 217, 220, 228
Fixed point, 122–124, 126, 128, 131 knowing that, knowing how, 12, 14, 16, 70,
Frankfurt, H., 1, 91 255–256
Frankfurt, S., 1, 67–69, 91 reflective, 1, 3, 5–6, 10, 15, 19–23, 25,
26–28, 48–49, 55, 82–87, 91–95,
97–114, 116–117, 119, 187, 203–204,
G
269
Gloor, J., 6, 97–117
Know thyself, 1, 11, 22–23
Goodman, N., 120 Korsgaard, K., 2, 84, 86, 98–99, 101–103,
Gratitude, 141, 156, 161, 163 105–106, 109–110, 112–114
Group agents, 5–6, 27–28, 82, 93–94
Guilt, 3, 7–8, 10, 24–27, 140, 144–147, L
151–152, 156–157, 159–160, 164–165, Lauria, F., 5, 59–77
193, 196, 254, 257, 267, 271 Lehrer, A., 131
Lehrer, K., 1, 6, 12, 15, 119–132
H Linguistic and social concepts, 4, 39
Helm, B., 24–25 Loop, 6, 19, 120, 123, 128–129, 131, 189
Hong, L., 125–126 Love, 3, 8, 27, 33, 35–36, 60–62, 68, 73–74,
Hume, D., 120, 131, 135, 139, 141, 164–165, 161–162, 171–190
167, 228
M
I Mental agency, 269
Mentalization, 186–189
Identity, 35, 51, 67–73, 83, 87–95, 109
Mirroring, 188
Immanent causation, 4
Modest sociality, 242, 246, 248
Indignation, 7–8, 26, 141, 156–157, 159–160, Moore, G., 16, 52, 263, 269
164–168, 197, 199
Moral agency, 158–160, 168
Individuation of mental states, 4, 46, 51, 55, 57 Moral judgment, 7, 135–144, 147–150, 152,
Injustice, 7, 76, 136, 141, 156–157, 160, 165 156–160, 168–169
Instrumentalism, 97 Moral obligation, 81, 166–169, 243
Intention (action-referential vs. propositional), Moral responsibility, 7, 106, 160, 165–167,
223 169
Index 279
Moran, R., 2, 46–48, 67, 255, 257–265, 267, Respect, 23, 27, 74, 76, 101–102, 104, 110,
269–270 112, 126, 136, 140, 156, 160, 194, 204,
Mulligan, K., 12–13, 21, 24–25, 70–71, 73, 77 208–210, 222, 239–240, 243, 254–255,
Murdoch, I., 183–184, 189 257–258, 261–262, 269
Rights, 8, 87, 89, 94–95, 140, 142, 150, 157
Rousseau, J. –J., 162
N
Normative beliefs, 73 S
Scheler, M., 61, 76, 205–206
O Schlick, M., 121
Obligation, 8, 10, 47, 81, 106, 156–158, Schmid, H. B., 215–235
166–169, 234, 237, 241–246, 248–251 Schmid, U., 10, 253–272
O’Brien, L., 9, 237–252 Self
-concept, 3–4, 33–38, 40–41, 271
O’Hear, A., 203–207, 210
-confidence, 226, 229–230, 234
Other minds, 8, 171–173, 175, 255 -efficacy, 216–217
-evaluation, 1–28, 33, 42, 77, 81–95,
P 97–117, 119–132, 147–148, 193, 196,
Page, S. E., 7, 125, 129 237–252, 253–272
-knowledge, 1–2, 8, 10–11, 19, 21–24, 26,
Pé-Curto, A., 5, 59–77
51, 98–99, 102, 171–190, 253–272
Person, 3–5, 8, 19–28, 33–38, 41–43, 47–51, -reference, 1–2, 11, 101
53–55, 59–60, 62–64, 66–67, 70, -representation, 1, 121, 128–129
72–77, 84, 91, 99, 108, 125, 131, -trust, 3, 6–7, 119–132, 229–230, 235
136–138, 143–144, 146, 148–150, Sentimentalism, 7, 135–152
156–157, 160, 162, 164, 166, 168, Shame and, 193–211
172–183, 199, 204, 218–220, 224, 228, audience, 196, 198–200, 203–204, 207–209
231, 233, 238, 241, 243, 249, 255, 257, autonomy, 105–107, 110, 112, 122
260–261, 270–271 change of perspective, 202–204, 206–207
Personal identity, 34 heteronomy, 9, 195, 197, 199
Personality, 22–23, 59–77, 155–157, 159–160, reputation, 3, 209–210
188 Shared intention, 10, 115–116, 132, 237,
Phenomenology, 9, 50, 103, 115, 197, 199, 241–251
204, 215–235, 237 Strawson, P. F., 7–8, 156, 160–169, 176, 255,
Practical consistency, 228 260
Prinz, J., 7, 135–152
T
Process (term), 6, 11–12, 14, 21, 46, 50–51,
68, 87, 120–122, 129–131, 138, 145, Taylor, G., 24, 203–204, 206–207
158, 189, 195, 199, 239, 253, 258–260, Traits, 5, 22–23, 59–63, 66–67, 72, 74–76,
262–263, 269, 271–272 204–205
Trust, 61–62, 116–117, 140, 186, 215, 217,
Psychoanalysis, 47, 171 234
Psychopath, 138, 155–156, 159, 168 Trying, 5, 54, 61, 65, 90, 101, 138, 145, 182,
220, 222, 224–226, 248
R
Rationality, 2–3, 5–7, 17, 48, 75, 79–132, U
219–221, 226–229, 233, 240, 259, 269 Understanding, 3–5, 8, 10, 19, 22, 39–41,
Reactive attitudes, 7–8, 24, 101, 156, 160–169 43, 47, 50–55, 59, 61, 74–77, 82, 97,
99, 101–102, 108, 116, 157–158, 166,
Recalcitrant emotions, 228–229
168–169, 171, 186–187, 194, 196,
Reddy, V., 53–54 198–199, 209, 225, 238, 248, 253–254,
Reflective reflexivity, 19–23, 25–28 260, 263, 267, 269
Resentment, 7–8, 26, 50, 155–169, 204 Utilitarianism, 82
280 Index
V Weights, 121–128, 131

