KR 6 KJ 6 RDTK

Panayot Butchvarov
Anthropocentrism in Philosophy
Brought to you by | Michigan State University

Authenticated
Download Date | 7/1/15 4:53 PM
Eide
Foundations of Ontology
Edited by
Javier Cumpa Arteseros, Jorge J. E. Gracia,
E. Jonathan Lowe, Peter Simons and Erwin Tegtmeier
Volume 8

Authenticated
Panayot Butchvarov
Anthropocentrism
in Philosophy
Realism, Antirealism, Semirealism

Authenticated
ISBN 978-1-61451-792-4
e-ISBN (PDF) 978-1-61451-849-5
e-ISBN (EPUB) 978-1-61451-947-8
ISSN 2198-1841
Library of Congress Cataloging-in-Publication Data

A CIP catalog record for this book has been applied for at the Library of Congress.
Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available on the Internet at http://dnb.dnb.de.
© 2015 Walter de Gruyter Inc., Boston/Berlin

Printing: CPI books GmbH, Leck
♾ Printed on acid-free paper
Printed in Germany
www.degruyter.com

Authenticated
For Karter, Kole, and Norah

Authenticated
Authenticated
Contents
Chapter One: Introduction 1
 Anthropocentrism 1
 A Glance at History 9
 Antirealism and its Varieties 15
 Logical Antirealism and Semirealism 20
Part One: Epistemology and Ethics Dehumanized
Chapter Two: Three Varieties of Epistemology 29

 Naturalistic Epistemology 29
 Subjective Epistemology 32
 Epistemology-as-Logic 42
Chapter Three: The Property Good 52

 Anthropocentrism and Conceptual Analysis in Ethics 52
 The Good and the World 59
 The Relevance of the Property Good 62
Chapter Four: Saying and Showing The Good 66

 The distinction explained 66
 Logic and the World 71
 The World and the Good 74
Part Two: Metaphysics Humanized
Chapter Five: The Role of Language in Cognition 85

 The empirical and the a priori question 85
 Philosophical opinions 91
 Scientific opinions 94
 Language in logical cognition 101
Chapter Six: Metaphysical Realism and Logical Antirealism 104

 The Logic of Realism 104
 Antirealism: ontological, cosmological, and logical 111
 The logical structure of the world 116
 Frege and Russell on Negation and Generality 122

Authenticated
VIII Contents
Chapter Seven: Logical Semirealism 130

 Ineffability 130
 Wittgenstein on Generality 133
 Bergmann on Generality 141
Chapter Eight: Generic Statements 151

 The Ubiquity of Generic Statements 151
 Facts, Generic Facts, and Realism 155
 The Irreducibility of Generic Statements 159
 Logical experiences 165
Chapter Nine: Facts and Truth 169

 Realism and antirealism regarding facts 169
 Semirealism regarding facts 178
 Truth 180
Part Three: Metaphysics Dehumanized
Chapter Ten: I and the World 189

 The paradox of antirealism 189
 First-person singular pronouns 192
 The Self 198
Chapter Eleven: We and the World 203

 Consciousness 203
 The I that is We and the We that is I 206
 Idealism 212
Chapter Twelve: Mind and the World 218

 Concepts, Properties, and Universals 218
 Solipsism and Pure Realism 225
 Philosophical method 232
 Philosophy without anthropocentrism 235
Bibliography 238
Index 244

Authenticated
Chapter One: Introduction
1 Anthropocentrism
Anthropocentrism, the belief that humans enjoy special, central, even cosmic
significance, is present in everyday thought as an attitude toward other animals
and the environment generally, and in religion as the Biblical teaching that hu-
mans alone were made in the image of God. “I am unable to believe that, in the
world as known, there is anything that I can value outside human beings, and, to
a much lesser extent, animals,” wrote Bertrand Russell.¹ Many think that such
anthropocentrism mars our relationship to other animals and the environment,
just as egocentrism mars our relationship to other humans. Speciesism, they
would say, is no more acceptable than is egoism, androcentrism, or ethnocen-
trism. Many also think that the anthropocentrism in religion mars our conception
of God. They would agree with Spinoza that, contrary to standard religious doc-
trine, “neither intellect nor will pertain to the nature of God,”² and that “God is
free from passions, nor is He affected with any emotion of joy or sorrow.”³ To at-
tribute to God human characteristics such as intellect, will, joy, or sorrow, they
would say, is to think of God as a sort of superhuman.
These instances of anthropocentrism are well-known and have been amply
discussed for centuries. They are not the topic of this book. Its topic lies deeper:
the anthropocentrism present, though seldom discussed or even acknowledged,
in philosophy, the discipline charged with our most fundamental thinking –
about knowledge (in epistemology), goodness (in ethics), and the world itself
(in metaphysics).
Ethics is commonly understood as concerned with human well-being, even
happiness, and epistemology with human knowledge, especially perception. But
these are empirical matters, investigated today in psychology and neuroscience,
philosophers generally lacking the qualifications or even inclination for empiri-
cal research. Ethics and epistemology remain anthropocentric even when con-
cerned only with language, because the language in question is surely human
and investigated properly in linguistics and lexicography. In metaphysics, an-
thropocentrism takes the form of antirealism, the orientation that has dominated
philosophy since Berkeley and especially Kant. Broadly understood, it claims
that the world depends, at least insofar as it is knowable, on our cognitive ca-
 Paul Arthur Schilpp, The Philosophy of Bertrand Russell (La Salle, IL: Open Court, 1989), 19 –20.
 Benedict de Spinoza, Ethics (New York: Hafner, 1963), Part One, Proposition VII, Note.
 Ethics, Part Five, Proposition XVII.
Brought to you by | University of Michigan

Authenticated
2 Chapter One: Introduction
pacities. The claim seems absurd if taken to mean, as it often is, that we, hu-
mans, “make the world.”
I shall argue that, if properly understood, epistemology is not about human
knowledge and ethics is not about the human good despite the fact that we all
desire the human good and respect human knowledge, and that metaphysics is
not about “us,” despite the tautology that we can know the world only as it can
be known by us. My argument will rest not on abstract and often enigmatic phil-
osophical premises but on specific and readily understandable truths.
Whatever the nature of the world may be, humans are only inhabitants of it.
The world can hardly depend on them. And knowledge of humans, like knowl-
edge of its other inhabitants, is credibly sought only by empirical, evidence-
based methods. But philosophy is not an empirical discipline, and its claims
are seldom supported with empirical evidence. Philosophers perform no experi-
ments, maintain no labs, use neither telescopes nor microscopes, embark on no
field trips. The moral to be drawn, however, is not that philosophers are experts
on nonempirical things or facts. If numbers are such things, it is mathemati-
cians, not philosophers, who specialize in them.
Concern with human beings, of course, is natural and morally expected of us
all. It is a professional concern, however, only for some: neuroscientists, psychol-
ogists, psychiatrists, economists, sociologists, anthropologists, demographers,
linguists, lexicographers, physicians. Aristotle did engage in biological investiga-
tions, but at the time biology was hardly a science. Today it is.
Philosophers’ willingness to assume authoritative stands on human beings
became especially incongruous when the experimental sciences devoted to the
study of humans emerged. For most of the history of philosophy and science,
if a topic did not belong in mathematics, astronomy, medicine, or theology, it
was dispatched to philosophy – there seemed to be no other place to put it.
Even today, in some institutions psychology is called “mental philosophy” and
physics “natural philosophy.” But neither is considered part of philosophy,
and few philosophers today can claim expertise in psychology or in physics.
I shall be concerned here with the ways anthropocentrism has affected epis-
temology, ethics, and metaphysics, arguing that it has no place in them, that all
three should be radically refocused. The same reasoning would apply, directly or
indirectly, to the other branches of philosophy – from the philosophy of art and
of science to political philosophy and the philosophy of education – but they will
not be discussed here. They all depend in part on theories developed in metaphy-
sics, epistemology, or ethics, but also on developments in fields like history, eco-
nomics, or psychology. Logic is an exception, for reasons to be explained shortly.
Suffice it to note here that, insofar as it belongs in philosophy rather than math-
ematics, it is a part of metaphysics.

Authenticated
1 Anthropocentrism 3
Some may say that not all of human nature is empirical, that humans also
have immortal nonphysical souls. But this is a matter of faith, not investigation,
empirical or not. Others may say that even if humans have no immortal souls
they have nonphysical minds, entirely distinct from both their brains and their
behavior. But there has been an empirical science investigating such minds:
the introspective psychology of James, Wundt, Titchener, and many others. To
be sure, it was largely unsuccessful, though not because its subject matter called
for nonempirical investigation – the introspective psychologists explicitly relied
on experience, often in collaboration with others, sometimes in “laboratories.”
Much the same can be said about continental phenomenology in its early stages,
which was a close relative of introspective psychology and was summed up in
Husserl’s slogan “We must go back to the things themselves,” back to what
we actually find before us, rather than what philosophical or scientific theory,
or even common sense, says is there.
In Husserl’s later works, and especially Heidegger’s and Sartre’s, phenomen-
ology evolved into a kind of metaphysics, similar to Kant’s transcendental ideal-
ism, Hegel’s absolute idealism, or even Nelson Goodman’s “irrealism.” Its chief
tenet became that the empirical world itself is in some sense human, “made by
us,” as Goodman put it. This was essentially the thesis of antirealism, in a very
broad sense of the term that would apply to Berkeley’s “immaterialism” as well
as to Kant’s, Hegel’s, and Goodman’s views. Much of this book will be devoted to
that thesis.
More likely today is to be told that in fact philosophical inquiries are not
about human beings, that they really are conceptual or linguistic. They are
about concepts or words, not about the things or facts, human or nonhuman,
those concepts or words stand for. For example, it would be said, in ethics phi-
losophers investigate the concept of happiness or the use of the word “happi-
ness,” not any facts about happiness, which indeed are usefully investigated
today by psychiatrists and pharmacologists, and in epistemology they investi-
gate the concept of perception or the use of “perceive,” not any facts about per-
ception, which for centuries have been investigated by psychologists and in re-
cent years also by neuroscientists. But surely the concepts and words in
question are themselves human, not platonic or divine, and thus are part of
an empirical subject matter. The investigation of them calls for observation
and sometimes experiment – as in psychology, neuroscience, linguistics, and
lexicography – not philosophical speculations, intuitions, a priori arguments,
analyses, or definitions.
That this is so is hardly news. It was powerfully argued more than half a century
ago by W.V. Quine in “Two Dogmas of Empiricism,” where he attacked philosophical
appeals to meanings. At about the same time Wittgenstein’s Philosophical Investiga-

Authenticated
tions was published, arguing in part that words in ordinary language are not used in
accordance with necessary and sufficient conditions, and therefore that their use
cannot be captured in definitions. In the same decade Gilbert Ryle castigated
“the confusion between a ‘use’, i.e., a way of operating with something, and a
‘usage’…. A usage is a custom, practice, fashion or vogue…. The method of discov-
ering linguistic usages are the methods of philologists.”⁴ Also in that decade, Chom-
sky began publishing articles and books that stressed the biological, largely inher-
ited, core of linguistic competence, and urged that the study of language employ the
standard methods of scientific research.
The traditional claim of philosophy to a distinctive place among the cogni-
tive disciplines has rested on its absolute fundamentality, supreme abstraction,
and unlimited scope. In these respects it surpasses even mathematics: one of its
topics is the subject matter of mathematics itself. Its scope includes that of phys-
ics and astronomy – space, time, and whatever is in them – but philosophy is
also concerned with anything that is not or might not be in space and time. Phi-
losophy presupposes nothing and conceals nothing. This is why philosophers
court paradox when preoccupied with things as concrete, literally “down to
earth,” as humans. The paradox is no less glaring than it would be if they
were preoccupied with cetaceans. If some do not see the paradox, the reason
presumably is that they are human. Had they been cetacean, they might have
been preoccupied with cetaceans.
The concern in philosophy with humans is not a trivial consequence of its
unlimited scope, of its interest in “all time and existence.” It is not the trivial ap-
plication to humans of general philosophical propositions, like the application
to humans of arithmetic by the Census Bureau or of physics by a pilot monitoring
takeoff weight. It is supposed to be a substantive concern. It may be woefully
misguided, but it is natural. The reason is obvious. Plumbers or philosophers,
we all are humans. We are deeply interested in ourselves and other humans.
We see ourselves as the center of the universe even when we know that we
are at its periphery. To suggest that philosophy should not be about humans,
that it ought to be in this sense “dehumanized,” may seem even offensive. “De-
humanized” is an ugly word, but it does capture literally and succinctly the aim
of the drastic change needed in philosophy – freedom from anthropocentrism –
just as “humanized” captures much of the current state of philosophy.
 Gilbert Ryle, “Ordinary Language,” Philosophical Review LXII (1951), 173 – 74. See also Gilbert
Ryle, “Use, Usage, and Meaning,” Proceedings of the Aristotelian Society Supplementary Volume
XXXV (1961).

Authenticated
A familiar thesis in ethics is that one ought to treat others “humanely,” as

ends, not just as means, perhaps even love them. The claims of kings and barons
to special dignity were rejected in the past by declaring the dignity of all hu-
mans. “Inhuman” and “inhumanity” are standard terms of condemnation, “hu-
mane” and “humanitarian” of approbation. Politicians’ handlers try to “human-
ize” their clients in order to get them elected. The “human condition” is a
perennial object of despair but sometimes of marvel. Works of art are often prais-
ed for their “human quality.” A favorable review of a recent novel emphasizes its
being “deeply human.” Many demand that space exploration be funded only if it
leads to cures for human diseases. A recent television commercial announces,
“The human element, nothing is more fundamental, nothing more elemental,”
and advocates adding it to the periodic table. A distinguished contemporary phi-
losopher writes of the “heart-breaking specialness” of the human.⁵
Indeed, we all think humans are special, and even feel their specialness. But
the reason is not that we think we are angelic and thus special in a way that, say,
cetaceans are not, or that we are intellectually, aesthetically, morally, etc., supe-
rior at least to all other terrestrial life. The plain reason is that we happen to be
human. We all desire and seek pleasure, happiness, well-being – for ourselves,
for those we love, often for strangers. So, ethics has devoted itself to investigat-
ing the human good, even happiness and pleasures, and the habits, actions, and
institutions conducive to it, rather than to the good of cetaceans, extraterrestri-
als, or angels. We all treasure our ability to see and hear, and to remember and
think. So, epistemology has devoted itself to investigating the nature and sources
of human knowledge, not cetacean, extraterrestrial, or angelic knowledge.
That human happiness and human knowledge are empirical matters, be-
longing today in the subject matter of developed empirical sciences, has been
pointed out repeatedly and eloquently by the proponents of naturalism, the
dominant orientation in current philosophy. Its most prominent defender,
W.V.O. Quine, called it “the recognition that it is within science itself, and not
in some prior philosophy, that reality is to be identified and described.”⁶
Quine limited the substantive claims in his “epistemology naturalized” to men-
tioning the role in cognition of what he called “surface irritations,” but wisely
left the investigation of these irritations to biology.⁷ Other naturalists, however,
seem to lack the courage of their convictions and continue to engage in “analyz-
 Cora Diamond, The Realistic Spirit: Wittgenstein, Philosophy and the Mind (Cambridge, MA:
MIT Press, 1991), 35.
 W.V.O. Quine, Theories and Things (Cambridge, MA: Harvard University Press, 1981), 21.
 Willard Quine, “Two Dogmas of Empiricism,” From a Logical Point of View (Cambridge, MA:
Harvard University Press, 1994).

Authenticated
ing concepts” or describing the “workings of our language,” as if concepts and

language were not themselves parts of nature and thus belonging in the province
of empirical science, not philosophy.
Anthropocentrism has been fueled by various assumptions, some true and
some false. We assume that – with the possible exception of gods, angels, or ex-
traterrestrials – humans alone are “rational,” capable of reasoning. We assume
also that only humans are capable of moral and aesthetic judgment, and per-
haps that only humans enjoy the moral and political status, “dignity,” of pos-
sessing “rights.” The monotheistic religions, which were the home of medieval
philosophy and also profoundly influenced modern philosophy, assure us of hu-
mans’ unique origin and special place in nature. Some of these assumptions and
assurances are matters of faith, e. g., that, though all things were created by God,
only man was created in God’s image. Some are essentially scientific but now
abandoned, e. g., geocentrism, the Ptolemaic view that the Earth is the center
of the Universe. Other assumptions are also essentially scientific but of unknown
and perhaps unknowable truth-value, e. g., that there is no intelligent life else-
where in the Universe. But many do seem to rest on scientific fact, e. g., that
no other terrestrial animals equal humans in intelligence. These assumptions
not only encourage our special interest in ourselves but also seem to justify it.
They help explain the presence of anthropocentrism in epistemology and ethics.
But antirealism, the form anthropocentrism assumes in metaphysics, is a
purely technical philosophical orientation, exemplified by positions ranging
from Berkeley’s “immaterialism” to Kant’s “transcendental” idealism, Hegel’s
“absolute” idealism, and recent positions such as Michael Dummett’s “antireal-
ism” and Nelson Goodman’s “irrealism.” The term “antirealism” may be too
strong to apply to all these positions, but there is no convenient alternative to
it. “Nonrealism” might be one, but it is too indefinite.
According to the metaphysical antirealist, reality depends, at least insofar as
it is knowable, on our ways of knowing it, our cognitive capacities – sense per-
ception, introspection, intellectual intuition, imagination, memory, recognition,
conceptualization, inductive and deductive reasoning, use of language and of
other symbolism. The rather stilted word “cognition,” rather than “knowledge,”
is needed to refer to the employment of these capacities, which leads to knowl-
edge when successful but to error when unsuccessful. So understood, antireal-
ism allows for the possibility of an unknowable reality, e. g., Kant’s “things-in-
themselves,” which is independent of our cognitive capacities, even if, as Good-
man claimed, it might not be “worth fighting for or against.”⁸
 Nelson Goodman, Ways of Worldmaking (Indianapolis: Hackett, 1978), 6.

Authenticated
The essentially Kantian claim that the world is “shaped,” “sculpted,” by our
cognitive faculties, and thus that it depends on us, who of course are humans,
amounts to the humanization of metaphysics. It rests on the tautology that we
have no cognitive “access” to the world except through our cognition of it,
that whatever we know, perceive, understand, believe, imagine, or say about
the world depends on our cognitive capacities. The world, of course, is cognized
also by nonhuman animals, as well as, perhaps, by extraterrestrials and angels,
but in order to know or understand and say even this we must rely on our cog-
nitive capacities, if only our imagination and language.
Anthropocentrism is paradoxical in all branches of philosophy, though for
different reasons. Epistemology and ethics claim expertise about what they
must and usually do regard as certain animals. But animals are part of the sub-
ject matter of the empirical sciences, not philosophy. Metaphysical antirealism
holds, however tacitly, that reality, the whole world – at least insofar as it is
knowable – depends on the cognitive faculties of those animals. But common
sense – the mature and thoughtful judgment we all share and all theorizing, sci-
entific or philosophical, begins with and must respect even if not accept – finds
such cosmological humanism, human creationism, bizarre.⁹ How could the
whole world, it asks, depend on some members of one of its planets’ fauna?
In the case of epistemology and ethics, anthropocentrism faces only the par-
adox of implying that philosophers, supposedly the spectators of all time and
existence, engage really in zoological investigations. In the case of metaphysics,
it faces the paradox of implying that the world itself is zoological. To avoid the
former paradox, we need only to redirect our efforts in epistemology and ethics.
But to avoid the latter paradox, we must do much more than redirection. We must
understand the first-person pronouns used in the formulation of antirealism as
impersonal and thus as not referring to humans. This would require a radical re-
thinking of their role. When used in philosophical contexts like Cartesian doubt
or the realism/antirealism debate, which question the existence of the world it-
self, consistency requires that “I” and “we” are not taken to refer to parts of that
world. The rethinking of the role of these pronouns would require also a radical
rethinking of the distinction between oneself and the world.
To redirect epistemology, ethics, and metaphysics away from anthropocen-
trism is not, of course, to abandon them. Metaphysics understood as ontology,
the listing and description of the most general kinds (“categories”) of entities
and the relations among them, would be unaffected. The realism/antirealism
 For a detailed and incisive discussion of common sense, see Noah Lemos, Common Sense
(Cambridge: Cambridge University Press, 2004).

Authenticated
issue is irrelevant to many metaphysical inquiries, especially those in pre-Kant-

ian philosophy. Also unaffected would be epistemology as tasked with the ap-
praisal of certain fundamental but nonformal inferences, for example, from
what has been the case to what will be the case,¹⁰ and ethics as the home of non-
anthropocentric theories of goodness like Plato’s, Aquinas’s, and Moore’s.¹¹
Although in epistemology and ethics anthropocentrism is natural and under-
standable, though indefensible, in metaphysics it is unnatural and almost incom-
prehensible, but at least as defensible as Kant’s transcendental idealism and 20th
century antirealism. In epistemology and ethics, we can reject anthropocentrism un-
qualifiedly, however painful this might be. In metaphysics, such rejection would be
a blunder. The reason is that it is present there as antirealism, which is made plau-
sible even if not entailed by a tautology and therefore truth. The tautology is that
whatever we know, perceive, understand, believe, imagine, or say about the
world depends on our cognitive capacities. So, in metaphysics we face the challenge
of finding a way to reject anthropocentrism without rejecting this tautology. For we
ought not to return to pre-Kantian metaphysics just in order to avoid anthropocen-
trism, much as we ought not to return to pre-Socratic philosophy just in order to
avoid Platonism. Yet human creationism is hardly acceptable. We must therefore in-
terpret antirealism as making no reference to humans.
Hence, the unusual dialectical structure of this book. Part One is devoted to
defense of the dehumanization of epistemology and ethics, Part Two to explan-
ation and provisional defense of the antirealist humanization of metaphysics,
and Part Three to the dehumanization of antirealist metaphysics. In Part One an-
thropocentrism in epistemology and ethics is rejected, in Part Two it is defended
in metaphysics, and in Part Three it is rejected in metaphysics. There is no need
for an explanation of the humanization of epistemology and ethics. The anthro-
pocentrism present in them is easily understood, and so is what motivates it. Not
so in the case of metaphysics. Its humanization was strictly a philosophical
event, obscure to virtually everyone outside philosophy and even to many pro-
fessional philosophers.
 See my “Epistemology Dehumanized,” in Quentin Smith, ed., Epistemology: New Essays (Ox-
ford University Press, 2008).
 See my “Ethics Dehumanized,” in Mark Timmons and Terry Horgan, eds., Metaethics After
Moore (Oxford University Press, 2006).

Authenticated
2 A Glance at History 9
2 A Glance at History
The history of anthropocentrism in philosophy is illuminating, though we can

devote to it here no more than a passing glance. It was preceded and encouraged
by the anthropocentrism already present in religion. According to Genesis, after
creating the heaven and the earth, including the earth’s flora and fauna, God cre-
ated man in the image of God. But Genesis attributed to God actions such as rest-
ing from work, speaking, and inflicting punishment. Thus it also appeared to de-
pict God in the image of man. Its anthropocentrism with respect to Creation
seemed accompanied by anthropomorphism with respect to the Creator. Cosmo-
logical anthropocentrism seemed to lead to theological anthropomorphism. This
personification of God became essential to Judaism, Christianity, and Islam.
Pre-Socratic philosophy focused on the cosmos – the heavens and the earth,
the four elements (earth, air, fire, and water), ensouled animals. The Socratic rev-
olution shifted the focus dramatically. It moved philosophy from concern with all
things to concern with just one. Its slogan became “Know thyself!” It did not go
unchallenged. The stand most distinctive of Plato’s philosophy was both anti-
preSocratic and anti-Socratic. The chief concern of philosophy, he taught, is nei-
ther the cosmos nor oneself, but rather what lies beyond both: the abstract en-
tities he called Forms (“Ideas,” eidos). Almost immediately, however, Aristotle
urged a return to the cosmos, away from both Socrates’ preoccupation with
the innermost and Plato’s preoccupation with the uttermost. Medieval philoso-
phers – Christian, Jewish, and Muslim – sometimes followed Plato, sometimes
Aristotle, but all took for granted the declaration in Genesis that humans were
made in the image of God.
Following the scientific revolution in the 16th and 17th century, when Copernicus,
Galileo, Boyle, and Newton became the authorities on space, time, and matter, phi-
losophers came to hold, in sharp contrast to their ancient and medieval predeces-
sors, that their concern was only with “minds” and “ideas.” The new physics com-
pelled them to adopt the “new way of ideas.” Anthropocentrism thus became firmly
established in philosophy, since it was not cetacean, extraterrestrial, or angelic
minds and ideas that attracted the attention of the early modern philosophers.
But their anthropocentrism consisted in preoccupation with just one part or aspect
of human beings: their minds and their ideas. It eventually led many to idealism,
the view that minds and ideas are all that there is.
Early in the 17th century, Descartes began his Meditations by arguing that “I
exist” was the only truth he could not doubt, presumably referring by “I” to himself,
Descartes, a certain human. Such reference, of course, would have been inconsis-
tent, since humans have bodies, the existence of which he still doubted. But after
his propaedeutic dalliance with epistemology, and thus with human matters like

Authenticated
the possibility and extent of human knowledge, Descartes focused on the existence
of God and the nature of mind and matter. This focus was metaphysical and non-
anthropocentric. It was shared later in the century by Spinoza and Leibniz. But it
was vigorously opposed by the British empiricists, who returned to Descartes’ epis-
temology and thus to anthropocentrism. The titles of their chief works spoke vol-
umes: An Essay Concerning the Human Understanding (Locke), The Principles of
Human Knowledge (Berkeley), A Treatise of Human Nature (Hume).
In 1781, less than half-a-century after the publication of Hume’s Treatise,
Kant rejected in his Critique of Pure Reason both continental metaphysics,
which he called dogmatism, and British epistemology, which he called empirical
idealism. He described his own philosophy as transcendental idealism, arguing,
for example, that space and time are not “actual entities” because they belong
only to “the subjective constitution” of the mind as “forms of appearances.”¹²
But Kant allowed for what he called “things-in-themselves,” entities independ-
ent of that subjective constitution of our minds, i. e., of our cognitive faculties.
Transcendental idealism is often described as part of the humanism character-
istic of the Enlightenment. But this is misleading, and so is the vague term “human-
ism.” Like any antirealist position, transcendental idealism does rest on the tautol-
ogy that our knowledge of the world depends on our cognitive faculties. On the
basis of this tautology, Kant went on to assert also that the world as knowable by
us depends on our cognitive faculties. But whether there is a world that is not know-
able was a further question. So was also the question whether there are parts or as-
pects of the knowable world that are not knowable.¹³ The latter question makes
Kant’s distinction between things-in-themselves and things-for-us appear less un-
reasonable, and his antirealism more plausible, by allowing that, even if only the
things we can know exist, these things may have parts or aspects that we cannot
know. Neither science nor common sense need disagree.
When Kant wrote that space and time belong only to the subjective constitution
of the mind, he was explicitly referring to humans. “We can … speak of space, ex-
tended things, and so on, only from the human standpoint,” he wrote, and ex-
plained that this is why space is “transcendentally ideal,” though also empirically
real in the sense that it is not an illusion, or mere fancy.¹⁴ But when he wrote
 Immanuel Kant, Critique of Pure Reason, trans. Paul Guyer and Allen W. Wood (Cambridge:
Cambridge University Press, 1998), B 37 ff.
 E.g., Critique of Pure Reason, B xxvii-xxviii.
 Critique of Pure Reason, B 42– 44. For the general distinction between transcendental and
empirical idealism, see A 369 – 370. Heidegger remarked that Kant’s “transcendental conception”
was “possible only on the basis of the subjectivity of man’s essence.” (“On the Essence of Truth,”
in D. F. Krell, Martin Heidegger: Basic Writings (New York: Harper & Row, 1977, 118.)

Authenticated
that space and time belong only to the subjective constitution of the mind, he could
not have meant Kant’s or any other human’s mind. For humans themselves, includ-
ing Kant, would be inhabitants of the transcendentally ideal space and thus tran-
scendentally ideal. To be sure, they also are empirically real: although Kant held
that our knowledge is limited to appearances, he made a sharp distinction between
appearance (Erscheinung) and illusion (Schein).¹⁵ Transcendental idealism was thus
anthropocentric insofar as humans are empirically real, but it was not anthropocen-
tric insofar as humans are transcendentally ideal. Nonetheless, it has been under-
stood by many as unqualifiedly anthropocentric. So have its immediate successors:
Fichte’s absolute idealism, as defended in The Vocation of Man, and Hegel’s abso-
lute idealism, according to which the Absolute achieves self-knowledge through the
human mind. We shall find that a nonanthropocentric reading of all three – Kant,
Fichte, and Hegel – is not only possible but plausible. To explain this reading would
be part of the challenge of dehumanizing antirealist metaphysics, which will be at-
tempted in Part Three.
Standard ethics, of course, is not concerned with humans – their well-being,
happiness, pleasure, etc. – as transcendentally ideal. It is concerned with them
only as empirically real. Kant called such ethics practical anthropology, which he
described as an empirical discipline, and reserved the term “metaphysics of mo-
rals” for what he considered the properly philosophical inquiry into morality.
Standard epistemology also regards humans – their cognitive capacities – as em-
pirically real. Kant did not have a special name for the empirical study of hu-
mans’ cognitive capacities, but presumably would have agreed today that it is
the task of what we call cognitive psychology and neuroscience. The properly
philosophical inquiry into our cognitive faculties, he thought, was exemplified
by his transcendental aesthetic and analytic.
Nevertheless, Kant certainly thought, as everyone does, that all disciplines –
the transcendental aesthetic and analytic as well as the physics of space and
time, the metaphysics of morals as well as practical anthropology – are anthro-
pocentric, indeed literally human, in the straightforward sense that they employ
human cognitive powers and thus are constrained by the demands and limita-
tions of those powers. Whether we contemplate all time and existence or just
snails and tomatoes, we cannot transcend our sense organs and brains, just as
we cannot get out of our skins. Understood as the tautology, on which antireal-
ism rests, that our knowledge of the world depends on our cognitive faculties,
this truth is not grounded in facts about human beings, just as the truth of
 Critique of Pure Reason, B 69.

Authenticated
the tautology “All humans are human” is not grounded in facts about humans.
Both enjoy the certainty of logic.
Antirealism became dominant after Kant in the form of straightforward ide-
alism. Early in the 19th century, Hegel declared that Spirit (mind, Geist) develops
or “unfolds” logically (“dialectically”) from “subjective spirit” (individual mental
states like sensations) to “objective spirit” (society as exemplified by the family,
customs and traditions (Sittlichkeit), the state, and institutions such as corpora-
tions and guilds, among which would be today’s academic disciplines), and
reaches its fulfillment in “absolute spirit,” the three stages of which are art, re-
ligion, and philosophy (in Hegel’s sense of philosophy as the perfect system of
knowledge). Thus, seen superficially, Hegel appeared to endorse not only meta-
physical anthropocentrism but also a sort of metaphysical anthropomorphism.
But his anthropocentrism involved a major epistemological innovation: a dra-
matic move from earlier philosophers’ cognitive individualism, the view of
knowledge as a personal achievement, to what, for the lack of a better term, I
shall call cognitive collectivism: the view of knowledge as a social, often literally
collaborative, achievement. It is especially evident today that the cognitive dis-
ciplines are inherently social and, at least to users of Wikipedia, that so is virtu-
ally all cognition beyond the infantile stage. Cognitive collectivism, of course,
need have no political implications. The state is not the only “collective.”
In the 20th century, Wittgenstein declared in Tractatus Logico-Philosophicus,
“I am my world,” seemingly referring, though if so inconsistently, to a human
being, namely himself.¹⁶ Heidegger focused his Being and Time on human
being, “the inquirer,” using Dasein as a technical term for it,¹⁷ though two dec-
ades later he insisted that “[e]very kind of anthropology and all subjectivity”
should be “left behind.”¹⁸ Sartre referred in Being and Nothingness to being-
for-itself as realité humaine, which he contrasted with being-in-itself ¹⁹, and
wrote the famous article “Existentialism is a Humanism,” to which Heidegger re-
sponded in the not less famous paper titled “Letter on Humanism.”
While the scientific revolution of the 16th and 177h century motivated philos-
ophers to confine their subject matter to minds and ideas, in the 20th century
 Ludwig Wittgenstein, Tractatus Logico-Philosophicus, trans. D. F. Pears and B. F. McGuinness

(London: Routledge, 1972), 5.63. In references to the text, I shall use Wittgenstein’s numerical
designations of sentences. All italics, upper-case letters, and parentheses in the quotations
will be Wittgenstein’s.
 Martin Heidegger, Being and Time (New York: HarperOne, Revised edition, 1962), 27.
 Martin Heidegger, “On the Essence of Truth,” section 9, Note, in Basic Writings, 138.
 Jean-Paul Sartre, Being and Nothingness, trans. Hazel E. Barnes (New York: Philosophical Li-
brary, 1956).

Authenticated
minds and ideas themselves became the subject matter of flourishing empirical
sciences – first psychology, then neuroscience – and so philosophers retreated
further. Instead of the way of ideas, they took this time the way of words, “the
linguistic turn,” claiming that their subject matter was language. Sometimes
they described their inquiries as conceptual, not factual, but by “concepts”
they did not mean the ideas or mental images of early modern philosophy.
They meant meanings or uses of words and syntactic structures, the “workings
of our language.”
But language and words are matters no less empirical than space and time,
or human minds and ideas, and they are investigated today by linguists, lexicog-
raphers, even computer scientists. Research in them requires meticulous empiri-
cal description and fruitful, empirically verifiable hypotheses, not “analyses”
and “definitions,” as in philosophical writings. Philosophers have no more spe-
cial insight into the workings of language than into the workings of matter or the
workings of the human mind. Instead of investigating space, time, and matter,
the 17th and 18th century philosophers settled for investigating mind. Instead of
investigating mind, the 20th century philosophers settled for investigating talk.
They remained open to the charge of inquiring into empirical matters. To be
sure, philosophers have a good ear for the nuances of some segments of speech,
as J. L. Austin famously did, but Austin insisted that such an ear was not a sub-
stitute for empirical research.²⁰
In ethics and epistemology some have eschewed the need for empirical in-
vestigation by taking “the deontic way,” concerning themselves with what
ought to be the case, or is right, justified, or valid, rather than with what is
the case. They have thus avoided encroaching on the empirical sciences,
which have no interest in deontic matters except to record and explain people’s
beliefs about them. The deontic way seems natural in ethics, given the latter’s
emphasis on imperatives. But a deontological ethics must ground its impera-
tives. Kant held that “the ground of obligation … must not be sought in the na-
ture of man or in the circumstances in which he is placed,” urging that “it is a
matter of the utmost necessity to work out for once a pure moral philosophy
completely cleansed of everything that can only be empirical and appropriate
to anthropology.”²¹ He appealed to the nonsensible world of noumena as the
ground of the “commands of practical reason.” Most contemporary deontologists
do not follow Kant, but they are aware that a ground is needed. A theory in eth-
 J. L. Austin, How to Do Things with Words (New York: Oxford University Press,1976).
 Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. H. J. Paton (New York:
Harper and Row, 1964), 57.

Authenticated
ics, of course, may be essentially an elaborate proposal of ways of acting, espe-

cially in politics, but to that extent it is not a cognitive contribution.
The deontic way is also natural in epistemology insofar as its concerns have
to do with the validity of certain inferences and thus are essentially logical. The
inferences in question are seldom deductive, and for this reason the need for
grounding them is especially evident. Much of epistemology has consisted in
the search for such a ground.
In contemporary ethics and epistemology the ground is often what philoso-
phers call “intuitions,” meaning usually their felt inclinations about what they
would or would not say is morally right or epistemically justified in various, usu-
ally imagined, circumstances, perhaps attempting to render them “coherent.”
Philosophers who shun the deontic way also often resort to such intuitions,
and for the same reason – to secure a ground for their claims. In both cases suc-
cess depends on the worth of the intuitions. If the appeal to them is merely a way
to avoid empirical investigation of the linguistic facts, it is likely to be minimal.
In his Foreword to the 2013 edition of Robert Nozick’s Anarchy, State, and Utopia,
Thomas Nagel writes of “a belief in the reality of the moral domain, as an area in
which there are real questions with right and wrong answers,” but adds that
“progress…toward discovering the right answers” could be made by “formulat-
ing hypotheses…and subjecting them to confirmation or disconfirmation by
the intuitive moral credibility of their various substantive consequences…The
method depends on taking seriously the evidential value of strong moral intu-
itions about particular cases, including imaginary cases.”²² Neither Hume nor
Kant, neither Plato nor Aristotle, neither Hegel nor Mill, would have allowed
that such intuitions enjoy evidential value. Nagel does not explain how and
why they might, though he does say that “[T]hese convictions form the deep
common element in…Rawls’ A Theory of Justice, Dworkin’s Taking Rights Serious-
ly, Walzer’s Just and Unjust Wars, Thomson’s “A Defense of Abortion,” and No-
zick’s Anarchy, State, and Utopia.“
When combined with antirealism, as it often was, the linguistic turn had a
startling implication: If the world is sculpted by our cognition of it and our cog-
nition is sculpted by our language, then the world is sculpted by our language.
Indeed, we do find Wittgenstein saying in Tractatus Logico-Philosophicus, “the
limits of my language mean the limits of my world.”²³ We also find Heidegger
writing in “Letter on Humanism” that “language is the house of being,”²⁴ and
 Robert Nozick, Anarchy, State, and Utopia (New York: Basic Books, 2013), pp. x-xiii.
 Tractatus, 5.6.
 Martin Heidegger, “Letter on Humanism,” in Basic Writings, 193.

Authenticated
3 Antirealism and its Varieties 15
in “The Origin of the Work of Art” that language is “the happening in which for
man beings first disclose themselves to him each time as beings.”²⁵ These asser-
tions, of course, require detailed discussion, some of which will be attempted
later. Suffice it here to note that the language to which Wittgenstein and Heideg-
ger were referring presumably was not cetacean, extraterrestrial, or angelic.
3 Antirealism and its Varieties
Antirealism is not limited to metaphysics. There is also antirealism in ethics, aes-

thetics, philosophy of science, philosophy of mathematics, and so on. Even in
the case of metaphysical antirealism, numerous qualifications, distinctions,
and explanations are needed. No metaphysical antirealist denies the reality of
everything, just as no metaphysical realist asserts the reality of everything, in-
cluding, say, the Easter Bunny. The solipsist says, “Only I exist,” not “Nothing
exists.” Berkeley denied that there are material objects, he called them “stupid
material substances,” but he insisted on the existence of minds and their
ideas. According to Kant, as we saw earlier, material objects are “transcendental-
ly ideal,” dependent on our cognitive faculties, but they are nonetheless “empir-
ically real,” not mere fancy. Bertrand Russell distinguished between existence
and subsistence: some things do not exist, yet they are not nothing – they sub-
sist; for example, material objects exist but universals only subsist. According to
Wittgenstein’s Tractatus, some things cannot be “said,” i. e., represented in lan-
guage, but they “show” themselves in what can be said. Among them, he held,
are those that matter most in logic, ethics, and religion.
Metaphysical antirealism rests on the proposition that our knowledge of the
world depends on our cognitive faculties – perception, conception, language.
This proposition and its antirealist implications are not unfamiliar or even wholly
unacceptable to science and common sense. We know that dogs can hear sounds
we cannot and that eagles can see things when we could not. We take for granted
that things look very different to a fish or a fly. If there are intelligent extraterres-
trial life-forms, their sense organs, concepts, and modes of reasoning presumably
would be fundamentally different from ours, and so would be the world as cog-
nized by them; space scientists puzzle over what pictures or symbols to engrave
on space probes that might reach such life-forms. Russell tried, somewhat naively,
to explain Kant’s transcendental idealism by saying: “If you always wore blue
 Martin Heidegger, “The Origin of the Work of Art,” in Poetry, Language, Thought (New York:
Harper&Row, 1971), 74, trans. Albert Hofstadter.

Authenticated
spectacles, you could be sure of seeing everything blue.”²⁶ Neuroscience unambig-

uously holds that our brain is not a tabula rasa on which the senses just write in-
formation – it processes that information. Before the insurance payout to the
leaseholder of the two World Trade Center Towers, it was necessary to decide
whether their destruction on 9/11/2001 counted as one or two events, a decision
obviously not dictated by any facts in the world.²⁷ Our beliefs about the history
of the United States depend on what history teachers have told us and what his-
tory books we have read. The difficulty, sometimes impossibility, of faithful trans-
lation from one language to another is notorious, especially in the case of poetry.
In science, “the observer effect” is a familiar phrase, meaning the changes that the
act of observing produces in what is observed. What is expressed in mathematical
symbolism may be inexpressible in a natural language, and vice versa. We often
find a picture of a happy child’s face or of a city devastated by a hurricane
worth more than a thousand words, and a caricature sometimes tells us what
no text could. Educated people do find the idea that the physical world is totally
unlike what physics and astronomy tell us idle, if not empty. What the world might
be “in itself,” i. e., as neither known nor knowable, does seem to many, as it did to
Goodman, not worth fighting for or against.
Of course, as a philosophical doctrine, metaphysical antirealism deserves
our interest only if argued. But the standard arguments have been densely ab-
stract, often enigmatic, and seldom persuasive. Even if, as Berkeley held, we
cannot know the objects we perceive unless we perceive them, it hardly follows
that they cannot exist unperceived. Kant argued that we can know things only as
they are “for us,” not as they are “in themselves,” but his premise – that for
knowledge to be possible, the objects of knowledge must conform to knowledge,
rather than knowledge to its objects – is hardly less obscure or more plausible
than his conclusion. Perhaps, as Michael Dummett argued two centuries later,
a realist interpretation of a sentence requires understanding what would count
as its conclusive verification or its conclusive falsification, and such understand-
ing is possible in the case of few if any sentences. But this is comprehensible
only to a few professional philosophers, and even they seldom find it persuasive.
Hilary Putnam argued for one of his several versions of antirealism by saying
that it “does not require us to find mysterious and supersensible objects behind
our language games that we actually play when language is working.”²⁸ But the
objects behind our language “games” need not be supersensible, and supersen-
 Bertrand Russell, A History of Western Philosophy (New York: Simon and Schuster, 1972),
707– 708.
 Steven Pinker, The Stuff of Thought (New York: Viking, 2007), 1.
 Hilary Putnam, Ethics without Ontology (Cambridge: Harvard University Press, 2004), 21– 22.

Authenticated
sible objects need not be mysterious (love and hatred are familiar but they are
not objects of the senses, whether as events or as dispositions). A clear and plau-
sible defense of antirealism, I suggest, must bypass the standard arguments. It
must start afresh, from specific and readily understandable truths, not abstract
and vague philosophical assumptions.
The reasoning underlying the standard arguments for antirealism, in Berke-
ley’s, Kant’s, as well as its recent versions, may be sketched as follows: (1) We
cognize only what we can, i. e., have the capacity to, cognize. This is a tautology.
But (2) that there is no reality, no world, independent of our cognitive capacities
does not follow from (1). What follows is another tautology: (3) that we cannot
cognize reality independently of our cognitive capacities. Antirealists often
argue on the basis of (1) for (2) probably because the negation of (2), namely,
Kant’s view (4) that there is a reality, “things-in-themselves,” which is independ-
ent of our cognitive capacities, seems to them idle. But there is at least one very
good reason for accepting (4), namely, that (2) seems to imply human creation-
ism, the proposition, presumably held by no one, that the whole world – from
the page you are reading now to the outermost known galaxies, and since the
Big Bang until the farthest conceivable future – depends for its existence and na-
ture on the minds, cognitive capacities, of humans, a certain species of animals
on one of its planets. Because of their forbidding level of abstraction, the stan-
dard arguments leave unclear both what they claim and what motivates them.
I shall not ignore these arguments, but my focus will be on certain specific
and readily understandable truths that lead to antirealism. Arguments for anti-
realism from such truths have the following form: (1) we cognize (perceive, un-
derstand, describe) the world as necessarily having a certain uncontroversial and
familiar specific feature. But it is obvious that (2) the world does not, perhaps
cannot, have that feature. Therefore, (3) the world as we cognize it, as it is
“for us,” is not as it is “in itself.”
The major defenders of antirealism, from Kant to Goodman and Putnam, of-
fered also arguments of this second kind. In defense of his obscure thesis of the
ideality of space, Kant pointed out that we can imagine only one space, and that
we can imagine it as empty but not as absent. Regarding the ideality of time, he
noted that all objects of sense, outer and inner, are necessarily in time, and that
time is necessarily one-dimensional. Regarding the ideality of causality, Kant ar-
gued that we necessarily conceive of the objects of sense perception as causally
related even though we do not perceive causal relations. Goodman dazzled his
readers with examples of features of the world that are best understood as
“made” by us, as how we perceive, conceive, or represent the world in language

Authenticated
or in art, not as how it is in itself.²⁹ We see the sun rising in the east, majestically
moving overhead, and setting in the west, but if educated we know that it is we,
not the sun, that is moving. The “fairness” of samples is a sacrosanct require-
ment both in science and business, but there are no objective criteria for it.
We often see the world, at least briefly, as radically different after we watch a
film, read a novel, or hear a symphony. We conclude at time t that all emeralds
are green because we have observed only green emeralds, but the same observa-
tions support with equal logical legitimacy also the conclusion that all emeralds
are grue, if “grue” applies to all things observed before t just in case they are
green, and to other things just in case they are blue. We reach the former conclu-
sion because “green,” not “grue,” is “entrenched in our linguistic practice.” Put-
nam pointed out that we can “count the objects in a room (a lamp, a table, a
chair, a ballpoint pen, and notebook) and come up with the answer ‘five’,”
but that if we also count their mereological sums and ignore the null object
then we come up with the answer ‘31.’³⁰ Such examples were often their most
persuasive arguments for antirealism.
The specific and readily understandable truths on which I shall focus are
quite different and less impressionistic. The first is that, as Wittgenstein pointed
out in the Tractatus, the world, if there is one, is the totality not of things but of
facts, in the robust Russellian sense of “fact.”³¹ (Its ordinary sense, in which
many speak even of “false facts,” is too vague to be of philosophical value.)
Most philosophers today would deny that there are such entities as Russellian
facts. So would common sense, since they are supposed to be entities categorial-
ly different from the individual objects, properties, or relations familiar to com-
mon sense. Nevertheless, the category of fact is essential to realism regarding the
world, which following etymology I shall call cosmological realism, even if not to
realism regarding individual things, which, again following etymology, I shall
call ontological realism. The reason is simple and obvious. If Jack admires Jill
but Jill does not admire Jack, what would distinguish the world in which this
is so from the world in which Jill admires Jack but Jack does not admire Jill,
the world in which they admire each other, and the world in which neither ad-
 Ways of Worldmaking, especially chapters I and V.

 Hilary Putnam and James Conant, Words and Life (Cambridge: Cambridge University Press,
1995), 308.
 The phrase “state of affairs” is often used as a substitute for “fact.” A distinction between
actual and possible states of affairs is made, true statements supposed to correspond to the for-
mer and false statements to the latter. “Fact” then, if used at all, is applied only to actual states
of affairs. For the sake of terminological simplicity, I will use only “fact” and ignore the question
whether false statements correspond to anything.

Authenticated
mires the other, if these worlds differed in no other respect? There would be no
answer if we supposed that there are only individuals, properties, and relations
in the world. Only the fact that Jack admires Jill but Jill does not admire Jack, not
their mere presence in the world, including even the presence of a relation of ad-
miration, would distinguish that world from the other three. Hence the conclu-
sion that the world is the totality of facts. There may be no such entities as
facts, but then there would be also no world, and cosmological antirealism
would win by default. Realists cannot consistently hold both that there is a
world and that there are no facts – robust, brute facts.
The second truth on which I shall focus is the nonexistence of “logical ob-
jects,” in Wittgenstein’s sense, to be explained in the next section.³² We shall
find that facts also should be counted as logical objects. Common sense
would readily agree that there are no such entities in the world as ifness and all-
ness, though it would be surprised that anyone should think otherwise. Indeed,
very few philosophers would disagree.
The third truth on which I shall focus is the obvious absence from the world
of facts that might correspond to what linguists call generic statements, usually
of the form “Fs are G,” as contrasted with universal statements, usually of the
form “All Fs are G.” Missourians believe that Iowa winters are severe, but not
that all are. Physicians believe that patients with prior strokes benefit from tak-
ing Lipitor, but they do not believe that all patients with prior strokes benefit
from taking Lipitor, nor do they confuse what they believe with their evidence,
say, that in a clinical trial 265 or 11.2 % of the patients who took Lipitor suffered
a stroke over five years, while 795 or 37 % of those who took a placebo did. Much
if not most of what we think we know about the world is properly expressed only
in generic statements. But we do not need abstruse argument to see that no dis-
tinctive facts correspond to them, that there are no generic facts, even if there
were facts that correspond to the related singular and universal statements.
 Frege had used the phrase “logical objects” for the objects of arithmetic in the context of his
project of reducing arithmetic to logic, a project continued later by Russell and Whitehead. Ber-
trand Russell used it in his posthumously published Theory of Knowledge: The 1913 Manuscript
(London & New York: Routledge, 1984), 97, claiming that we must be “acquainted” with logical
objects in order to understand “logical terms” such as “particulars, universals, relations, dual
complexes, predicates.” Wittgenstein read Russell’s manuscript and criticized it severely. Several
years later he asserted in Tractatus Logico-Philosophicus, “There are no logical objects” (4.441).

Authenticated
4 Logical Antirealism and Semirealism
Stated briefly and crudely, the antirealist thesis – that the world, at least insofar
as it is knowable, depends on us – would convince no one. It gains plausibility,
however, if we agree with Wittgenstein that “there are no logical objects”³³ even
though “logic pervades the world,”³⁴ because the world must have a logical, not
just spatiotemporal, physical, or causal, structure. Wittgenstein meant that logi-
cal expressions, in particular, the sentential connectives (“not,” “and,” “or,” “if…
then”) and the quantifiers (“all,” “some”), stand for nothing in the world. Yet,
they are indispensable for any cognition beyond that of babes. I shall call all cog-
nition that requires their use logical. It would include but should not be con-
fused with the cognition pursued by logic. Statements expressing logical cogni-
tion usually contain also nonlogical expressions, while those in logic do not. “All
men are mortal” includes the logical expression “all,” but it is not a statement of
logic. Few, if any, statements do not include logical expressions. Hence the
power of what I shall call logical antirealism.³⁵
It is a far-reaching version of metaphysical antirealism, yet almost everyone
would find it plausible. If a realist interpretation of a true statement is one that
pairs the statement with a fact, in the Russellian sense of “fact,” i. e., an entity
that “makes” the statement true but is categorially different from anything men-
tioned in the statement, then almost everyone would find plausible an antirealist
interpretation of general statements, whether universal (e. g., “All men are mor-
tal”), particular (e. g., “Some men are mortal,” “At least one man is mortal”), or
generic (e. g., “Men are mortal”), and of compound or molecular statements,
whether negative (“This page is not blue”), conditional (“If this page is white
then the next page is white”), disjunctive (“Either this page is white or it is
blue”), or conjunctive (“This page is both white and rectangular”). Few believe
that there are in the world universal, particular, generic, negative, conditional,
disjunctive, or conjunctive facts, even if they believe that there are “atomic”
facts, e. g., the fact to which the statement “This page is white” might corre-
spond. Even fewer believe that there are in the world entities such as allness, not-
ness, ifness, orness, andness, which might be the defining constituents of those
facts. Yet, as Sartre eloquently argued, negation plays an essential role in cogni-
tion; his striking example was seeing that the person you expected to meet in a
café is not there. And Gustav Bergmann pointed out that there would be no laws
 Tractatus, 4.441.
 Tractatus, 5.61.
 See my “Metaphysical Realism and Logical Antirealism,” in Richard Gale, ed., Guide to Met-
aphysics (Oxford: Blackwell, 2002).

Authenticated
4 Logical Antirealism and Semirealism 21
of nature if generality, meaning universal facts and something the quantifier all
stands for that makes them universal, were not in the world, for most laws of na-
ture are universal statements.³⁶
Indeed, facts themselves may be counted as “logical objects,” in the sense
that, like the sentential connectives and the quantifiers, they are required by log-
ical, not empirical, considerations. If so, then the declarative sentences express-
ing them would also be logical expressions. Indeed, we do not learn sentences as
we learn names of things, we learn to make sentences. This is why Wittgenstein
counted the concept of fact as formal, like the concept of object, and thus unsuit-
ed for literal application.³⁷ This is why he denied that there can be representation
in language of the logic of facts.
If there are no logical objects in the world, then the world has no logical
structure and therefore there is no world. Spatiotemporal/physical structures
and even individual things may involve logical structure, but whether this so
is far from evident. They are not facts and thus involve neither propositional con-
nectives nor quantifiers in the way facts, and the sentences expressing them, do.
They can be subjected to chemical analysis but not to logical analysis. Logical
antirealism does not deny the reality of spatiotemporal/physical structures or
of individual things. It denies only the reality of logical structure, of “logical ob-
jects.” This is why it is more moderate and far more plausible than, say, Kant’s or
Goodman’s antirealism. It has perhaps the same metaphysical bite, but it does so
in a principled fashion.
Logical antirealism is not the only species of metaphysical antirealism. There
is one corresponding to each species of cognition. The ancient and most familiar
version is perceptual antirealism. It denies that there “really” are objects corre-
sponding to what we seem to perceive. Berkeley’s “to be is to be perceived” was
its brief but memorable slogan. But perceptual antirealism need not be so dras-
tic. It can be limited to only some features of perceived objects, say, their “sec-
ondary qualities,” such as color, and then it would represent the view held by
contemporary science and most educated people: we seem to see colors but in
the world there is only the light that initiates vision.
More innovative is the species of metaphysical antirealism we may call con-
ceptual –objects, perceived or not, depend on our understanding of them, on the
concepts we possess. Conceptual antirealism was Kant’s major contribution to
philosophy. It can be limited to just some concepts and thus gain greater plau-
 See Gustav Bergmann, New Foundations of Ontology (Madison: University of Wisconsin

Press, 1992, ed. by William Heald), 173. Heald’s introduction is obligatory reading for all inter-
ested in Bergmann’s philosophy. He has also included an invaluable glossary.
 Tractatus, 4.1272.

Authenticated
sibility. It can be limited to what Kant called the pure concepts of the under-
standing, such as causality. It can be limited especially plausibly to the concepts
expressed by logical expressions, and then it would coincide with logical antire-
alism. I have repeatedly argued elsewhere that at least the concept of identity
cannot be given a realist interpretation (we hardly perceive a relation of identity
or suppose that its existence is somehow hidden), but virtually no cognition, not
even simple recognition, is possible unless the concept of identity is applicable,
even if not explicitly applied.
The version of conceptual antirealism that became characteristic of 20th cen-
tury philosophy after it took the linguistic turn is linguistic antirealism. If cogni-
tion necessarily involves language, then insofar as it is cognizable the world is
dependent on language, if not on the particular characteristics of the language
we speak, then on the characteristics that all languages share. Conceptual anti-
realism would entail linguistic antirealism if concepts are words or syntactic
structures, rather than “ideas in the mind.” And if limited to logical expressions
and syntactic structures, it would be also a version of logical antirealism. But not
all versions of logical antirealism need be linguistic. We may allow, as Kant did,
for the possibility of logical concepts that are purely psychological.
Logical antirealism is the most plausible version of metaphysical antireal-
ism. Few would disagree with Wittgenstein that there are no logical objects –
no items, no fragments of the world, that correspond to the propositional con-
nectives and the quantifiers, and thus no distinctive facts that correspond to
compound and general statements. No such items and facts belong in what
can be “said,” and so they would not count as ordinary denizens of reality.
But, Wittgenstein also held, they can be “shown.” Wittgenstein’s distinction be-
tween saying and showing thus introduced a position that is neither unqualified
logical realism, like that of Frege and Russell, nor unqualified logical antireal-
ism. We may call it semirealism.³⁸ In chapter 4 I shall argue that the distinction
between saying and showing has a clear and noncontroversial application to or-
dinary, not just “logical,” pictures, and that Wittgenstein’s picture theory of
meaning on which the distinction rested was no more than a sophisticated ver-
sion of the traditional view that meaning and thought involve mental represen-
tations, i. e., ideas, perhaps mental images.
Before Kant, the tautology on which antirealism rests – that our knowledge of
the world depends on our cognitive faculties – may have led to skepticism. But anti-
realism should not be confused with skepticism, though they are similar in some
 Semirealism should not be confused with what Simon Blackburn calls “quasi-realism” in Es-
says in Quasi-Realism (Oxford: Oxford University Press, 1993).

Authenticated
respects. Both are rooted in the commonsense belief that “things may not really be
what they seem to be.” Both appeal to the proposition that “we can never discover
what things really are because any discovery would still be a discovery only of what
some things seem to be.” The skeptic concludes that we can never know what
things really are. The antirealist concludes that there is no distinction between
what things really are and what they seem to be. This is why, left unexplained, anti-
realism is paradoxical, while skepticism is at most outrageous.³⁹
Antirealism is a metaphysical view, skepticism belongs in epistemology. We
shall see in chapter 2, however, that much of traditional epistemology can be un-
derstood as logic, insofar as its chief concern has been the validity of certain in-
ferences. If the home of skepticism is such epistemology-as-logic, it is no more
anthropocentric than is logic. But in its usual formulations skepticism is overtly
and unabashedly anthropocentric. It concerns the limits of human perception
and understanding. And today the proper study of these limits belongs in psy-
chology and especially neuroscience. Whether certain brain states are the out-
come of external stimulation, and if they are to what extent and in what way
they represent anything external, is hardly to be determined by armchair specu-
lation. Moreover, if what I called cognitive collectivism is accepted, the epistemo-
logical question “What do I know?” would be replaced by the question “What do
we know?” and traditional skepticism would become less implausible. We may
still hold that there are things that, say, physics does not know, perhaps cannot
know, but our reasons would rest on certain facts about the scientific discipline
of physics and bear little resemblance to Cartesian skeptical concerns. Metaphys-
ical antirealism would also become less implausible. Surely the nature of the
spatiotemporal world insofar as it is cognized is what the disciplines of physics,
astronomy, and biology say it is.
Metaphysical antirealism enjoys little public celebration, but its indirect and
usually unnoticed influence on nonphilosophical thought has been enormous. A
noteworthy example is Kuhn’s important account of the history of science as involv-
ing relativity to shifting “paradigms.” Less admirable are the fashionable but care-
less and unphilosophical relativisms that insist that truth is relative to era, culture,
race, gender, ethnic origin, or even just personal “belief system.” Indeed, metaphys-
ical antirealism is a form of relativism, but it must not be confused with any of
these. It acknowledges only relativity to being human. This is why metaphysical an-
tirealism is not a sort of subjectivism. What our cognitive faculties deliver can count
as objective in the straightforward sense that, in principle, it can be and often is the
same for all humans, and personal divergence from it is what we count as subjec-
 See my Skepticism about the External World (New York: Oxford University Press, 1998).

Authenticated
tive. This is why metaphysical antirealism can allow for a sharp distinction between
objective truth and personal opinion or fancy.
We know that there are fairly advanced nonhuman cognizers, e. g., whales
and chimpanzees. There may be also extraterrestrial cognizers far more ad-
vanced than humans. The world cognized by whales is relative to cetacean cog-
nition. The world cognized by chimpanzees is relative to simian cognition. The
world cognized by extraterrestrials would be relative to their cognition, The anti-
realist holds that the world cognized by humans is relative to human cognition.
That relativity is biologically inescapable. By contrast, relativity to era, culture,
race, gender, ethnic origin, or personal belief is not. We cannot literally transport
ourselves to an era in the past, but we can and often do transcend the present by,
for example, viewing what past architects designed and past builders built, read-
ing what past authors wrote, and today even hearing recordings of what past
singers sang. We cannot change our ethnic origin, culture, race, or gender, but
we can transcend it by communicating, and often agreeing, with people of a dif-
ferent origin, culture, race, or gender. We can do so because, since we belong to
the same species, we share roughly the same cognitive faculties. This is why rel-
ativity to era, culture, race, gender, or ethnic origin is quite unlike relativity to
being human. As to relativity to personal belief system, it would differ from
the triviality that one believes what one believes only if one’s beliefs really con-
stituted a system, if one did not hold contradictory beliefs, and if there were gen-
uine criteria for membership in the system. But don’t we sometimes believe the
premises and reject the conclusion of a valid argument? And is the belief, for ex-
ample, that yoghurt is better than sour cream really part of my system of beliefs?
One’s beliefs are more likely to form not a system but just a “multitude”: “Do I
contradict myself? Very well then I contradict myself. (I am large, I contain multi-
tudes),” Walt Whitman wrote.
I mention in this book, and sometimes discuss in detail, the views of various
philosophers, from Plato, Kant, and Hegel to Moore, Wittgenstein, and Sartre. No
philosophical discussion should ignore the historical context of its topic. But this
is not a book in the history of philosophy. I do not engage in exegesis for its own
sake. Some readers may be surprised by the considerable attention accorded to
Hegel, Moore, and the early Wittgenstein (often with quotations rather than para-
phrase so the readers can judge for themselves what I say about them). But Hegel’s
insistence on the cognitive priority of society over the individual suggests, when gen-
eralized, the radical rethinking of the distinction between oneself and the world that
is needed in an antirealist but nonanthropocentric metaphysics. Moore’s ethics re-
mains the modern paradigm of a nonanthropocentric ethics, and his account of the
nature of consciousness is invaluable for the formulation of a defensible antirealism
in metaphysics. And Wittgenstein’s distinction between saying and showing provid-

Authenticated
ed a strikingly original third alternative in the realism/antirealism debate. It has not

been sufficiently appreciated, perhaps because, like Hegel’s and Moore’s views, it is
found too “obscure” and “difficult.”
Several chapters are rooted in previously published articles: “Metaphysical
Realism and Logical Antirealism,” in Richard Gale, ed., Guide to Metaphysics
(Blackwell, 2002); “Saying and Showing the Good,” in Heather Dyke, ed.,
Time and Ethics: Essays at the Intersection (Kluwer Academic Publishers,
2003); “Ethics Dehumanized,” in Mark Timmons and Terry Horgan, eds., Meta-
ethics After Moore (Oxford University Press, 2006); “Bergmann and Wittgenstein
on Generality,” Metaphysica, Vol. 7, No. 1, April 2006; “Epistemology Dehuman-
ized,” in Quentin Smith, ed., Epistemology: New Essays (Oxford University Press,
2008); “Facts,” in Javier C. Arteseros, ed., Studies in the Ontology of Reinhardt
Grossmann, (Frankfurt: Ontos Verlag, 2010); and “Generic Statements and Anti-
realism,” in Logos & Episteme, I, 1 (2010).

Authenticated
Authenticated
Chapter Two: Three Varieties of Epistemology
1 Naturalistic Epistemology
The subject matter of both epistemology and ethics traditionally has been consid-
ered human – the knowledge and the good, respectively, that humans, not ceta-
ceans or angels, seek and enjoy or lack. This, I argued in the Introduction, is a mis-
take. Its correction calls for refocusing these branches of philosophy, their
dehumanization. Such refocusing is more easily accomplished in the case of episte-
mology. Throughout its history, it has wrestled mainly with issues concerning the
validity of certain inferences, hardly a matter to be settled by zoological considera-
tions. In effect, epistemology has often been a sort of logic. In this chapter I shall
attempt to make clear how epistemology-as-logic differs from naturalistic epistemol-
ogy, which celebrates the primacy of zoological considerations, as well as from sub-
jective, Cartesian, epistemology, which is logically incompatible with zoological con-
siderations but thereby also lacks subject matter altogether.
The case for dehumanizing epistemology is best understood in the context of
the important differences among these varieties of epistemology. (Fundamental
disagreements in epistemology, as elsewhere in philosophy, often arise from dif-
ferences of interest, not genuine conflict.) Naturalistic epistemology is explicitly
anthropocentric, humanized; the “natural” objects it considers are not cetacean
or simian. To be consistent, subjective epistemology cannot be anthropocentric,
though its practitioners are seldom aware that this is so. It would beg the ques-
tion against the skeptic if it takes itself to be about any one or several humans.
Epistemology-as-logic is as nonanthropocentric as logic; it is, of course, a human
endeavor, but humans are not its subject matter. All three have been with us at
least since Socrates. My chief concern in this chapter will be with epistemology-
as-logic, but it is naturalistic and subjective epistemology that have represented
the standard conception of epistemology.
It may seem obvious that epistemology should be naturalistic. Its name is a
synonym of “theory of knowledge,” the knowledge in question surely is that of
humans, and humans are parts of nature, of its fauna. Epistemology naturalized
is thus epistemology humanized: it is about humans. Not only does it ignore
gods, angels, and extraterrestrials, it ignores also chimpanzees, whales, and
bats. But thereby it also lacks the supreme generality and abstraction distinctive
of philosophy. Humans already belong in the subject matter of several special
sciences that seek detailed information about them, including their perception,
conceptualization, and reasoning. This is why naturalistic epistemology is only
programmatic. The substantive work is done by biology and psychology. As I

Authenticated
30 Chapter Two: Three Varieties of Epistemology
mentioned in the Introduction, Quine, who took up the case for “epistemology
naturalized,” often mentioned the role in cognition of what he called “surface
irritations,” but did not himself investigate these irritations. Naturalistic episte-
mology is focused on human matters even when straying into talk about nonhu-
man biological and computational states. They are of interest to it mainly for the
light they might cast on the epistemic faculties and states of humans. The intrin-
sic interest of such nonhuman states is of course indisputable, but they even
more obviously belong in the province of the empirical sciences.
Naturalistic epistemology may be only programmatic, but its pedigree is im-
pressive. Much of Aristotle’s epistemology was naturalistic. When he described
the parts and functions of the soul, he was doing in principle what biologists
and psychologists do today. And the rationale of the program seems impeccable.
Humans, obviously, are parts of nature, they are not heavenly spirits even if they
possess immaterial souls. But this is also why the proper investigation of them
and of their epistemic faculties and states is empirical and belongs in the natural
sciences. It would be strange to propose today investigating any part of nature
nonempirically. Even if the human epistemic faculties and states were faculties
and states of immortal souls, a genuine investigation of them would still be em-
pirical. A human immortal soul is still the soul of a human being, a certain an-
imal. Both Plato and Aquinas would have agreed. Much the same can be said
about the suggestion that they are faculties and states of “persons.” Surely the
persons in question are animals, not gods.
If human beings, including their epistemic faculties and states, belong in the
subject matter of disciplines other than philosophy, the obvious question is what
room is left for naturalistic epistemology. We saw in the Introduction that concern
over this question may explain the shift to the view of philosophy as just “concep-
tual,” not “factual” – neither about natural facts nor about nonnatural facts, but
about concepts or words. Hence its preoccupation with “definitions,” “analyses,”
and “elucidations” of the “workings of our language.” But if the concepts and
words are in nature – presumably in human minds and languages – they, too, lie
outside philosophers’ professional competence: there is psychology and neuro-
science, as well as linguistics and lexicography. (If concepts are not in nature,
e.g., if they are Platonic Forms, then they should be of no concern to naturalistic
epistemology.) The investigation of brain-states and words calls not for “definitions”
or “analyses,” to be tested by “intuitions,” but for meticulous empirical descriptions
and fruitful hypotheses, tested by standard scientific methods. The very idea of aim-
ing at definitions or analyses of brain-states is foreign to neuroscience. As to words,
more than half a century ago Wittgenstein pointed out, as I noted in the Introduc-
tion, that they are not used in accordance with necessary and sufficient conditions,
and thus their uses cannot be captured in definitions.

Authenticated
1 Naturalistic Epistemology 31
Perhaps the most familiar project in recent epistemology was the search for a
definition of knowledge, which preoccupied it from the 1960s through the 1980s. It
was born in the late 1950s, when A. J. Ayer’s Problem of Knowledge and R. M. Chis-
holm’s Perceiving appeared. But the project was out-of-date already at its birth.
Thirty years earlier Wittgenstein had written: “If I was asked what knowledge
is, I would list items of knowledge and add ‘and suchlike.’ There is no common
element to be found in all of them, because there isn’t one.”⁴⁰ Linguists and lex-
icographers of course agreed. A famous paper by Edmund Gettier, a student of
Wittgenstein’s disciple Norman Malcolm, argued the point in the early 60s. But
few of those who wrote the thousands of pages devoted to discussion of his
paper seemed aware that, whatever its author’s intentions might have been, it
called not for greater diligence, sophistication, or imagination in the project of de-
fining knowledge but for its abandonment.
Epistemology is the theory of knowledge, but the word “knowledge” stands
either for a disciplinary, essentially social achievement, such as grammar (Aris-
totle’s favorite example), astronomy, and arithmetic, or for a personal achieve-
ment. The study of knowledge as a social achievement belongs in the history
of science and the sociology of knowledge. Investigation of it would be, of
course, naturalistic, essentially historical and sociological. But epistemologists
usually have been interested in knowledge only as a personal achievement.
And this interest may assume one of two very different forms. I may ask whether,
how, and what knowledge is possible for me, a certain human. If so, my episte-
mological endeavor would remain anthropocentric and therefore would count as
naturalistic. It would be objective, though rather narrow in subject matter. But I
may ask instead whether, how, and what knowledge is possible for me in ab-
straction from the fact that I am human and ignoring the question of whether,
how, and what knowledge is possible for other humans. This would be the ques-
tion that a philosophical skeptic who respects consistency would ask, especially
if it concerns knowledge of the existence of an “external” world consists of rocks,
trees, stars, as well as human bodies such as yours and mine. We may describe
an epistemology limited to this question as subjective.
 MS 302, “Diktat für Schlick” 1931– 33. Quoted by David Stern, “Sociology of Science, Rule
Following and Forms of Life,” in M. Heidelberger and F. Stadler, eds., History of Philosophy and
Science, 347.

Authenticated
2 Subjective Epistemology
While the subject matter of naturalistic epistemology, whether human knowl-

edge or human concepts or language, lacks the generality expected of philoso-
phy, subjective epistemology seems to lack generality altogether. It seems to
be baldly and bleakly about only one person, myself, hardly a topic of philo-
sophical or scientific interest, whoever and whatever I might be. But it cannot
presuppose, explicitly or implicitly, my humanity, including the existence of
my body, my language, and my place on earth or in history, for then it would
be a naturalistic, though incompetent, epistemology limited to just one “sample”
of humanity. The subjectivity of subjective epistemology does not consist in its
being about a solitary human being.
The term “subjective” is not a synonym of “mental” or “mentalistic.” If minds
are immaterial, my study of your mind might be as objective as my study of my
mind. And if I could study my brain only as my brain, the study would be subjec-
tive and form no part of biology. Quine’s rejection of Cartesian epistemology on
naturalistic grounds was both too narrow and too wide. It was too narrow because
what was characteristic of Descartes’ epistemology was not its subject matter, pre-
sumably what he called a “thinking thing,” but his exclusive use of first-person
indexicals in its defining initial stages. In his argument “I think, therefore I am”
Descartes did not argue for the existence of Descartes, a Frenchman who of course
possessed a material body even if also an immaterial soul. To be consistent, Des-
cartes could only use “I,” as of course he did. One need not be naturalistically,
“scientifically,” minded in order to reject a theory dependent on the exclusive
use of first-person indexicals. Quine’s rejection of Cartesian epistemology was
also too wide because, as R. M. Chisholm pointed out, in its initial stages, includ-
ing the proof of “I exist,” Cartesian epistemology was consistent with a naturalis-
tic, even materialist, ontology. ⁴¹ For all Descartes knew when he was initially stat-
ing the proof, the pronoun “I” in it might have referred to a material, not a
“thinking,” thing. The truth is that, pace Descartes and almost all other philoso-
phers, his cogito had no ontological content at all.
Subjective epistemology is not a capricious narrowing of the subject matter
of naturalistic epistemology from all humans to just one. If it were, it would not
be subjective, though also it would be of no philosophical interest. Subjective
epistemology arose, most notably with Descartes, Berkeley, and Hume, as a dis-
tinct variety of epistemology in order to face the skeptical challenge. The skeptic
 Roderick M. Chisholm, A Realistic Theory of Categories (Oxford: Cambridge University Press,

1996), 99 – 105. Chisholm speculated that the mind might be “microscopic part” of the brain.

Authenticated
2 Subjective Epistemology 33
cannot assume that he or she is human, since being human involves having a
body, which is a part of the material world the existence of which the skeptic
doubts. Therefore, the subjective epistemologist also cannot make this assump-
tion when attempting to answer the skeptic. Subjective epistemologists cannot
consistently write, even though the best of them did, books titled “An Essay Con-
cerning Human Understanding,” “A Treatise Concerning the Principles of Human
Knowledge,” or “A Treatise of Human Nature.” If true to their titles, such books
would beg the question against the skeptic.
Subjective epistemology is essentially Cartesian, though it was anticipated
by the Greek skeptics, especially Sextus Empiricus. Its raison d’être was the proj-
ect of answering the skeptic. Had it succeeded, its mission would have been ac-
complished, and there would have been room left only for naturalistic epistemol-
ogy and epistemology-as-logic.
Naturalistic epistemology does not beg the question against the skeptic by
taking its subject matter to be humans because it is not concerned with the skep-
tic’s question; Aristotle was no more concerned with skepticism than was Quine.
Indeed, though a “theory of knowledge,” naturalistic epistemology need have lit-
tle concern with knowledge itself. It is best understood as concerned with cogni-
tion, that is, the employment of our perceptual, conceptual, and verbal faculties,
which may lead to knowledge but would be of interest even when it does not. For
subjective epistemology, however, since it is mainly an attempt to answer the
skeptic, only knowledge matters. Alleged cognitive states such as justified belief
are at best images of knowledge. We seek them in the hope of finding something
still worth having when knowledge is absent or impossible. The skeptic about the
external world would not be answered by being told that one would be justified
to believe that there is an external world. One reason is that the issue is too im-
portant for anything less than knowledge – it concerns the existence also of
other people, including those we see daily and love. Another reason is the im-
propriety of such uses of “justified” and “believe.”
As used in epistemology, “justified” is a technical term, of obscure meaning and
uncertain reference, indeed often explicitly introduced as primitive. In everyday
talk, it is a deontic term, usually a synonym of “just” or “right,” and thus “justified
belief” is a solecism. For it is actions that are justified or unjustified, and beliefs are
not actions. If told that the phrase stands for belief resulting from “reliable process-
es,” this would be a verbal stipulation, and would also involve the use of “reliable,”
another technical term that has required extensive explanation.
Even the word “belief,” whether or not prefaced by an adjective like “justi-
fied,” is seldom used in epistemology with clear sense and reference. In every-
day talk, it may refer to a behavioral disposition, moreover a multi-track one,
manifested in a great variety of events, e. g., in what one says, what one does

Authenticated
to oneself or to one’s spouse, children, colleagues, friends, and strangers, what

one buys, sells, or reads, what one eats or drinks, etc. A proper translation then
of “S believes that p” into logical notation would include all sorts of counterfac-
tuals and would be too cumbersome to bother stating. It certainly would have
neither the form “B(S,p)” nor the form “B(S,x,y…, R),” where S stands for the be-
liever, p for what is believed, and x, y, and R for constituents of p. If we take be-
lief to be a brain state, process, or condition, then the translation of “S believes
that p” into logical notation, were it possible, would need to be provided by neu-
rologists, though they would have little use for it. And if we take belief to be an
irreducibly mental state or condition, plain respect for the phenomenological
facts would tell us that beliefs so understood are as rare as experiences of
faith, commitment, or strong conviction seem to be.
The truth is that in ordinary discourse “S believes that p” (e. g., “S believes
that Jones owns a Ford,” when uttered in the course of office chatter) functions
roughly as a synonym of the colloquial “S thinks that p,” and is no more enlight-
ening or needed in philosophy than is the latter. And “belief” is used more or
less as a synonym of “opinion,”view,” “stand,” etc., as in “The belief that …..
originated in…..” Neither the verb nor the noun usually expresses religious
faith or some other commitment or conviction that may have psychological real-
ity and thus deserve the interest of psychology and, perhaps, philosophy. For ex-
ample, it would be absurd for me to say now that I believe that I am typing, let
alone that I am justified in believing that I am typing.
The word “justified,” whether applied to such phantom beliefs or to the sen-
tences, statements, assertions, judgments, hypotheses, etc., supposedly express-
ing them, is often prefaced by the adverb “epistemically.” But this renders its use
in epistemology even less appropriate because the noun corresponding to that
adverb in Greek is episteme and in English “knowledge,” exactly the word we
tried to avoid. We resort to the phrase “justified belief” when aware that we
have no knowledge but also hope that we are not entirely ignorant. But it is
not the natural, traditional word for describing what we might have in such
cases. It is the word “evidence” that comes closest to serving this function –
in science, courts of law, and careful everyday discourse.
The root of “evidence,” however, is the same as that of “evident.” When
what is evident is a proposition, it is evident if it is or can readily be seen, wheth-
er literally or metaphorically, to be true. “Seeing is believing,” the saying goes,
but what it usually means is that seeing is knowing. Hence, the traditional ac-
count of knowledge as apprehension, intuition, awareness, or acquaintance,

Authenticated
and the existence of the so-called strong sense of “know,” roughly that of cer-
tainty, as contrasted with its weak sense of some sort of true belief.⁴²
A proposition is sometimes evident because it is “seen” to be true by itself,
i. e., to be self-evident. More often it is evident because it is seen to follow from
one or more other propositions that are evident.⁴³ We seldom say, however, that
the latter are evidence for the former. We seldom call the premises of a valid de-
ductive argument evidence for its conclusion, even if the premises are evident
and the validity of the argument is itself evident, as it would be if its form
were as simple as, say, that of modus ponens. Rather, we speak of evidence
when what we want to know is neither self-evident nor seen to follow from any-
thing evident, yet we think or hope that something else “supports” the proposi-
tion in some other manner. It is then that skepticism is born and so is the sub-
jective epistemology that attempts to answer it.
J. L Austin wrote, “The situation in which I would properly be said to have
evidence for the statement that some animal is a pig is that, for example, in
which the beast itself is not actually on view, but I can see plenty of pig-like
marks on the ground outside its retreat. If I find a few buckets of pig-food,
that’s more evidence, and the noises and the smell may provide better evidence
still. But if the animal then emerges and stands there plainly in view, there is no
longer any question of collecting evidence; its coming into view doesn’t provide
me with more evidence that it’s a pig, I can now just see that it is, the question is
settled.”⁴⁴ Some would say, however, that when the animal comes in view it is
self-evident that the animal is a pig. Taken literally, of course, the term “self-evi-
dent” is a pleonasm, as “self-seen” and “self-visible” would be. But it does serve
to mark the important difference between what is evident in virtue of being seen
to be true by itself and what is evident in virtue of being seen to follow from one
or more other propositions that are seen to be true. Even in a modus ponens argu-
ment with self-evident premises, the conclusion, if not self-evident itself, would
not be evident unless seen to follow from the premises, i. e., unless the validity of
the argument is self-evident. The conclusion would be evident only if its relation
to the premises was evident.
 The distinction between a weak and a strong sense of “know” was made by Norman Mal-
colm in “Knowledge and Belief,” included in Knowledge and Certainty (Englewood Cliffs, NJ:
Prentice-Hall, 1963), 58 – 72
 Compare G. E. Moore’s explanation, to be discussed in chapter 3, of what he meant by calling
the fundamental propositions of ethics self-evident, in Principia Ethica (Cambridge: Cambridge
University Press, 1903), Preface.
 J. L. Austin, Sense and Sensibilia (Oxford: Clarendon Press, 1962), 115. Original italics.

Authenticated
Skepticism begins by noting that usually what we want to know is neither

self-evident nor made evident by anything that is self-evident. And subjective
epistemology usually tries to avoid the disconcerting implications of this fact
by appealing to something else that it hopes is relevant to what we want to
know. We may call it “evidence,” although, even if it is itself self-evident, it
does not make what we want to know evident. This is how Austin used “evi-
dence” in his example of the pig: “a few buckets of pig-food,” “the noises and
the smell.” Religion and the law are the noteworthy home of such uses of the
word, which are often exquisitely circumspect. They are also common, even if
less circumspect, in the lab and the street. The notion they express is a degen-
erate offspring of the notion of the evident. But, though degenerate, it is usually
understandable and also harmless. We may need to know, not merely believe,
that God exists, yet we may be aware that we do not. So we look for “evidence”
of his existence. In the courtroom, a verdict of guilt or innocence may be man-
datory, though neither guilt nor innocence is likely to be self-evident or made
evident by anything that is self-evident. So we look for something else we
hope is relevant to guilt or innocence, and call it “evidence,” whether “beyond
reasonable doubt” or not, and whether just “circumstantial” or not.
If we think that such degenerate evidence is “strong” enough, we may even
say that we know that for which we take it to be evidence, thus making also a
degenerate use of “know.” Saying this would make explicit the reason we ap-
pealed to it in the first place, i. e., our desire for truth, and it may seem to anoint
the appeal. How strong the evidence must be, however, is never made clear be-
cause, given the sort of reasons that lead us to appeal to it, this cannot be made
clear. As cognitive beings, we seek knowledge because it is truth we want, not
such “evidence,” even if we were to dress it up with phrases like “epistemic
probability” and “epistemic justification.” The idea that evidence comes in de-
grees and that possession of it yields an approximation to knowledge, something
worth having even though knowledge is absent or impossible, may suggest to
some that there can be such a thing as an approximation also of truth. But
while truth may be incomplete, irrelevant, or misleading, there cannot be two-
thirds or 86 % truth.
Nonetheless, though as cognitive beings it is truth and therefore knowledge
that we seek, we are not purely cognitive beings. The degenerate uses of “evi-
dence” and “know” in religion and the courts of law are defensible. They are
often needed even in the lab and the street. To go about our business we must
think of certain judgments as final, settled, even if we soon revisit them. There
are practical reasons in religion, the court of law, the lab, or the street for resort-
ing to a degenerate notion of evidence. But no such reasons exist in epistemol-
ogy, which is neither a religion or a courtroom nor a lab or the street. Our con-

Authenticated
cerns in it are purely cognitive. This is why the degenerate notion of evidence is
not harmless in epistemology. It gives rise to the illusion that knowledge is rel-
atively easy to achieve, or at best that what knowledge requires is merely the
limit, even if only ideal, of a range of degrees of “evidence,” of “epistemic prob-
ability,” or of “epistemic justification,” and that what falls short of that limit
would nonetheless suffice. But for what it might “suffice” is unclear, since prac-
tical considerations are now irrelevant. Unsurprisingly, it has not sufficed for an-
swering the skeptic.
In everyday life and thought, the degenerate notions of evidence and evi-
dence provide a way of clearing our epistemic conscience. They are analogous
in this respect to the degenerate notion in ethics of “subjective duty,” which pro-
vides a way of clearing our moral conscience. The weak sense of “know” is anal-
ogous to the weak sense of “ought” introduced by the notion of subjective duty.
If ignorant, as we often are, of our objective duty, of what we ought to do, we may
settle for doing what we think, perhaps feel, we ought to do, our subjective duty.
We may even insist that one always ought to do what one thinks or feels one
ought to do. However, just as our concern as cognitive beings is with truth,
our concern as moral beings is with doing what we really ought to do. The
weak senses of “know” and “ought” are natural, in view of the scarcity of
cases in which we can use “know” and “ought” in their proper, strong senses.
There is no need for legislation against them. But we are deeply aware of the dif-
ference when facing matters of major importance, and then we stay faithful to
the strong senses. We do not usually say that we know we will be alive tomorrow
and thus that we need not pay our life insurance premium today, regardless of
how healthy and safe we think or feel we are today. Serious people buy fire in-
surance even though they have never had a fire. And we do not say that children
ought to sacrifice their lives if they think or feel they ought to do it.
We can now understand why the strong sense of “know,” which requires
that what we say we know is self-evident or seen to follow from what is self-evi-
dent, has been central in subjective epistemology. The attraction of religion is
that it promises certainty about matters of ultimate concern, not mere probabil-
ity. A religious person would not be satisfied if told that God probably exists. A
missionary does not win converts by assuring them that they would be justified
in believing, or that it is probable, that there is God. The attraction of subjective
epistemology is that it seeks certainty about matters of ultimate concern where
nothing less suffices. Its main topic was the existence of material things – of
the earth and the sun, of your body and mine. Among its topics were also the
existence of other minds and the validity of induction. It would be jejune in ev-
eryday life to say that we are only justified in believing, or that it is only prob-
able, that we have bodies, that others have thoughts and feelings, or that the

Authenticated
past tells us anything about the future. Only a philosopher might be satisfied if
told that other people only probably exist. And it would be outrageous in epis-
temology to suppose that by saying that the external world probably exists we
are genuinely answering the skeptic. The focus on epistemic justification or epis-
temic probability, rather than knowledge, came about precisely when epistemol-
ogists concluded that we cannot have genuine knowledge of such matters, yet
remained unwilling to accept skepticism.
The concept of knowledge played a central role in Cartesian epistemology.
But Descartes’s principal question was not “What is knowledge?” This question
had been asked before and usually answered, briefly and informally but suffi-
ciently, in the same way – knowledge is apprehension, grasping, getting hold
of the truth and then steadfastly holding on to it. The Cartesian question that in-
augurated subjective epistemology was rather whether we have knowledge of
anything, especially of an external material world. Cartesian epistemology
began by taking skepticism seriously, hoping to refute it. And for this reason
it was essential that the Cartesian epistemologist ask whether I have knowledge,
that is, to employ an indexical, the first-person pronoun, rather than a name like
“Descartes” or a definite description like “the author of the Meditations.” For had
the Cartesian epistemologist done the latter, the skeptic would have complained
of question-begging. One who questions the existence of the external material
world questions also the existence of human beings, including Descartes,
since they, or at least their bodies, are parts of that world.
In the First Meditation, Descartes could not have used “I” to refer to the
Frenchman named Descartes, since that Frenchman was part of the world the ex-
istence of which Descartes was to prove later, in the Second Meditation. When he
employed the argument “I think, therefore I am,” Descartes could not have been
referring even to the “thinking thing” that in the Second Meditation he conclud-
ed he was. He could not have said he was referring to his thinking thing, for how
would he have answered the question, which thinking thing it was? He could not
have said, “my thinking thing,” for “my” is the possessive adjective correspond-
ing to “I.” Louis XIII also was, or had, a thinking thing, but Descartes did not
suppose he was proving the existence of that thinking thing and thus of Louis
XIII. Nor could he infer from “I think” just “There is a thinking,” as some
have suggested. Was it Descartes’s or Louis XIII’s thinking? If it was no one’s,
there might be thousands of such orphaned thinkings. The existence of which
one was Descartes inferring? Epistemological ventures seldom benefit from on-
tological adventures. To confront the skeptic without begging the question Des-
cartes needed to begin his inquiry by renouncing claims to any subject matter.
He could refer to nothing even when using “I.” Subjective epistemology must
lay claim to no subject matter when attempting to refute skepticism.