Values, 1–3, 5, 7–8, 11, 15, 18, 20–28, 59, Winnicott, D. W., 171, 184
71–77, 81, 95, 126, 139, 142, 147–152, Wittgenstein, L., 8, 40, 171–179, 181,
194–195, 201, 205–210, 271–272 184–185, 255, 261
Wollheim, R., 193, 197–199, 201, 209
W
Wagner, C., 121 Z
Wanton, 91–93, 251 Ziv, A. K., 1–28, 269

Ziv, Lehrer, Schmid (Eds.) (2011) - Self-Evaluation. Affective and Social Grounds of Intentionality

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Ziv, Lehrer, Schmid (Eds.) (2011) - Self-Evaluation. Affective and Social Grounds of Intentionality

Hochgeladen von

Copyright:

Verfügbare Formate

Self-Evaluation

PHILOSOPHICAL STUDIES SERIES

Founded by Wilfrid S. Sellars and Keith Lehrer

Stephen Hetherington, The University of New South Wales, Sydney, Australia

Senior Advisory Editor

Keith Lehrer, University of Arizona, Tucson, AZ, U.S.A.

Board of Consulting Editors

Lynne Rudder Baker, University of Massachusetts, Amherst, MA, U.S.A.

For further volumes:

ISBN 978-94-007-1265-2 e-ISBN 978-94-007-1266-9

Library of Congress Control Number: 2011927921

© Springer Science+Business Media B.V. 2011

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Self-evaluation is a term recently reassessed in academic philosophy by the joint

1 Self-Evaluation – Philosophical Perspectives . . . . . . . . . . . . . 1

Part I Evaluative and Self-Evaluative Attitudes

Part II Self-Evaluation and Rationality

Part III Self-Evaluative Emotions

11 Is Shame a Social Emotion? . . . . . . . . . . . . . . . . . . . . . . 193

Part IV Evaluating the Social Self

Lynne Rudder Baker Department of Philosophy, University of Massachusetts,

Carol Rovane Department of Philosophy, Columbia University, New York, NY,

Anita Konzelmann Ziv

Self-evaluation is not a technical term in philosophy.1 Rather, it is common to

A. Konzelmann Ziv (B)

A. Konzelmann Ziv et al. (eds.), Self-Evaluation, Philosophical Studies Series 116, 1

1.1.1 Evaluative and Self-Evaluative Attitudes

1.1.2 Self-Evaluation and Rationality

1.1.3 Self-Evaluative Emotions

presuppose specific relations between individuals, qualified by specific expectations

1.1.4 Evaluating the Social Self

regards herself as subject to evaluation, as a failure in certain circumstances and a

Herewith I conclude the first part of this introduction to philosophical per-

1.2.1 Evaluation and Knowledge

5 My view of judging as asserting propositionally structured “thoughts” or “contents” is not com-

specified sense of reflective knowledge, or discursive knowledge, referring not only

1.2.2 The Reflexive Character of Self-Evaluation

itself in a loop-like movement. Self-repairing flight control or self-cleaning ovens

14 A systematic and historically-informed discussion of views on the reference of “self” can be

illness. Contrary to completed actions, propositional thought or objective images,

i. Externality: To the extent that self-evaluation is reflective reflexivity that

Invoking externalism is a rough-and-ready way to express the view that self-

22 For a recent discussion of (non)compatibility between externalism and knowledge internalism

1.2.3 Affective and Social Dimensions of Self-Evaluation

According to axiological emotivism, a variety of subjectivism, value statements do

28 Objective self-evaluation is not to be confused with “gratuitous self-assertions” (Diggory 1966,

Lynne Rudder Baker

Self-directed and self-evaluative attitudes are often connected to one’s social

A. Konzelmann Ziv et al. (eds.), Self-Evaluation, Philosophical Studies Series 116, 33

2.1 Self-Directed Attitudes

possible. There is an enormous range of self-directed attitudes – from self-disgust

2.1.1 The Simple Argument

2.2 Defense of First Premise of the Simple Argument

Thus, the first premise of the Simple Argument is true.

2.3 Defense of the Second Premise of the Simple Argument

(2a) x has a robust first-person perspective only if x has a self-concept.

4 My notion of having a concept deliberately collapses Peacocke’s distinction between attribution

Here is a somewhat crisper characterization of L&S concepts:

of which is not usually thought of as dependent on social and linguistic prac-

6 A fugue is a polyphonic composition with a short subject harmonized by rules of counterpoint. A

2.4 A Moral About Individuals

Implicit in Roderick Chisholm’s (1964/2002) argument in favor of the idea of

A. Konzelmann Ziv et al. (eds.), Self-Evaluation, Philosophical Studies Series 116, 45