Authenticated
Descartes probably did not see that he faced these difficulties because all
along he thought he was “directly aware” of a thinking thing and its ideas.
But even if we ignore the question about which thinking thing and ideas
those were, “his” or Louis XIII’s, yet another question, also fundamental but ig-
nored by Descartes, can be asked. If a necessary condition of awareness is that
its object exists, then the skeptic would ask whether Descartes was really aware
of a thinking thing and its ideas, just as the skeptic asks whether we really per-
ceive bodies when we think we perceive them. And if the existence of its object is
not a necessary condition of the awareness, then the skeptic of course would
question the cogency of the inference from the occurrence of the awareness to
the existence of the thinking thing and its ideas. Descartes thought he might
be deceived by God or an evil demon regarding 3+2=5, but did not see that if
this is so then he might be deceived also about what he thought he was aware of.
This, we may note, vitiates also Descartes’ several inferences from the exis-
tence of his idea of God to God’s existence. Couldn’t God, or an evil demon, de-
ceive him into thinking that he had that idea? Perhaps he did not really have it.
The failure to see that in this way the skeptic could question any appeal to
awareness vitiates also the familiar appeals in post-Cartesian subjective episte-
mology to “intuition,” “immediate experience,” or “direct acquaintance.” Plato
pointed out the poverty of such appeals in the Theatetus, and so did Hegel,⁴⁵
but they remain common in philosophy.
Thus, while naturalistic epistemology has a subject matter too limited to be
philosophical, subjective epistemology appears to have no subject matter at all.
Its raison d’être is to meet the challenge of skepticism. Otherwise, there would be
no rationale for distinguishing it from naturalistic epistemology, albeit it would
be a naturalistic epistemology concerned, inexplicably, with just one natural ob-
ject, just one human being – oneself. To remain subjective, subjective epistemol-
ogy must refer to the “self” only by means of indexical expressions such as “I.”
To both have a subject matter and not beg the question against the skeptic, it
must be satisfied with a subject matter that is an entity that can be referred to
only with indexicals. But would anything be an entity if it could be referred to
only with an indexical? To suppose that there could be thinkers who are only
Is borders on incoherence, just as to suppose that there could be times and pla-
ces that are only nows and heres borders on incoherence. Even to say this has
required use of the grammatical monstrosities “Is,” “nows” and “heres.”
 See, for example, Hegel’s Logic, trans. William Wallace (Oxford: Clarendon Press, 1975), Part
Five.

Authenticated
Before his optimistic inferences to the existence of God, Descartes seemed to

have only himself, to be in a state of absolute solitude. But in fact he did not
have even himself, whether as the writer of the Meditations or just as a certain
thinking thing. He had nothing. Subjective epistemology is dependent on the
use of “I” as a dangling pronoun, a pronoun without an antecedent noun. It
is like a geography of here or a history of now that in principle is unable to
say where is here or when is now.
This is why subjective epistemology is often described as epistemology from the
first-person perspective. But there are no such entities as first persons, second per-
sons, or third persons, there are only first-person, second-person, or third-person
pronouns. The vague and much abused term “perspective” can be misleading,
but it is helpful because there are no obvious alternatives to the adjective “perspec-
tival.” There is an alternative to the noun “perspective,” however, that is less affect-
ed, namely, “view,” as long as we understand this term not in the optical but its or-
dinary broad sense. So understood, “view” is a synonym of “cognition.” The “first
person,” we may then say, is only a perspective, a view, a cognition. It is in this
sense that subjective epistemology is only perspectival. There is no entity that is
just the first person, and so the subject matter of subjective epistemology is not
that privileged entity – the first person is not a person. It is only a view, a cognition.
I shall have much more to say on this topic in Chapter Ten.
Nonetheless, though subjective epistemology lacks a subject matter and is
only perspectival, the rationale for it is impeccable. Lack of subject matter
does not imply unimportance. The idea of a geography of here or a history of
now that is unable to say where is here or when is now does seem absurd. But
the idea of traveling from here and now without being able to say from where
or when does not. A journey must begin somewhere and at some time, even if
we cannot say where and when. Subjective epistemology may be an epistemol-
ogy of pronouns without nouns, but to get nouns we must, so to speak, begin
with pronouns. The first-person pronoun is indispensable, not because of what
it refers to but because of the role it serves in the initiation of cognition. It is es-
sential to all talk and thought, and thus to all inquiry.
In normal contexts, to heedfully assert the sentence p one must be willing to
assert “It is true that p.” But to heedfully assert “It is true that p,” one must be will-
ing to assert “I know that p” (rather than the very different “I think that p” or even
“It is probable that p”). Indeed, one must be willing to assert “I know that p” in
order to heedfully assert “He (she, Jack, the expert) knows that p.” I can say that
Jack knows when the train will leave even if I cannot say that I know, but I cannot
say he knows that it will leave at 5 p.m. if I cannot say that I know this. In any in-
quiry, one must begin with the first-person view, with the use, however implicit, of
“I,” even if only in judgments, implicit or not, such as “I’ll look for it in the bush,”

Authenticated
“I’ll ask Jill,” or “I’ll check the dictionary.” This is a proposition neither of physics
nor of metaphysics. It’s like “Every journey must begin somewhere,” not like “Every
journey must begin in Woodbury.” Yet the proposition enjoys the abstraction char-
acteristic of philosophy and bequeaths it to subjective epistemology. “I think” must
be able to accompany all of our representations, Kant held, even though, as Sartre
later argued, it seldom actually does. Russell wrote, “When you are considering any
sort of theory of knowledge, you are more or less tied to a certain unavoidable sub-
jectivity, because you are not concerned simply with the question what is true of the
world, but ‘What can I know of the world?’…You cannot go outside yourself and
consider abstractly whether the things that appear to you to be true are true.”⁴⁶ Rus-
sell may have been wrong in thinking that there is an “inside” to be contrasted with
an “outside,” but his grasp of the rationale for subjective epistemology was firm.
Unless one opens one’s eyes and looks, one does not see. Unless there is a view,
nothing is seen.
This is why the allure of the subjective turn that Cartesian epistemology ini-
tiated is ever-present. It would be sad if subjective epistemology were all there
were to epistemology, but outrageous to deny its essential insight. As a theory
it is futile and usually misguided, yet it is as indispensable and unavoidable
as one’s awareness that to get anywhere one must start somewhere and that
to see anything one must look. The mistake is to suppose that subjective episte-
mology is about me, even if there is such an entity as me, whether a human
being or a mere thinking thing. It is the mistake of supposing that subjective
epistemology has a subject matter and thus that it is a cognitive discipline, a
theory of something, presumably knowledge or cognition, when in fact it only
draws attention to what is the necessary entry into any subject matter and serves
as the prelude to any discipline. Subjective epistemology must use “I,” or a syn-
onym of it, yet it can refer with it to nothing, not because there is nothing to refer
to but because of the very nature of its project. That project was not a mistake.
Antirealism is a metaphysical, not epistemological, theory, but it shares with
subjective epistemology the peculiar feature I have just described. Its thesis is
that the world, at least insofar as it is perceived, understood, or described, depends
on our powers of perception, conceptualization, and speech. This is not a zoological
proposition, however. It is not about humans, even though we are humans. We can
now see better how it should be understood. Indeed, the proposition is not about
humans, but neither is it about nonhumans. It is not about entities at all. Rather,
it is about the necessary conditions of all thought and talk about entities. As
such, it is intimately related to subjective epistemology, as intimately as Kant’s phi-
 Bertrand Russell, The Philosophy of Logical Atomism (Chicago & La Salle: Open Court, 1985), 37.

Authenticated
losophy was related to Hume’s, and can cast further light on it. That Hume’s skep-
ticism led to Kant’s transcendental idealism was not just an event in Kant’s personal
life. It manifested a turning-point in the history of philosophy, just as more than a
century earlier Descartes’ method of doubt had manifested its turn to epistemology.
Hegel’s absolute idealism manifested a third turning-point. The subjective episte-
mologist makes essential use of the first-person singular pronoun “I,” but Hegel
saw that it is the first-person plural pronoun “we” that is essential to full-fledged
cognition. The self-centered focus on the conditions that my cognition, my heedful
thought and talk, must satisfy was broadened as well as deepened by Hegel’s focus
on the conditions that our cognition, our heedful thought and talk must satisfy. To
heedfully assert p, one must indeed be willing to assert “I know that p.” But this is
only the necessary prelude to full-fledged cognition, which would be expressed by
“We know that p.”
3 Epistemology-as-Logic
Subjective epistemology does not have a subject matter. Naturalistic epistemol-

ogy does, but its subject matter is human and thus lacks the supreme abstraction
and generality characteristic of philosophy. The third variety of epistemology,
epistemology-as-logic, has a subject matter that exceeds the bounds of the spe-
cial sciences. Like formal logic, it is unambiguously “dehumanized” and belongs
in philosophy. All three varieties of epistemology, however, are defensible within
the limits of their very different yet not incompatible concerns. Their differences
call for mindful distinctions, not mindless quarrel.
In the Introduction I noted that in ethics some have eschewed the need for
empirical investigation by taking “the deontic way,” focusing on what we ought
to or at least are morally permitted to do. In the present chapter we saw that
some have taken the deontic way in epistemology by focusing on what we
ought to or at least would be “justified” to believe. They have thus avoided com-
peting with the empirical sciences. Whatever its merits in in these cases, the de-
ontic way is natural in epistemology insofar as its concerns have to do with the
validity of certain inferences and thus are essentially logical. Logic, the “art of
reasoning,” is commonly said to tell us what may and what may not be inferred
from a given proposition or set of propositions, clearly a deontic matter. The in-
ferences of interest in epistemology are seldom deductive, and for this reason the
need for grounding judgments about their validity is especially evident. Much of
traditional epistemology has consisted in the search for such grounding.
Contrary to what textbooks sometimes say, formal logic is concerned not
with inferences as actions, presumably human and thus properly studied by psy-

Authenticated
3 Epistemology-as-Logic 43
chology, but with their formal validity, the relation of the truth-value of the prem-
ises to the truth-value of the conclusion, in particular, the formal consistency of
the conjunction of the premises and the negation of the conclusion. Its general
subject matter thus consists of alethic relations, in the broad and etymologically
proper sense of relations between propositions with respect to their truth-value.
If some propositions, or at least sentences, are neither true nor false, as a con-
sequence of the truth-value of other propositions or sentences, this fact too
would belong in its subject matter.
Formal logic thus exemplifies the generality and abstraction definitory of
philosophy to the highest and purest degree. This is why Aristotle assigned
“the principles of the syllogism,” especially that of noncontradiction, to “the sci-
ence of being qua being.” This is why Frege wrote, “Just as ‘beautiful’ points the
way for aesthetics and ‘good’ for ethics, so do words like ‘true’ for logic…[I]t falls
to logic to discern the laws of truth…The Bedeutung [reference, meaning] of the
word ‘true’ is spelled out in the laws of truth.”⁴⁷ Elsewhere, Frege explained:
“What is distinctive about my conception of logic is that I begin by giving
pride of place to the content of the word ‘true’ …”⁴⁸ If metaphysics is the science
of being qua being, logic may be said to be the science of being qua truth, ethics
of being qua goodness, and aesthetics of being qua beauty. Indeed, all four –
Being, Truth, Goodness, and Beauty – belonged among what the medievals
called transcendentalia.
Epistemology-as-logic differs from formal logic by focusing on the validity –
legitimacy, cogency, worth – of certain nonformal inferences, but its subject mat-
ter, like that of formal logic, consists of alethic relations, in particular, the rela-
tion of the truth-value of the premises of the nonformal inference to the truth-
value of its conclusion. It too enjoys the level of generality and abstraction char-
acteristic of philosophy. Like formal logic, it is concerned not with inferences as
human actions but with the alethic relations they exemplify. Unlike subjective
epistemology, it does not lack subject matter, it is not just perspectival. And un-
like naturalistic epistemology, which does have a subject matter, it is not just
programmatic. Of course, epistemology-as-logic does apply to human matters,
just as formal logic does. But it is not about them. There is nothing puzzling
about this. Arithmetic also applies to humans, as well as to bats and stars,
but it is neither about humans nor about bats or stars. It is about numbers.
In attempting to answer the skeptic, subjective epistemology hoped to find
cogent inferences, formal or nonformal, from premises it deemed known to be
 Gottlob Frege, “Thought,” in M. Beaney, The Frege Reader (Oxford: Blackwell), 325 – 326.
 Beaney, 36.

Authenticated
true, even if as minimal as “I think.” Epistemology-as-logic, however, does not

ask whether the premises of such an inference are true, nor does it agonize if
the inference is not deductive. It is free from obsession with skepticism, just
as naturalistic epistemology enjoys such freedom. In this it follows the lead of
its older and more mature sibling, formal logic. In evaluating a deductive argu-
ment, formal logic is not concerned with the truth of its premises. And, we may
note in passing, it also does not fret that, even if they are true, the formal validity
of the argument might not suffice for the truth of its conclusion because God
might deceive us about logic just as he might deceive us about arithmetic, the
latter a possibility Descartes did worry about.
Epistemology-as-logic may aim at providing eventually a general theory of
the alethic relations exemplified in nonformal inferences. But it began by exam-
ining particular kinds of such inference, just as formal logic began with the ex-
amination of Aristotelian syllogisms and only later, mainly through Frege’s work,
offered a general theory of the alethic relations they exemplify. Inferences involv-
ing probability, induction, and abduction are standard topics in subjective epis-
temology. They would also be topics in epistemology-as-logic, though in abstrac-
tion from possible use against skepticism, which was the reason for subjective
epistemologists’ interest in them. Appeals to probability, of course, have failed
to answer the skeptic, who would either deny that mere probability suffices or
question the truth of one or more of the premises. Nonetheless, the calculus
of probability remains an established discipline of some distinction. Appeals
to induction and abduction, notoriously, also have failed to answer the skeptic.
But they remain standard topics in the philosophy of science, which seldom
strays into Cartesian doubts.
But of special interest to epistemology-as-logic would be relations of nonfor-
mal entailment. That there are such relations is usually acknowledged independ-
ently of epistemological concerns. A standard example is the entailment of being
colored by being red, and anyone who, like Kant, regards mathematical truths as
necessary but “synthetic,” i. e., not reducible to logical truths, allows also for
nonformal entailments in mathematics. Neither mathematicians nor philoso-
phers of mathematics worry that God might be deceiving us about 7 + 5 = 12.
An inventory, much less detailed discussion, of all nonformal alethic rela-
tions is neither possible nor needed here. I shall limit myself to an especially im-
portant one: the relation of presupposition. Bringing to light and focusing on
that relation was the turning-point in the development of epistemology beyond
its subjective stage. If we call subjective epistemology Cartesian, then epistemol-
ogy-as-logic, insofar it focuses on the relation of presupposition, may be called
Kantian. Hence the application of Kant’s term “transcendental” to recent argu-
ments from presupposition. Kant’s project in the Critique of Pure Reason can

Authenticated
be described as discovering the presuppositions of the several levels of cogni-

tion, though instead of “presupposition” Kant used “condition” and would
have described the project as discovering the conditions that the several levels
of cognition must satisfy. But epistemology-as-logic need not adopt Kant’s essen-
tially mentalistic approach to epistemology or any of his specific doctrines.
The relation of presupposition became a major, though also controversial,
topic in 20th century philosophy because of Strawson’s criticism of Russell’s theo-
ry of descriptions.⁴⁹ Strawson gave as examples the presupposition of “There is a
king of France” by “The king of France is wise,” and the presupposition of “He is
not dead” by “He cares about it.” (“Does he care about it? He neither cares nor
doesn’t care; he is dead.”)⁵⁰ If France does not have a king, neither “The king of
France is wise” nor “The king of France is not wise” is true. A proposition p pre-
supposes a proposition q, according to Strawson, when both “If p then q” and “If
not-p then q” are true, in other words, when q is a necessary condition of both
the truth and the falsity of p, thus of p’s being a coherent proposition at all, even
though p does not formally entail q and not-p does not formally entail q.⁵¹ In the
example of the king of France, the presupposition manifests itself in our dismiss-
ing as confused, not inconsistent, both anyone who said that the king of France
is wise but denied that France has a king and anyone who said that the king of
France is not wise but denied that France has a king. Our judgment of anyone
who said, “He cares about it but he is dead,” would be similar.
Presupposition is neither formal entailment nor a probabilistic, inductive, or
abductive relation. This is why appeals to it have seemed to provide answers to
skepticism entirely different from the usual answers. The latter are almost cer-
tainly either formally invalid or contain premises the skeptic finds as question-
able as the conclusion. The anti-skeptic’s predicament has been that to answer
the skeptic one must assume more than the skeptic would allow, but if one as-
sumes less then the answer does not follow from the assumptions. In appeals to
presupposition, however, the consequence of denying the presupposed proposi-
tion is not the falsity but the incoherence of the proposition that presupposes it,
whether a trifling incoherence as in the example of the king of France, or a deep
one as in the examples I shall sketch shortly.
 P. F. Strawson, “On Referring” (Mind, 1950) and Introduction to Logical Theory (London: Me-
thuen, 1952), 175 – 179. In the former, more influential work, Strawson did not use the term “pre-
supposition,” and wrote instead of “some sense of ‘imply’” that is “not equivalent to ‘entails’ or
‘logically implies.’”
 Introduction to Logical Theory, 18.
 Introduction to Logical Theory, 175.

Authenticated
Some have said that presupposition is merely a feature of language, just “inter-
nal” or “pragmatic,” not “logical” or “semantic,” as if pervasive features of lan-
guage are ever merely features of language. Aristotle defended the principle of non-
contradiction not by trying to infer it from “more certain” propositions, but by
showing that it is presupposed even by any reasoning intended to cast doubt on
that principle. Russell repeatedly pointed out that the proposition “what follows
from a true proposition is true” is primitive and presupposed by all deductive rea-
soning.⁵² That the natural sciences are rife with presuppositions has always been
evident. Physics presupposes, it does not discover, the existence of a spatiotemporal
world. Psychiatrists presuppose, they do not discover, that it is not evil spirits that
cause mental illness. The examples from Strawson that I have mentioned may be of
little intrinsic interest but this cannot be said of those in his major metaphysical
work, Individuals, or his book on Kant, The Bounds of Sense. And certainly it cannot
be said of the examples in Kant’s own works.
Kant defended important but controversial philosophical propositions on
the ground that they are presupposed by other propositions that are not contro-
versial. A simple example is the presupposition that the objects of sense percep-
tion (“outer sense”) are in space. “This page is white” would not be true if this
page were not in space, but neither would “This page is not white” be true if this
page were not in space. Kant’s argument that morality presupposes freedom is
another and famous example. Freedom seems to be presupposed by all genuine
actions, moral, immoral, or nonmoral. It is what seems to distinguish actions
from mere movements. A no less famous but more difficult example is Kant’s ar-
gument that objective order in time presupposes causal necessity. It is complex
and not to be dealt with lightly, whether in agreement or disagreement, but we
need not go into its details to get a glimpse of it. If we ask whether Jack met Jill
before or after she moved to town, the answer would depend in part on reason-
ing about when and where he could have met her. When such questions and their
answers really matter (as they often do in courts of law), it would be foolhardy to
rely on memory impressions, what Kant called the subjective play of fancy. As
these examples show, what is presupposed need not be a single proposition,
just as a deductive proof ordinarily does not rest on a single premise. It might
even be a system of propositions, and what presupposes it might itself be
such a system. This is why the philosophically interesting examples of presuppo-
 Bertrand Russell, The Problems of Philosophy (Oxford: Oxford University Press, 1953), 71– 77.
See Alfred North Whitehead and Bertrand Russell, Principia Mathematica to *56 (Cambridge:
University Press, 1962), 1.1.

Authenticated
sition seldom have the simple structure of the examples about the present king
of France and the man who is dead. Here are three other examples.
The first is a presupposition especially relevant to traditional epistemology:
the existence of a material world. It assumes many forms, and is neither simple
nor obvious. G. E. Moore noted that in doubting the existence of the material
world Descartes would have had to doubt the existence of other philosophers,
past and present, including those he had read, heard, argued with, and whose
works and views were the context of his doubt, whether through agreement or
disagreement.⁵³ Philosophers are human beings, therefore inhabitants, parts,
of the material world. The history of philosophy is not a history of angels. Des-
cartes could not have taken his doubt seriously as a philosophical doubt if he had
considered what would be the case with respect to his own doubt if there were
not a material world. If genuinely philosophical, skepticism about the material
world questions its own existence.
The historical context of philosophical thinking, such as Descartes’ doubt, is
essential to it, however original the thinking may be. It is even more obviously
essential to it than, as contemporary essentialists have argued, the biological ori-
gin of an organism is essential, “metaphysically necessary,” to that organism.
The “historicity” of a philosophical view is no more a contingent fact than the
historicity of a political event. Both bear necessary relations to their past. Neither
Cartesian epistemology nor Democratic or Republican politics in the 21st century
would be comprehensible if stripped of such relations. The skeptic questions
what makes it possible for skepticism to be the philosophical view it is: its
roots in what some other philosophers have held. It would not exist if those phi-
losophers had not existed, and thus if the material world did not exist. Des-
cartes’ methodological doubt would not have occurred if the proposition he
doubted had not been true.
Indeed, the very language Descartes used to develop and explain his doubt
would not have existed. And employment of language is essential to philosoph-
ical thought, even if rudimentary thoughts are possible without language. Phil-
osophical thought, whether superior or mediocre, involves argumentation, good
or bad, which has a fairly complex structure, distinct premises and conclusions,
each with its own structure, and logical connections rooted directly or indirectly
in that structure. The terms employed in the argumentation are chosen usually
with deliberation and discretion from a fairly extensive and often highly techni-
cal lexicon. Any language employed in philosophy, say, Descartes’ French or
 G. E. Moore, “A Defense of Common Sense,” in Philosophical Papers (London: Allem &

Unwin, 1959), 32– 60. I discuss the argument in Skepticism about the External World.

Authenticated
Latin, also involves phonemes and inscriptions, and has been shaped by a
human community. All three – phonemes, inscriptions, and the human com-
munity – are parts of the material world.
Could the argumentation be stated and explained in a private language, in
the minimal, not necessarily Wittgenstein’s, sense of a language created by
the philosopher alone without reliance on a public language, as in devising a
secret code? Surely such a private language, even if possible, would be too prim-
itive. The reader is invited to attempt constructing a fragment of one and then
translating a paragraph from Descartes’ Meditations into it. Writing philosophy
is not like recording one’s aches and pains. A private language for the latter, per-
haps invented by a hypochondriac, might be possible, but to say that the sophis-
ticated language employed by the philosophical skeptic could be such as lan-
guage would be mere posturing. Might the argument take place just in the
skeptic’s thought, without use of language, public or private? Even if some
thought without language, e. g., recalling a sensation, were possible, to suppose
that philosophical thought might be such thought would be like supposing that
we can understand differential equations without using symbols. The skepticism
considered in Descartes’ First Meditation was not a tipsy sailor’s declaration,
“Maybe I know nothing.” It was a professional, serious and informed, philosoph-
ical view. This is why we still take it seriously. Of course, that philosophical skep-
ticism about the material world questions its own existence does not entail that
it is false. It does not make it self-contradictory. But it does make it deeply inco-
herent. If the material world did not exist, then it itself would not exist.
The second example of philosophically interesting presupposition can be
found in Sartre’s strikingly original defense of the existence of other minds.
One’s acceptance of the “Other” is not discursive, he pointed out. It is presup-
posed by many of one’s own psychological states, it is essential to them. Sartre
dwelt at length on the experience of shame when looking through a keyhole but
seeming to hear footsteps. It is “an immediate shudder that runs through me
from head to foot without any discursive preparation.” It is “shame of oneself be-
fore the Other,” even if I know that no one is actually looking at me.⁵⁴ To say that
what is presupposed here is only the possibility of being the object of another’s
look would be to deny that the experience is genuine shame. There is almost al-
ways such possibility. At any rate, the interesting sort of skepticism about the ex-
istence of other minds, one that is not just a trivial consequence of skepticism
about the existence of other human bodies, questions even that possibility.
For it questions the very intelligibility of there being anyone “other than myself.”
 Being and Nothingness, 22. Italics in original.

Authenticated
It denies that we can even “conceive” of another mind, somewhat as Berkeley

denied that we can conceive of unperceived bodies. Such a skeptic owes us an
account of that “immediate shudder that runs through me from head to foot.”
My third example of presupposition belongs to an even deeper level. We may
call it conceptual presupposition. What is presupposed is not a proposition but a
particular understanding of a certain concept. For example, all discussions,
skeptical and anti-skeptical, of the existence of bodies presuppose a particular
understanding of the concept of existence. Therefore, to ask whether bodies
exist presupposes some answer to the question of what it is for a body to
exist, and ultimately of what it is for something to exist. Is existence a property?
If it is not, then what are skepticism and anti-skepticism about? Would it be in-
telligible to speak of the truth or falsity of any sentences asserting or denying ex-
istence, and thus of knowledge or ignorance of their truth? Would such senten-
ces express genuine propositions?
Standard epistemology, whether naturalistic or subjective, provides little
guidance on these questions. It usually takes for granted Kant’s view that exis-
tence is not a real predicate, meaning that it is not a property, a “determination,”
of a thing, res, but fails to consider the epistemological implications of the
view.⁵⁵ Kant himself wrote that the existence of a thing has to do “only with
the question whether [the] thing is given to us in such a way that the perception
of it could in any case precede the concept.”⁵⁶ This was not Berkeley’s “to be is to
be perceived.” It was closer to Mill’s “matter is a permanent possibility of sensa-
tions.” Kant held that to be actual (wirklich) a thing must “stand…in accordance
with the laws of empirical progression.”⁵⁷ We cannot go here into the details of
these laws, but like any other laws they would involve, according to Kant, appli-
cations of the pure concepts of the understanding. If so, the actuality or exis-
tence of a thing, like its causality, was for him not so much a matter of discovery
of empirical fact as the imposition of a concept. “Exists” expresses not a property
we find in a thing, but rather a conception of the thing that we contribute.⁵⁸ If,
following etymology, we use “object” for anything that is or can be “placed be-
fore or presented to the eyes or other senses” (Oxford English Dictionary) or to
consciousness generally, the “mind,” we can use “entity,” again following ety-
mology, for any object to which the concept of existence applies. Obviously,
there are objects, “things,” that do not exist, e. g., the Easter Bunny or a child’s
favorite but imaginary friend Jack. Objects and entities may not differ with re-
 For a succinct statement of Kant’s view, see Critique of Pure Reason A 598/B 626.
 Critique of Pure Reason A 225/B 272.
 Critique of Pure Reason A 493/B 521.
 I develop this view in detail in Skepticism about the External World.

Authenticated
spect to aesthetic qualities, but it is entities that are of interest in cognition, as

children soon learn. This does not mean that how we apply the concept of exis-
tence is mere caprice, any more than that how we apply the concept of causality
is one. What is meant is closer to what Nelson Goodman meant by the “entrench-
ment” of the predicate “green,” which the predicate “grue” lacks, though for
Kant the application of the concept of existence was grounded in the activity
of our faculty of the understanding, not in our linguistic practices. Clearly,
Kant’s view of existence required that skepticism and subjective epistemology,
insofar as they concern the existence of bodies, be drastically rethought – or
just altogether bypassed by him, as in fact they were.
Is existence just what the existential quantifier expresses, the “satisfaction”
of a propositional function, as Russell argued and most contemporary epistemol-
ogists take for granted?⁵⁹ To suggest that it is would be a nonstarter. Whether the
propositional function “x is a horse” is satisfied depends on what we allow as
values of the variable x. Is “x is a horse” satisfied by both Secretariat and Pega-
sus, or only by Secretariat? If we say the latter, our reason would be that Pegasus
does not exist, but now in a sense of “exist” obviously other than yet presup-
posed by Russell’s. It is its ordinary sense, which we employ in saying that
the Easter Bunny and the child’s friend Jack do not exist.
In any genuine case of inference involving presupposition, there is a natural
desire to think of it as a case of formal entailment, since this is the sort of entail-
ment we find most familiar and understand best. And when we see that the con-
ditional corresponding to the inference is not a tautology we are tempted to de-
clare the inference invalid. Or, if we still find it compelling, we are tempted to
change the conditional so that it becomes a tautology, and bless the change
by calling it an “analysis” or “translation.” This is what Russell did in his theory
of definite descriptions. He saw that if the present king of France is wise then, of
course, France has a king. He also saw, however, that this is not a tautology. So,
he proceeded to “translate” it into one. In the philosophically substantive cases,
however, like those Kant and Sartre made familiar, no such analyses or transla-
tions are plausible. The proposition “If there is objective order in time, then there
is causal necessity,” for example, obviously is not a tautology, and to try to
change it into one by “analysis” would hardly be a task worth attempting. So,
the remaining option is to just deny the proposition.
When the presupposition is conceptual, however, there is no such third op-
tion. For example, the skeptic can deny, however implausibly, that skepticism,
 Bertrand Russell, Introduction to Mathematical Philosophy (New York: Simon and Schuster,
1971), 164.

Authenticated
like all philosophical thought, presupposes the existence of the material world,
that objective order in time presupposes causal necessity, or that the phenomen-
on of shame presupposes the Other’s look. But the skeptic is not likely to ques-
tion the presupposition of a particular understanding of the concept of existence
when doubting or affirming the existence of an external world. An answer,
whether affirmative or negative, to a question presupposes understanding the
question. The skeptic cannot deny the central place of the concept of existence
in any discussion of what does or does not exist, or of what we can or cannot
know to exist. Berkeley’s “to be is to be perceived” was not a casual remark
made when defending his immaterialism.

Authenticated
Chapter Three: The Property Good
1 Anthropocentrism and Conceptual Analysis in Ethics
Although ethics usually has concerned itself explicitly with human matters, there
have been noteworthy exceptions. Plato’s theory of the Form of the Good in The Re-
public is a famous example. In the 18th century, as we saw in the Introduction, Kant
distinguished sharply between the metaphysics of morals and the empirical disci-
pline of “practical anthropology.” In the 20th century, G.E. Moore offered in Principia
Ethica a detailed exposition and defense of an ethics concerned with what he called
the nonnatural property good. It is well-known but usually vigorously rejected. And
fifteen years later Wittgenstein outlined in Tractatus Logico-Philosophicus a subtle
conception of ethics as concerned with the world, not oneself or other humans. It
has received little attention, unlike the rest of the book. Plato, Kant, Moore, and
Wittgenstein provided both the rationale and the outline of a dehumanized ethics,
though they differed much in other respects. In this chapter I shall consider Moore’s
ethics, and Wittgenstein’s in Chapter Four. My aim will not be to offer exegesis but to
provide detailed examples of dehumanized ethics at its best and explain why and
how they avoided anthropocentrism.
In Principia Ethica Moore proposed that good (I shall follow his use of the ad-
jective instead of the noun “goodness”) is a simple, indefinable, nonnatural prop-
erty, and that this property is the proper subject matter of ethics. The book was pub-
lished in 1903, and became a signpost in the philosophy of the following 100 years.
It may still be too early to judge how 20th century philosophy ended, but its begin-
ning was remarkable. Russell’s Principles of Mathematics also appeared in 1903, the
first volume of Husserl’s Logical Investigations in 1900 – 01, and four of William
James’s major philosophical books in 1902– 09. Wittgenstein’s Tractatus Logico-Phil-
osophicus was written between 1914 and 1918. There was not a significant difference,
except perhaps in style and temperament, between Anglo-American and European
philosophers. The analytic/continental schism came much later. In the Preface of
Principia Moore wrote that his ethics was closest to Franz Brentano’s. Both Russell
and Husserl began as mathematicians. Frege was a German philosopher-mathema-
tician to whom, by their own admission, Russell and Wittgenstein were heavily in-
debted. Russell studied and discussed Meinong in detail. James was admired in
both Britain and Europe, influenced Husserl and Wittgenstein, and was the subject
of articles by Moore and Russell.
In Principia Ethica Moore dehumanized ethics even more clearly than Plato had
done. But the book also inaugurated what has come to be known as analytic ethics,
the sort of ethics characteristic of 20th century analytic philosophy. Analytic ethics

Authenticated
1 Anthropocentrism and Conceptual Analysis in Ethics 53
began largely in opposition to Moore’s thesis that the property good is indefinable.
But Moore had made clear that he had no interest in what he called verbal, and the
tradition calls nominal, definitions. They are the business of lexicography, he wrote.
Yet it was just such definitions that analytic philosophers sought, sometimes calling
them “analyses.” The most familiar example, noted in chapter 2, comes from ana-
lytic epistemology, not ethics: the definitions of “S knows that p” that preoccupied
epistemologists from the late 1960s to the early 1980s. The definitions sought and
offered by analytic philosophers were not even lexicographical, definitions that re-
cord lexical facts and are tested by empirical investigation of speech and writing.
Rather, they recorded their authors’ impressions of lexical facts, and were tested
by the authors’ “intuitions” about what we would or would not say, usually in
some hypothetical situation. The question “How do I know what we would or
would not say in that situation?” was usually ignored. For it could be answered
properly only by appealing to what others have said in similar situations, and
thus making an appeal, however amateurish, to lexical fact. Even the Oxford English
Dictionary is valuable mainly for the examples of usage it lists, not the definitions it
tries to distill from them.
Searches for such definitions became alien to contemporary philosophy of
language because of the three trailblazing developments in it more than half a
century ago that were mentioned in the Introduction. The first was Quine’s arti-
cle “Two Dogmas of Empiricism,” published in 1951, which attacked appeals in
philosophy to meanings. It is widely accepted today, though often only pro
forma. Phrases such as “conceptual question,” “conceptual content,” and “con-
ceptual connection” still abound in the literature, with the noun “concept” ex-
plicitly used for the meaning or use of a word. The second development, also
often accepted but just pro forma, was Wittgenstein’s relentless argument in Phil-
osophical Investigations, posthumously published in 1953, that words are not
used in accordance with necessary and sufficient conditions. He gave “game”
as an example, but the argument applies also to “good,” “right,” “reason,”
“know,” “exist,” and other denizens of the philosopher’s lexicon that, like
“game,” are everyday words, not technical terms introduced as abbreviations
of multi-clause descriptions. The third development was Chomsky’s linguistics,
first proposed in that same decade. It marked a striking advance by insisting
on the biological basis of linguistic competence and the use in the study of lan-
guage of the standard methods of scientific research.
The kind of definition Moore did allow in philosophy was an account of the
constitution of what is defined, a listing of its parts. It was closer to what the phil-
osophical tradition calls real definition, though it mentioned not genus and differ-
entia but parts. Such a definition can be called an analysis, in a sense reasonably
similar to that employed in chemistry. In later years analyses were offered mainly of

Authenticated
54 Chapter Three: The Property Good
facts and propositions, which were taken to be nonlinguistic entities categorially dif-
ferent from those chemistry analyzes. Their analysis was intended to reveal logical
form, and for this reason was called logical analysis. It was in such analyses that
analytic philosophy took root, beginning in 1905 with Russell’s theory of definite de-
scriptions and perhaps culminating in Moore’s claim two decades later, in “A De-
fense of Common Sense,” that he knew the proposition “This is a hand” to be
true but did not know how to analyze it. In Principia Ethica, however, Moore’s ex-
ample was the definition of a horse and consisted of an anatomical inventory.
Moore in effect agreed with Kant that “in matters of morality it is always real defi-
nitions that must be sought.”⁶⁰
Despite its inattention to what he meant by “definition,” analytic ethics did
begin and develop in relation to Moore’s ethics. But discussions of Principia sel-
dom ventured beyond Chapter I, which alone is included in most anthologies.
Usually ignored was the crucial Preface, where Moore explained what he
meant by “intuition” and “self-evidence,” and thus what anyone calling him
an intuitionist and foundationalist ought to mean. Also usually ignored were
the beginning of Chapter 2, where he explained what he meant by “natural”
and “nonnatural,” thus what anyone calling his ethics nonnaturalistic ought
to mean, and Chapter V, where he explained his theory of right on the basis of
the theory offered in Chapter I.
By “intuitions,” Moore wrote, he meant self-evident propositions, and “nothing
whatever as to the manner or origin of our cognition of them.”⁶¹ And a self-evident
proposition, he explained, is one that is evident but not by virtue of inference from
other propositions.⁶² Moore did not say what he meant by “evident,” perhaps think-
ing it unnecessary. I suggested in the previous chapter that a proposition is evident
if it is or can readily be seen to be true, either literally or metaphorically. Therefore, it
may also be said to be known, in the serious and traditional sense of “know,” which
sharply distinguished between knowledge and belief or opinion. I also noted that
the noun “evidence,” as used in a court of law or in the lab, has a wider meaning,
but the same root. Moore described as “self-evident,” if evident at all, those prop-
ositions in ethics that state “what kind of things ought to exist for their own sakes,”
i.e., are intrinsically good.
As to the meaning he attached to calling something “natural,” Moore wrote
he meant that the thing is in time. Therefore, what is nonnatural is what is not in
time. This is not an idiosyncratic sense of “natural.” To be sure, we usually think
 Immanuel Kant, Logic, trans. Robert S. Hartman and Wolfgang Schwartz (New York: Dover,
1974), 144.
 G. E. Moore, Principia Ethica, Revised Edition (Cambridge: University Press, 1993), 36.
 Principia Ethica, 193.

Authenticated
of the things in nature as being in both time and space. But there might be things
that unquestionably are in nature but not in space because they are not physical.
The obvious example would be irreducibly mental states, the existence of which
most psychologists at the time accepted. They would not be in space but they
would be in time. (This is why in later philosophy the paradigm of nonnatural
entities were abstract entities such as numbers.) The mere fact that a thing
falls outside the subject matter of physics does not make it nonnatural. If we de-
fine a natural thing as one belonging in the province of the natural sciences, we
would need a noncircular account of what is meant by “natural sciences,” as
Moore was doubtless aware and for this reason did not offer such a definition.
But the fact is that the distinction between the natural and the nonnatural did
not play a central role in his book, though the phrase “naturalistic fallacy”
did. As Moore made clear in the also usually ignored Chapter IV, which was de-
voted to what he called metaphysical ethics, even ethical theories concerned
with the “supersensible” committed the fallacy. The fallacy was merely a case
of confusing two things: the property good and some other property.
In Chapter I of Principia Ethica Moore argued that the property good is non-
natural and simple, therefore indefinable if defining something is listing its
parts, that almost all earlier ethical theories had committed the naturalistic fal-
lacy of confusing it with some other property, and that they could be refuted with
what later was called the open-question argument, which in effect encouraged
the reader to pay close attention to the property the theory confuses with the
property good in order to see that they are two properties, not one.
It may be worth mentioning that Moore’s contemporaries in the Society of
Apostles and the Bloomsbury Circle, who included Russell, Keynes, and Virginia
Woolf, found more important, not these metaphilosophical generalities, but the
substantive view, defended in Chapter VI, that personal affection (love, friend-
ship) and aesthetic appreciation (contemplation of beauty, in art and in natural
objects, human and nonhuman) are the greatest goods. In contrast with Kant’s
view that a good will is the only thing that is unconditionally good and Mill’s
that pleasure alone is good, Moore held that love and the contemplation of beau-
ty are the Ideal. But he rejected Sidgwick’s contention that nothing “appears to
possess this quality of goodness out of relation to human existence, or at least to
some consciousness or feeling.”⁶³ This may be what prompted Keynes to rate
Moore higher than Plato.
 Principia Ethica, 133.

Authenticated
Chapter VI has also been ignored by analytic ethics, which has focused in-
stead on the preliminary discussions in Chapter I, especially the “objectivity of
value” it took Moore to be defending there. But, as Brian Hutchinson points
out in a recent book, “Moore never even entertained doubts about the objectivity
of value.” Hutchinson acknowledges that for us this may be “a mystery difficult
to fathom,” but suggests that the mystery “is to be savored rather than solved.”⁶⁴
While the central tenet of Moore’s theory of good was that good is a simple,
indefinable, and nonnatural property, the central tenet of his theory of right was
that the right action, i. e., duty, in a particular situation, is the action that “will
cause more good to exist in the Universe than any possible alternative,”⁶⁵ that “is
the best thing to do,” what “together with its consequences presents a greater
sum of intrinsic value than any possible alternatives,” either because it “itself
has greater intrinsic value than any alternative” or because “the balance of in-
trinsic value” of its consequences does, so that “more good or less evil will
exist in the world” if it is adopted.⁶⁶ “Cause” and “produce” are used in the
broad sense of “contribute,” since the action might be the best thing to do be-
cause of its own goodness or because of its “organic,” not causal, relationships.
Moore has been called a utilitarian, but unlike Bentham’s, Mill’s, or Sidgwick’s
utilitarianism, Moore’s presupposed a theory of good that places no limits on
what items might be intrinsic goods, thus allowing that some may be actions.
Moore’s theory of right may be called cosmological. It tells us that we ought
to do what would be best, all things in the Universe considered. It is not idiosyn-
cratic. It accords, for example, with Aquinas’s first principle of natural law:
“Good is to be done and promoted, and evil is to be avoided.”⁶⁷ It does imply,
as Moore noted, that justice is not to be done if the heavens should fall – unless,
he wryly added, “by the doing of justice the Universe gains more than it loses by
the falling of the heavens.”⁶⁸ As we shall see, the ethical views of Russell and
Wittgenstein, the other two founders of analytic philosophy, were also nonnatur-
alistic and cosmological, perhaps because of Moore’s influence. But, with the ex-
ception of H. A. Prichard, a philosopher of unsurpassed acuity, and W. D. Ross,
whose terminology and distinctions are still found indispensable, later Anglo-
American ethics diverged in both respects. They are related. If ethics is natural-
 Brian Hutchinson, G.E. Moore’s Ethical Theory: Resistance and Reconciliation (New York:
Cambridge University Press, 2001).
 Principia Ethica, 198.
 Principia Ethica, 76 – 77.
 Thomas Aquinas, Summa Theologica, Part Two, Question 94, Article
 Principia Ethica, 197.

Authenticated
istic, it is not likely to be cosmological. And if it is cosmological, it is not likely to

be naturalistic.
Naturalistic ethics is almost certainly ethics humanized: it is about humans,
not cats or bats. So it is not cosmological. Not only does it ignore the good of the
universe, it ignores that of extraterrestrials, angels, and gods, if there are any,
and usually also that of nonhuman animals, plants, and rivers. Thus it lacks
the supreme generality and abstraction distinctive of philosophy and alone jus-
tifying its existence alongside the other cognitive disciplines. A cosmological
ethics can be expected, of course, to have application to humans, just as chemis-
try and mathematics can. But this makes none of them about humans.
Of course, we are human, thus interested in ourselves and other humans.
But this is not a reason for making ethics be about us, any more than it is a rea-
son for making astronomy to be about us. Indeed, as I noted in the Introduction,
there is a special, deep, and often misunderstood sense in which humans may be
cosmically central – namely, that leading to Kant’s transcendental idealism and
its recent versions, e. g., in Nelson Goodman’s and Hilary Putnam’s works. In that
special sense, Kant, Putnam, and Goodman may be said to have humanized even
astronomy. But they did not hold that astronomy is about humans.
If humans are natural objects, a species of animal, we can hardly expect to
have special philosophical knowledge of them, just as we can hardly expect to
have special philosophical knowledge of stars or bats. Accounts of human
well-being do not belong in philosophy, just as accounts of human anatomy
and human evolution do not. They belong, I have argued, in the empirical scien-
ces. The argument does not depend on a narrow use of the words “natural” and
“science.” If mental states are not reducible to physical states, there could still be
a natural science of them, in Moore’s sense of “natural” and the traditional
sense of “science,” in which history and political geography are social sciences.
In fact there was such a science in Moore’s time – namely, introspective psychol-
ogy. The argument did depend, however, on the fact that qualification for re-
search in empirical matters requires special training, for example, in chemical
analysis or the use of telescopes. Nothing analogous with respect to humans oc-
curs in philosophy seminar rooms.
How to achieve happiness, in the ordinary sense, recognized by both Kant
and Mill, of enjoyment or satisfaction of our needs and desires, has been a
stock question in ethics, with both Epicurus and Plato offering advice, but argu-
ably the invention of aspirin and contraceptives, tractors and pesticides, air con-
ditioning and spreadsheets, answered it more effectively. In Buddhist ethics,
sadly but realistically, suffering seems the primary concern, not pleasure, as
in Western ethics. But Indian utilitarians who hope to learn from Americans
how to reduce suffering presumably go to American colleges of agriculture

Authenticated
and public health, not American philosophy departments. One may ask, indig-
nantly, what about loftier goods, not Bentham’s but certainly Plato’s and
Kant’s, such as justice, authenticity, salvation? Especially in India, a deeply re-
ligious country, they are often thought far more important. But these loftier
goods call for nonzoological considerations. Neither psychology nor lexicogra-
phy would illuminate them.
Philosophers who avow allegiance to naturalistic ethics do write about some
loftier goods, at least about justice. But, as we saw in the Introduction, they are
more likely to adopt a conception of ethics, far removed from both naturalism
and nonnaturalism, as a “conceptual,” not “factual,” discipline. This allows
them to avoid both commitment to nonnatural facts and responsibility for com-
petence regarding natural facts. But if concepts are in nature – presumably in
human minds, brains, or languages – they too lie outside philosophers’ compe-
tence: there is psychology (introspective or not), neuroscience, as well as linguis-
tics and scholarly lexicography. If they are not in nature, Moore’s venture into
the nonnatural was at least straightforward. It was also not dated, though its crit-
ics often call it “obsolete.” Like the 17th century way of ideas, conceptual analysis
perhaps went out of date when Kant pointed out in 1787 that our business is not
merely to analyze concepts but to extend our knowledge.⁶⁹
Like post-Gettier analytic epistemology, post-Moorean analytic ethics was
unfazed by misgivings such as Quine’s about meanings. It clung to conceptual,
in fact linguistic, analysis. It went through several stages. The first began in Vien-
na, soon after the publication of Wittgenstein’s Tractatus and with some person-
al involvement by him. Ethical statements were rejected as nonsense, or at least
as lacking cognitive sense. The subtlety of Wittgenstein’s verbally similar posi-
tion, however, was missed altogether. The second was to offer a positive charac-
terization: they are expressions of emotion. But the rich literature already in ex-
istence on the emotions in psychology (from James to Arnold) and
phenomenology (from Meinong to Sartre) was ignored, even though it seems
to show that the emotions are not, as the emotivists thought, self-contained sub-
jective episodes, Humean “impressions of reflexion,” but rather intentional
states, directed upon objects, with character dependent on that of their objects,
and thus in principle cognitive. The third stage, perhaps motivated by the expe-
rience of World War II, which made both the outright rejection and the emotivist
interpretation of ethical statements appear jejune, was to claim that they express
a special “moral point of view,” something psychologically no less genuine than
the emotions but supposedly less subjective, and that their function is to “guide,
 Critique of Pure Reason, B 18.

Authenticated
2 The Good and the World 59
not goad.” In effect, this third stage consisted in denying ethical statements a
full-fledged, unqualifiedly cognitive status, yet conceding that their function is
not merely that of imperatives or exclamations. In all three stages it was taken
for granted that the job of ethics is to describe the content of ethical concepts,
or the meanings or uses of ethical words, the “workings” of ethical discourse.
We cannot give people what really interests them, namely, an ethics that says
what they should do, Moore’s heirs held, but we can give them an ethics that
says what they mean, a “metaethics.” This was the message even of the more re-
cent fallback positions of “projectionist” antirealism and “supervenience” real-
ism, where the focus remained metaethical, not substantive. Few worried that
the very idea of telling people what they mean seems paradoxical.
2 The Good and the World
By taking the conceptual turn analytic ethics did not provide a genuine alterna-
tive to “ethics humanized.” The alternative provided by Moore remained. Let us
consider some of its details. I suggested earlier that the role of the indefinability
and nonnaturalness of the property good was relatively minor, given Moore’s ex-
planation of what he meant by them. Less familiar is that Moore proposed a cri-
terion, a test, for determining whether a thing exemplifies that property, whether
it is intrinsically good.
The criterion was “the method of isolation.” It consisted in asking whether a
world – a whole world – that contains the thing but otherwise was just like a
world that lacks it would be better.⁷⁰ The two worlds might be wholly inanimate,
and even considered “apart from any possible contemplation…by human be-
ings.” The focal good is that of the world, not that of the human or sentient
parts of it. It is independent even of possible human consciousness. An essential
part of Moore’s ethics was the principle of organic wholes: “the value of a whole
may be different from the sum of the values of its parts.” The method of isolation
suggests that the world itself is an organic whole.
In addition to (1) being good intrinsically, independently of anything else, by
exemplifying the property good, a thing (object, action, state, property) may be
said to be good because (2) it has a totality of consequences that exemplify that
property, or because (3) it noncausally enhances the goodness of the organic uni-
ties or wholes of which it is an element. Its overall goodness thus depends on the
actual or possible goodness of the world, the “Universe.”
 Principia Ethica, 135– 136, 143 – 147, 236 – 238, 245 – 247.

Authenticated
Pleasure is a familiar example of something that is good in sense (1), surgery

of something good in sense (2), and the presence of attractive surroundings
might be an example of something that is good in sense (3). A thing could be
said to be good overall only if considered in all three respects. In the case of ac-
tions, consequentialism ignores sense (1), deontological ethics ignores sense (2),
and sense (3) is ignored by both. Acknowledging all three, as Moore did, gives
both consequentialism and deontological ethics their due. In particular, conse-
quentialism is given its due because the goodness or badness of an action overall
(in W. D. Ross’s terminology, if not meaning, its being actually, not just prima
facie, right) would depend, in part, on the goodness or badness of its consequen-
ces – all of them, for anything less would be morally unacceptable.
All the consequences of an action and all the organic wholes to which it belongs
are relevant to its being good or bad overall. The action will have consequences even
after the human species has become extinct, a fact that thoughtful environmental-
ists would not find irrelevant. An explicit provision of Moore’s method of isolation
was that the two hypothetical worlds might be supposed to be unpopulated, even
considered “apart from any possible contemplation…by human beings,” as Moore
did when applying the method to beauty.⁷¹ The action belongs to organic wholes
possessing parts, both near and remote, past and future, with an indeterminate
range, and thus requiring reference to all space and time.
Moore defined duty in Principia Ethica as the action that will cause more good
to exist in the Universe than any possible alternative, but he also described what “is
good in itself or has intrinsic value” as what “ought to exist for its own sake.”⁷² In
his later work Ethics he repeated that “it is always our duty to do what will have the
best possible consequences,” but denied that this is “a mere tautology.”⁷³ In the
preface to the second edition of Principia Moore wrote that he had used “good”
in a sense that bears an “extremely important relation to the conceptions of
‘right’ and ‘wrong.’”⁷⁴ Clearly, what he offered was not a stipulative definition,
but neither was it a lexicographical report. This is what we should expect.
The connection between good and right cannot be trivially definitional, but it
also cannot be just happenstantial. And even if good were a natural property,
right would remain nonnatural if understood in terms of good in the manner
Moore proposed. For it would still involve reference to all the consequences of an
action and all the organic wholes of which it is a part – to all space and time,
 Principia Ethica, 133. The propriety of Moore’s use of beauty as an example here is astutely
discussed by Robert Audi in “Intrinsic Value and Reasons for Action,” in Metaethics after Moore.
 Principia Ethica, 34.
 G. E. Moore, Ethics (Oxford: Oxford University Press, 1912) 73.
 Principia Ethica, 4.

Authenticated
2 The Good and the World 61
the whole world. But these totalities of consequences and organic wholes, indeed
the world itself, would not be natural objects in Moore’s sense. They would not
be natural because they would not be in time, even if they consisted only of things
that are in time. They may be not natural also in a larger sense. As we shall see in
the next chapter, Wittgenstein held that, although sentences about such totalities
show something, indeed the “higher” (das Höhere), they say nothing.
The idea that the rightness of an action involves reference to the whole world
is not purely philosophical. It is supported by scrupulous moral thought, which
sets no time or place beyond which it cares not what happens. Some Americans
do care about the floods in Bangladesh, and many people, wherever they may
be, care about the climate on earth a century from now. Authentic environmen-
talists do not say that when humans become extinct, whales and prairie grass
might as well. Many believe honesty would be owed to, and expected of, even
extraterrestrials, angels, and gods, should they exist.
I mentioned earlier that Russell’s and Wittgenstein’s views were also nonna-
turalistic and cosmological, perhaps partly because of Moore’s influence. I am
interested here not in their historical connections, but in the light they may
cast on each other. In 1903, when Principia was published, Bertrand Russell
wrote, “Man’s true freedom … [lies] in the determination to worship only the
God created by our own love of the good.”⁷⁵ In 1914, after two years of intense
discussions with Wittgenstein, he attributed to “the ethical work of Spinoza…
the very highest significance,” as “an indication of some new way of feeling to-
wards life and the world.”⁷⁶ This new way of feeling, Russell added, lay outside
the scope of “the scientific method.” Hegel wrote that “to be a follower of Spi-
noza is the essential commencement of all philosophy.”⁷⁷ Spinoza defined God
as “substance consisting of infinite attributes, each one of which expresses eter-
nal and infinite essence,” and declared that “love towards God is the highest
good which we can seek.”
The method of isolation implies important and illuminating, but seldom no-
ticed, similarities of Moore’s views also to Kant’s and Wittgenstein’s. For Kant,
“the ultimate end of the pure use of our reason” was ethical, but he resolved
to “[keep] as close as possible to the transcendental and [to set] aside entirely
what might…be psychological, i. e., empirical,”⁷⁸ since “the metaphysic of ethics
 Bertrand Russell, “A Free Man’s Worship,” in Mysticism and Logic (London: Allen & Unwin,
1917), 50.
 Bertrand Russell, “Scientific Method in Philosophy,” in Mysticism and Logic, 109.
 Hegel’s Lectures on the History of Philosophy, (London & New York: Routledge and Kegan
Paul ; Humanities Press, 1974), 257.
 Critique of Pure Reason, A 797/B 825-A 801/B 829.

Authenticated
is really the pure morality, which is not grounded on any anthropology.”⁷⁹ Witt-
genstein’s position will be discussed in chapter 4. Suffice it here to point out
that, since Moore’s theory of right action involves reference to the world as a
whole (duty is “that action, which will cause more good to exist in the Universe
than any possible alternative”), he might have agreed with Wittgenstein that in
some sense ethics concerns the “limits of the world,” not its contents, that it is
“transcendental.”⁸⁰ It may be worth noting that as a student Moore studied Kant
assiduously, even attending a course on Kant in Germany, and that he and Witt-
genstein enjoyed a close relationship.
Moore and Russell were not Hegelians, but they grew in a philosophical cul-
ture dominated by Hegel’s philosophy. Wittgenstein had close ties to both. Spi-
noza’s Substance, Hegel’s Absolute, and Moore’s, Russell’s, and Wittgenstein’s
worlds were very different, but all five were taken to be reality as a whole.
That Russell and Wittgenstein may have shared a nonanthropocentric view
of ethics such as Moore’s should not be surprising. The view was not entirely
novel. Plato held that the philosophic life culminated in a glimpse of the
Form of the Good, which he held to be indescribable. Aquinas placed Good in
the company of Being, One, Truth, and Beauty, the so-called transcendentalia,
which were said to range across the categories, i. e., the highest genera, thus
to lack even the status of categories of things in the world, much less the status
of things, and therefore to be indefinable per genus et differentiam. In philosoph-
ical theology God had been described as a being of infinite goodness that is the
source and measure of all other goodness, earthly and unearthly. And Kant, as if
using words from Wittgenstein’s Tractatus, wrote that, unlike what he called
practical anthropology, moral thought is concerned, not with what happens,
but with what ought to happen, even if it never happens.⁸¹
3 The Relevance of the Property Good
It is often asked whether a dehumanized ethics such as Moore’s could be “rele-

vant” to action. The question is as ancient as Aristotle’s complaint about the “in-
efficacy” of Plato’s Form of the Good. But it is ambiguous. It may be asking (1)
whether Moore’s property good could motivate actions by itself, (2) whether
one could be motivated by cognitive states of which it is an object, by itself or
 Critique of Pure Reason, A 841– 842/B 869 – 70.

 Tractatus, 6.13, 6.421.
 Critique of Pure Reason, A 802/B 830.

Authenticated
3 The Relevance of the Property Good 63
as a constituent of a state of affairs, or (3) whether there could be such cognitive

states in the first place.
The idea of being motivated to action by Moore’s property good indeed has
doubtful coherence, but only because the idea of being motivated by any prop-
erty seems incoherent, for purely logical, not ethical or psychological, reasons. A
sentence of the form “FRx” would be ill-formed. But it is not incoherent to ask
whether one can be motivated by cognitive states of which the property good,
or a state of affairs that includes it as constituent, is the object. Could such cog-
nitive states be “reason-giving”? Could they be “action-guiding”? The question is
difficult to answer because those who specialize in such matters, namely, psy-
chologists working on motivation, still know too little, perhaps because of the
moral and social constraints on serious experimentation with human subjects.
Hume did announce that reason is, and ought to be, the slave of the passions,
but we may wonder how he knew this. Perhaps Hume could be excused for
thinking that “passions” are discoverable through introspection. But his own
views on causation should have kept him from thinking that their being motives,
i. e., their motivating, could be discovered in such manner.
Some have denied that there can be cognitive states of which the property
good, or a state of affairs that includes it as constituent, is the object on the
grounds that the property good could have no causal efficacy. They usually
rely on causal or quasi-causal theories of knowledge, which are accepted either
because they seem “scientific” or because of thought-experiments about what we
would or would not say in imaginary situations. Some even deny the cognitive
status of arithmetic if it is taken to be about abstract, nonspatiotemporal entities,
since they assume that causation is a spatiotemporal relation. But this is to as-
sume that the nature of causality is clearer than the nature of knowledge, or at
least that we understand the former better than the latter.
Many, of course, just deny that there is such a property as Moore’s property
good. Elsewhere I have suggested that it should be considered a generic property,
though one on a very high level of generality.⁸² It is the genus of which Moore’s
personal affection and aesthetic appreciation, Aristotle’s eudaimonia, Mill’s
pleasure, and other goods reasonably proposed by reasonable people would
be species. This is why we cannot “see” it in the way we can see a shade of yel-
low. Neither can we see color, the genus of which a shade of yellow is a species,
in that way, though color is a generic property on a much lower level of general-
ity than goodness. Perhaps generic properties have no causal powers by them-
selves but their species do. A tire’s having shape helps no car roll but its
 Skepticism in Ethics (Bloomington & London: Indiana University Press, 1989).

Authenticated
being round does. A traffic light’s being colored stops no driver but its being red
sometimes does. If Moore’s ideal goods – personal affection and aesthetic appre-
ciation – are species of goodness, perhaps they have causal powers. If they do,
as seems obviously to be the case, but their genus does not, the reason may be
that the latter is a generic property, like shape or color.
Whether or not this suggestion is right, those who deny that we can have
cognitive access to Moore’s property good on the grounds that it would have
no causal efficacy need to pay more attention to such metaphysical details.
They also owe us answers to questions like those explored by David Armstrong⁸³
and Evan Fales.⁸⁴ The first is what is causation? Is it, as Armstrong and Fales
argue, a relation between universals, properties? If not, is it nonetheless ground-
ed in such a relation? And are properties universals in the first place, or are they
rather particulars, “tropes,” perhaps both as Moore in fact held?⁸⁵ How would
the arguments against the cognitive accessibility of Moore’s property good
read in the case of each possibility? According to Fales, there must be properties
we can identify independently of their causal powers if a vicious infinite regress
is to be avoidable. His examples are the properties characterizing the content of
sense perception, though his ultimate concern is with properties in physics.⁸⁶ But
if some properties can be identified without reference to their causal powers, so
might Moore’s property good, whether or not it has causal powers.
Needless to say, I shall not attempt to answer these questions here, but an-
swers, detailed and carefully worked out, are needed. Otherwise it is difficult to
take seriously the complaint that Moore’s property good can have no causal effi-
cacy, that we could have no cognitive access to it, that therefore that it would be
irrelevant to action. Appealing to “naturalism” or “the scientific point of view,”
much less to “our intuitions,” is not enough. Without such answers the com-
plaint might be like the 17th century natural philosophers’ complaint that Newton
appealed to occult and immaterial gravitational forces, rather than to intelligible
and robust bumping, or like H. A. Prichard’s complaint that Einstein’s theory of
relativity was unintelligible because it relied on a non-Euclidean geometry. How
a body could “motivate” another body at a distance was incomprehensible to
those 17th century natural philosophers, and Prichard could form no mental
image of a non-Euclidean space.
 David M. Armstrong, Universals and Scientific Realism (Cambridge: Cambridge University

Press, 1978).
 Evan Fales, Causality and Universals (London: Routledge, 1990).
 G. E. Moore, “Identity,” Proceedings of the Aristotelian Society, n.s., v. I, 1900 – 1901.
 Causality and Universals, 221– 224.

Authenticated
3 The Relevance of the Property Good 65
Therefore, I shall ignore the complaints about the relevance of Moore’s prop-
erty good to action. But another question does arise. It is both legitimate and
deep. By requiring reference to the whole world in judgments of duty, Moore
could tell us nothing specific about how we ought to act in a particular situation.
This is why he virtually admitted that, unlike his theory of good, his theory of
right was profoundly skeptical. There might be an action we ought to do, but
as a matter of empirical fact, not philosophical theory, we could not know
what it is. In view of the mind-boggling range of its consequences and organic
relationships, even probability statements about it could not be seriously
made. Radical moral skepticism thus seems inevitable.
But Moore’s moral skepticism does not lead to amoralism. His position is
often described as “ideal utilitarianism” because it was not mere consequential-
ism. An action may be intrinsically good even if it ought not to be done, even if it
did not make the world better. In Ross’s terminology, if not meaning, it may be a
prima facie duty even if not an actual duty. This is why respect for the good may
continue to inform one’s actions. Such respect would be akin to love, whether
practical or pathological, not to calculation. It would have as its object the intrin-
sic goodness of the action, its being a “prima facie duty,” even if not an actual
duty. Only a part of the world, not the whole world, can be loved.
I am not suggesting an inference, which surely would be specious, from the in-
trinsic goodness of an action to its rightness. No claim is being made that the former
makes the latter “probable,” or that it “justifies” or is a “reason” for the action. In
Hutcheson’s useful terminology, if not meaning, it is at most an “exciting reason,”
not a “justifying reason.” If in acting one is motivated and guided only by respect for
the good, yet only the intrinsic goodness of an action is intellectually visible, then
one is motivated and guided only by respect for the intrinsic goodness of the action.
One has no knowledge of the totality of its consequences and organic relationships,
indeed one has even no genuine conception of that totality.
Thus, qua agent, the ideal utilitarian can only be a deontologist, not a con-
sequentialist. We have to settle for right-minding, even if it does not coincide
with right-acting. This is why Moore’s ideal utilitarianism was not inimical to
moral common sense, which views with distaste the spirit of calculation, of
cost-benefit analysis, that ordinary consequentialism cultivates. The ideal utili-
tarian has no “justifying reasons” but plenty of “exciting reasons” for doing
good particular actions, namely, their goodness. Thus Moore’s dehumanized eth-
ics may be seen as the marriage – of love, not convenience – of the two great
ways of moral thinking: the utilitarian and the deontological.

Authenticated
Chapter Four: Saying and Showing The Good
1 The distinction explained
In Tractatus Logico-Philosophicus Wittgenstein made the case for a dehumanized

ethics more briefly but also more powerfully than Moore had done fifteen years
earlier. He knew Moore well. And Principia Ethica had been published nine years
before he met him. But he did not owe his views about ethics to Moore, even if
they influenced him in some respects. They involved tying the notions of ethics
to the notion of the world as a whole. As we saw in the previous chapter, this had
been done also by G. E. Moore, key features of whose ethics were the notions of
worlds “considered in isolation” and of the totality of the consequences of an ac-
tion. But Wittgenstein’s ethics was largely based on his views in logic, in partic-
ular, that the world is the totality of facts (Tractatus 1.1), that “fact” is a formal
concept (4.1272), that “world” therefore is also a formal concept, that the meta-
physical subject or philosophical self does not exist (5.633), that therefore there
is no philosophical distinction to be made between me or my life and the world
and ethical statements can only be about the world, and that like all sentences
employing a formal concept they say nothing even if they show much.
Wittgenstein’s first stay at Cambridge was in 1912– 14. Moore’s ethics was a
central topic of discussion in Britain at the time, not only among philosophers
but intellectuals generally, especially in the Bloomsbury Circle. Both Moore
and Wittgenstein were members of the Society of Apostles, together with Russell,
Whitehead, and Keynes. Wittgenstein had read Principia Ethica. In a 1912 letter to
Russell, he wrote that he hadn’t liked book, because Moore repeated “himself
dozens of times.”⁸⁷ Wittgenstein also attended Moore’s lectures. They camped to-
gether in Norway, Wittgenstein dictating philosophical notes to Moore. Moore
might well have agreed with Wittgenstein that by requiring reference to the
world as a whole ethics is about “the limits of the world,” not its human inhab-
itants. It would be surprising if they did not notice the similarity of their views on
this deepest level, despite their many differences, philosophical and personal.
At a meeting of the Apostles in 1912 Wittgenstein heard Moore’s paper “Is
Conversion Possible?” which Moore had first read to the Apostles in 1900
while working on Principia. That Moore read the paper again suggests he had
not abandoned its ideas. Moral conversion, he said in it, “is not unlike religious
conversion,” even though it “is not necessarily connected with any religious
ideas.” It is “both a great good in itself and it secures all other goods which de-
 Letters to Russell, Keynes and Moore, ed. G.H. von Wright (Oxford: Blackwell, 1974), 9.

Authenticated
1 The distinction explained 67
pend on one’s own mind alone….You see ‘life steadily and whole’ and can feel
neither desire nor fear of what you see to be bad in it.”⁸⁸
At about the same time, Wittgenstein wrote in his Notebooks: “To believe in a
God means to understand the question about the meaning of life…to see that the
facts of the world are not the end of the matter…to see that life has a meaning,”⁸⁹
and later, in the Tractatus: “The sense [Sinn] of the world must lie outside the
world” (6.41). The sense (or meaning) of the world is not something in the
world because it is the sense of the whole world. He went on to remark that
“It is not how things are in the world that is mystical, but that it exists” (6.44).
Indeed, even if we refrain from calling the world mystical, we should acknowl-
edge that it is mysterious. (It was one of the three spurious objects of knowledge
Kant thought required treatment in the transcendental dialectic, the other two
being the Self and God.) The reason the world is mysterious is logical, not mawkish
or cabalistic. It is not that the world is too big or too unlike what we take it to be. Not
its size or content, but its logical/ontological category, or rather its failing to fall in
any category, is what makes it mysterious. We may say that the world is everything,
but this would only acknowledge its peculiarity. For to speak of everything is to em-
ploy the predicate “is a thing” or “is a fact,” depending on whether we think the
world is the totality of things or of facts. Both predicates, Wittgenstein held, express
only formal concepts, corresponding to formal or internal properties, and thus the
sentences in which they occur say nothing, though they show much. This is why
Wittgenstein held that genuine propositions about it, and thus ethical propositions
as he understood them, are impossible.
Contrary to received opinion, neither this distinction between saying and
showing nor the picture theory of meaning and thought on which it is based
is idiosyncratic or obscure. The distinction has a straightforward, noncontrover-
sial application even to ordinary pictures, say, paintings and photographs, in-
deed to representations generally. And the picture theory is merely a subtler ver-
sion of the traditional theory of meaning and thought, which was unabashedly
representational, “pictorial”: thought involves “ideas,” often explicitly under-
stood as mental images or pictures, and the meaning of an expression is what
it stands for.
The distinction between what can be said and what can only be shown pro-
vided a welcome alternative to the stark choice between realism and antirealism,
in logic, metaphysics, ethics, even the philosophy of religion. Wittgenstein
 Tom Reagan, Bloomsbury’s Prophet (Philadelphia: Temple University Press, 1986), 144.
 Ludwig Wittgenstein, Notebooks, 1914 – 1916, trans. G. E. M. Anscombe (New York: Harper &
Row, 1961), 74.

Authenticated
68 Chapter Four: Saying and Showing The Good
thought it concerned “the cardinal problem of philosophy.” Many of his readers

find it obscurantist. Tough-minded philosophers ask, “What are those things
that can only be shown? “How can there be such things?” But these questions
miss the point of the distinction, namely, to provide an alternative to both real-
ism and antirealism. What only shows itself is not a part of reality, like Secretar-
iat or oxygen. But neither is it unreal, like Pegasus or phlogiston. This is why in
the Introduction I called Wittgenstein’s position semirealism, distinguishing it
from both standard realism and standard antirealism.
The distinction becomes less puzzling when stated as the distinction, famil-
iar to kindergarteners as well as film producers, between telling and showing –
“show and tell.” Kindergarteners and film producers might or might not agree
that some things can only be shown, but they would agree unanimously that
showing is often much more effective than telling and sometimes alone possible.
Of course, they would not have Wittgenstein in mind. But Wittgenstein’s picture
theory of meaning does suggest a reasonably clear account of what it is to be
something that can be told or said, namely, to be capable of being pictured, rep-
resented, whether in an ordinary picture such as a painting or photograph, or in
a “logical picture” such as a sentence, and of what it is to be something that can-
not be told or said but could be shown, namely, to be capable of being shown in
a picture but incapable of being pictured. That there are “things” of the latter
sort should be noncontroversial.
In a painting, much is shown that is not and cannot be pictured by the paint-
ing or any part of it. For example, the painting may represent a tree next to a
barn, each represented by a part of the painting, and the spatial relation between
the parts of the painting that represent the tree and the barn would represent
their relation of being next to each other. But nothing in the painting represents
that relation’s being a relation, nothing “says” that their being next to each other
is a relation (rather than, say, a shape or color). Yet the painting shows this, in-
deed must show it in order to represent what it does represent. What it shows
cannot be denied as one might deny, for example, that the painting is a portrait
of Churchill. The absence from the painting of what it only shows would not be
like Churchill’s absence. Of course, paintings do not consist of words, and sen-
tences are only “logical” pictures. But like all pictures, physical or mental, paint-
ings are logical pictures, though not all logical pictures are paintings.
The picture theory of meaning was offered in the Tractatus first as an account of
thought in general: “A logical picture of facts is a thought [Gedanke]” (3). Only then
did Wittgenstein also offer it as an account of sentential meaning: “In a proposition
[Satz] a thought finds an expression that can be perceived by the senses” (3.1). “I call
the sign with which we express a thought a propositional sign” (3.12).

Authenticated
1 The distinction explained 69
The theory was in accord with the received, traditional doctrine that thinking
consists in operating with “ideas,” “mental representations,” perhaps even “mental
images.” It implied, as that doctrine also did, the no less traditional doctrine that
the truth of a judgment consists in correspondence to reality, to be established, if
possible, by comparison. But Wittgenstein applied these traditional doctrines to lan-
guage. He did so by proposing an unusual but not implausible conception of declar-
ative sentences as logical pictures. He thus made the distinction between thinking
and speaking seem insignificant. It was a “linguistic turn.”
To think of something, traditional philosophy of mind held, is to represent it
in the mind. Notoriously, this was most plausible and least unclear in the case of
the representations called mental images. But to speak of something, Wittgen-
stein held, is also to represent it in a picture, though the picture would not be
a painting or a photograph, or even a mental image. It would be a logical pic-
ture – a “propositional sign,” a sentence. Wittgenstein in effect strikingly broad-
ened the traditional conception of a picture.
A sentence (“propositional sign”) is a logical picture because it “depicts”
what it says by sharing with it logical form, rather than, say, shape or color,
as paintings, photographs, and perhaps mental images do. But Wittgenstein’s
view of a sentence was also strikingly broad, it was not the grammarian’s:
“The essence of a propositional sign is very clearly seen if we imagine one com-
posed of spatial objects (such as tables, chairs, and books) instead of written
signs” (3.1431). It would be unwieldy but possible to use tables, chairs, and
books instead of words to refer to objects, and to use configurations of tables,
chairs, and books instead of sentences to make statements. A sentence could
also be composed of mental objects. In a letter to Russell, Wittgenstein wrote
that a thought consists of “psychic constituents that have the same sort of rela-
tion to reality as words,” though he added: “What those constituents are I don’t
know”⁹⁰ This is why Wittgenstein’s often used “thought” [Gedanke] and “propo-
sition” [Satz] interchangeably. The linguistic turn that was implied by his asser-
tion “The limits of my language mean the limits of my world,” therefore, must not
be confused with the linguistic turn that became fashionable later, even though
it was initiated by the Tractatus. If our cognitive access to reality consists in “rep-
resenting” it, but the representations need not be more than logical, then wheth-
er the access is psychological or linguistic becomes irrelevant. This was the lin-
guistic turn that Wittgenstein took. Descartes, Locke, and Kant accepted the first
part of the antecedent of this conditional, but the second part did not even occur
 Letters to Russell, Keynes and Moore, 7.

Authenticated
to them. Had it done so, they might have accepted both, and then the history of
modern philosophy would have been dramatically different.
Wittgenstein’s distinction between “saying” and “showing” may be initially
mystifying, but one of its applications is familiar even to beginning students of
logic. That a statement is logically true, a tautology, i. e., that its negation is a
contradiction, often “shows” itself and is immediately “seen” in its logical
form, without reference to what the statement is about. But that a statement is
true in some other way, presumably empirically, does not show itself, its truth
cannot be just seen in its form – we must also know what it is about, what it
“says,” by attending to what the descriptive expressions in it stand for.
Although in accord with the traditional views of thought as representation
and truth as correspondence, Wittgenstein’s picture theory of meaning and his
distinction between saying and showing were a major, far-reaching revision of
those views. He accepted them only on the level of atomic sentences, where in-
deed alone they are plausible. But this is an extraordinarily primitive level. Sen-
tences on the highest levels, like those of logic and ethics, allow only showing,
not saying, he held. Even ordinary molecular and general sentences fail to be
pictures in Wittgenstein’s or indeed any view. A sentence of the form “If p
then q” is not itself a picture, even if p and q are. Not surprisingly, in his later
work Wittgenstein abandoned the picture theory of meaning and the correspond-
ence theory of truth. They are not defective, but their applicability is limited. This
is true, however, also of the traditional theory of thought and meaning: no phys-
ical or mental representation can be made of what a sentence of the form “If p
then q” says, let alone the sentences of logic and ethics
We can now understand the Vienna positivists’ passionate opposition to
what can only be shown, especially the ethical. No picture of any kind – phys-
ical, mental, or merely logical – can be made of it. No painting can literally de-
pict the goodness of a person or the rightness of an action. It follows that what
can only be shown is not observable, since presumably anything observable can,
at least in principle, be pictured, physically or mentally, and therefore also log-
ically. The traditional empiricists denied that there are unobservable entities –
we cannot have “ideas” of them, since ideas are copies of sensory impressions.
The 20th century logical empiricists denied that there are things that only show
themselves, and in particular that there are “ethical objects.” Their most familiar
claims about ethics did appear to coincide with Wittgenstein’s. If propositions
are pictures, then there can be no propositions of ethics. The ethical cannot
be said. But Wittgenstein held that it can be shown. He avoided unbridled real-
ism in ethics sufficiently to inspire logical positivism, yet he also avoided unbri-
dled antirealism sufficiently to protect the ethical.

Authenticated
2 Logic and the World 71
Wittgenstein used “shown” (zeigt), of course, as a metaphor. He resorted oc-

casionally to other terms, as in Tractatus 4.121: “Propositions cannot represent
[kann nicht darstellen] logical form: it is mirrored [spiegelt sich] in them….They
display it [Er weist sie auf].” And “say” (sagen), even if not used metaphorically,
is sufficiently vague in everyday language, German as well as English, to allow
for a variety of uses, as does also the closely related in meaning “tell.” A sen-
tence says or tells us something, but so do a list of names, a thermometer,
and a human face. Rudolf Carnap chided Wittgenstein for writing a whole
book and then concluding that what he had said in it could not be said. But a
simple way of avoiding the appearance of paradox would be to introduce, in
place of “say,” the two verbs “say1” and “say2,” explaining that what is said1
is what according to Wittgenstein is said, and what is said2 is what according
to him is only shown. For example, the sentence “This page is white” says1
that this page is white but says2 that the page is an individual object. Then Witt-
genstein’s otherwise puzzling conclusion in the Tractatus would be that what he
had said2 in the book could not have been said1. The difference would be ex-
plained sufficiently as the difference between what can be pictured and what
cannot. There can be a picture of what the sentence “This page is white”
says1, perhaps a photograph of it, but there can be no picture of what “This
page is an object” says2.
2 Logic and the World
In the Tractatus, as the title makes explicit, Wittgenstein was mainly concerned
with logic. No explanation is needed, therefore, of the brevity of his remarks
about ethics. Even in metaphysics, his focus was on its most abstract level,
that of logical form. Not only would nothing be perceived or thought in a
world without logical form, nothing would be a world without it. For nothing
that violated or could not be captured by logic could be a world. Even a world
consisting solely of immaterial or nonspatiotemporal objects must conform to
logic. Wittgenstein’s logical and metaphysical views would have remained un-
changed if he had supposed that his simple objects were angels. Unlike the log-
ical antirealist, he did not deny that there is logical form in the world. Rather, he
drew attention to the radical difference between logical form and the things that
exhibit it by pointing out that, unlike them, it cannot be pictured, though it can
be shown: “There are, indeed, things that cannot be put into words. They make
themselves manifest. They are what is mystical” (6.522).
If we take for granted that truth is correspondence to fact, as Wittgenstein
did, then we must take for granted also that there is something in the world

Authenticated
to which the logical form of a true sentence corresponds, viz. the logical form of
the fact it asserts. Otherwise, there would not be a fit sufficiently specific and
definite for truth. Yet the logical form of the fact is quite unlike the things the
sentence is about, viz. the constituents of the fact. It is invisible, as is the logical
form of the sentence, even if those things as well the sentence and its grammat-
ical form are visible. But it is also quite unlike the usual examples of what is in-
visible. The subject-predicate form would be present also in the putative facts as-
serted by the theological sentence “God is wise” and the mathematical sentence
“3 is an even number,” and while God and numbers are invisible there is an ob-
vious difference between the reason they are invisible and the reason logical
form is invisible. The page and color that the sentence “This page is white” is
about are visible, and so is its surface grammar, but its logical form is not. In
the case of “3 is an even number,” both the logical form of the sentence and
what the sentence is about are invisible.
In the Introduction I called Wittgenstein’s position logical semirealism in
order to distinguish it from both logical realism, which cheerfully allows for
such statements as “This page is an individual object,” and logical antirealism,
which no less cheerfully dismisses them. It is a sophisticated view. It should not
be confused with mere denials of reality. Its aim is not to fight superstitions or
fairy tales. Denying that some things can be said is not like denying that there
is such a horse as Pegasus or such a substance as phlogiston. What is denied
is that they are like Secretariat or oxygen. This is not to assert that they are
like Pegasus or phlogiston. It is to assert that they are different from all four.
Wittgenstein applied his semirealism chiefly to what he called “logical ob-
jects.” But it also had important implications for ethics – for what might be
called “ethical objects.” Indeed, shortly after completing the Tractatus, he
wrote that the point, the meaning, of the book was “an ethical one.”⁹¹ In this
chapter I am mainly concerned with this ethical “point.” But to understand it,
we must pay some attention to the logical/ontological views on which it rests,
though much more will be said about them in chapters 6 and 7.
The point of the Tractatus may have been ethical, but Wittgenstein declared
that its “fundamental idea” was that the “logical constants” (the sentential con-
nectives like “not,” “and,” “or,” “if…then”) and the quantifiers (“all,” “some”)
stand for nothing in the world.⁹² More obviously relevant to ethics was that
both “object” (“thing”) and “fact” were included in Wittgenstein’s list of formal
 Paul Engelmann, Letters from Ludwig Wittgenstein, 143 – 144. Cf. David G. Stern, Wittgenstein
on Mind and Language (New York: Oxford University Press, 1995), 8, 70 – 74.
 Tractatus, 4.0312.

Authenticated
2 Logic and the World 73
concepts (4.1272). It follows that the distinction between saying and showing ap-
plies also to the concept “world,” since the world is presumably either the total-
ity of things or the totality of facts. It is “nonsensical” to speak of all facts or of all
objects. We cannot speak of “totalities” determined by properties that can only
be shown, i. e., formal properties. Therefore, we also cannot speak of the world.
The totality of facts is the totality determined by the one-place predicate “is a
fact,” and the totality of objects is the totality determined by the one-place predicate
“is an object.” But being a fact and being an object are formal properties, which can
only be shown. “Object,” “complex,” “fact,” “function,” “number” signify formal
concepts, represented in logical notation by variables, for example, the pseudo-con-
cept object by the variable “x” (4.1272). The properties they appear to stand for are
formal, internal, such that it is unthinkable that what they are attributed to should
not possess them (4.123). For this reason it would be just as nonsensical to assert
that something has a formal property as to deny it (4.124).
The statement “This page is white,” for example, does say something. What
it says can be pictured literally, in a painting or a photograph. But the putative
statement “This page is an (individual) object” does not, for it presupposes what
it purports to say, its having sense, in particular the use of “this page” as its sub-
ject, depends on its being true.⁹³ Yet it is not gibberish.
The motivation behind the two most common reactions to Wittgenstein’s dis-
tinction between saying and showing was plain, though often tacit, empiricism.
The first holds that what only shows itself is not anything at all, since neither
logical nor ethical objects can be observed. Prima facie, there is little to be
said in favor of this view. Surely, what only shows itself, in logic or in ethics,
is not like Pegasus or phlogiston. Its claims to a status in reality would remain
even if they could not be fully met. The second, less straightforward, reaction is
to hold that what only shows itself cannot be said for reasons of surface gram-
mar. It resembles the common interpretation of Frege’s claim that the concept
horse is not a concept: “is a horse” is a grammatical predicate, not subject.
For example, Warren Goldfarb writes, “All we are doing [in speaking of logical
form] is noting that names have to be put together in one way or another in
order to make sentences.”⁹⁴ Of course, Wittgenstein was “noting” this, but it
was not all that he was doing. He was also trying to explain why we put
names together, why we need sentences rather than just names.
 In Principia Mathematica an individual is defined as anything that is neither a proposition

nor a function (9.13).
 Warren Goldfarb, “Metaphysics and Nonsense,” Journal of Philosophical Research XXII
(1997), 66.

Authenticated
Even when something can be said with a sentence, why could it not be said
with a list of names? This would be an awkward way of saying something, but
why is it not a genuine way? Lists of names sometimes do seem to say, “tell,”
something, e. g., in inventories. Wittgenstein’s answer was that we need senten-
ces, not just names, because the world is the totality of facts, not of things. It was
a metaphysical, not grammatical, answer. A world in which Jack is the father of
Jim and a world in which Jim is the father of Jack may contain the same things
and thus the same inventory, but they are different worlds. And why is it that
only some ways of putting words together say anything, i. e., count as well-
formed sentences? Wittgenstein’s answer was that the formal properties of
what names stand for, i. e, objects, allow only some configurations of objects
into states of affairs and thus only some logical pictures.
Perhaps there are better answers to these questions, but answers are needed
and few other than Wittgenstein’s have been given. The etymology of the word
“reality” deserves attention by both realists and antirealists. The root of “reality”
is res, the Latin for “thing,” and no argument is needed that “logical things”
such as negation and “ethical things” such as goodness are not things like Sec-
retariat or oxygen. To this extent logical and moral realism are unacceptable. But
also no argument should be needed that they are not things like Pegasus or phlo-
giston. To this extent logical and moral antirealism are unacceptable.
3 The World and the Good
Duty, Moore held, is the action that “will cause more good to exist in the Uni-
verse than any possible alternative.”⁹⁵ However, as we noted in the previous
chapter, a sophisticated but sensible moral realism like Moore’s would allows
that the action itself might be intrinsically good (prima facie right, in W. D.
Ross’s terminology). This would be a sufficient reason for doing it even if we
do not know what its consequences would be. Doing justice would be bad
(“wrong”) if the heavens should fall, though presumably it would remain intrins-
ically good. To say that justice is to be done even if the heavens should fall would
be moral posturing, not moral thinking. As Kant noted, Frederick the Great’s
committing suicide, if captured by the enemy, in order to protect his country
from extortion, might be good (“right”), even if suicide is bad (“wrong”) in itself.
Nevertheless, such moral realism does remain beholden to the future. For it
still enjoins us to take into account all the consequences of an action, since gen-
 Principia Ethica, 198.

Authenticated
3 The World and the Good 75
uinely moral thought sets no date and no place beyond which what happens
would not “matter.” Today’s small children would be retiring more than half-
a-century from now, but if we can we ought to help now to make their retirement
possible. The disastrous floods in Bangladesh occur thousands of miles from
Missouri, but Missourians who can ought to provide aid. The moral realism
just described, however, inherits the major defects of ordinary consequentialism.
The first is epistemological: we cannot know what we ought to do because
we cannot know all the consequences of our actions; we cannot even make seri-
ous probability judgments about such an indefinite, possibly infinite, totality.
The problem is familiar, it was discussed at length by Sidgwick. It arises because
of commonsense considerations, not philosophical theories. It renders literal
cognitivism questionable. I have discussed it in detail elsewhere.⁹⁶ The second
major defect of ordinary consequentialism is metaphysical and not familiar:
with respect to such a totality, realism, not just literal cognitivism, is questiona-
ble. Wittgenstein was the first to see this.
Wittgenstein described the concerns of ethics and religion as “the higher” (das
Höhere). But his reason was logical, nor ethical or religious: they are about the limits
of the world and thus exceed the limits of what is sayable. A logical category, such
as object or fact, might not seem to be something “higher” in the way the concerns
of ethics and religion do, but it does enjoy highest generality.
Whether the world is the totality of facts or of objects, “world” is a formal
concept. But that something falls under a formal concept cannot be said. Accord-
ing to Wittgenstein, ethical statements involve putative reference to the world.
Therefore, they attempt to say what cannot be said. They say nothing, even if
they show much. The controversy between moral realism and moral antirealism
thus becomes a special case of the controversy between metaphysical realism
and metaphysical antirealism. If ethics involves putative statements employing
formal concepts about the totality of objects or of facts, that is, about the
world, as both Moore and Wittgenstein believed, then according to Wittgenstein,
though not Moore, both moral realism and moral antirealism must be rejected.
Contemporary antirealism in ethics usually begins by rejecting Moore’s view
that goodness is a nonnatural property. But, as I pointed out in the previous
chapter, even if goodness were a natural property, rightness might still be non-
natural. For, as understood by Moore it involves reference to all the consequen-
ces of an action and all the organic wholes to which it belongs. If Wittgenstein is
right, such reference would be impossible. There could be no genuine statements
about such a totality, just as there could be no genuine statements be about the
 Cf. Skepticism in Ethics.

Authenticated
totalities of objects and of facts. Thus, in a different though no less important

sense of the word, those totalities could be said to be nonnatural. Yet rightness,
natural or nonnatural, is the raison d’être of ethics as a discipline insofar as it is
about moral appraisal of action.
In the Tractatus Wittgenstein wrote, “[I]t is impossible for there to be prop-
ositions of ethics” (6.42). But what he meant was not at all what his positivist
successors thought – indeed, it was just the opposite. He went on to explain:
“Propositions can express nothing that is higher….Ethics cannot be put into
words. Ethics is transcendental” (6.42– 6.421). So was also logic: “Logic is not
a body of doctrine, but a mirror-image of the world. Logic is transcendental”
(6.13). Wittgenstein took this to follow from the very nature of logic: “The prop-
ositions of logic describe the scaffolding of the world, or rather they represent it.
They have no ‘subject-matter’” (6.124). Propositions cannot represent logical
form. But they show it. Wittgenstein was not a realist, whether in logic or in eth-
ics, but neither was he an antirealist. He was a semirealist.
The difference between goodness or rightness and oxygen or yellow color
that moral antirealism insists on, even if grossly misunderstands, has been
amply demonstrated in its long history. It is also felt by anyone who asks the fa-
miliar and seductive but logically puzzling question, “Why should I do what I
ought to do?” Few agree with moral realism that goodness or rightness is at
all like oxygen or yellow. Yet, that the goodness and rightness are like Pegasus
and phlogiston is also seldom taken seriously, except by enfants terrible and phil-
osophes subtile. Few are unable to understand the difference between what
ought to be the case and what is the case. Few think of it as resembling the dif-
ference between phlogiston and oxygen.
Moral antirealists are seldom sensitive to the distinctions that led Wittgen-
stein to logical and then to moral semirealism. In this respect they differ radical-
ly from the philosophical roots they share with him in Kant’s transcendental ide-
alism. Kant proposed a unified account of the “empirically real” and the
“transcendentally ideal,” and he vigorously defended the need for acknowledg-
ing both. Nothing comparable in scope or depth can be found in standard moral
antirealism. Perhaps the reason is the tacit prejudice that the distinction between
reality and nonreality does not admit of refinement. Meinong called it prejudice
in favor of the actual. By “the actual” he meant, as Moore did, what is in time.
Early in the twentieth century, philosophers on both sides of the English
channel, including Moore, Russell, Meinong and Husserl, routinely distinguish-
ed between being and existence, meaning by the latter what Meinong meant by
“actuality,” and unhesitatingly asserting, for example, that relations and num-
bers have being, but denying that they exist, that they are in time. What exists,
they held, is in time. What has being but is not in time does not exist, though

Authenticated
some said that it “subsists.” Indeed, usually we speak of the existence of spatio-
temporal items, not of items such as relations or numbers. But, unless we are
philosophers, we do not say that relations or numbers do not exist, e. g., that
there is no such relation as fatherhood or such a number as 5. We usually just
ignore the issue and say nothing. Wittgenstein wrote the Tractatus when the dis-
tinction between being and existence was familiar, and undoubtedly he was sen-
sitive to the reasons for making it. It is not the same as his distinction between
what can be said and what can only be shown, but resembles it in motivation.
Contemporary antirealists ignore both distinctions, and assume that being, exis-
tence, reality, and actuality are the same, all expressed by the particular (“exis-
tential”) quantifier. Wittgenstein in effect showed that this assumption is not so
much wrong as primitive. He was particularly sensitive to its failure to fit the spe-
cial status of “the logical” and “the ethical.” As Kant pointed out, ethics is con-
cerned not with what does happen, but with what ought to happen, even if the
latter has never happened and will never happen. Wittgenstein made essentially
the same point by saying that the ethical is not in the world. It certainly is not the
sort of thing that can be observed and pictured in ordinary physical or mental
pictures. But, going beyond anything that Kant or anyone else had held, Wittgen-
stein concluded that it cannot be pictured even in logical pictures, that it cannot
be said: “Ethics cannot be put into words.” If this conclusion was mystical, its
mysticism was grounded in logic.
In Tractatus 5.62 Wittgenstein wrote, “what the solipsist means is quite cor-
rect; only it cannot be said, but makes itself manifest.” The sentence was preced-
ed by “The world and life are one” (5.621) and that “I am my world” (5.63), and
was followed by “There is no such thing as the subject that thinks or entertains
Ideas.” (5.631). In Part Three I shall have much to say about these seemingly
alarming statements. Suffice it here to observe that Hume and Sartre also denied
that there is what Wittgenstein called “the subject that thinks or entertains
Ideas.” If they are right, then the world and life may indeed be said to be one.
And if by “value” is meant the sense or meaning of life, value would not be in
the world (6.41), because the sense of the world would not be the sense of any
“subject’s” life. Of course, Wittgenstein did not deny that there were human be-
ings, whose lives may be said to have a sense. The Tractatus was not about them.
It was about the world and its logic. Clearly, few if any of the usual specific or
informative statements can be made about ethical issues if the world is so under-
stood, nor can useful examples can be given. No one should be surprised by the
obscurity of Wittgenstein’s ethics. It could not have been clear.
Wittgenstein reached the strikingly Spinozistic conclusion that to ask about
value and thus the sense of the world requires “view[ing] the world sub specie
aeterni,” as well as “feeling” it as a limited whole,” which, he added, is some-

Authenticated
thing “mystical” (6.45). Ethics does ask what makes life good, worth living, but
“the good life is the world seen sub specie aeternitatis”⁹⁷ This is why Wittgenstein
also wrote: “If the good or bad exercise of the will does alter the world, it can
alter only the limits of the world, not the facts – not what can be expressed
by means of language. In short the effect must be that it becomes an altogether
different world. It must, so to speak, wax and wane as a whole…” (6.43).
Wittgenstein’s point was not that an action, an “exercise of the will,” does
not alter the world, which of course it does. Nor was it that actions are ethically
irrelevant. His point was that the value, the goodness or rightness, of an action
does not consist in its producing some particular event or events in the world, as
standard consequentialism holds; rather, it consists in the world itself becoming
different – as a whole, at its limits. It is in this sense that “How things are in the
world is a matter of complete indifference for what is higher…” (6.432). Realiza-
tion of value, whether goodness or rightness, consists, not in the occurrence in
the world of some particular event, but in the world itself becoming different,
at its limits, in its waxing and waning as a whole (6.43). However, all this can
only be shown. There cannot be ethical propositions. The reason is not that,
as Wittgenstein’s early followers thought, there is nothing for such propositions
to be about. The reason is that what they purport to say cannot be said, though it
can be shown (6.42).
“The sense [Sinn] of the world must lie outside the world… For all that happens
and is the case is accidental” (6.41).” The sense of the world is not something in the
world, because it is the sense of the whole world. This is why it may be said to “lie
outside the world,” at its “limits.” Wittgenstein’s assertion that value is not anything
that happens because it is not accidental need not mean more than what in the
Foundations of the Metaphysics of Morals Kant called the third fundamental propo-
sition of morality: duty is the necessity of acting from respect for the law. Duty is
what ought to happen, “must” happen. What happens is already in the world,
what ought to happen is not yet, or perhaps ever.
In the Notebooks, Wittgenstein had written: “To believe in a God means to un-
derstand the question about the meaning of life… to see that life has a meaning.”⁹⁸
The sense or meaning of life is the ultimate value. It constitutes “[t]he solution of the
riddle of life,” but that solution “lies outside space and time” (6.4312). It is not an
item in the world because it is the sense of the whole world. What Wittgenstein
called the riddle of life presumably concerns the sense or meaning of life. This, in-
deed, was a central topic in traditional ethics. To ask about ultimate value is to ask
 Notebooks, 83e. Wittgenstein adds: “the work of art is the object seen sub specie aeternitatis.”
 Notebooks, 74.

Authenticated
about the meaning of life, what makes life worth living. (Plato’s Republic began with
Cephalus’s claiming that what makes life worth living is not pleasure, as commonly
thought, but justice.) One who asks about the meaning of one’s life sometimes
phrases the question as asking about “the sense of it all.” Life can hardly be
fully meaningful in a meaningless world. Indeed, that the world exists at all, that
there is something rather than nothing, has been for some the ultimate object of
joy or sorrow, and certainly of wonder.
Much later, in 1929, Wittgenstein tried to explain: “What is good is also divine.
Queer as it sounds, that sums up my ethics. Only something supernatural can ex-
press the Supernatural.”⁹⁹ This “supernatural” was as central to Wittgenstein’s eth-
ics as the “supersensible” was to Kant’s. Elsewhere, also in 1929, he wrote, “[Attri-
butions of] absolute value are nonsensical but their nonsensicality [is] their very
essence… [A]ll I wanted to do with [those attributions] was to go beyond the
world and that is to say beyond significant language.”¹⁰⁰ Wittgenstein meant, how-
ever, not that such attributions would be gibberish, but that they would not be log-
ical pictures, in the sense required by his theory of meaning.
Wittgenstein’s remarks about ethics left most of his readers bewildered, and his
claim that there is something “mystical,” which cannot be expressed, was unpalat-
able to his heirs in analytic philosophy. But that there is what he called “the mys-
tical” is also a consequence of his fundamental views about logic, and these cannot
be easily rejected. The existence of the (“actual”) world, indeed of something, may
not be mystical, but it is a basic presupposition of logic. And how to express exis-
tence (with a predicate or just the existential quantifier?) has remained deeply con-
troversial since Anselm proposed his ontological argument for the existence of God.
Noteworthy signposts in the ongoing discussion have been Kant’s denial that exis-
tence is a “real predicate,” Meinong’s theory of objects, Russell’s vehement attack
on it, and present-day quandaries about what it is for one world to be “actual”
and the rest just “possible.” Children find the difference between what exists and
what does not, say, between their real and imaginary playmates, simple and easy
to comprehend, but philosophers do not.
 Ludwig Wittgenstein, Culture and Value, edited by G. H. von Wright and translated by Peter
Winch (Chicago: University of Chicago Press, 1980), 3e.
 Ludwig Wittgenstein, Philosophical Occasions 1912 – 1951 (Indianapolis, IN: Hackett, 1993),
40, 44.

Authenticated
Authenticated
Part Two: Metaphysics Humanized
It is solely from the human standpoint that we can speak of space, of extended things
Kant

Authenticated
Authenticated
I have argued that insofar as they are acceptable, epistemology is about certain
nonformal inferences and ethics is about the world. The case of metaphysics,
however, is different and much more complex.
In the Introduction I suggested that while anthropocentrism in epistemology
and ethics is natural and understandable but indefensible, in metaphysics it is
unnatural and almost incomprehensible but at least as defensible as Kant’s tran-
scendental idealism and contemporary antirealism. In metaphysics it is due not
to humans’ natural interest in themselves, as it is in epistemology and ethics, but
to purely philosophical considerations. The present Part is devoted to making
these considerations clear. In effect, it explains and defends antirealism, the hu-
manization of metaphysics. For there can be no justification for just rejecting it
and returning to pre-Kantian metaphysics. Yet, antirealism seems to lead, how-
ever tacitly, to the absurdity of human creationism. The explanation of how anti-
realism may avoid anthropocentrism, and thus how antirealist metaphysics may
be dehumanized, will be our task in Part Three.
The subject matter of metaphysics is traditionally described as the world or
reality. According to the Oxford English Dictionary, “world” usually signifies “The
material universe; the cosmos,” or just a planet, perhaps Earth, but it applies
also to “Any state or realm of existence.” In the latter sense it is a synonym of
“reality,” in the OED sense of “the aggregate of real things or existences.” Reality
may include, in addition to the spatiotemporal world, abstract, nonspatiotempo-
ral entities such as Platonic universals and numbers, as well as God, who by
being its creator is not in the spatiotemporal world. In discussions of our
topic, however, the terms “world” and “reality” are usually interchangeable.
The term “realism” applies to the metaphysical view that the world is inde-
pendent of our faculties of cognition, in the contemporary sense in which cogni-
tion includes perceiving and understanding (conceptualizing) as well as describ-
ing (by writing or speaking). I have used the term “antirealism” for the opposite
view. Its versions are as diverse as Berkeley’s “immaterialism,” Kant’s “transcen-
dental idealism,” Hegel’s “absolute idealism,” and recent positions such as Hi-
lary Putnam’s “internal realism,” Michael Dummett’s “antirealism,” and “Nelson
Goodman’s “irrealism.” Metaphysical realism and antirealism are usually under-
stood as concerned with all reality, and therefore should be distinguished from
realism and antirealism regarding just an alleged part of it, e. g., moral values
(moral realism/antirealism), abstract entities (Platonic realism/antirealism),
the theoretical entities of physics (scientific realism/antirealism), and so on.
My concern in this and the next Part is with metaphysical realism and anti-
realism. But, as we saw in the Introduction, even within them, important distinc-
tions are needed. One can be a realist regarding “things” but an antirealist re-
garding “facts” and thus regarding the “world,” if the world is the totality of

Authenticated
84 Part Two: Metaphysics Humanized
facts, not of things. One can be a realist regarding the spatiotemporal/physical

and causal structure of the world but antirealist regarding its logical structure:
One may believe that there are physical objects but deny that there are logical
objects. Moreover, asserting and denying the reality of an item are not the
only options. Bertrand Russell held that some things do not exist but subsist,
and Wittgenstein, as we saw in the previous chapter, that some things cannot
be said but show themselves.

Authenticated
Chapter Five: The Role of Language in Cognition
1 The empirical and the a priori question
The most readily understood and least controversial version of metaphysical an-
tirealism today is linguistic antirealism. It has dominated contemporary philos-
ophy. Its immediate appeal rests on the plausible belief that cognition, or at least
thought, is impossible without language. The present chapter will be devoted to
linguistic antirealism insofar as it rests on this belief. The next two chapters will
consider the more technical considerations that may lead to it.
Unlike Kant’s transcendental idealism, which focused on the role of our fac-
ulties of sensibility and understanding in shaping cognition and thus the world
as cognized, linguistic antirealism focuses on the role of our language faculty, or
“organ” as Steven Pinker has called it. Contemporary antirealists thus avoid
Kant’s questionable appeal to inhabitants of consciousness such as concepts,
let alone sensations and ideas. For, as David Armstrong has remarked, “Con-
cepts [understood as mental items] are a more mysterious sort of entity than lin-
guistic expressions.”¹⁰¹ The language faculty may in some sense be “mental,”
but its role in cognition can be understood and defended, as indeed it was by
Wittgenstein, Ryle, Austin, and Strawson, without appeal to any manifestations
of its work in consciousness. Linguistic antirealism thus seems to avoid the ab-
surd implication that the world is an inhabitant of one or several humans’ con-
sciousness. Moreover, even if the human mental faculties are in some sense
properties of the human brain, the role of the language faculty in cognition
can be understood and defended without appealing to neuroscience. Linguistic
antirealism thus seems to avoid also the absurd implication that the world is in
one or several humans’ brains.
Of course, it would be no less absurd to hold that the world is in language,
that it is linguistic. But in Part Three we shall find that this seeming implication
of linguistic antirealism can be avoided. Wittgenstein did write: “The limits of my
world are the limits of my language,” but he also wrote, “I am my world,” which
makes no mention of language. And Heidegger insisted that it is language, not
we, that “speaks.” These are obscure pronouncements, to which I shall return.
They suggest that linguistic antirealism can be understood in such a way that
it involves no reference to language as a human, zoological, phenomenon. But
there would have been little motivation for accepting it if human cognition
were not believed to be essentially or at least importantly linguistic, even though
 David Armstrong, Universals and Scientific Realism, 25.

Authenticated
86 Chapter Five: The Role of Language in Cognition
this is obviously an empirical topic and neither Wittgenstein nor Heidegger

thought of himself as engaged in empirical research. I shall therefore begin
the present Part by considering the role of language in cognition.
The linguistic turn was the central event in both Anglo-American and conti-
nental 20th century philosophy. According to Michael Dummett, after Frege phi-
losophy of language replaced epistemology as the central branch of philosophy,
just as epistemology had replaced metaphysics after Descartes. The phrase “lin-
guistic turn” was coined by Gustav Bergmann and popularized by Richard Rorty
in order to characterize recent Anglo-American philosophy, but as we saw it ap-
plies also to much of contemporary continental philosophy. The turn began with
the publication of Wittgenstein’s Tractatus Logico-Philosophicus after World War
I and the birth shortly thereafter of logical positivism in Austria, but it hastened
with the migration in the following two decades of Wittgenstein and most of the
positivists to Great Britain and especially the United States. In addition to claim-
ing, as the 19th century positivists had done, that all knowledge is empirical and
in principle scientific, the logical positivists held that philosophy, which obvious-
ly is not an empirical science, nonetheless does have a subject matter: language.
To distinguish philosophy from the empirical disciplines of linguistics and lexi-
cography, however, they identified that subject matter more specifically as what
they called the logic of language. For logic, of course, has always been a branch
of philosophy, and a paradigm of an a priori discipline.
But the linguistic turn probably would not have been taken were it not for
the assumption, often tacit, that cognition, or at least thought, without language
is impossible. The philosophers who took the turn considered this assumption
true as a matter of empirical fact, as indeed it seems to be, except, as we shall
find in section 4 of this chapter, in the crucial case of what I have called logical
cognition. Yet few attended to the relevant empirical facts. For example, a highly
relevant fact, to which I shall return, is the apparent occurrence of cognition of
numbers, or at least of “numerosity,” in prelinguistic children and even some
nonhuman animals. No armchair philosophy can discuss it responsibly, yet ob-
viously it is an empirical datum crucial for the general question whether cogni-
tion without language is possible. And the ultimate value of the linguistic turn,
indeed much of the subsequent course of Anglo-American philosophy, did seem
to depend on how that question should be answered.
I have said repeatedly that philosophy cannot competently investigate em-
pirical facts, and this applies also to the facts about the dependence of cognition
on language. But it does not mean that philosophy should be unaware of them.
After all, they were the larger context of the linguistic turn and thus of the vari-
eties of antirealism most common today. Some of these empirical facts are so ob-
vious as to require no special competence or investigation, e. g., that mathemat-

Authenticated
1 The empirical and the a priori question 87
ics requires symbols. But other relevant facts are not obvious, including the al-
leged fact that speakers of different human languages perceive and think of the
world differently. In this chapter we shall look at what has been said about these
issues, first by philosophers and then by linguists and psychologists. I shall not
attempt to provide a comprehensive account. For that a separate book, indeed
several books, would be needed. And we must keep in mind throughout that
the linguistic turn and linguistic antirealism are logically independent of each
other, even though the former encouraged the adoption of the latter. Linguistic
antirealism is a metaphysical position. The linguistic turn was essentially the
adoption of a philosophical method.
Unfortunately, the discussions by philosophers of the question whether cogni-
tion without language is possible are seldom detailed, and the discussions of it by
scientists are seldom useful. Philosophers have relied usually on speculations or, at
best, exercises in introspection. Scientists have been hampered by the virtual impos-
sibility, moral and social, of serious experimentation with children before and after
they acquire language. That the most advanced, the “hard,” sciences – physics,
chemistry, biology – are experimental, not just empirical, while the “soft” sciences –
economics, sociology, political science, and much of psychology – are not, is not an
accident. Experiments vastly augment not only the number but, more importantly,
the variety of relevant empirical data and thus make detection of significant, rather
than merely coincidental, correlations much easier.
That there is a connection between cognition and language was hinted by
the historically influential use of logos in Greek for both reason and word. The
definition that man is a rational animal could be understood also as saying
that man is a speaking animal. This is why the question whether cognition with-
out language is possible has often been taken to ask whether thought without
language is possible. The first definition of “thought” in the Oxford English Dic-
tionary reads as follows: “The action or process of thinking; mental action or ac-
tivity in general, esp. that of the intellect; exercise of the mental faculty; forma-
tion and arrangement of ideas in the mind.” We are also told that “think” is “The
most general verb to express internal mental activity, excluding mere perception
of external things or passive reception of ideas.”
As the OED implies, “thought” and “think” have both a wide sense, that of
mental action or activity in general, and a narrow sense, that of mental action or
activity of the intellect. They have these two senses also in philosophy, though
philosophers, especially Brentano and Husserl, usually add that the action or ac-
tivity is “intentional,” directed upon an object, real or not, and describe being
conscious of something as an “act.” Descartes used the words in their wide
sense. After he offered the argument “I think [cogito, pense], therefore I am,”
he explained, “What is a thinking thing [res cogitans, une chose qui pense]? It

Authenticated
is a thing that doubts, understands, conceives, affirms, denies, wills, refuses;

that imagines also, and perceives.” Clearly, by “thought” Descartes meant
what later philosophers, in particular, Brentano, Husserl, and Sartre, meant by
“consciousness.” Philosophical theories of thought in this sense are essentially
philosophies of mind. For example, soon after Descartes offered his argument,
the British empiricists proposed theories of mental activity as a “succession”
or “association” of discrete “ideas,” whether abstract ideas (roughly the “con-
cepts” of later philosophy), ideas of “sensation” (later called sense-data), or
ideas of the imagination (later called mental images).
Descartes drew a sharp distinction between a “thinking thing” and the
human body. A thinking thing is “simple,” it has no parts. The body, in contrast,
is a paradigm of an entity that has parts, “organs” – an “anatomy,” in the mod-
ern literal as well as the etymologically pregnant sense. Indeed, one of the stan-
dard arguments for the immortality of the soul was that it is simple and therefore
incapable of “corruption,” “disintegration,” “falling apart,” which was the com-
mon conception of how a thing ceases to exist.
Later philosophers more commonly used “thought” in its narrow sense, that
of mental action or activity of the intellect. In that sense, thought is the level of
cognition expressed in Greek by noēsis and in English by “understanding,” “con-
ceptualization,” or “judgment.” Unlike a neonate’s sense perception, which
seems to be much like that of nonhuman animals, thought understood as activity
of the intellect involves the application of concepts and seems distinctively
human. Indeed, in the 17th century John Locke declared that “brutes think
not,” on the grounds that they are incapable of entertaining “abstract ideas”
and using language.
In the 18th century, Kant rejected the atomism of the empiricists’ view of
mental activity as a succession of ideas by pointing out the synthesizing function
of the “understanding” (Verstand). In the 19th century, Franz Brentano distin-
guished among presentation, judgment, and phenomena of interest. Judgment
necessarily involves presentation, because one must be aware of what the judg-
ment is about (whether by sensing it or through a concept). And judgment makes
possible the characteristically mental acts of acceptance or rejection, belief or
disbelief.¹⁰² In the 20th century, H. H. Price followed Kant by regarding thinking
as the employment of concepts, even in recognition, and argued that a concept is
essentially a principle of classification.
 Franz Brentano, Psychology from an Empirical Standpoint (London: Routledge, 1973), trans.
by A.C. Rancurello, D.B. Terrell, and L. McAlister. 2nd ed., intr. by Peter Simons, 1995.

Authenticated
1 The empirical and the a priori question 89
Unlike primitive sense perception and imagination, a judgment has as its object
a proposition, what is “judged” to be true or false. And propositions usually are un-
derstood in philosophy as the primary vehicles of truth-value. Judgments, sentences,
beliefs, opinions, etc., are said to be true or false in virtue of the propositions they
“express.” It may be said, however, that while a declarative sentence expresses a
proposition, the assertion of that sentence expresses a judgment. Frege used
“thought” (Gedanke) in a sense close to that of “proposition.” Thoughts, he held,
belong in a “third realm,” neither mental nor physical, and thus they are not judg-
ments, which presumably are mental.¹⁰³
If a judgment is expressed by the assertion of a declarative sentence, it can
be expected to have parts and a structure analogous to those of a sentence. In-
deed, Jerry Fodor has argued that there is a “language of thought,” thus attrib-
uting to thought the sort of structure characteristic of language, though he re-
gards thought to be a state or activity of the brain.¹⁰⁴ And Gustav Bergmann
analyzed all mental acts as “complexes,” facts or states of affairs, each consist-
ing of a momentary particular and two attributes that the particular exemplifies:
a “species” that determines the kind of mental act it is (e. g., perceiving, remem-
bering, imagining) and a “proposition” (sometimes called “text” by Bergmann)
that determines which state of affairs is its object (“intention”). He held that
even the objects of ordinary sense perception and imagination are states of af-
fairs, and thus require sentences for their expression in the “ideal language.”¹⁰⁵
In this book the question whether there is thought without language would
have a clearer and more promising focus if we ask, instead, whether there is cog-
nition without language. For it is the term “cognition,” not “mind” or even “judg-
ment,” that captures literally what is relevant to the epistemological and meta-
physical issues that give rise to antirealism. Yet, even then, the question would
remain too broad, partly because the affirmative answer seems obvious. Neo-
nates do enjoy sense perception, surely a level of cognition, however rudimen-
tary, months or years before they learn to speak; they also communicate, though
not by speaking but often by crying. Later, though still before being able to
speak, they enjoy recognition, e. g., of Mother, a primitive level of conceptual
cognition, and perhaps they even engage in rudimentary reasoning, e. g., that
crying leads to being fed. We shall find in section 4 that only logical cognition
can be considered clearly impossible without language, because only there the
impossibility is logical.
 “Thought,” in Beaney, The Frege Reader.

 Jerry Fodor, The Language of Thought (Cambridge: Harvard, 1975).
 Gustav Bergmann, Meaning and Existence (Madison, University of Wisconsin Press, 1959).
See also New Foundations of Ontology.

Authenticated
Since language is an empirical subject matter, the study of its role in cogni-
tion should also be empirical. In keeping with the stance explained in the Intro-
duction regarding philosophical inquiries into empirical matters, the present
chapter will offer no theories about language. It will be limited to a brief review
of what others have said. Careless observation and incautious generalization
from one’s own experience can be grossly misleading.
A useful discussion, scientific or philosophical, of the question whether cog-
nition without language is possible must be specific regarding the respects in
which it might or might not be possible. It must also be specific regarding the
kinds of cognition, language, and possibility that are at issue. Elementary and
obviously relevant distinctions must be made. Many of the most familiar views
tend to be intolerably general and vague. Surely, there are fundamental differen-
ces between sense perception and calculation, between the language of babes
and the language of bards, and between causal impossibility and logical impos-
sibility. Judgments about the causal possibility or impossibility of cognition with-
out language are especially hazardous. One may be unable to balance one’s
checkbook without talking to oneself, but other people can do it. When driving
in a strange city, one may be compelled to issue to oneself instructions like “Turn
left on 3rd street, then right at second stop-light,” but other people might not. The
professional activities of some people, especially philosophers, consist almost
entirely of talking, reading, and writing. (Sartre titled his autobiography
Words.) But the activities of farming, fishing, and acrobatics do not.
Language might not be causally necessary even for some relatively advanced
levels of cognition, such as designing a house and planning a trip. But surely it is
necessary for doing physics or mathematics. A fly’s perceptual cognition does
not require language, but any conceptual cognition that whales perhaps engage
in might require it. In the case of extraterrestrial life forms, angels, or God, lan-
guage might not be necessary for any cognition. To be sure, a human being who
lacks a language cannot do physics or mathematics, but surely God can, and for
all we know so can intelligent beings outside our solar system. We may not un-
derstand what such cognition might be like, but neither can we form auditory
images of the high-frequency sounds that dogs but not humans can hear. Hu-
mans certainly need language, symbols, for cognition in physics and mathemat-
ics, but this may be due to their limited cognitive powers. A god might cognize
directly, for example, the ultimate constitution of matter, without relying on lan-
guage or any symbols, including those of mathematics.
The truth is that, so far, there is neither sufficient knowledge nor conceptual
maturity in psychology or biology to judge responsibly whether thought or cog-
nition is causally possible without language. But judgments about whether some

Authenticated
2 Philosophical opinions 91
forms of it are logically possible can be made in philosophy. This question is not
empirical, it is properly philosophical. We will come to it in section 4.
2 Philosophical opinions
Let us briefly review what philosophers have said about the role of language in
cognition. In the Cratylus Plato sharply distinguished between names and the
things they name, considered whether names are arbitrary and conventional,
and argued that the study of names is inferior to the study of the things
named. But, at the beginning of modern philosophy, Bacon wrote, “Men believe
that their reason governs words; but it is also true that words react on the under-
standing; and this it is that has rendered philosophy and the sciences sophistical
and inactive.”¹⁰⁶ Hobbes dwelt on language in detail, arguing that “‘true’ and
‘false’ are attributes of speech, not of things.”¹⁰⁷ Locke held that brutes abstract
not, because they talk not: “the power of abstracting… [is] an excellency which
the faculties of brutes do by no means attain to.”¹⁰⁸ Leibniz wrote, “Nothing ex-
ists in the intellect that was not before in the tongue – except intelligence itself.”
Hegel remarked that “[t]he forms of thought are, in the first instance, displayed
and stored in human language.”¹⁰⁹ This last opinion was accepted by most 20th
century philosophers. As we have seen, Wittgenstein declared that “the limits of
the language … mean the limits of my world,” and Heidegger announced that
“Language is the house of being.”
But in Thinking and Experience, published in 1953, H. H. Price wrote, “it is
sometimes supposed that no intellectual activity of any kind can occur without
the use of words. This is not true of recognition… Recognition is a prelinguistic
process in the sense that it is not dependent on the use of words… [W]ords them-
selves have to be recognized.”¹¹⁰ Price’s view was plain common sense. Recogni-
tion is a level, indeed a fundamental level, of all cognition, primitive or ad-
vanced. It is present in children long before they acquire language. In adults it
often occurs when no name or even description of what is recognized is availa-
 Works, Novum Organum, Aphorisms Concerning the Interpretation of Nature and the King-
dom of Man, LIX.
 Leviathan, Part I, Chapter IV.
 John Locke, An Essay Concerning Human Understanding Vol. 1 (London: Dent, 1960), 126.
But in his Treatise of Human Nature, Hume disagreed: “no truth appears to me more evident,
than that beasts are endowd with thought and reason as well as men.”
 Science of Logic, trans. A. V. Miller (London: George Allen, 1969), 31.
 Thinking and Experience, 36 – 38.

Authenticated
ble, for example when recognizing, “reading,” facial expressions.¹¹¹ And creative
work in music or painting, as well as in mathematics and advanced theoretical
physics, only minimally involves the use of a natural language, though in math-
ematics and physics it does require mathematical symbols. Price felt obliged to
say what I quoted because of the dominance at the time of the philosophy of or-
dinary language, according to extreme versions of which, he wrote, “an intelli-
gent being…must always be talking to himself or to others.”
Indeed, in the Philosophical Investigations, which by coincidence was pub-
lished (posthumously) also in 1953, Wittgenstein wrote, “When I think in lan-
guage, there aren’t ‘meanings’ going through my mind in addition to the verbal
expressions: the language is itself the vehicle of thought…..”¹¹² Expounding what
he took to be Wittgenstein’s view, Renford Bambrough argued in 1960 that ob-
jects called by the same name are distinguishable from other objects just by
the fact that they are uniquely called by that name: they neither have a common
property nor bear to each other resemblances except for what Wittgenstein had
called family resemblances.¹¹³ In 1973 Michael Dummett wrote that “to possess a
concept is to be a master of a certain fragment of language,”¹¹⁴ and in 1975 that
“Only with Frege was the proper object of philosophy finally established: name-
ly, first, that the goal of philosophy is the analysis of the structure of thought;
secondly, that the study of thought is to be sharply distinguished from the psy-
chological process of thinking; and, finally, that the only proper method for an-
alyzing thought consists in the analysis of language.”¹¹⁵ In 1982 we find Donald
Davidson declaring that “a creature cannot have a thought unless it has lan-
guage.”¹¹⁶ In 1995 Quine wrote that “thought, as John B. Watson claimed, is pri-
marily incipient speech,” though Quine allowed that in the case of artists, acro-
bats, and engineers, sometimes they are “thinking with nonverbal muscles.”¹¹⁷
However, after thus paying homage to muscular thought, Quine declared that be-
 Noam Chomsky, Rules and Representations, 248. Chomsky appeals to innate structures and
speaks of a possible “universal grammar of faces.”
 Ludwig Wittgenstein, Philosophical Investigations, trans. G.E.M. Anscombe, (Oxford: Black-
well, 1991), 329 – 330.
 J. R. Bambrough, “Universals and Family Resemblances,” Proceedings of the Aristotelian So-
ciety, LXI (1960 – 61): 207– 223.
 Michael Dummett, Truth and Other Enigmas (Cambridge: Harvard University Press, 1978),
438.
 Truth and Other Enigmas 458.
 Donald Davidson, “Rational Animals,” in Subjective, Intersubjective, Objective (Oxford: Ox-
ford University Press, 2001), 100.
 W.V. Quine, From Stimulus to Science (Cambridge, Mass.: Harvard University Press, 1995),
88 – 89.

Authenticated
2 Philosophical opinions 93
liefs, meanings, ideas, properties, and propositions are all entia non grata, add-
ing that “believing, doubting, hoping, expecting, regretting, all continue alive
and well, and their objects, by my lights, are sentences.”¹¹⁸ Even Nelson Good-
man, who for years had held that what he called “symbol systems” need not
be linguistic, slipped in 1988 and denied that “there is a readymade world be-
yond discourse.”¹¹⁹
Nevertheless, analytic (though perhaps not continental) philosophy soon
went beyond these extremist views. In 1992 we find Dummett himself asserting
that “linguistic practice is no more sacrosanct, no more certain to achieve the
ends at which it is aimed, no more immune to criticism or proposals for revision,
than our social, economic or political practice.”¹²⁰ Though intended as an attack
on the philosophy of ordinary language (especially J. L. Austin’s), Dummett’s as-
sertion had independent philosophical significance by tacitly implying that criti-
cism of language presupposes cognition that is not dependent on language.
It is noteworthy that, except perhaps for Quine’s casual reference to Watson,
these opinions about language were offered without mention of any empirical
data or of the relevant empirical sciences.
Some have come to believe that, though all cognition, from simple percep-
tion to mathematical reasoning, is necessarily representational, symbolic, the
representations need not belong to a natural language – they can belong to a
“language of thought,” in Jerry Fodor’s phrase.¹²¹ For Fodor this was just a phil-
osophical hypothesis, though he thought it has the support of neuroscience. Tim
van Gelder has explained: “Contemporary orthodoxy maintains that it [cogni-
tion] is computation: the mind is a special kind of computer, and cognitive proc-
esses are the rule-governed manipulation of internal symbolic representa-
tions.”¹²² But then he drew the natural conclusion “that because the cognitive
system traffics only in symbolic representations, the human body and the phys-
ical environment can be dropped out of consideration; it is possible to study the
cognitive system as an autonomous, bodiless, and wordless system whose func-
tion is to transform input representations into output representations.”¹²³ If cog-
 From Stimulus to Science, 93.

 Nelson Goodman and Catherine Z. Elgin, Reconceptions in Philosophy (Indianapolis & Lon-
don: Hackett, 1988), 154. But elsewhere, including this same work (see vii, 9, 11, 155, and section
VIII, 3), Goodman points out that what he calls “symbol systems” need not be linguistic, that
they could be, e. g., notational or pictorial.
 Michael Dummett, The Logical Basis of Metaphysics (London: Duckworth, 1991), 214– 215.
 The Language of Thought (Thomas Y. Crowell Co., 1975).
 Tim van Gelder, “What Might Cognition Be, If Not Computation?” The Journal of Philosophy,
Volume XII, number 7 (July 1995), 345.
 van Gelder, 373.

Authenticated
nition is what such a cognitive system does, cognition does not involve language,
even if some neural states are described as “symbolic,” though neither is it also
directed upon objects and thus it is not cognition of reality – or of anything. At
most these neural states would be like the characters employed by a computer,
but as John Searle had shown in 1980, the idea that they mean anything other
than what the programmer means by them is grossly mistaken.
3 Scientific opinions
Let us now briefly review what linguists and psychologists have said about the
role of language in cognition. More than a century after Locke announced that
brutes abstract not, the German linguist Wilhelm von Humboldt wrote, “Lan-
guage is the formative organ of thought…. Thought and language are therefore
one and inseparable from each other.”¹²⁴ Humboldt offered no evidence in sup-
port of these opinions. He did speculate, however, that existing languages differ
from each other markedly in regard to “perfection,” which he attributed to na-
tional, ethnic, and racial differences among their speakers.
Early in the 20th century, Ferdinand de Saussure wrote (or was reputed to
have said – the text was reconstructed from lecture notes): “Psychologically, set-
ting aside its expression in words, our thought is simply a vague, shapeless
mass. Philosophers and linguists have always agreed that were it not for
signs, we should be incapable of differentiating any two ideas in a clear and con-
stant way. In itself, thought is like a swirling cloud, where no shape is intrinsi-
cally determinate. No ideas are established in advance, and nothing is distinct,
before the introduction of linguistic structure.”¹²⁵ Saussure did not say how he
(or the philosophers and linguists to whose authority he appealed) learned all
this. For example, did he study human infants before they learned to speak?
William James famously announced that the world perceived by an infant is a
“blooming buzzing confusion,” but he also failed to say how he knew this.¹²⁶ James
did think, however, that thought without language is “perfectly possible.” In sup-
port, he quoted at great length a deaf-mute instructor, “Mr. Ballard,” who was re-
ported to have written: “It was during those delightful rides, two or three years be-
fore my initiation into the rudiments of written language, that I began to ask myself
 Wilhelm von Humboldt, On Language, (Cambridge: Cambridge University Press, 1988), 54–
55.
 Ferdinand de Saussure, Course in General Linguistics (Chicago and LaSalle, Ill.: Open Court,
1986), 111 [155].
 William James, The Principles of Psychology (New York: Henry Holt, 1890), 46.

Authenticated
3 Scientific opinions 95
the question: How came the world into being? When this question occurred to my
mind, I set myself to thinking it over a long time.” Concerning this passage, Wittgen-
stein later wrote, “Are you sure – one would like to ask – that this is the correct
translation of your wordless thought into words? Do I want to say that the writer’s
memory deceives him? I don’t even know if I should say that. These recollections are
a queer memory phenomenon, – and I do not know what conclusions one can draw
from them about the past of the man who recounts them.”¹²⁷
Despite Wittgenstein’s misgivings, such reports of personal experience, if
truthful, are useful. They must be read with caution, but the evidence they pro-
vide is unavailable elsewhere. For example, if Helen Keller’s acquisition of lan-
guage was indeed as she and her teacher Anne Mansfield Sullivan claimed, then
it was an extraordinary example of how, despite the absence of the usual sort of
exposure to language, one can acquire a rich, indeed stylistically first-rate, lin-
guistic proficiency. Yet Keller’s editor did write: “Philosophers have tried to
find out what was her conception of abstract ideas before she learned a lan-
guage. If she had any conception, there is no way of discovering it now; for
she cannot remember, and obviously there was no record at the time. She had
no conception of God before she had the word ‘God.’”¹²⁸
But another report of personal experience, by the mathematician Kalvis M.
Jansons, is instructive and may serve as an example of what is needed: “From
an early age I found that many things were easier to think about without lan-
guage. This usually, but not always, meant thinking in terms of pictures and
was particularly true when trying to make or understand intricate mechanisms.
… To me, abstract pictures and diagrams feel more important than words.”¹²⁹
Not all reports of personal experience on our topic are as credible as Jan-
sons’. This is especially true of some familiar claims by leading psychologists,
which must be understood as reports of personal experience since no other evi-
dence is cited. For example, J. B. Watson wrote that “thinking is merely talking,
but talking with concealed musculature.”¹³⁰ Leonard Bloomfeld held that think-
ing is talking to ourselves, “suppressing the sound-producing movements and
replacing them by very slight inaudible ones.”¹³¹ (I should mention that today
 Philosophical Investigations, 34.

 Helen Keller, My Story, ed. John Albert Macy, 234. Mark Twain is reported to have said that
the two most interesting characters of the nineteenth century were Napoleon and Helen Keller
(The Principles of Psychology, 225).
 Kalvis M. Jansons, “A personal view of dyslexis and of thought without language,” in L. in
Thought without Language, ed. L. Weiskrantz (Oxford: Clarendon Press, 1988), 502– 503.
 J. B. Watson and W. McDougal, The Battle of Behavior (New York: Norton, 1929), 33.
 Leonard Bloomfield, Language (Chicago: University of Chicago Press, 1933), 28.

Authenticated
there is evidence that thinking is not prevented by drugs that suppress muscular
activity.) B. F. Skinner wrote, “thought is simply behavior – verbal or nonverbal,
covert or overt.”¹³² Edward Sapir appealed to “the frequent experience of fatigue
in the speech organs” after “intensive thinking.”¹³³ This, presumably, was a re-
port of personal experience. But elsewhere Sapir went much farther: “Human be-
ings do not live in the objective world alone…..but are very much at the mercy of
the particular language which has become the medium of expression for their
society…..We see and hear and otherwise experience very largely as we do be-
cause the language habits of our community predispose certain choices of inter-
pretation. “¹³⁴ This was a crude version of linguistic antirealism, but offered as a
scientific hypothesis and then left unsupported by empirical evidence. In 2008,
field linguists discovered a “new” language, Koro, spoken in some villages in
northeastern India.¹³⁵ One of the linguists was reported as saying that languages
like Koro “construe reality in very different ways…They uniquely code knowledge
of the natural world in ways that cannot be translated into a major language.”
But we are not told how the linguist knew this.
Sapir’s student, Benjamin Lee Whorf, did engage in extensive empirical
work on Indian languages. But he cited no particular evidence when he wrote,
“Thinking…follows a network of trails laid down in the given language…The in-
dividual is utterly unaware of this organization and is constrained completely
within its unbreakable bonds.”¹³⁶ Whorf thought that he and a Hopi could not
discuss the “same” world. According to him, “[S]egmentation of nature is an as-
pect of grammar…We cut up and organize the spread and flow of events as we
do, largely because, through our mother tongue, we are parties to an agreement
to do so, not because nature itself is segmented in exactly that way for all to
see.”¹³⁷ But, again, no evidence is provided.¹³⁸ Perhaps it was natural that
 B.F. Skinner, Verbal behavior (New York: Appleton-Century-Crofts, 1957), 449.
 Edward Sapir, Language (New York: Harcourt, Brace & World, 1921), 19.
 Edward Sapir, Language, Culture and Personality (ed. David G. Mandelbaum, Berkeley &
Los Angeles: University of California Press, 1963), 16. (Reprinted from “The Status of Linguistics
as a Science,” Language 5, 209.) See also A. H. Bloom, The Linguistic Shaping of Thought: A Study
in the Impact of language on Thinking in China and the West; Laurence Erbaum, 1981; Richard B.
Brandt,
 New York Times, October 11, 2010.
 Benjamin Lee Whorf, Language, Thought and Reality: Selected writings of Benjamin Lee
Whorf, ed. John B. Carroll, (New York: John Wiley and Sons, 1956), 226. Zellig Harris described
Whorf’s view as “occupational imperialism for linguistics,” in ‘Distributed Structure’, in The
Structure of Language: Readings in the Philosophy of Language, eds., Jerry A. Fodor and Jerrold
J. Katz (1965), 38, n. 1.
 Language, Thought and Reality, 240, reprinted from Language 5 (1929): 207– 214).

Authenticated
what came to be known as the Sapir-Whorf hypothesis was offered in American

linguistics. It was in America that non-Indo-European languages were plentiful
and readily available for study: there were more than 100 American Indian lan-
guages. In this respect American linguistics had an important advantage over
European linguistics.
Watson, Bloomfield, Sapir, Whorf, and Skinner were the most influential sci-
entists on our topic in the first half of the 20th century. This may explain in part
the unbridled linguisticism characteristic of American philosophy during that
period. Respect for science was characteristic of the logical positivism dominant
then. The supposedly scientific view offered by the psychologists seemed identi-
cal with the philosophical view that Hilary Putnam described as holding that
“objects arise out of discourse, rather than being prior to discourse.”¹³⁹ But
were Watson, Bloomfield, Sapir, Whorf, and Skinner right? It is generally agreed
that to answer this question on empirical grounds and with confidence we would
need to study the prelinguistic child’s cognition. And such studies have been no-
toriously unsuccessful, probably because serious experimentation with children,
e. g., delaying a child’s learning a language, would be grossly unethical. Less
alarming experiments tend to be inconclusive, perhaps because small children,
not just infants, seldom cooperate with or even take seriously the investigators,
who often happen to be their parents. This seems true, for example, in the case
of the theorizing about early cognition and language acquisition by Bärbel In-
helder and Jean Piaget, which tends to be little more than speculation.¹⁴⁰
The most important development in linguistics, of course, took place in the
second half of the 20th century. It was due to Noam Chomsky. Chomsky is not
only a major linguist, he is also a philosopher, whose views (especially in his
later writings) on scientific methodology and the mind-body problem are sophis-
ticated and subtle. But it is his work in linguistics that is of special relevance to
the topic of this chapter. Its central thesis was anticipated by Humboldt: “Since
the natural disposition to language is universal in man, and everyone must pos-
sess the key to the understanding of all languages, it follows automatically that
 But Whorf did add: “Our Indian languages show that with a suitable grammar we may have
intelligent sentences that cannot be broken into subjects and predicates…When we come to
Nootka, the sentence without subject or predicate is the only type…Nootka has no parts of
speech; the simplest utterance is a sentence, treating of some event or event-complex” (Lan-
guage, Thought and Reality, 242).
 Hilary Putnam, Realism and Reason. Philosophical Papers. Volume 3 (Cambridge: Cam-
bridge University Press, 1983), xvi.
 For example, Bärbel Inhelder and Jean Piaget, The Growth of Logical Thinking (New York:
Basic Books, 1958), trans. Anne Parsons and Stanley Milgram.

Authenticated
the form of all languages must be essentially the same, and always achieve the
universal purpose.”¹⁴¹ This thesis is essentially Kantian in spirit, and it has been
defended throughout Chomsky’s works.
Discussions of Chomsky’s linguistics often bog down in what he rightly re-
gards as misguided puzzling over whether what he calls universal grammar
has “psychological reality,” rather than just utility as a theoretical hypothesis.
Surely he is right that there is no principled difference between the propriety
of hypotheses in linguistics and the propriety of hypotheses in physics.¹⁴² Chom-
sky has always held that the status of innate linguistic competence is ultimately
biological.¹⁴³ The “innate structures” to which he appeals are not accessible to
consciousness, presumably because they are in the brain.¹⁴⁴
Chomsky’s nativism is not what is argued in this book, if for no other reason
than that, unlike what a philosophical book can properly attempt, it is essential-
ly scientific and subject to empirical confirmation. But it does imply that what I
have called the logical structure of language is not learned from experience.
Chomsky has claimed, for example, that “the familiar [quantifier-variable] nota-
tion is ‘read off of’ the logical form that is the mental representation for natural
language.”¹⁴⁵ But this is offered as a substantive scientific hypothesis. And the
mental representation in question is ultimately identified with a state or feature
of the brain. The hypothesis is hardly philosophical, even though, as Chomsky
makes clear, it is proposed on a very high level of abstraction.
Chomsky’s nativism is not as novel as many take it to be. It would not seem
innovative to anyone familiar with Aristotle’s distinction between first actuality
and second actuality; one of Aristotle’s applications of it was to knowledge of
grammar. Nelson Goodman wrote about Chomsky’s view: “until the term ‘innate
 Wilhelm von Humboldt, On Language, 215.

 See, for example, his Rules and Representations, chapters 3 and 5, and “Language and Na-
ture.”
 In Rules and Representations, Chomsky wrote: “What many linguists call ‘universal gram-
mar’ may be regarded as a theory of innate mechanisms, an underlying biological matrix that
provides a framework within which the growth of language proceeds.” (187). For more recent
examples, see his “Language and Nature,” Mind, 104, 1– 60; Powers and Prospects (London:
Pluto, 1996); “Comments: Galen Strawson, Mental Reality”, Philosophy and Phenomenological
Research, vol. LVIII, no. 2 (June 1998). Chomsky often uses the phrase “mind/brain,” but
when questioned about what the slash in it means, he admits that this is a problem and disa-
vows any inclination to accept reductionism or eliminative materialism. See Noam Chomsky,
Language and Thought (Wakefield, Rhode Island & London: Moyer Bell, 1993), 79 – 87; the ques-
tion was asked by James H. Schwartz, Language and Thought, 71– 77.
 Rules and Representations, 244.
 Rules and Representations, 165.

Authenticated
idea’ is applied, what is advocated is the rather trivial truth that the mind has
certain faculties, tendencies, limitations.”¹⁴⁶ But even if this truth is trivial, its
importance was neglected in philosophy and psychology until Kant drew atten-
tion to it.
Indeed, Chomsky’s appeal to innate psychological structures is a sort of bio-
logical “Kantianism,” a biological transcendental idealism.¹⁴⁷ His application of
it is not restricted to universal grammar. He applied it to perception: “we know
that the visual system of a mammal will interpret visual stimulations in terms of
straight lines, angles, motions, and three-dimensional objects” ¹⁴⁸ He applies it
also to conceptualization, in ways Kant would have found surprising: “[H]uman
nature gives us the concept “climb” for free. That is, the concept “climb” is just
part of the way in which we are able to interpret experience available to us before
we even have the experience. That is probably true for most concepts that have
words for them in language. This is the way we learn language. We simply learn
the label that goes with the preexisting concept. So in other words, it is as if the
child, prior to any experience, has a long list of concepts like “climb,” and then
the child is looking at the world to figure out which sound goes with the concept.
We know that the child figures it out with only a very small number of presen-
tations of the sound.”¹⁴⁹
The role of language in cognition was discussed by Chomsky’s student Ste-
ven Pinker in his widely read book The Language Instinct. He wrote, “Sometimes
it is not easy to find any words that properly convey a thought. When we hear or
read, we usually remember the gist, not the exact words, so there has to be such
a thing as a gist that is not the same as a bunch of words. And if thoughts de-
pended on words, how could a new word ever be coined? How could a child
learn a word to begin with? How could translation from one language to another
be possible?¹⁵⁰
This is plain good sense. We may wince, however, when Pinker goes on to claim
that psychologists have shown that babies can do arithmetic: “The developmental
psychologist Karen Wynn has recently shown that five-month-old babies can do a
simple form of mental arithmetic. She used a technique common in infant percep-
tion research. Show a baby a bunch of objects long enough, and the baby gets bored
and looks away; change the scene, and if the baby notices the difference, he or she
will regain interest. The methodology has shown that babies as young as five days
 Nelson Goodman, Problems and Projects (New York: Bobbs-Merrill: 1972), 74.
 Cf. Rules and Representations, 251.
 Language and Problems of Knowledge, 171.
 Language and Problems of Knowledge, 191.
 Steven Pinker, The Language Instinct (Harvard University Press, 2007), 57– 58.

Authenticated
[sic] old are sensitive to number. In one experiment, an experimenter bores a baby
with an object, then occludes the object with an opaque screen. When the screen is
removed, if the same object is present, the babies look for a little while, then get
bored again. But if, through invisible subterfuge, two or three objects have ended
up there, the surprised babies stare longer.”¹⁵¹
Wynn’s experiments had obvious merits, but Pinker’s interpretation of their re-
sults is naïve. The naiveté is not philosophical but scientific. It fits Mark Twain’s de-
scription of science as an endeavor in which “one gets such wholesome returns of
conjecture out of such trifling investment of fact.” It is an exaggerated abductive in-
ference from meager and ambiguous data, not much better than Benjamin Whorf’s
inference from what he knew about Indian languages that thought depends on lan-
guage, an inference Pinker justly criticizes. To claim, on the basis of experimental
results like Wynn’s, that babies have knowledge of arithmetic is like claiming that
since birds know how to fly they have knowledge of aerodynamics. Indeed, another
scientist, Alan M. Leslie, has written of “[Baillargéon’s] important discovery about
the young infant’s understanding of mechanics,” on the basis of an experiment sim-
ilar to Wynn’s.¹⁵² In the same collection of papers we also find Gabriel Horn arguing
that thought does not require language on the grounds that mice “distinguish self
from other objects” and thus are aware of their bodies because they (generally) don’t
get stuck in holes too small for them to get through.¹⁵³ Such opinions in psychology
bring to mind recent reports of discoveries in genetics of a “gambling-gene” and an
“alcoholism-gene.”
But we should not complain inordinately about them. Despite the exagger-
ated conclusions drawn, the underlying facts are important in their own right.
While crediting babies with cognition of arithmetic at best ignores the nature
of arithmetic, there is no doubt that any prelinguistic cognition of what in the
scientific literature is called “numerosity” would be relevant to the question
about the dependence of cognition on language. More recent experiments, sim-
ilar to Wynn’s but involving monkeys, suggest that monkeys also are capable of
cognition of numerosities.
Pinker writes: “Grammar offers a clear refutation of the empiricist doctrine
that there is nothing in the mind that was not first in the senses.”¹⁵⁴ This implies
that syntactical, and therefore presumably also logical, form corresponds to
nothing perceived or otherwise experienced in the world. But Pinker’s claim
 The Language Instinct, 124. Karen Wynn’s piece appeared in Nature 358 (1992), 749 – 750.
 Thought Without Language, 194. Baillargéon’s works are cited at the end of Leslie’s.
 Gabriel Horn, “Thought without Language in Birds,” in L. Weiskrantz, Thought without Lan-
guage (Oxford: Clarendon Press, 1988), 299.
 The Language Instinct, 77.

Authenticated
4 Language in logical cognition 101
that therefore the “seat” of the language instinct is in the brain casts little light
on the status or nature of cognition. To be told that logical cognition is grounded,
not in perceived objects in the world, but in a structure of the brain is no more
illuminating than to be told that it is grounded in evolution.
4 Language in logical cognition
Our brief survey of what philosophers and scientists have said about the general
question of the role of language in cognition suggests that it has received at best
inconclusive treatment. But there is a much more specific question that can be
answered with reasonable confidence. It concerns the role of language in logical
cognition. To answer that question requires no empirical research and remains
properly a task of philosophy, the home of logic.
Wittgenstein held that the logical constants do not “represent,” that there
are no “logical objects.” In the Introduction, I suggested that this is true of all
logical expressions – variables, quantifiers, sentential operators, the verb “to
be” in its senses of identity, existence, and predication, and the symbols or syn-
tactic structures expressing these senses, as well as of declarative sentences
themselves – and that we count as logical objects what all logical expressions
so understood might be taken to represent. I called the kind of cognition that re-
quires logical expressions for its expression “logical cognition.” It includes but
must not be confused with the sort of cognition pursued by logic. The scope
of the latter is much narrower. Statements expressing logical cognition usually
contain also nonlogical expressions, while those in logic contain only logical ex-
pressions. I suggested that we call the former logical statements and the latter
statements of logic. For example, “All men are mortal” is a logical statement be-
cause it includes the logical expression “all,” but it is not a statement of logic.
Statements of logic are purely formal and usually employ only technical sym-
bols. An example would be (($x) Φx ≡ ~ ("x) ~ Φx)), i. e., “Something is Φ if
and only if it is not the case that nothing is Φ.”
I shall discuss logical expressions, logical statements, and logical cognition
in detail later. What is relevant here is that if logical expressions do not “repre-
sent,” i. e., if they stand for nothing, then logical cognition would indeed seem
impossible without language, language would seem to be “all there is to it” since
there would be nothing else that might be pertinent. We would have no “access”
to what logical cognition is about except through language because it is not
about anything. In this respect logical cognition is dramatically different from,
say, perception, where in addition to talking about its objects we can also per-
ceive them. Not even God could know that all men are mortal by “perceiving”

Authenticated
all men. God might perceive all individual things, perhaps an infinity of them,
but it is not by perception that he could know that they are all the individual
things that there are: this is not something that can be perceived because
there is nothing, perceivable or unperceivable, that the word “all” stands for.
(I shall leave it to theology to tell us whether or how then God would know
that they are all the individual things that there are.)
If the logical expressions do not “represent,” then a limited but far reaching
linguistic species of conceptual antirealism, the species I called logical antireal-
ism, ought to be accepted. Any advanced cognition requires statements that
make essential use of logical expressions. If logical expressions stand for noth-
ing, then those statements have no extralinguistic significance even if they con-
tain nonlogical components that do.
Many expressions, e. g., in fiction, also stand for nothing, but this is why
they are not taken to serve a cognitive function. The function of logical expres-
sions, however, is unquestionably cognitive. The structure of a sentence de-
pends, at least in part, on the logical expressions the sentence contains. And
it is relevant to cognition because it is relevant to the truth-value of the sentence.
We may call the role of logical expressions in cognition “transcendental,” in
a sense related to Kant’s yet different in that it applies to language, not mental
items or faculties. Indeed, the linguistic turn has often been described as a tran-
scendental turn. If limited to logical cognition, it is more measured, discerning,
cautious, and therefore more plausible than the linguistic turn exemplified in the
statements quoted earlier from Wittgenstein and Heidegger. The limits of my lan-
guage may mean the limits of my world, but surely there is more to the world
than language. Language may be the house of being, but surely it is not also
the furniture in that house.
Logical antirealism rests on philosophical, a priori, not empirical consider-
ations. That logic is an a priori discipline has seldom been questioned. To be
sure, in recent years doubts have been expressed, most notably by Albert Casu-
llo, that any knowledge is a priori.¹⁵⁵ But the fact remains that questions in logic
are usually discussed and settled without making empirical appeals and certain-
ly without engaging in empirical research.
Despite the extensive debates over the general issue of realism/antirealism,
little has been said about the specific issue of logical realism/antirealism, even
though all advanced cognition involves cognition of logical structure. Perhaps
this neglect has been due to the comforting thought that logic, which is a branch
 Albert Casullo, A Priori Justification (New York: Oxford University Press, 2004). But see also
Laurence BonJour, In Defense of Pure Reason. (London: Cambridge University Press, 1998).

Authenticated
4 Language in logical cognition 103
of philosophy so advanced that it is often considered a separate discipline, has

already dealt with and settled the issue. This, of course, is an illusion. Indeed, in
the Tractatus, Wittgenstein held that the logical constants do not represent, and
offered his truth tables to explain them. But we shall see in the next chapter that
Russell felt compelled to allow for the existence of irreducibly negative and gen-
eral facts, which surely would count as logical objects. He showed that the at-
tempts to reduce negative facts to “incompatibility” and general facts to conjunc-
tions or disjunctions of singular facts are unsuccessful.

Authenticated
Chapter Six:
Metaphysical Realism and Logical Antirealism
1 The Logic of Realism
In Chapter Five I considered the belief that cognition, or at least “thought,” in-
volves language. This belief encouraged and remains characteristic of the version
of metaphysical antirealism that came to dominate 20th century philosophy when
it took the linguistic turn. Linguistic antirealism is the most plausible version of
metaphysical antirealism. Language is indispensable at least in the case of log-
ical cognition, the sort of cognition that involves notions that are characteristi-
cally logical, such as negation and generality. Antirealism with respect to logical
cognition, which I called logical antirealism, is the most plausible version of lin-
guistic antirealism.
Metaphysical antirealism is best understood in the context of the develop-
ment of its opposite, metaphysical realism. The latter’s guiding principle may
be put as follows: acknowledge as being there what must be there if cognition
and truth as correspondence are to be possible. This principle governed the con-
struction of the ontological inventories, categorial schemes, of Aristotle, Frege,
Russell, and Bergmann. I select these philosophers because they exemplify espe-
cially clearly the role of the principle in the development of realist ontology.
In the Categories Aristotle used the notions of said of and present in as prim-
itive, and with unsurpassed elegance proposed the following ontological inven-
tory: (1) items said of but not present in something else (what he called “secon-
dary substances,” i. e., substance universals, such as cat), (2) items present in
but not said of something else (particular “accidents,” in any of the nine catego-
ries of accident, e. g., this cat’s whiteness, its white color), (3) items both present
in and said of something else (universal accidents, e. g., the color white), and (4)
items neither said of nor present in something else (what he called primary sub-
stances, i. e., particular substances, such as this cat).
Frege’s inventory included (1) objects (roughly, what we call particulars or
individual things), (2) first-level functions (he called them “concepts” but in a
nonpsychological sense closer to that of “property” in current philosophy),
which take objects as arguments and yield truth or falsity as values, (3) sec-
ond-level functions, expressed by what we call quantifiers, which also yield
truth or falsity as values but take first-level functions, not objects, as arguments,

Authenticated
1 The Logic of Realism 105
(4) thoughts, roughly what we call “propositions,” the vehicles of truth-value,

and (5) functions that take thoughts as arguments and yield compound thoughts,
what the propositional connectives express.¹⁵⁶
Russell’s inventory, at least at the time of The Philosophy of Logical Atomism,
included (1) particulars, (2) universals (properties), (3) atomic facts, (4) negative
facts, thus at least one kind of molecular facts, and (5) general facts. His notion
of fact was briefly explained in the Introduction.
Bergmann proposed a much richer inventory, which is why he is especially
relevant here. It included (1) particulars, (2) universals, (3) the “ultimate sorts” of
particularity and universality, in virtue of which an item is a particular or a uni-
versal, (4) facts (also called complexes), which might be only “potential,” possi-
ble but not actual, (5) what the propositional connectives express, (6) what the
quantifiers express, (7) the “modes” of actuality and potentiality, which “per-
vade” facts and render them actual or potential, and (8) the three nexus of exem-
plification, set-membership, and meaning.¹⁵⁷
The logic of the development of realism that these inventories exemplify now
emerges. All four include the category of particulars. Even sense data, with
which, a la Berkeley, Russell toyed, are particulars. The privileged status of
the category is due presumably to the fact that particulars are the standard ex-
amples of what is perceived or imagined. This may be why some opponents of
antirealism (perhaps motivated by the title of Nelson Goodman’s Ways of World-
making) object to it by saying, “Surely there were a sun and an earth before there
were human beings and thus, for all we know, before there were minds, or lan-
guage and concepts.” But the antirealist claims, not that the sun and the earth
were created by our thought or language (Goodman was not a sort of creationist),
but that they cannot be perceived, conceived, and described independently of
our perception, thought, and language – which is hardly more than, in Hilary
Putnam’s phrase, a virtual tautology.¹⁵⁸
Such opponents of antirealism presumably would be willing to say (1) that
the sun and the earth are particulars, (2) that they have properties, (3) that the
sun exemplifies the property of being a star and the earth the property of
being a planet, (4) that it is a fact that there was a sun and an earth before
there were human beings, (5) that the sun is not a planet and the earth not a
 See “Concept and Object,” “Function and Object,” “Thought,” and “Negative Thoughts,” in
Beaney, The Frege Reader.
 See New Foundations of Ontology. Also “An Ontological Inventory,” Journal of the British So-
ciety of Phenomenology and Existentialism, 1974, and Logic and Reality (Madison: University of
Wisconsin Press, 1960).
 Hilary Putnam, “The Dewey Lectures,” Journal of Philosophy, 1994, 513.

Authenticated
106 Chapter Six: Metaphysical Realism and Logical Antirealism
star, (6) that they and other stars and planets were all that there was before there
were human beings, (7) that they actually, not just possibly, existed then, (8) that
the sun and the earth constitute a pair of which they are the members, and (9)
that some things, mental or verbal, mean, refer to, the sun and the earth.
If the opponents of antirealism refuse to make such additional assertions,
and especially to acknowledge items that might correspond to the italicized
words in them, what would be their justification? Might they be merely display-
ing, however unwittingly, the continuing hold of 17th and 18th century empiricism,
in particular its confusion of knowing with perceiving and of thinking with imag-
ining? We are comfortable with particulars because they come to mind first when
asked what we perceive or can imagine. But they are hardly all that we can know
or think. The opponents of antirealism who confidently assert the existence of
the sun and the earth ignore the question whether the sun and the earth
would have existed if there were not also items corresponding to the italicized
words. At least they owe us a serious discussion of how sparse their ontology
could be. That it cannot limit itself to such items as the sun and the earth –
i. e., to particulars – became evident in the logical development of realism
when it left the terra firma of particulars.
The first move beyond particulars was to introduce the category of proper-
ties, a move with which Plato dazzled philosophers. It was almost inevitable be-
cause we think and speak not only of particulars but also of what they are. The
move to Fregean thoughts and Russellian or Bergmannian facts came more than
two millennia later but seemed also inevitable. We speak in sentences, not lists of
names. If the description of the world requires sentences, what items in the world
require this and might be said to correspond to sentences? As we saw in the In-
troduction with the example of Jack’s admiring Jill but Jill’s not admiring Jack,
surely they are not the items that correspond to names and predicates. At
least atomic facts must be allowed. But a further move to molecular, even if
only negative, and then to general facts also seemed required. Atomic sentences
are woefully inadequate for any cognition that is at all advanced. Any language
beyond that of babes requires molecular, at least negative, as well as general
sentences, whether universal or particular. And so a move to negation and gen-
erality themselves, i. e., to what makes negative facts negative and general facts
general, seemed needed. This move was made explicitly by Frege and Bergmann,
as well as by Russell in Theory of Knowledge ¹⁵⁹ though not in The Philosophy of
Logical Atomism. Bergmann even included in his inventory particularity, univer-
 Bertrand Russell, Theory of Knowledge: The 1913 Manuscript (London: Allen & Unwin,
1984), 99.

Authenticated
sality, actuality, and potentiality, as what allows us to distinguish respectively be-

tween particulars and universals and between actual and merely possible facts,
as well as exemplification, set-membership, and meaning, which respectively “tie”
particulars to their properties, members of sets to their sets, and thoughts to
what they are thoughts of. Like Meinong, Bergmann has often been accused
by other realists of “going too far,” but those realists do not explain where
and how the line should be drawn. The usual half-hearted attempts to rescue re-
alism from “excesses” by adopting some sort of reductionism, for example, of
negative facts to incompatibility and of general facts to conjunctions or disjunc-
tions of atomic facts, were shown by Russell in The Philosophy of Logical Atom-
ism to fail because they presuppose what they claim to “reduce.”
Aristotle’s, Frege’s, Russell’s, and Bergmann’s ontological inventories are in
many ways similar, despite their major differences. All four include particulars
and properties, though Frege called particulars “objects” and properties “con-
cepts.” Frege’s, Russell’s, and Bergmann’s include a category of items that corre-
spond to sentences, though Aristotle’s does not. Frege called these items
thoughts, Russell and Bergmann called them facts. Fregean thoughts were
said by Frege to belong in a “third realm,” which is neither mental nor physical,
but this would be true also of Russellian and Bergmannian facts. A fact may in-
clude constituents that are mental or physical, but in no clear sense can itself be
called mental (admiration may be something mental, but is the fact that Jack ad-
mires Jill also mental, in whose mind would it be?) or physical (what are the
weight, size, and location of the fact that Jack admires Jill?) This is why facts
are sometimes said to be only in logical space, just as Fregean thoughts were
said to be in a third realm. Frege’s, Russell’s, and Bergmann’s inventories includ-
ed negative and general facts, but only Frege’s and Bergmann’s included nega-
tion and generality themselves. Except in the abandoned Theory of Knowledge,
Russell committed himself only to negative and general facts, not to what
makes them negative or general. And only Bergmann followed the logic of real-
ism further by including particularity, universality, exemplification, actuality, po-
tentiality, set-membership, and meaning.
Realists were required by the realist principle to acknowledge such addition-
al “items,” though only Bergmann acknowledged all of them. The reason for the
requirement was that we can hardly suppose that our thought or language cor-
responds to the actual, objective world of particulars if we do not suppose that
most, if not all, of those additional items are also in it. There would not be a suf-
ficiently specific, definite fit between thought or language and the world without
them. For example, the sentence “Jack admires Jill” could not, if true, correspond
just to Jack and Jill, or even to Jack, Jim, and the relation of admiring, since the
sentence “Jill admires Jack” also would correspond to those things.

Authenticated
The motivation behind antirealism is now easy to discern. As Russell said

about disjunctive facts, it is not plausible that somewhere “in the actual objec-
tive world” there are all those things “going about” that the realist principle
seems to require.¹⁶⁰ It is implausible because, except for particulars and perhaps
properties, the “things” in question are all logical, in the broad sense of “logical”
explained in the Introduction. It becomes tempting, therefore, to conclude that in
the case of such things there are only the words or syntactic structures supposed
to correspond to them.
But Wittgenstein made evident in the Tractatus that this conclusion is at
least hasty. By distinguishing between saying and showing, he proposed a
third alternative, what I called semirealism. I argued in chapter 4 that the distinc-
tion is neither idiosyncratic nor obscure. It has an obvious application even to
ordinary pictures. I also argued that the related picture theory of meaning and
thought conforms to the traditional view that thought involves representations,
“ideas,” perhaps even mental images or “pictures.”
Wittgenstein’s rejection of realism in logic was unequivocal and explicit:
“My fundamental idea is that the ‘logical constants’ are not representatives
[nicht vertreten]; that there can be no representatives of the logic of facts”
(4.0312). But his rejection of antirealism in logic was also unequivocal and ex-
plicit: “…is it really possible that in logic I should have to deal with forms
that I can invent? What I have to deal with must be that which makes it possible
for me to invent them” (5.555).¹⁶¹ Even though the logical constants represent
nothing, there must be something that makes our use of them possible and
thus is presupposed by it.
Wittgenstein went to some lengths telling us what that “something” is. It in-
cluded the existence of objects, which are (1) the meaning (reference, Bedeutung)
of names, (2) the values of the individual variables in general propositions, in-
cluding “the laws of logic,” (3) and the constituents of atomic facts (Sachver-
halte, “states of affairs”), and thus (4) the “substance of the world.”¹⁶² It includ-
 The Philosophy of Logical Atomism, 71.

 “In logic nothing is accidental: if a thing can occur in a state of affairs, the possibility of the
state of affairs must be written into the thing itself…” (2.012). “…We have said that some things
are arbitrary in the symbols that we use and that some things are not. In logic it is only the latter
that express: but that means that logic is not a field in which we express what we wish with the
help of signs, but rather one in which the nature of the absolutely necessary signs speaks for
itself [zeichnen sich aus]…” (6.124).
 “The propositions of logic…presuppose that names have meaning and elementary propo-
sitions sense; and that is their connexion with the world….” (6.124). “A name means an object.
The object is its meaning….” (3.203).

Authenticated
ed the senses of atomic propositions,¹⁶³ the existence or nonexistence of the atom-

ic facts such propositions assert,¹⁶⁴ and thus their truth or falsity. ¹⁶⁵ It included
the truth-possibilities of atomic propositions, the combinations of which are
made explicit in the truth tables for the molecular propositions composed of el-
ementary propositions. It included the totality of the singular propositions that
general propositions show but do not assert.¹⁶⁶ It included formal concepts
like those of being an object and being a fact. It included therefore also the total-
ity of objects and the totality of facts, i. e., the world itself. All these can be
“shown,” but not “said.”
The central doctrine in the Tractatus was that what logic is about is inex-
pressible. This motivates the antirealist reading of the book, for example, Cora
Diamond’s¹⁶⁷ and Warren Goldfarb’s,¹⁶⁸since one might reasonably conclude
that what cannot be said lacks reality. But the Tractatus also tells us that
what cannot be said nonetheless can be shown or, as Wittgenstein sometimes
writes, e. g., in 5.5561 and 6.522, that it can show itself (zeigt sich). One might
then no less reasonably conclude that what cannot be said but can be shown
is part of reality. After all, it is not like Pegasus or phlogiston. It does show itself.
This motivates the realist reading, for example David Pears’¹⁶⁹ and P. M. S. Hack-
er’s.¹⁷⁰ Wittgenstein, however, avoided both antirealism and realism. He did
write “There are, indeed, things that cannot be put into words. They make them-
selves manifest [Es gibt allerdings Unaussprechliches],” but his next sentence was
“They are what is mystical [Dies ‘zeigt’ sich, es ist das Mystische]” (6.522). These
“things” are not of concern just to “soft” disciplines like ethics and religion. They
include logical form and formal properties, the subject matter of logic. We cannot
say of an item that it is an object, a complex, fact, function, number, or the total-
ity of objects or of facts. Therefore, we cannot even say of the world that it is a
world. But what we cannot say shows itself in what we can say. It is not nothing.
 “A proposition shows [zeigt sich] its sense. A proposition shows [zeigt] how things stand if it
is true. And it says that they do so stand…” (4.022).
 “The existence and non-existence of states of affairs is reality” (2.06). “The simplest kind of
proposition, an elementary proposition, asserts the existence of a state of affairs” (4.21).
 “4.25 If an elementary proposition is true, the state of affairs exists: if an elementary prop-
osition is false, the state of affairs does not exist.”
 Letters to Russell, Keynes and Moore, 71.
 Cora Diamond, The Realistic Spirit: Wittgenstein, Philosophy and the Mind (Cambridge, MA:
MIT Press, 1991).
 Warren Goldfarb, “Metaphysics and Nonsense,” Journal of Philosophical Research XXII
(1997).
 David Pears, The False Prison (Oxford: Oxford University Press, 1987).
 M.S. Hacker, Insight and Illusion (Oxford: Clarendon, 1972), 20 – 24.

Authenticated
Surely, the world is “something.” Surely, so is what Wittgenstein called “the

higher,” in ethics or in logic. Yet, neither is it straightforwardly a part of reality.
If it were, it would not be inexpressible.
Cora Diamond calls such a reading of the Tractatus a “chickening-out,”¹⁷¹
and Warren Goldfarb calls it “irresolute.”¹⁷² But I doubt they would say that log-
ical form is like Pegasus or phlogiston. Wittgenstein is explicit that in some sense
it is in “reality” [Wirklichkeit]: “Propositions can represent the whole of reality,
but they cannot represent what they must have in common with reality in
order to be able to represent it–logical form” (4.12), “Propositions show the log-
ical form of reality. They display it” (4.121). Logical form can only be shown, not
because it is nothing but because its presence in both language and reality is
what makes any saying possible. Saying presupposes showing. Something is
said only if something else is shown.
But while logical form is not like Pegasus or phlogiston, neither is it like Secre-
tariat or oxygen. Although Wittgenstein was not an antirealist, he also was not a re-
alist. This should not surprise us. In both his early and his later works he avoided
simplistic distinctions: “And yet you again and again reach the conclusion that the
sensation itself is nothing. – Not at all. It is not a something, but not a nothing ei-
ther!”¹⁷³ Wittgenstein’s distinction between saying and showing amounted to a rejec-
tion of standard realism, but it did not amount to acceptance of standard antireal-
ism. In this respect, he was not unlike his distant precursor Kant, who described his
position as transcendental idealism but also empirical realism.¹⁷⁴
Had Wittgenstein provided an explicit inventory of what according to him is
in the world, which he did not because the inventory would have employed for-
mal concepts, it would have been sparse. At most, it seems, it would have con-
tained two categories: objects and facts. Moreover, it is not clear whether the cat-
egory of objects included both particulars and properties. Indeed, it is not clear
that Wittgenstein allowed this distinction. Nor is it clear that he allowed for mo-
lecular and general facts, rather than just for atomic facts, though he did seem to
allow for negative facts: “We call the existence of states of affairs a positive fact,
and their non-existence a negative fact” (2.06).
 The Realistic Spirit, e. g., 181, 194 ff..

 Warren Goldfarb, “Metaphysics and Nonsense,” Journal of Philosophical Research XXII
(1997), 66. See also, in the same issue, Cora Diamond, “Realism and Resolution: Reply to Warren
Goldfarb and Sabina Lovibond,” 75 – 86.
 Philosophical Investigations, 304.
 Critique of Pure Reason, A 369 – 370.

Authenticated
2 Antirealism: ontological, cosmological, and logical 111
Frege also allowed for what cannot be said, though not quite in Wittgen-
stein’s sense.¹⁷⁵ One of his ontological categories, as have seen, was that of con-
cepts. But we cannot speak about concepts: “The concept horse is not a con-
cept,” Frege wrote. What expresses a Fregean concept is a grammatical
predicate, and “the concept horse” is not a predicate. The relevant predicate is
“is a horse,” but it cannot serve as the grammatical subject of a well-formed sen-
tence, including a sentence of the form “x is a concept.” Yet, of course, there is
such a concept. Some objects are horses. Frege did not deny that there are con-
cepts. He was an unmitigated realist. But neither did he explain their peculiar
status. This is why his readers have found his assertion that the concept horse
is not a concept confusing.
2 Antirealism: ontological, cosmological, and logical
According to metaphysical realism, the existence and nature of things (res),

reality, the world, is independent of our cognition of them, in the broad contem-
porary sense of “cognition” that includes sense perception, introspection, intel-
lectual intuition, imagination, memory, recognition, conceptualization, induc-
tive and deductive reasoning, the use of language and other symbolism.
Metaphysical antirealism denies this. I call both “metaphysical” in order to dis-
tinguish them from realism and antirealism in ethics, aesthetics, the philosophy
of science, the philosophy of mathematics, and other branches of philosophy,
where realism and antirealism are restricted to a specific subject matter.
The thesis of metaphysical antirealism is that the world insofar as it is know-
able by us depends on our capacities and ways of knowing, our cognitive facul-
ties. The inclusion of “insofar as it is knowable” does not produce the sort of
equivocation present, for example, in the sentence “As king of England, John
was subject to no one,” which is true but compatible with “As duke of Aquitaine,
John was subject to the king of France.” We know both that John was king of
England and that he was duke of Aquitaine. But we know the world only insofar
as it is knowable. And this is a tautology, unlike “As king of England, John was
subject to no one.”
The standard argument for the thesis of metaphysical antirealism, in both
the Kantian and its more recent versions, was sketched in the Introduction as fol-
 Cf. P.T. Geach, “Saying and Showing in Frege and Wittgenstein,” in Essays on Wittgenstein
in Honor of G.H. von Wright, ed. Jakko Hintikka, Acta Philosophica Fennica 28 (Amsterdam: North
Holland, 1976).

Authenticated
lows: (1) We cognize only what we have the capacity to cognize. This is a tautol-
ogy. Therefore, (2) there is no reality, no world, that is independent of our cogni-
tive capacities. But (2) does not follow from (1). What follows is another tautol-
ogy, (3) that we cannot cognize reality independently of our cognitive
capacities. Contemporary antirealists argue on the basis of (1) for (2), not for
(3), probably because the negation of (2), Kant’s view (4) that there is a reality,
“things-in-themselves,” which is independent of our cognitive capacities, seems
to them idle. But there is a very good reason for (4), namely, that (2) implies a
sort of cosmic humanism, human creationism, the proposition that the world de-
pends on certain members of one of its planets’ fauna.
Metaphysical antirealism comes in many varieties, as different as Berkeley’s
subjective idealism, Kant’s transcendental idealism, Hegel’s absolute idealism,
Wittgenstein’s logical antirealism, Heidegger’s phenomenology of being-in-the-
world, as well as, in more recent philosophy, Michael Dummett’s antirealism
and Nelson Goodman’s “irrealism.” Berkeley held that the existence of the things
we perceive is dependent on our perception of them, Kant held that their nature
is dependent also on our understanding, and Wittgenstein, Heidegger, and
Goodman stressed their dependence on our language.
In one form or another, metaphysical antirealism has dominated modern
philosophy at least since Berkeley, even though it is metaphysical realism that
remains the bedrock of everyday and scientific thinking. But, as we have seen,
numerous though seldom made distinctions within it are needed. And the rea-
sons for accepting it have seldom been stated in detail. Usually they have con-
sisted in abstract, obscure generalities such as “Nothing can be conceived that
cannot be perceived” and “For knowledge to be possible, objects must conform
to knowledge,” which are no less controversial than what they are reasons for.
In the Introduction I distinguished antirealism from skepticism and relativism. I
also distinguished the several varieties of antirealism: ontological, cosmological,
perceptual, conceptual, linguistic, and logical. One can be an antirealist but not a
skeptic, and one can be a skeptic but not an antirealist. One can be an antirealist
but not a relativist. One can be a cosmological but not an ontological antirealist.
One can question the reality of logical objects such as facts, and thus of the
world conceived as the totality of facts, but not the reality of things, e.g., animals,
vegetables, and minerals. And one may be neither a realist nor an antirealist regard-
ing some items, e.g., those that subsist (Russell) or those that can only be shown
(Wittgenstein), hence the need to acknowledge what I called semirealism.
Within both cosmological and ontological antirealism, we need also the dis-
tinction between perceptual and conceptual antirealism. Since a concept may be
understood as either a purely mental, psychological, and therefore subjective
item, or as a linguistic, public, and therefore objective one, we must distinguish

Authenticated
between Kant’s psychological version and the 20th century linguistic versions of
conceptual antirealism. Further distinctions are needed within the variety of con-
ceptual antirealism that I called logical. As we shall see, one may accept a realist
interpretation of singular statements, a semirealist interpretation of compound
and universal statements, and an antirealist interpretation of generic statements.
The distinction between ontological and cosmological realism/antirealism is
of special importance. If the world is the totality of facts, not of things, then one
who rejects the category of fact also rejects cosmological realism, realism with re-
spect to the world, but not necessarily ontological realism, realism with respect to
things. Even if the world is not the totality of facts, it is certainly not just the to-
tality of things, a mere collection of unrelated items, a whole without structure. It
must still have a logical structure, and thus its reality is subject to question by
the logical antirealist.
These distinctions may seem fussbudgety, academic fretwork, but they are
needed because each variety of metaphysical antirealism calls for different consid-
erations. The realism/antirealism debate for too long has appeared to allow only two
alternatives: all and nothing. It has been needlessly vague because those crucial dis-
tinctions have not been made. One aim of this book is to show that the issue is far
more complex than participants in the debate usually suppose.
I argued in the Introduction that a clear and plausible defense of antirealism
should shun purely abstract arguments and appeal to specific and readily under-
standable truths. One such truth is that logical expressions, i. e., the expressions
distinctive of logic, play an essential role in all developed talk and thought but
almost certainly stand for nothing in the world. It is the truth on which logical
antirealism rests. Standard examples of such expressions are the sentential op-
erators: “not” (“~”), “and” (“•”), “or” (“v”), “if…then…” (“⊃”); the quantifiers:
“all” (“"”), “some” (“$”); and the verb “to be” in its senses of identity (“=”), ex-
istence (“there is,” “exists”), and predication (“is,” “are”), even when the latter is
expressed only by syntactic order. “Is” is a logical expression in both “Socrates is
human” and “God is,” though standard logic would express the former not with
a separate sign but by the juxtaposition of the subject and predicate terms (e. g.,
“Hs,” “H” standing for “is human” and “s” for “Socrates”), and would either ig-
nore the latter as ill-formed or translate statements in which it occurs by employ-
ing the particular quantifier “some.” No abstract argument or obscure philo-
sophical principle is needed to convince nonphilosophers as well as most
philosophers that, though essential to any cognition above that of infants,
such expressions do not stand for any items, that in the world there is no
such object, for example, as not, all, is, or and.
Frege used the phrase “logical objects” for the objects of arithmetic in the
context of his project of reducing arithmetic to logic, a project continued later

Authenticated
by Russell and Whitehead.¹⁷⁶ Wittgenstein used it for the entities a logical realist
may think are required for understanding the sentential connectives. In the In-
troduction I applied the phrase also to facts, in Russell’s sense. Clearly, it is ap-
plicable to “thoughts” in Frege’s sense. Both Russellian facts and Fregean
thoughts are categories of entities accepted because of broadly logical consider-
ations. Both were explicitly introduced as the category of entities that declarative
sentences stand for or express. I have accordingly also counted declarative sen-
tences as logical expressions, facts being the logical “objects” to which, if true,
they supposedly correspond. One of the central concerns of logic has been the
analysis of sentences with respect to their logical form. A basic presupposition
of logical analysis is that sentences that are unlike in surface grammar may
share the same logical form. In analysis we begin with the surface grammar of
the sentence but search for what it must have in common with other sentences,
especially those in other languages, if they are to have the same truth-value and
the same implications. That common feature is the logical form of the sentence.
Its representation would require logical expressions.
The logical realist holds that at least some logical expressions correspond to
entities. As we saw in the previous section, Frege and Russell (at one central
stage of his philosophy), were clearly logical realists. So seemed to be Gustav
Bergmann, though in the next chapter we will find that this was not at all
clear. Wittgenstein, in Tractatus Logico-Philosophicus and his later works, ap-
peared to be a logical antirealist, but we have seen that his position would be
better described as semirealism. Logical antirealism was characteristic, even if
only tacitly, of traditional philosophy. For example, Kant wrote that logic ab-
stracts “from all objects of knowledge,” that it has “itself alone and its form”
to deal with, and that the concern of logic is to give “an exhaustive exposition
and a strict proof of the formal rules of all thought.”¹⁷⁷ In effect, he acknowl-
edged the chief thesis of logical antirealism: there are no logical objects even
though logic is present in all thought. All thought, even when it does not have
objects, must conform to logic, but logic has no objects. This could have been
an argument for transcendental idealism additional to those Kant explicitly of-
fered, but he did not present it as such.
Logical antirealism leads directly to cosmological antirealism if “fact” is a
logical expression and the world is the totality of facts. Perhaps animals, vege-
tables, and minerals also have logical structure, but this is not evident as it is
 Gottlob Frege, The Basic Laws of Arithmetic (Los Angeles & Berkeley: University of Califor-
nia Press, 1987), ii, § 147.
 Critique of Pure Reason, B ix.

Authenticated
in the case of a fact. Thought and talk about individual things do usually employ
logical concepts and expressions, but only phenomenalist theories, which ana-
lyze statements about material objects as statements about actual and possible
experiences, explicitly attribute logical structure to individual things. Logical an-
tirealism does not entail ontological antirealism because individual things have
no logical structure, they are not facts and thus involve neither propositional
connectives nor quantifiers in the way facts, and the sentences expressing
them, do. They can be subjected to chemical analysis but not to logical analysis.
The belief that they could be was generated by Wittgenstein’s claim that all com-
plexity is propositional and had been encouraged by Mill’s view that a physical
object is nothing but the ‘permanent possibility of experience,’ i. e., nothing but
its actual and possible appearances, what one would sense in all relevant cir-
cumstances. Much later, Price showed that these would be unlimited in number,
and consequently statements about the object could not be translated into a con-
junction of statements about its appearances.¹⁷⁸
An antirealism that is only logical and also only cosmological would be so
moderate as to be virtually part of common sense, the thoughtful nonphilosoph-
ical judgment that all theorizing must respect even if not accept. Common sense
would rebel against the postulation of objects corresponding to the sentential
connectives. It is satisfied with counting them as syncategorematic, mere ancil-
laries in assertions, incapable of use as subjects or as predicates. Common sense
could be easily convinced that sentences, the expressions Russell took to stand
for facts, are also “syncategorematic.” They, too, function as neither names nor
predicates. Facts were not on Aristotle’s or any other list of categories before Rus-
sell and Wittgenstein included them in theirs. It would occur only to philoso-
phers that the sentential connectives and sentences themselves might stand
for anything in the world. The position of semirealism, of course, would be too
technical for common sense to judge.
Only a few philosophers, most notably Russell, have thought that there are
“general facts,” which correspond to universal sentences, as well as “molecular
facts,” which correspond to molecular, especially negative, sentences. Frege, who to-
gether with Russell was the target of Wittgenstein’s attack on logical realism,¹⁷⁹ ab-
horred Russell’s category of fact (on the grounds that to say that p is a fact is to say
no more than that it is true that p) but he did appeal to the no less questionable
category of objective entities he called “thoughts,” which are expressed by senten-
 Price, H. H., Perception, (London: Methuen, 2nd ed., 1950).

 “[T]he words ‘complex’, ‘fact’, ‘function’, ‘number’ … all signify formal concepts, and are
represented in conceptual notation by variables, not by functions or classes (as Frege and Rus-
sell believed)” (Tractatus 4.1272).

Authenticated
ces. Frege also accepted the existence of general and negative thoughts, as well as of
“compound thoughts” such as conjunctions, disjunctions, and conditionals. His vig-
orous arguments, as well as Russell’s, for the irreducibility of general and negative
statements were, in effect, arguments for logical realism.
To avoid confusion, in this chapter I shall often follow Frege’s, Russell’s, and
Wittgenstein’s practice of using “general” instead of “universal,” even though
particular (“existential”) statements and generic statements (to be discussed in
chapter 8) are also general. None of them thought that true generic statements
stand for generic facts. They did not even consider them.
3 The logical structure of the world
The subject matter of metaphysics is said to be being, reality, what exists, or just
“the world” as in Goodman’s Ways of Worldmaking. Its first question is, what is
real, what kind of entities exist? Various answers have been given, e. g., that only
material entities exist (Hobbes), that only mental entities exist (Berkeley), that in
addition to material and mental entities also abstract entities (Plato) or God
(Aquinas) exist. But however we answer the question, we take for granted that
the world has a structure, that it is not a mere collection of isolated items.
What kind of structure is fundamental, absolutely necessary, to the world?
Not a causal structure: Hume denied that there are causal connections, except
in the bland sense of spatiotemporal correlations. Not a physical or even just
spatial structure: the dualist holds that in addition to material entities there
are irreducibly nonphysical mental entities such as thoughts and feelings, and
the idealist even holds that everything is mental. Both deny that mental entities
are in space: they lack geometrical characteristics and do not enter in relations
such as two-miles-from. Nor is even a temporal structure absolutely necessary.
The Platonist holds that there are abstract entities, e. g., numbers, which are
not in time – they do not enter in relations such as two-years-earlier-than. The
theist holds that the spatiotemporal world and thus presumably space and
time themselves were created by God.
The fundamental structure of the world, which though seldom mentioned or
even considered is denied by no one, is logical. That a world must have a logical
structure is a requirement even more basic than that a world must be “logically
possible,” if the latter means that it must not “involve contradiction,” for even a
contradiction has logical structure, often as simple as that of “p and not-p.” As-
tronomy, physics, chemistry, and biology focus on the causal, spatial, temporal,
and physical structure of the world. Philosophy, at least as it was understood by
Frege, Russell, and Wittgenstein – the founders of contemporary logic and ana-

Authenticated
3 The logical structure of the world 117
lytic philosophy – focuses on its logical structure. If the first question of meta-
physics is what is real or exists, its second question is what kind of structure
it must have. And surely the answer is that it must have a logical structure. Ar-
istotle held that the “science of being qua being,” i. e., metaphysics, begins with
the study of the principles of the “syllogism,” i. e., logic.¹⁸⁰
What is meant by “logical structure”? The answer lies in what is meant by
“logic.” Logic is concerned with the relations between sentences (propositions,
statements, judgments) that hold in virtue of their logical form. And the logical
form of a sentence is best understood as what is left when we completely abstract
from its subject matter, what it is about. According to Russell’s canonical account, in
respect to logical form sentences (propositions) are either atomic, compound, or
general.¹⁸¹ Essential to discerning the logical form of a sentence are the logical ex-
pressions it contains. To say that the world must have a logical structure entails that
any adequate description of it must employ such expressions. And we have no con-
ception of a world that is not, at least in principle, describable.
Nothing would count as a world if it did not allow for atomic statements,
e. g., “This page is white.” Nothing would count as a world if it did not allow
for compound statements, e. g., “This page is not blue” and “If this page is
white then it is not blue.” And nothing would count as a world if it did not
allow for general statements, e. g., “All winters in Minnesota are severe” and
“There is water on Mars.” It is natural therefore for the logical realist to see
the logical structure of the world as providing room for atomic facts about the
properties and relations of individual things, for compound facts, most notably,
facts about what is not the case (negative facts), what is the case only if some-
thing else is the case (conditional facts), for two facts being both the case (con-
junctive facts), for either one or the other of two facts being the case (disjunctive
facts), and for general, either universal or particular, facts.
The logical realist may go further. Since nothing would be a world if it did not
allow for saying of a thing what it is, that is, for predication, the logical realist may
hold that there is in the world a relation or nexus of exemplification or instantiation.
Since nothing would be a world if it did not allow for distinguishing one thing from
another, the logical realist may hold that there is negation in the world. And since
nothing would be a world if it did not allow for generalization, the logical realist
may hold that there is generality in the world. The logical antirealist disagrees, in-
sisting that, insofar as it is knowable, the world has a logical structure only because
we describe it with a language that contains logical expressions.
 Aristotle, Metaphysics, 1005 b 7– 35.

 Principia Mathematica, Introduction.

Authenticated
Logical expressions are parts of sentences. Sentences are the vehicles for the
description of the world. I have suggested that we may consider sentences them-
selves as logical expressions, and the facts supposed to make them true as log-
ical objects. The logical antirealist will argue, however, that a fact is indistin-
guishable from the sentence it supposedly makes true. A fact is a complex
entity with a structure or form that seems to image the structure or form of a sen-
tence. For example, Russell’s requirement that an atomic fact “contain” particu-
lars and properties or relations corresponds to the grammatical requirement that
a simple sentence contain a subject and a predicate, and his distinction between
atomic and molecular facts corresponds to the grammatical distinction between
simple and compound sentences. Indeed, the former correspondence was explic-
itly acknowledged in Wittgenstein’s doctrine that atomic sentences are logical
pictures of atomic facts. It is fairly clear that when describing the ontological
characteristics of facts and their logical structure or form Russell relied on the
grammatical characteristics of sentences. Of course, he was concerned not
with the “surface” grammatical form of a sentence but with its logical form.
But one arrives at the latter only by beginning with the former. Logic is con-
cerned with what sometimes is called logical grammar, not with ordinary gram-
mar, but we have little if any conception of logical grammar apart from our con-
ception of ordinary grammar.
The central role of sentences was explicitly acknowledged in Frege’s explan-
ation of his technical notion of thought (Gedanke) as the “sense” (Sinn) of a sen-
tence. Even though he held that thoughts belong neither in the physical nor in
the mental world, but rather “in a third realm,”¹⁸² Frege wrote, “The world of
thoughts has a model in the world of sentences, expressions, words, signs. To
the structure of the thought there corresponds the compounding of words into
a sentence.”¹⁸³ Russell’s and Wittgenstein’s heavy indebtedness to Frege in log-
ical theory was freely acknowledged by both. Like Fregean thoughts, Russell’s
and Wittgenstein’s facts, at least those that are atomic, also were understood
as analogous to, even “pictured” by, sentences.
Although the German word Gedanke is a synonym of the English “thought,”
as used by Frege it has no ordinary translation in English. “Proposition” might
be least misleading, as long as we think of a proposition as an objective item dis-
tinct both from the sentences in the various languages that express it and from
the ideas and judgments we entertain about it. If so, there would be no clear dif-
ference between a true Fregean thought and a Russellian fact. This is why Frege
 Beaney, The Frege Reader, 337.

 The Frege Reader, 351.

Authenticated
wrote, “What is a fact? A fact is a thought that is true.”¹⁸⁴ His view that thoughts
are neither mental nor physical may seem mysterious, but as we saw earlier so
might Russell’s and Wittgenstein’s facts. Neither propositions nor facts are
“things” that might enter in spatial relations, or mental states that might enter
in temporal relations. Neither the proposition nor the fact that this table is
two feet from me at any other distance from me, and neither is a mental
image or state that is earlier or later than some other mental image or state.
In logic we speak usually of propositions, sometimes of sentences, not of facts.
This is a symptom of our ambiguous conception of its subject matter, our uneasy
attempt to straddle the perceived chasm between the logic of sentences, which is
about language – a certain behavior of some members of our planet’s fauna and
thus essentially a zoological subject matter – and the logic of facts, which suppos-
edly is about the world and thus, as Plato might have said, a more appropriate sub-
ject matter for “spectators of all time and all existence.” According to logical real-
ism, the chasm is real. According to logical antirealism, it is an illusion. The logic
of the world, the antirealist holds, is not distinguishable from the logic of words,
and Frege’s thoughts, as well as Russell’s or Wittgenstein’s facts, are merely hypo-
statized sentences, shadows that sentences cast upon the world. I shall have much
more to say about the topic of facts in chapter 9.
The general thesis of metaphysical antirealism now appears more plausible.
What makes a world a world, rather than a congeries of things, is what requires
sentences, not mere lists of names, for our description of it, namely, a logical
structure. But the only conception we have of logical structure is that of the log-
ical structure of sentences. This is why in speaking of a world we must appeal to
the category of facts. This is why Wittgenstein unhesitatingly wrote that the
world is the totality of facts, not of things. Sentences, of course, are parts of lan-
guage, and their logical structure is a feature of language. But language is some-
thing human, “ours.” We have no genuine conception of a language that is both
nonhuman and in principle untranslatable into a human language. Therefore,
insofar as we can conceive of the logical structure of the world, and thus of
the world as a world, they depend on us, they are ours, human. Of course,
that the only conception of logical structure we have is that of sentences does
not entail that the world does not have logical structure independently of lan-
guage. But, surely, it is a good reason for reaching that conclusion. For it does
entail that our cognition of the world, insofar as it involves logical concepts
and employs logical expressions, depends on language.
 The Frege Reader, 34.

Authenticated
Many twentieth century philosophers have held that all cognition, not just that
involving logical concepts, is dependent on language. Earlier I cited Wittgenstein’s
assertion that “The limits of my language mean the limits of my world,” Heidegger’s
that “language is the house of being” and that “language is the happening in which
for man beings first disclose themselves to him each time as beings,” and Quine’s
that “thought, as John B. Watson claimed, is primarily incipient speech.” These
opinions exemplify the linguistic turn philosophy took in the twentieth century. It
was more fundamental than its earlier “turns,” such as the Platonic turn to abstract
entities, the theological turn in the Middle Ages, and the idealist turn in the 18th cen-
tury. For it applies to everything we think is real, including abstract entities, God,
and minds. It led to linguistic antirealism.
But the linguistic antirealism I have sketched is not as extreme as that im-
plied by the quotations from Wittgenstein, Heidegger, and Quine. It does not
hold indiscriminately, as a matter of general principle, that the world as we cog-
nize it, as it is “for us,” depends on language. It restricts the dependence to log-
ical expressions. It avoids the highhandedness, often characteristic of philoso-
phy, which the quotations seem to exemplify. It does not hold that all
cognition depends on language. It is only a logical antirealism, even though
its implications are far-reaching.
It is thus a limited, moderate, and for this reason more plausible antirealism. It
does not offer a general and a priori answer to the question whether cognition de-
pends on language. This question, as we saw in Chapter Five, is mainly empirical, to
be answered by neuroscientists, psychologists, and linguists. And the proper answer
may well vary according to what kind of objects of cognition we have in mind. Rocks
are different from headaches, both are different from electrons, and all three are
very different from numbers. It would be rash to suppose that what is true of our
cognition of some is true of our cognition of all. Rock-climbing is not introspection,
and neither is physics. There are also many, very different kinds of cognition. Some
may depend on language while others do not. Surely, there is cognition in the form
of sense perception, enjoyed by infants and nonhuman animals, which does not in-
volve language. Recognition is a higher form of cognition, but it too is found in in-
fants and nonhuman animals. And driving a car involves specialized cognition that
only superficially, if at all, finds or needs expression in language. Creative work in
music or painting only minimally involves talking. Worldly people, especially in the
law, politics, and diplomacy, rely heavily on “reading” facial expressions, which lan-
guage notoriously fails to capture.
We achieve greater specificity and a sharper focus when we ask, Does logical
cognition have a distinctive subject matter, is it, at least in part, about anything dis-
tinctively logical? If not, then to that extent logical cognition would appear to be
nothing but use of language, since it would lack the feature essential to other

Authenticated
kinds of cognition: a subject matter, things they are about, which make them, at
least in part, the cognitions they are. Seeing a cat differs from seeing a dog partly
because cats differ from dogs. Seeing a cat differs from hearing a cat partly because
colors and shapes differ from sounds. It would not follow, however, that all there is
to logical cognition is language. I used the phrase “distinctive subject matter” pre-
cisely because, outside logic, sentences containing logical expressions do have a
subject matter, namely, what is denoted by the nonlogical expressions, the “descrip-
tive words,” that they also contain. But our question is whether such sentences have
a subject matter insofar as they are logical. The logical structure of the world may be
supplied by language, but the world does not have only a logical structure: it also
contains the entities that are structured. The tendency to forget this may explain why
philosophers seldom see that logical antirealism, which most of them accept, at
least tacitly, is a version of metaphysical antirealism. They say, for example, “Of
course ‘All humans are mortal’ is about the world – it is about humans.” They over-
look the fact that while what the sentence says depends on more than the logical
expressions “all” and “are,” it does depend on them. Its truth-value depends on
both the nonlogical and the logical expressions in it.
In contrast, the statements of logic employ only logical expressions. If logical
expressions stand for nothing, then the statements of logic have no extralinguis-
tic significance – a consequence commonly, though rashly, accepted. But all cog-
nition beyond that of babes, not just the cognition expressible in logicians’ state-
ments, must be expressible with statements employing expressions such as the
verb “to be” (especially in its sense of predication), the adverb “not,” sentential
connectives like “or,” and the quantifiers “all” and “some,” even if, unlike the
statements of logic, they employ also nonlogical, descriptive, expressions. All
developed human cognition is logical, in this broad sense. All statements, not
just the statements of logic, employ logical expressions. A language that lacked
them would be either in principle untranslatable or as primitive as baby “talk.”
Therefore, no developed human cognition is possible without language. This
does not entail but is a good reason for holding that the world, insofar as it is
the object at least of developed cognition, cannot have a character fundamental-
ly different from what we humans take it to have.
What I mean by developed cognition need not require higher education – it
would include the cognition expressed by “All sheep eat grass.” If a shepherd
did not know that all sheep eat grass, there would be precious little the shepherd
knew. If we could not say that all sheep eat grass, there would be precious little
we could say. In chapter 8 we shall see that we are more likely to make the ge-
neric statement “Sheep eat grass,” and then the antirealist implications of our
reliance on generality would be even more striking because obviously there
are no facts that are just generic. For now, however, I shall remain within the fa-

Authenticated
miliar context of universal statements. In that context, while the words “sheep,”
“eat,” and “grass” do stand for things in the world, “all” clearly does not. There
is no such entity in the world as all. It would not be an individual thing, nor a
property or relation of an individual thing. This is why the phrase “entity in the
world as all,” which I just used, is a grammatical outrage. It follows that there
are also no distinctive, universal, facts to which statements making essential
use of “all” might correspond. The sentence “All sheep eat grass” stands for
nothing distinctive in the world.
If we take advantage of the familiarity in philosophy of Kant’s term “tran-
scendental,” we may say that logical concepts and expressions are transcenden-
tal. They are essential at least to any developed cognition of the world, but stand
for nothing in the world. This would be a way of stating the thesis of logical anti-
realism. It would hardly be news. Philosophers and grammarians have common-
ly called the logical expressions syncategorematic. But they have offered little ex-
planation or argument, and few have seen the metaphysical implications.
4 Frege and Russell on Negation and Generality
I shall take the words “not” and “all” as paradigms of logical expressions. They
are fundamental not only to logic but all developed cognition. They express, re-
spectively, the logical concepts of negation and generality (more accurately, uni-
versality). In this chapter I shall offer a preliminary discussion of both, mostly in
relation to Frege’s and Russell’s classic defense of logical realism regarding
them. Wittgenstein’s logical antirealism/semirealism, his “fundamental thought”
that the logical constants do not represent, was mainly intended as a rejection of
their views. Chapter 7 will be devoted to a fuller discussion of generality, but in
relation to Wittgenstein’s and Bergmann’s classic defense of logical antirealism,
or at least of logical semirealism. Both the thesis of logical realism and the thesis
of logical antirealism would be sufficiently established if made plausible with re-
spect to negation and generality.
Regarding generality, Frege wrote, “It is surely clear that when anyone uses
the sentence ‘all men are mortal’ he does not want to assert something about
some Chief Akpanya, of whom perhaps he has never heard.”¹⁸⁵ The sentence
is not equivalent to the conjunction of its singular substitution instances, one
of which might be “Chief Akpanya is mortal.” Frege classified what “all” express-
 Peter Geach and Max Black, Translations from the Philosophical Works of Gottlob Frege (Ox-
ford: Blackwell, 1970) 83.

Authenticated
4 Frege and Russell on Negation and Generality 123
es as an “unsaturated” entity that is a second-level function, “saturated” by first-

level functions like those expressed by “x is a man” and “x is mortal,” which in
turn are saturated by individual objects such as Chief Akpanya.¹⁸⁶ A function is
unsaturated, or incomplete, in that it requires “attachment” to something else,
much as the grammatically incomplete expression “is human” requires attach-
ment to a grammatically complete expression such as “Chief Akpanya.” For ex-
ample, in the thesis of materialism, “("x)(x is material),” i. e., “Everything is ma-
terial,” the universal quantifier “"x” represents a second-level function, and “x
is material” represents the first-level function that completes it. The latter func-
tion is also incomplete, but what completes it is an individual thing, say, this
page. Frege’s terminology is awkward, but what matters for our purposes here
is his explicit acknowledgement that there is an entity to which the word “all”
corresponds. To be sure, it is an “incomplete” entity, but so is also the property
of being human. Frege called a first-level function a concept (Begriff), but like
many of his contemporaries he did not mean by “concept” anything mental.
He meant what “property” means in current philosophy.
Frege also held that there are negative thoughts, in his technical sense of
“thought” I explained earlier, since “for every thought there is a contradictory
thought” which “appears as made up of that thought and negation,” though
not as “mutually independent” parts. “The thought does not, by its make-up,
stand in any need of completion; it is self-sufficient. Negation on the other
hand needs to be completed by a thought. The two components are…quite differ-
ent in kind….One completes, the other is completed. And it is by this completion
that the whole is kept together.”¹⁸⁷ Negation is incomplete, just as the second-
level function expressed by “all” is incomplete. The difference is that negation
is completed by a complete entity, viz. a thought, while a second-level function
is completed by an incomplete entity, a first-level function. But negation is not
something mental or subjective. It is not “the act of denial.”¹⁸⁸ Nor is it a kind
of judging: “Negation does not belong to the act of judging, but is a constituent
of a thought.”¹⁸⁹ It is an objective part of a no less objective entity, viz. a negative
thought. The sentence “It is not the case that this page is blue” consists of two
parts: “it is not the case that” and “this page is blue.” The later could stand by
itself, the former could not. But what the whole sentence says is no more psycho-
logical or subjective than what “This page is blue” says.
 “Function and Object” and “Concept and Object,” both in Beaney.
 Beaney, 358.
 Beaney, 358.
 Beaney., 363.

Authenticated
Russell too was a realist about generality and negation, though he argued for
the existence of general and negative facts, not Fregean thoughts. Regarding
generality, he wrote, “When you have taken all the particular men that there
are, and found each one of them severally to be mortal, it is definitely a new
fact that all men are mortal.” For, “In order to arrive [by “complete induction”]
at the general statement ‘All men are mortal’, you must already have the general
statement ‘All men are among those I have enumerated.’” True general state-
ments, such as “All men are mortal,” stand for “general facts.” Therefore,
“there are general facts.” Moreover, because “You cannot ever arrive at a general
fact by inference from particular facts, however numerous,” “there must be prim-
itive knowledge” of some general facts.¹⁹⁰ There is therefore “the necessity of ad-
mitting general facts.”¹⁹¹
Russell also argued that “there are negative facts” and that “negativeness is
an ultimate.”¹⁹² He wrote, “Usually it is said that, when we deny something, we
are really asserting something else which is incompatible with what we deny. If
we say ‘roses are not blue,’ we mean ‘roses are white or red or yellow.’ But such a
view will not bear a moment’s scrutiny… The only reason we can deny ‘the table
is square’ by ‘the table is round’ is that what is round is not square. And this has
to be a fact, though just as negative as the fact that this table is not square.”¹⁹³
In discussions of our topic one often finds, in addition to logical realism and
logical antirealism, the supposedly third option of logical reductionism. For ex-
ample, in the case of universal statements, the realist holds that the universal
quantifier “"x,” or the word “all,” stands for a real entity, whether a Fregean sec-
ond-level function or some other kind of logical object. The antirealist denies
this. The reductionist, also finding the reality of such an entity implausible, “an-
alyzes” universal statements as the conjunctions, and particular statements as
the disjunctions, of the singular statements that instantiate them, claiming
that this is what they “really” say. Of course, the word “and,” or the symbol
“•,” which is essential to a conjunction, and the word “or,” or the symbol “v,”
are also logical expressions, and the reality of any logical objects they might
stand for is no less dubious. In any case, universal and particular statements
are not equivalent to such conjunction and disjunctions. Gustav Bergmann
took this to be obvious: “What can be said with the quantifiers cannot be said
 The Philosophy of Logical Atomism, 101– 103. Also in R. C. Marsh, ed., Logic and Knowledge
(London: Allen & Unwin, 1956), 235 – 236.
 Marsh, 289.
 Marsh, 215 – 216.
 Marsh, 288. Italics in original.

Authenticated
without them….Consider (1) ‘(x)G(x)’ and (2) ‘G(a1). G(a2)… G(aN).’ (1) implies (2).
(2) does not imply (1).”¹⁹⁴ \
Sometimes the reductionist appeals implicitly to the very same concepts that
are to be explained, and thus is no longer properly called “reductionist.” This is
case with the claim that the sentential connectives are “defined” by the corre-
sponding truth-tables and that this is “all there is” to what they mean. For exam-
ple, not-p is said to be “merely a truth function” of p, on the grounds that not-p is
true if and only if p is false. But we have no grasp of falsity except as the neg-
ation of truth, regardless of what theory of truth we hold, and thus the reduction-
ist is implicitly appealing to negation when offering that explanation of nega-
tion. As Russell remarked, it is “extremely difficult to say what exactly
happens when you make a positive assertion that is false, unless you are
going to admit negative facts.”¹⁹⁵ In his earlier article “On Denoting” Russell
had claimed that a universal statement such as (x) ϕx means that ϕx is always
true, and that ($x)ϕx means that ϕx is sometimes true. He pointed out repeatedly
that “always true” and “sometimes true” express primitive, indefinable notions.
But, if so, we could say the same about “all” and “some,” and avoid puzzling
attempts to say what they “mean.”
Reductionism is often little more than a front for antirealism (for example,
the “reduction” of mental states to brain states is little more than the denial
of the reality of mental states). But, as we just saw, in the case of logic it
seems to merely replace some logical statements with other logical statements,
e. g., universal statements with conjunctions of singular statements even though
a realist view of the sentential operators, including conjunction, is no more plau-
sible than a realist view of the quantifiers.
This is often overlooked because of the common assumption that necessary
equivalence is identity. If you are more comfortable with one of the sides of the
equivalence, you may be tempted to suppose that what the other side says is
“really nothing more” than what the side you prefer says. But, clearly, this is a
mistake. It is necessarily the case that p if and only if p or p, but it would be
silly, or at lest unmotivated, to speak of reducing either one to the other, of say-
ing for example that what p says is really nothing more than what p or p says. In
mathematics, to use Kant’s example, it is necessarily the case that 7+5=12, but we
would not want to say that one side of the identity sign is reducible to the other.
 Gustav Bergmann, “Generality and Existence,” in Logic and Reality (Madison: University of
Wisconsin Press, 1964), 69. Originally published in Theoria, 28 (1962), 1– 26.
 Marsh, 214.

Authenticated
Frege’s and Russell’s views on generality and negation were explicit espous-
als of both logical realism and logical nonreductionism. According to Russell
general and negative facts are objective constituents of the world, not reducible
to any other kinds of facts. It would seem that what makes them general or neg-
ative must also be in the world, though Russell vacillated about the matter. Frege
did not. He was explicit: general thoughts involve a second-level function, neg-
ative thoughts include negation as one of their two components. One would
think that Russell’s general and negative facts must also contain components
represented, respectively, by the quantifier and the negation sign. For, if they
did not, how would general facts differ from the corresponding singular facts
and how would negative facts differ from the corresponding positive facts? If
there is an item in the world represented by “"x (Φx)” as well as an item repre-
sented by “Φa,” as Russell held, how would the two differ if not in virtue of
“something” in one that is not in the other, presumably the presence in what
“"x (Φx)” represents of what “"x” represents? Similarly, if there is an item in
the world represented by “~ p,” as well as an item represented by “p,” how
would they differ if not in virtue of “something” in one of them that is not in
the other, presumably what “~” represents? For, according to Russell, the differ-
ences must be in the facts, in the world, not in our language or our minds. Ac-
cording to Frege, the presence of a second-level function in general thoughts is
essential to the truth-values of general sentences and the presence of negation in
negative thoughts is essential to the truth-values of negative sentences. Surely,
the presence in Russell’s general facts of something expressed by the quantifier
is essential to the truth-values of general sentences, and the presence of negation
in Russell’s negative facts is essential to the truth-values of negative sentences.
If, as Russell held, the truth of a statement depends on what is in the world, then
what is essential to its truth presumably must also be in the world.
The logical realist faces two tasks. The first, perhaps accomplished by Frege and
Russell, is to refute logical reductionism, the view that “upon analysis” the question
of the reality of logical objects does not arise because the logical constants that ap-
pear to stand for them have been “analyzed away.” By showing that at least in the
case of negation and generality this is not so Frege and Russell forced antirealists to
shun the comfort of slogans like “A negative statement is just the corresponding
false affirmative statement,” “Universal statements are just the conjunctions, and
particular statements the disjunctions, of the singular statements that instantiate
them,” or “Molecular statements are just truth-functions of their components.”
The second task the logical realist faces, however, is to convince us that, since log-
ical reductionism is false, the logical constants do represent entities in the world.
The plausibility of Frege’s and Russell’s views attached to their nonreductionism,
not to their realism, which remained implausible.

Authenticated
The idea that “all” and “some” correspond to entities has seldom been enter-
tained. As we shall see in Chapter Seven, Gustav Bergmann, one of the few philos-
ophers who thought deeply about this topic, did write: “Each quantifier represents
something which is sometimes presented. Had it never been presented, we would
not know what the quantifier meant.”¹⁹⁶ But this is an (abductive) argument for a
statement of phenomenological observation, a statement about what is “presented,”
not a report of that phenomenological observation, which is what it should have
been, if Bergmann were right. And surely such a report would have been false.
But Bergmann at least was aware of what would be necessary if logical realism
is to be defended. We shall see in Chapter Seven that so was Wittgenstein, and
that in his later work Bergmann revised his uncompromisingly realist stand.
Even if Frege’s and Russell’s arguments against reductionism in the case of
negative sentences are accepted, antirealism about negation would remain com-
pelling.¹⁹⁷ That negation corresponds to nothing in the world seems almost a tau-
tology. Sartre claimed that it is consciousness that “introduces” negation into the
world precisely because he believed that it is not “already” in the world.¹⁹⁸ The
words “no” and “not” are learned early in childhood to signal the absence or
nonexistence of a thing, to refuse an offered object, or to reject what one is
told, and later to deny the truth of what one hears or reads. But surely they
do not stand for any entity such as absence or nonexistence. Antirealism
about generality would also remain compelling, for much the same reasons.
In denying that logical expressions stand for entities, the logical antirealist
is not just denying an application of the simplistic “Fido”– Fido principle, ac-
cording to which every word is a name. What is denied is the natural, not at
all simplistic even if ultimately mistaken, assumption that if a word is to serve
a cognitive role then it must relate to something in the world in a specific, dis-
tinct, and comprehensible way, even if it does not name it, that there must be
something in what is cognized that grounds its cognitive role even if it is not
named by the word, and that this “something” is accessible to us, if not directly
in perception, like colors and books, then indirectly in sophisticated thought,
like quarks and perhaps God. The logical antirealist claims that in the case of
logical expressions none of this is true.
This claim should not be confused with the much weaker claim, to which chap-
ter 5 was devoted, that language is causally necessary for cognition. Presumably, a
human being who lacks a language cannot have detailed knowledge of astronomy
 Gustav Bergmann, Logic and Reality (Madison: University of Wisconsin Press, 1964), 70.
 But see E. Peterson, “Real Logic in Philosophy,” The Monist 60, 2, 1986, and “Logic Knowl-
edge,” The Monist 72, 1, 1989.
 Jean-Paul Sartre, Being and Nothingness, 21– 45.

Authenticated
or of medieval history. But surely God can, and for all we know so can intelligent
life-forms in outer space. We may not understand what their knowledge would be
like, but then neither can we visualize a non-Euclidean space or have auditory im-
ages of the high-frequency sounds that dogs but not humans can hear. The same
may be said about humdrum cognitions like that expressed in an inventory of
the thousands of chairs in a large building. We cannot make the inventory without
using language or other symbols. But, surely, God can.
Could the thought expressed in a universal statement like “All my toys are
upstairs” be entertained by a child before learning the word “all”? Perhaps it
could, if it were just a collection of several singular thoughts, e. g., that the
doll is upstairs, that the ball is upstairs, and that the whistle is upstairs. But a
mere collection of thoughts is not even what a conjunctive statement, a state-
ment requiring the sentential connective “and.” And even if it were, it certainly
is not what the universal statement, the statement employing “all,” expresses. As
Russell would have pointed out, the conjunction must include as additional con-
junct also the universal statement “These are all the toys I have.” A universal
statement is “made true” not just by the “atomic facts” corresponding to its sin-
gular substitution instances, but also by “the further fact about the world that
those are all the [relevant] atomic facts…[which] is just as much an objective
fact about the world as any of them are.”¹⁹⁹
The linguisticism of the position we have reached so far regarding negative
and universal statements appears obvious. It does fit the usual formulations of
linguistic antirealism. But these formulations make no distinction between onto-
logical and cosmological realism/antirealism, much less between metaphysical
and logical realism/antirealism or between antirealism and semirealism. For ex-
ample, Michael Dummett writes, “[U]ntil we have achieved an understanding of
our language, in terms of which we apprehend the world, and without which,
therefore, there is for us no world, so long will our understanding of everything
else be imperfect.”²⁰⁰ To attempt to “strip thought of its linguistic clothing,” he
also says, is to confuse it with “its subjective inner accompaniments.”²⁰¹ Dum-
mett may be right, but surely his judgment is rashly abstract. As we have
seen, there are important differences in respect of our “apprehending” them be-
tween the world and plants, animals, or minerals, between this book and all
books, between this page and its being white, and as we shall see in Chap-
 Marsh, 235 – 36.

 Michael Dummett, “The Justification of Deduction,” in Truth and Other Enigmas, 311.
 Truth and Other Enigmas, 44.

Authenticated
ter Eight between winters in Iowa being severe and all winters in Iowa being se-
vere.
Cognition of the world involves confrontation with nonlinguistic items, in sense
perception, introspection, perhaps also intellectual and even mystical intuition.
Such confrontation need not involve language. A neonate’s exercise of normal
sight does not. We need not deny the independent reality of the items thus confront-
ed. Thus we need not deny that we have cognition that is unmediated by language.
But it would be cognition at its most primitive level. Since it would not involve state-
ments, it could not even appear to have facts as its objects and therefore, if the
world is the totality of facts, to count as cognition of a world.

Authenticated
Chapter Seven: Logical Semirealism
1 Ineffability
In the previous chapter I offered a preliminary account of logical antirealism with

respect to negation and generality. One need not be a metaphysical antirealist to
doubt their existence as entities in the world. Distaste for such entities has led to
reductionism, the attempt to show, often in the context of the realism/antirealism
debate, that negative and general statements do not really say what they seem to
say. In the previous chapter we saw how Frege and Russell dealt with this attempt.
Reductionism is far less plausible in the case of “not” and “all” than in the case of
sentential connectives like “and,” “or,” and “only if.” Later analytic philosophers
provided little detailed discussion of negation, leaving that task to phenomenolo-
gists like Heidegger and Sartre. Surprisingly, they also had little to say also about
generality. An exception in this respect was Wittgenstein. So was Gustav Bergmann.
I shall focus in this chapter on generality, because of its obvious, direct, relevance to
cosmological antirealism.
Wittgenstein usually called universal statements just “general,” as Russell
had done. Sometimes, so did Bergmann. In discussing their views, I will too.
All three took for granted that particular statements are equivalent to negations
of negative universal statements (($x) Fx ≡ ~ ("x) ~ Fx)), and thus to need no
separate discussion, though as we shall see Bergmann vacillated on this point
in his early work. None considered the status of generic statements, statements
like “Winters in Iowa are severe” and “Patients with prior strokes benefit from
taking Lipitor,” the ubiquity of which provides the case for antirealism with de-
cisive support. Generic statements will be the topic of chapter 8. The present
chapter is devoted to universal statements, statements like “All winters in
Iowa are severe” and “All patients with prior strokes benefit from taking Lipitor.”
The logical expressions for generality are of special relevance to cosmolog-
ical antirealism, antirealism about the world. Antirealism regarding the other
logical expressions may lead also to ontological antirealism, antirealism about
things. But I shall not consider this possibility here. I want to present here a ver-
sion of antirealism that is measured, discerning, and as cautious and uncontro-
versial as possible. Cosmological antirealism is such a version. It questions only
the reality of the world understood as the totality of facts. It does not question
the reality of the world understood as the totality of things. It does not question
the reality of “animals, vegetables, or minerals,” and thus it remains safely in
accord with common sense, which would find Wittgenstein’s notion of the
word as the totality of facts, not of things, too philosophical to fathom.

Authenticated
1 Ineffability 131
A serious discussion of general statements must focus on what Wittgenstein

and Bergmann said about them. With the exception of Frege and Russell, considered
in the previous chapter, there is little of metaphysical import to be found elsewhere.
According to Bergmann, the universal and the particular quantifiers stand for items
with which we are sometimes “presented,” i.e., of which we are sometimes aware.
Few have agreed with him. Few had agreed even with Russell’s weaker claim that at
least there are general facts, of some of which we have primitive knowledge. Berg-
mann modified his view later, moving closer to the position Wittgenstein had held.
Though adamant in opposing Russell’s logical realism, Wittgenstein did not merely
deny that there are general facts. He also offered a positive view, which Bergmann’s
later view resembled in important respects. Both represented, with unparalleled so-
phistication, the middle ground between simplistic realism and simplistic antireal-
ism that I have called semirealism.
Of course, general statements had been a central subject matter of logic since
Aristotle’s syllogistic. But when Frege invented modern quantification theory, they
became also a fundamental concern of metaphysics. Contemporary logicians and
metaphysicians seldom ask what, if anything, general statements correspond to
in the world. But, as we saw in the previous chapter, the founders of analytic phi-
losophy, Frege and Russell, did, and the question was a major theme also in Witt-
genstein’s and Bergmann’s works. All five – Aristotle, Frege, Russell, Wittgenstein,
and Bergmann – were aware that there could be no genuine world if there were
not generality in the world. This is why all five also saw the tie of the topic to
what later became known as the realism/antirealism issue, even though for Aristotle
generality involved universals, for Frege general “thoughts,” and for Russell, Witt-
genstein, and Bergmann general “facts” or “complexes.”
Frege held that general statements express the saturation of second-level
functions by first-level functions; Russell, that if true they stand for general
facts; Wittgenstein, that they involve matters that can only be “shown,” not
“said”; and Bergmann, that they involve the entities generality and existence,
for which the universal and the particular quantifiers respectively stand.
Like Frege and Russell, Bergmann rejected the facile reductionist view of
universal statements as the disguised conjunctions, and of particular statements
as the disguised disjunctions, of their singular substitution instances. In his early
paper “Generality and Existence” he argued that, like “individuality, universality,
and exemplification,” generality and existence belong to the “world’s form.”
One is “presented” with them, but they do not “exist” – rather, they “subsist.”²⁰²
 Gustav Bergmann, “Generality and Existence,” in Logic and Reality (Madison: University of
Wisconsin Press, 1964), 69. Originally published in Theoria, 28 (1962), 1– 26.

Authenticated
132 Chapter Seven: Logical Semirealism
Thus Bergmann already seemed inclined toward only semirealism regarding gen-
erality, not realism though also not antirealism. He used “existence” in two
senses: the sense of “some,” which the particular quantifier expresses, and
the sense of “exist” in which the world’s form does not exist. Later, he expressed
regret over the ambiguity. It is absent from the posthumously published New
Foundations of Ontology, where Bergmann’s views received their most developed
and detailed formulation.
“Generality and Existence” had been preceded by “Ineffability, Ontology, and
Method.”²⁰³ Bergmann described the two articles as “materially one.” The first
topic of “Ineffability, Ontology, and Method” was the “ineffability” of individuality
and exemplification. He wrote, “When I know that this is a green spot, I know also
that (1) the spot is an individual, (2) the color is a character, and (3) the former ex-
emplifies the latter (and not, perhaps, the latter the former). How could I know all
this if it were not, in some sense, presented to me?”²⁰⁴ But what was thus presented
could not be also represented, at least not without futility. “Looking at a name…I
know…even if I do not know which thing it has been attached to as a label…the
kind of thing, whether individual or character, to which it has been or could be at-
tached.”²⁰⁵ Saying that it is an individual or that it is a character would tell one
nothing, for it presupposes what it purports to say. Its even having sense depends
on its being true. Yet the sentence is not gibberish.
Bergmann went on to say that a certain name “is on the lips of every likely read-
er,” but that he would not mention it because he did not “on this occasion wish to
make assertions about the reading of a notoriously difficult text.”²⁰⁶ The name of
course was Wittgenstein’s, and the text was Tractatus Logico-Philosophicus. Wittgen-
stein had written, “In a certain sense we can talk about formal properties of objects
and states of affairs… It is impossible, however, to assert by means of propositions
that such internal properties and relations obtain: rather, this makes itself manifest
in the propositions that represent the relevant states of affairs” (4.122). As we saw in
Chapter Four, by “formal” or “internal” property Wittgenstein meant such properties
as being an object and being a fact, “external property” being reserved for what
Bergmann meant by “character,” e.g., color or shape. Statements about an object
say what external properties it has. Formal properties, however, cannot be properly
predicated, “said,” though they can be shown.
The similarity of Bergmann’s views in “Ineffability, Ontology, and Method” and
“Generality and Existence” to Wittgenstein’s views in the Tractatus was obvious, as
 Philosophical Review, 69 (1960), 18 – 40, also included in Logic and Reality.
 Logic and Reality, 47.
 Logic and Reality, 49 – 51.
 Logic and Reality, 50.

Authenticated
2 Wittgenstein on Generality 133
Bergmann readily acknowledged. It involved, however tacitly, Wittgenstein’s distinc-

tion between “saying” and “showing,” which Wittgenstein later called the main con-
tention in the Tractatus since it concerned “the cardinal problem of philosophy.” In
Chapter Four I noted that some interpreters, for example, Cora Diamond and Warren
Goldfarb, deny that according to the Tractatus there is anything that can be shown
but cannot be said. I pointed out, however, that Wittgenstein held that what only
shows itself is “the higher” (das Höhere). Surely, he did not think that the higher
is nothing. To say that Socrates is an individual, rather than, say, a property or a
relation, is not to make a remark about Socrates’s properties, but neither is it to
say nothing. To speak of the sense of life is not like speaking of the length of
human life, but it is hardly speaking of nothing.
The reception of Wittgenstein’s and of Bergmann’s views in the philosophi-
cal community were similar, perhaps because both referred to the ineffable. Both
dealt with metaphysical questions that few philosophers had even considered,
and offered answers of which no previous philosophers had been aware. Berg-
mann’s critics complained that his philosophy is a Meinongian jungle, or they
just avowed that they found it “too difficult.” The critics of Wittgenstein’s Trac-
tatus disparaged it as “too metaphysical,” or they just interpreted it in terms of
the much later Philosophical Investigations – they found misery in Wittgenstein’s
glory, and glory in Wittgenstein’s misery, as Bergmann sarcastically put it.²⁰⁷
The ineffable, what only shows itself, is not nothing. It does show itself. One
is inclined to ask, If it is not nothing, then what kind of being or reality does it
have, what is its “ontological status”? Natural though this question may be, it
presupposes the sharp and unilluminating distinction between what is real
and what is not real, which Wittgenstein attempted to bypass with his distinction
between saying and showing. I have already discussed the distinction in other
contexts, and here I shall consider the details of how it affected his account
of generality in the Tractatus and later works.
2 Wittgenstein on Generality
In Tractatus 5, Wittgenstein wrote that “A proposition is a truth-function of ele-

mentary propositions.” He had explained that “The simplest kind of proposition,
an elementary proposition, asserts the existence of a state of affairs” (4.21), and
that “It is obvious that the analysis of propositions must bring us to elementary
 Bergmann used these terms in “The Glory and the Misery of Ludwig Wittgenstein,” Rivista
di Filosofia, 52, 1961, 587– 606, Italian translation, included in Logic and Reality.

Authenticated
propositions…” (4.221). (In the Introduction to the Second Edition of Principia

Mathematica, Russell wrote that “Atomic and molecular propositions together
are ‘elementary propositions’.”²⁰⁸ It seems to follow that a general proposition
is also a truth-function, presumably of its singular (“elementary”) substitution
instances. And so, in a letter to Wittgenstein written in 1919, Russell objected:
“[In an account of general (universal) propositions in terms of elementary prop-
ositions,] it is necessary also to be given the proposition that all elementary prop
[ositions] are given.”²⁰⁹ As we saw in the previous chapter, Russell had argued
that, for example, the universal proposition “All men are mortal” is not equiva-
lent to “Socrates is a man and Plato is a man and Aristotle is a man and…,” un-
less another universal proposition such as “All men have been enumerated” or
“These are all the men there are” is added to the conjunction. If “All men are
mortal” is translated as “(x) (if x is a man then x is mortal),” which seems to
say that every object is such that if it is a man then it is mortal, Russell’s objec-
tion would be that it is not equivalent to the conjunction of all singular state-
ments of the form “if x is a man then x is mortal” unless the proposition that
all objects have been enumerated or taken into account is added to the conjunc-
tion. The thesis of materialism, “(x) (x is material),” would not be equivalent to
the conjunction of all statements of the form “x is material,” unless we added
that we have considered all objects, all values of the variable x.
Wittgenstein’s reply to Russell’s objection was vehement. He did not deny Rus-
sell’s logical point, which amounted to reminding us that the rule of universal in-
stantiation is not a rule of equivalence. Rather, Wittgenstein wrote, “There is no
such proposition [as that all elementary propositions are given]! That all elementary
propositions are given is shown by there being none having an elementary sense
which is not given…” And he added: “I’m afraid you [Russell] haven’t really got
hold of my main contention, to which the whole business of logical prop[osition]s
is only a corollary. The main point is the theory of what can be expressed [gesagt,
said] by prop[osition]s – i.e., by language – (and, which comes to the same, what
can be thought) and what can not be expressed by prop[osition]s, but only shown
(gezeigt)”; which, I believe, is the cardinal problem of philosophy.”²¹⁰
Wittgenstein in effect pointed out the implications of his distinction between
saying and showing for the notion of the totality of elementary propositions. As
 Alfred North Whitehead and Bertrand Russell, Principia Mathematica to *56 (Cambridge:
University Press, 1962), xvii.
 Russell: the Journal of the Bertrand Russell Archives 10, 2, 107– 109.
 Letters to Russell, Keynes and Moore, 71. In Principia Mathematica (Part I, Summary), an el-
ementary proposition is defined as one that “contains no reference, explicit or implicit, to any
totality.”

Authenticated
we saw in chapter 4, the distinction is especially clear when applied to formal or

internal properties, such as being an object. These properties can only be shown.
This is why we cannot speak of the “totalities” determined by them: it is nonsen-
sical to speak of the total number of objects or facts, because “object” and “fact”
signify formal concepts (4.1272). If “fact” is a formal concept, surely so is “prop-
osition,” and therefore also “elementary proposition,” since propositions are log-
ical pictures of facts
There is no such proposition as Russell’s “all elementary propositions are
given” because “proposition” is a formal concept, like “object” and “fact.” In his ob-
jection, Russell seemed to take for granted what later was called the “substitutional”
interpretation of quantification, according to which, roughly, general statements
refer to all the elementary statements that are their substitution instances. According
to the alternative, more intuitive, “objectual” interpretation of quantification, gener-
al statements refer to all values of the individual variable they contain, that is, all
objects.²¹¹ Whether the two interpretations in fact require such reference is a debat-
able question, but we need not consider it here. What matters is that even if Russell
had taken for granted the objectual interpretation, Wittgenstein’s reply would have
been similar. Russell’s objection would have been that the proposition “all objects
are given” must be given, and Wittgenstein would have replied that there is no
such proposition because “object” signifies a formal concept.
This has to be understood with some care. It does not mean that there are no
universal propositions. What is meant is that a universal proposition is not about
propositions or about objects, and thus does not say anything about all proposi-
tions or all objects, because, “proposition” and “object” being formal concepts,
no proposition could be about propositions or objects. For example, the logically
harmless “(x) (x is material)” must not be confused with “(x) (if x is an object
then x is material),” which employs the formal concept “object” and therefore
says nothing. Indeed, this is readily appreciated in common discourse. It is
how we ordinarily understand universal propositions. We find it natural to say
that “All men are mortal” is about all men, not, as logic suggests, that it is
about all individual things, including the moon, my pen, your feet, etc., saying
about each that if it is a man then it is mortal. Saying the latter belongs only in
the classroom. “All men are mortal” does not include the predicate “is an ob-
ject,” the only predicates in it are “is a man” and “is mortal.” If stated as “(x)
(if x is a man then x is mortal),” the only expression in it other than those pred-
 The classic discussion of the two interpretations of quantification is Ruth Barcan Marcus’s,
in “Interpreting Quantification,” Inquiry 5, 1962, 252– 259.

Authenticated
icates would be the variable “x,” which is neither a name nor a predicate. It is,
rather, “the proper sign for the pseudo-concept object” (4.1272).
Wittgenstein’s reply to Russell appealed, not to some “analysis” of general
propositions that was different from Russell’s, but to the formal status of the prop-
erty of being a proposition, and thus of the property of being an elementary prop-
osition. Like Russell, he refused to analyze a universal proposition as the conjunc-
tion, the truth-functional product, of its singular substitution instances: “I dissociate
the concept all from truth-functions…” (Tractatus 5.521). On the other hand, he did
not deny that, “in a certain sense,” a universal proposition is equivalent to that con-
junction: “Indeed the understanding of general propositions palpably depends on
the understanding of elementary propositions” (4.411).
Are 4.411 and 5.521 consistent? The answer is provided by the three sentences
that immediately follow 5.521: “What is peculiar to the generality-sign is first,
that it indicates a logical prototype, and secondly, that it gives prominence to
constants” (5.522), “The generality-sign occurs as an argument” (5.523), and “If
objects are given, then at the same time we are given all objects. If elementary
propositions are given, then at the same time all elementary propositions are
given…” (5.524). The occurrence of the variable “x” in “(x) Fx” “gives” objects
in the sense that the variable ranges over objects – it is objects that it admits
as values, rather than, say, properties or propositions. The generality-sign in
“(x) Fx” is “(x) (…x),” the form of “(x) Fx” in virtue of which it is a general prop-
osition.²¹² It may indeed be said to indicate a “logical prototype.” And it may be
said to occur in the proposition as an argument because it is part of what deter-
mines the truth-value of “(x) Fx,” just as a name put in place of the variable x in
Fx would be part of what determines the truth-value of the resulting singular
proposition. The statement “(x) fx” is a truth-function of its substitution instan-
ces, not because it is replaceable by the conjunction “fa . fb . fc ….,” but because
its truth depends, in a straightforward, literal, sense, on their truth. But this only
shows itself. It is not and cannot be said.
If the variable “x” “gives” objects in the sense that it takes objects as its val-
ues, the generality-sign “(x) (…x)” may be said to give all objects. A universal
proposition is thus, “in a certain sense,” about all objects (all values of “x”),
but this is not something it says, it is something it shows. The claim that
“There are objects” is a pseudo-proposition does not mean that there are no uni-
versal propositions. For such propositions are not about objects, even though the
 G.E.M. Anscombe (Introduction to Wittgenstein’s Tractatus [London: Hutchinson University

Library, 1959], 145) and Robert Fogelin (Wittgenstein [London: Routledge, 1987, second edition],
65) think that the generality-sign in “(x) Fx” is the variable x.

Authenticated
variable in them is understood as ranging over objects. They do not contain the
predicate “x is an object.”
This is why a general proposition might be said to “refer” to all objects, if we
accept the objectual interpretation, or to all elementary propositions, if we ac-
cept the substitutional interpretation. But such “reference” would consist in
showing, not saying. Since “object” is a formal concept, “(x) fx” does not say
that all objects are f. Nor does it say that all elementary propositions of the
form “fx” are true, since “elementary proposition” is also a formal concept.
All propositions of the form “fx” may be said to be shown by that form, in the
straightforward sense that it is their form – this is why “(x) fx” is a general prop-
osition. In a no less straightforward sense, the truth of “(x) fx” depends on there
not being a proposition of the form “fx” that is false. But “(x) fx” does not say
that there is no proposition of the form “fx” that is false. It is not a proposition
about propositions. Since “proposition” is a formal concept, there can be no
proposition about all propositions of the form “fx” or of any other form.
One of Russell’s complaints in his letter to Wittgenstein was that “it is awk-
ward to be unable to speak of [the negation of all the values of the propositional
variable ].” Wittgenstein replied: “This touches the cardinal question of what can
be expressed by a prop[osition] and what can’t be expressed, but only shown. I
can’t explain it at length here. Just think that, what you want to say by the ap-
parent prop[ositin] ‘there are 2 things’ is shown by there being two names
which have different meanings….e.g., φ(a, b)…doesn’t say that there are two
things, it says something quite different; but whether it’s true or false, it
SHOWS what you want to express by saying: ‘there are 2 things.’” Then Wittgen-
stein added: “I suppose you [Russell] didn’t understand the way, how I separate
in the old notation of generality what is in it truth-function and what is pure gen-
erality. A general prop[osition] is a truth-function of all PROP[OSITION]S of a cer-
tain form…I suppose you don’t understand the notation [for the values of the
propositional variable ]. It does not mean ‘for all values of .”²¹³
I suggest that “what is truth function” in “(x) fx” is what is expressed by “(x)
…,” and that “what is pure generality” is what is expressed by “fx.” All propo-
sitions of the form “fx” may be said to be shown by that form, in the straightfor-
ward sense that it is their form – this is the “pure generality” in “(x) fx.” And in a
no less straightforward sense, the truth of “(x) fx” depends on there not being a
proposition of the form “fx” that is false – this is the “truth function” in “(x) fx.”
But “(x) fx” does not say that there is no proposition of the form “fx” that is false.
It is not a proposition about propositions. Since “proposition” is a formal con-
 Letters to Russell, Keynes and Moore, 73.

Authenticated
cept, there can be no proposition about propositions of the form “fx” or of any
other form.
While the distinction between saying and showing has a reasonably clear
and important application to propositions of the forms “x is an object” and
“next-to is a relation,” how it applies to other, more complicated cases is less
clear though not less important. This is certainly true of its application to general
propositions. To understand it better, we may take advantage of the notion of
presupposition that Strawson proposed decades later and we examined in chap-
ter 2, and assume, at least provisionally, that presupposing something includes
implicitly reference to it. Then, if we accept the objectual interpretation of quan-
tification, we can hold that, though “(x) Fx” does not refer to objects and thus
does not say that all objects are F, it may be taken to presuppose such reference
and thus to refer to objects implicitly. It is “(x) (if x is an object then x is F),” not
“(x) Fx,” that explicitly mentions and thus refers to objects, and for this reason it
is inadmissible. The point may become clearer if we return to the statement “All
men are mortal.” Translated as “(x) (if x is a man then x is mortal),” it does not
say that all objects are such that if they are men then they are mortal, but it may
be taken to presuppose this since the variable x ranges over all objects and thus
to refer implicitly to all objects. What “All men are mortal” does say is just that
all men are mortal. It is about men, not about objects. Were we to accept the sub-
stitutional interpretation of quantification, we could hold that “(x) Fx” does not
say that all elementary propositions of the form “Fx” are true, but it can be taken
to presuppose this and thus to implicitly refer to all elementary propositions. The
actual statement “All men are mortal” does not say that all propositions of the
form “if x is a man then x is mortal” are true, but it may be taken to presuppose
this. What it does say is just that all men are mortal. It is about men, not about
elementary propositions.
After his return to Cambridge in 1929, Wittgenstein continued working on the
topic of generality. But what, according to the Tractatus, was shown by a general
proposition, whether universal or particular, in Philosophical Remarks, written
shortly after his return though published posthumously, was now described as
possibilities left open by the proposition, like an incomplete or partially indefi-
nite picture: “The general proposition ‘I see a circle on a red background’ ap-
pears simply to be a proposition which leaves possibilities open. A sort of incom-
plete picture. A portrait in which, e. g., the eyes have not been painted in. But
what would this generality have to do with the totality of objects?”²¹⁴ Also: “If
 Ludwig Wittgenstein, Philosophical Remarks (Chicago: University of Chicago Press, 1975, ed.

Authenticated
I give a correct description of a visual field in which three red circles stand on a
green ground, it surely won’t take the form of saying ‘($x (x, y, z): x is circular
and red and y is circular and red, etc. etc.’… That is, in such a case, you may
say: Certainly I know that three things have the property φ, but I don’t know
which; and you can’t say this in the case of the three circles.”²¹⁵
A couple of years later, in Philosophical Grammar, also published posthumous-
ly, Wittgenstein returned to the topic. Regarding the position he had defended in the
Tractatus, he wrote: “My view about general propositions was that ($x). φx is a log-
ical sum and that though its terms are not enumerated here, they are capable of
being enumerated….For if they can’t be enumerated we don’t have a logical
sum….Of course, the explanation of ($x). φx as a logical sum is indefensible….
[But] it is correct that ($x). φx behaves in some ways like a logical sum and (x).
φx like a product; indeed for one use of the words ‘all’ and ‘some’ my old explan-
ation is correct, – for instance for ‘all the primary colours occur in this picture’… But
for cases like ‘all men die before they are 200 years old’ my explanation is not cor-
rect.”²¹⁶ I take Wittgenstein to mean that the sentence about the primary colors
would be an exception because “primary color” is an abbreviation, say, of “red,
green, or blue,” and so the sentence would be an abbreviation of “red, green,
and blue occur in this picture,” not a general statement.
Wittgenstein also wrote, “If I say ‘there is a black circle in the square’, it al-
ways seems to me that here again I have something simple in mind, and don’t
have to think of different possible positions or sizes of the circle. And yet one
may say: if there is a circle in the square, it must be somewhere and have
some size. But in any case there cannot be any question of my thinking in ad-
vance of all the possible positions and sizes….I would like to say that in the prop-
osition ‘there is a black circle in the square’ the particular positions are not men-
tioned at all. In the picture I don’t see the position, I disregard it….”²¹⁷ When
asserting the general proposition “There is a black circle in the square,” one
does not mention the other possible positions and sizes the circle might have
had, one does not think of them, they are disregarded. Indeed, one does not
mention even the position it does have. Of course, the circle does have a posi-
tion, one of an indefinite number of possible positions, but none is mentioned
Rush Rhees, tr. Raymond Hargreaves and Roger White), 115. For the origin of the text, see the
Editor’s Note.
 Philosophical Remarks, 136.
 Ludwig Wittgenstein, Philosophical Grammar (Berkeley and Los Angeles: University of Cal-
ifornia Press, 1974, 268), composed after 1929 but published posthumously (for the origin of this
text, see the editor’s Note in Editing).
 Philosophical Grammar, 259.

Authenticated
in “There is a black circle in the square.” Wittgenstein did not explain these re-
marks in detail. Nevertheless, they do fit what he had said, not in the Tractatus
but in his 1919 correspondence with Russell: the possible positions of the circle
are not “said” but “show” themselves.
In Wittgenstein’s earlier terminology, which he no longer employed, we
might say that the possible positions of the circle are not “said” but “show”
themselves. According to Russellian logic, the universal statement “(x) Φx”
says that all individual objects are Φ, that everything is Φ, and that ($x). Φx
says that some individual objects are Φ, whichever they might happen to be.
But in the Tractatus Wittgenstein had held that “it is nonsensical to speak of
the total number of objects,” indeed of objects at all, since “object” is a formal
concept. Now, in Philosophical Grammar, he makes the further claim that general
statements are often, perhaps usually, not even intended to be used in accord-
ance with Russellian logic. The statement “There is a circle in this square” is
not intended to say anything about all objects, not even about all objects that
are in the square. In effect, Wittgenstein suggests that the general particular (“ex-
istential”) statement “There is a circle in the square” and the general universal
statement “There are only two things that are circles in this square” are better
understood as analogous to the singular statement “This circle is in this square,”
rather than to quantified statements containing a variable ranging over all indi-
vidual objects, or even over just all circles in the square. But how then do the
general statements differ from the singular statement?
We find no answer in Philosophical Grammar. The answer Wittgenstein had
provided in his 1919 reply to Russell was that the generality of a general state-
ment consists not in what it says but in what it shows. He insisted that our
use or understanding of general statements is far removed from what Russell’s
Principia Mathematica tells us. We do not use “There is a circle in the square”
to say something about all things, or even about all circles, viz., that some are
in the square. We certainly do not use it to say that it is not the case that no cir-
cles are in the square.
But In Philosophical Grammar Wittgenstein expressed doubts about the very
propriety of representing ordinary general propositions in the canonical forms of
Principia Mathematica. “[A] proposition like ‘there are two circles in this square’
is rendered as ‘there is no object that has the property of being a circle in this
square without being the circle a or the circle b’….[T]he Russellian notation
here gives an appearance of exactitude which makes people believe the prob-
lems are solved by putting the proposition into the Russellian form.”²¹⁸
 Philosophical Grammar, 265.

Authenticated
3 Bergmann on Generality 141
Wittgenstein’s misgivings in Philosophical Grammar about the Russellian in-

terpretation of general statements were motivated by his growing concern with
how we actually talk. In Remarks on the Foundations of Mathematics he wrote:
“‘Mathematical logic’ has completely deformed the thinking of mathematicians
and of philosophers, by setting up a superficial interpretation of the forms of
our everyday language as an analysis of the structure of facts.”²¹⁹ He was on
his way to the Philosophical Investigations. In the next chapter I will argue that
“mathematical logic” also ignores the fact that in actual talk and thought
most generalizations are what linguists call generic statements, i. e., statements
of the form “Fs are Gs,” rather than universal statements, of the form “All Fs are
Gs,” and that even universal statements are usually intended only as generic.
3 Bergmann on Generality
In “Generality and Existence” Bergmann had used an example similar to Witt-

genstein’s example in Philosophical Grammar: being “presented” with a square
inside a circle. Bergmann asked, With what else must I be presented when I
say “This square is the only one inside this circle”? He pointed out that the tran-
scription of the statement would be “F(a, b) . (x) [(x = a) v ~ F (x, b)],” with “(x),”
“a,” “b,” and “F” standing respectively for the universal quantifier, this square,
this circle, and the relation of being inside.²²⁰ Bergmann answered the question
by saying that when presented with a single square inside a circle he was also
presented with generality and existence. He was presented with generality in
seeing that the square was the only square in the circle, and with existence in
seeing that there is a square in the circle. They are the entities that the universal
quantifier, “(x)” or the word “all,” and the particular quantifier, “($x),” the word
“some,” or the phrase “there is at least one,” respectively represent. Therefore,
there are such entities as generality and existence, though they “subsist,” rather
than “exist.”²²¹
In New Foundations of Ontology Bergmann continued to hold that he was
presented with generality and existence, though now he called them the univer-
sal and the particular quantifiers. He went far beyond “Generality and Exis-
tence” by offering a much more complex account of quantification, which still
 Ludwig Wittgenstein, Remarks on the Foundations of Mathematics (Cambridge: MIT Press,
1994), 300.
 Logic and Reality, 71. Philosophical Grammar was published 22 years after “Generality and
Existence.”
 Logic and Reality, 68, 70.

Authenticated
resembled Wittgenstein’s but was a part of a rich, all-encompassing ontology,

which Wittgenstein never attempted. Bergmann renounced his earlier distinction
between existence and subsistence, subscribing now to the seeming paradox
that “whatever is thinkable exists.”²²² Yet he acknowledged that “the differences
among some of the several existents…are very great indeed…momentous, or
enormous,”²²³ thus acknowledging the rationale for the distinction.
Bergmann’s assertion that everything thinkable exists should be no more
surprising than Meinong’s assertion that “there are things of which it is true
that there are no such things,” but it is less confusing. There is a golden moun-
tain (Meinong’s famous example), it has being, it exists, Bergmann would say,
but of course it is fundamentally different from, say, the Rocky Mountains.
Like the latter, it is a “complex” of facts, but unlike the Rocky Mountains it is
“pervaded” by the “mode of potentiality,” not the “mode of actuality.” Berg-
mann’s critics, like Meinong’s, seem to attach magic significance to the words
“exist” and “being.” But, important though they are, still they are words, con-
ventional signs, and their ordinary use might not be suited for the purposes of
ontology. What matters is that we can think and talk about, even describe in
some detail a golden mountain east of Denver, just as we can think, talk
about, and describe the Rocky Mountains west of Denver. What we must not
do, of course, is to think that the former is an actual mountain.
Bergmann began his account of generality in New Foundations of Ontology
by denying that variables, whether free or bound, stand for anything.²²⁴ This
was not surprising in itself, but it did require surprising changes in the analysis
of general statements. We find Bergmann now saying that the universal quanti-
fier in a general statement stands for a function, to be represented by the sign
“\/” but without attaching a variable such as “x.” The function takes as argu-
ment a “2-tuple” that consists of the individual thing in the singular fact asserted
by some singular substitution instance of the general statement and that singu-
lar fact itself. The value of the function is the general fact asserted by the state-
ment.²²⁵ If, for example, the statement is “All f1’s are f2’s,” the 2-tuple might be
<a, f1(a) ⊃ f2(a)>. Bergmann used “2-tuple,” instead of “pair,” because in New
Foundations he also offered a highly original account of sets that prohibits cas-
ual uses of set-theoretical terms.
Perhaps most surprisingly, however, he insisted there that the conscious state or
awareness of a general fact, which he calls the “referent” of the corresponding gen-
 New Foundations, 61.

 New Foundations, 43.
 New Foundations, 64.
 New Foundations, 167– 168.

Authenticated
eral statement, also includes an “auxiliary act” of consciousness, the “intention” (in-
tentional object) of which is the sentence itself, the words used in making the state-
ment. Bergmann called this intention “the text of the awareness.”²²⁶ “I cannot
‘think’ any generality such as, say, all-men-are-mortal, without at the same time
‘thinking’ the words ‘all-men-are-mortal.’….One cannot believe, or doubt, or remem-
ber, and so on, any generality without also perceiving the appropriate words.”²²⁷ In-
deed, Bergmann went on to say that “all awarenesses, except primary Perceivings
and Imaginings (and undoubtedly some ‘Feelings’), are inseparable from their
texts. That…not only gives language its due without giving it too much; it also reas-
suringly recovers the sound core in a large body of recent and contemporary
thought…from Watson to Wittgenstein.”²²⁸
The fact that all f1’s are f2’s is “built” by the function \/ “not just from one
argument but, indifferently, from an indefinite number of alternative argu-
ments…from <a, f1(a) ⊃ f2(a)>, from <b, f1(b) ⊃ f2(b)>, and <c, f1(c) ⊃ f2(c)>, and
so on. In the text of [the awareness], however…there is no cue to this multiplic-
ity.”²²⁹ Bergmann gave “all green (things) are square” as an example. It is the text
of an awareness that has as its object the general fact, presumably not actual,
that all green (things) are square. The 2-tuples that the function \/ takes as argu-
ments, e. g. <this, if this is green then this is square>, are not mentioned in the
general statement, there is no “cue” in it to their “multiplicity.” But they all are
essential to the general fact. That fact would not be actual if the singular facts in
the 2-tuples were not all actual: “all f1’s are f2’s” would not be true if all its sin-
gular substitution instances were not true. From which of them the function \/
builds the general fact is ontologically indifferent. But it might not be psycholog-
ically indifferent, since the speaker or hearer of the general sentence must at
least in principle be able to perceive or imagine one of them. I shall return to
this latter point.
Thus the analysis of “all f1’s are f2’s,” i. e. its ontological analysis, “is not,
conventionally … (x) [f1(x) ⊃ f2(x)], but, rather, alternatively and indifferently…
\/ [<a, f1 (a) ⊃ f2 (a)] > or any of its variants; indifferently because all those var-
iants are one and not many.”²³⁰ The “variants” of \/ [(a, f1(a) ⊃ f2(a)], of course,
are \/ [<b, f1(b) ⊃ f2(b)>], \/ [<c, f1(c) ⊃ f2(c)>], and so on. Each is an alternative
analysis of the one and same general fact. We may note that in standard logic
it is also indifferent, unless the context requires otherwise, whether we symbol-
 New Foundations, 208.

 New Foundations, 204.
 New Foundations, 234.
 New Foundations, 235.
 New Foundations, 202.

Authenticated
ize “all f1’s are f2’s” as “(x) [ f1(x) ⊃ f2(x)], “(y) [ f1(y) ⊃ f2(y)], or “(z) [ f1(z) ⊃ f2(z)].
But standard logic uses variables, which according to Bergmann represent noth-
ing and therefore have no place in ontological analysis.
It may seem that Bergmann’s insistence that the arguments that the quantifier
\/ takes, say, in the case “all f1’s are f2’s,” are 2-tuples is an unnecessary complica-
tion, but the reasons for it are compelling. What else could those arguments be? Not
f1(x) ⊃ f2(x), because it contains variables. Nor the properties f1 and f2 themselves.
According to Bergmann’s “principle of acquaintance,” one cannot be presented
with f1 and f2 except when they are exemplified.²³¹ But even if one could, being pre-
sented with them as well as with the quantifier \/ would hardly count as being pre-
sented with the fact that all f1’s are f2’s, or indeed with any fact: the three presenta-
tions might be unrelated. They might be unrelated even if f1 and f2 are exemplified:
one might be presented with them as well as with the quantifier by virtue of being
presented with the fact, say, that all f2’s are f1’s.
Without the singular fact that is one of the terms of the 2-tuple, there would
be no relevant conscious state or awareness at all when one makes the general
statement, for there would be nothing relevant for one to be aware of. Could the
quantifier take as argument the singular fact f1 (a) ⊃ f2 (a), rather than the 2-tuple
<a, f1 (a) ⊃ f2 (a)>? No, because even if \/ could take f1 (a) ⊃ f2 (a) as argument, its
value would not be a general fact. The notation must also make explicit which
constituent of the singular fact the quantifier operates on, just as in standard log-
ical notation it must be explicit which variable the quantifier binds. If variables
are not used, this can be explicit only if the quantifier, so to speak, “brings” the
constituent “out of” the singular fact, while also “retaining” the singular fact.
The constituent and the singular fact must both be explicitly in the argument
that the quantifier takes, and this amounts to saying that the argument must
be the 2-tuple of which they are the members. Bergmann expressed the point
by saying that the individual is the “target” of the quantifier, while the singular
fact is its “scope.” In the case of the statement “all green (things) are square,”
the target might be any particular perceived or imagined object, even your
hand, which, if the statement were true, would indeed be square if it is green.
There are important similarities between Bergmann’s account of generality in
New Foundations of Ontology and each of Wittgenstein’s stands on the topic we
have considered – in the Tractatus, in his 1919 letter to Russell, in Philosophical Re-
marks, and in Philosophical Grammar –, though there are also obvious differences.
The singular substitution instance of the general statement that stands for the sin-
gular fact which Bergmann calls the scope of the quantifier is not asserted, it is not
 New Foundations, 65.

Authenticated
“said,” yet it must be, so to speak, in the background, if the statement is to express
a relevant conscious state. We may say that it must “show” itself. For it is the sin-
gular substitution instance that provides the general statement with its target and
scope, both of which must, in some sense, be “presented” or “given,” though of
course not as they would be if what was asserted was the singular, rather than
the general, statement. We may say that the 2-tuple from which, as its argument,
the quantifier “builds” the general fact must also only show itself. Indeed, we
may say that the whole indefinite number of alternative arguments from which
the quantifier builds indifferently the general fact show themselves. They must
be “there,” in the background, presupposed, like the indefinite number of possible
positions of the circle on a red background in the example Wittgenstein gave in
Philosophical Remarks. Like those positions, the alternative arguments may be
thought of as “possibilities left open,” neither enumerated nor capable of being
enumerated, but the speaker and hearer knowing that they are there, though not
which they are. Thus, the general statement may be said to be “indefinite,” “an in-
complete picture, like a portrait in which, e. g., the eyes have not been painted in.”
There can be no question of thinking in advance of all the alternative arguments
the quantifier may indifferently take, they are not mentioned at all, they are unseen
and disregarded. Yet they are there, presupposed, like the different possible posi-
tions and sizes of the circle in the square that the statement “there is a black circle
in the square” allows for, even though when making the statement the speaker has
“something simple in mind.”
In Philosophical Remarks Wittgenstein denied that the general proposition “I see
a circle on a red background” has anything to do with “the totality of objects,” but
in the Tractatus he could have said that it shows that totality. I have suggested that
when Bergmann writes that what he calls the scope of the quantifier is not asserted,
yet must be in the background if the statement is to express a relevant conscious
state, we may say that it must “show” itself, and also that what he calls the alter-
native arguments from which the quantifier builds the general fact may be said
to show themselves. If we did say this, would “show” have the sense that it had
in the Tractatus, or in Wittgenstein’s assertion in his 1919 letter to Russell that
while a universal statement does not say that all elementary (singular) propositions
are given, this is shown by there being none having an elementary sense which is
not given? The truth is that Wittgenstein did not explain that sense, just as Berg-
mann did not explain, for example, the sense of his term “presented.” The reason,
in both cases, presumably was that they thought what they meant was too basic to
allow for explanation.
According to Bergmann, one must be presented with what he called the
quantifier, i. e., generality, what “(x)” or “\/” stands for, and Wittgenstein pre-
sumably would have disagreed. “There are no ‘logical objects’,” he wrote, thus

Authenticated
announcing his break with the logical realism of Frege and Russell, though, as
we have seen, he did not adopt logical antirealism. But Wittgenstein might
have agreed that the thoughtful use of a general sentences involves awareness,
even if only peripheral and unfocused, of the sentence itself, what Bergmann
called the “text,” whether by seeing, hearing, or imagining it. Bergmann held
that this is a phenomenological, or as he also put it, anthropocentric, even an-
thropological, fact – this is how we humans think and speak.
But Bergmann also offered a detailed account of this fact, which Wittgen-
stein did not. Bergmann explained that the text is needed in order to close the
“phenomenological distance” between what is presented to us when we thought-
fully make a general statement and the “assay,” ontological analysis, of what the
statement says, “the gap between what the text of an awareness may lead one to
expect, on the one hand, and the assay in fact proposed for its referent, on the
other.”²³² The text is “fused,” “absorbed,” into “the nontext,” he wrote,²³³ it has
“fusing power.”²³⁴ The general fact that all f1’s are f2’s is built by \/ from an in-
definite number of alternative arguments, from <a, f1(a) ⊃ f2(a)>, <b, f1(b) ⊃ f2(b)>,
<c, f1(c) ⊃ f2(c)>, and so on, but there is no “cue” to this multiplicity in the sen-
tence “all f1’s are f2’s,” nor of course in its transcription, whether the convention-
al “(x)f1(x) ⊃ f2(x)” or Bergmann’s “\/ <a, f1(a) ⊃ f2(a)>.”
This goes unnoticed because of the “fusing” power of the sentence. On no
account of generality does a true general statement contain explicit reference
to the multiplicity of what makes it true. Whatever account we accept, we
must rely on the statement to serve, so to speak, as proxy for that multiplicity.
A merit of Bergmann’s account is that it makes explicit what all accounts of gen-
erality must admit, namely, that when saying, e. g., “all green things are square,”
we could, as he pithily puts it, in principle also say regarding any individual
thing “if this were green then this would be square.”²³⁵ Saying the latter
would differ from saying the former only by making explicit that the assertion
is a thoughtful one, not a mere utterance, that one actually has something rele-
vant in mind when making it. In the old empiricist terminology, it would make
explicit the presence “before the mind” of an “idea,” whether of “sensation” or
“imagination.” In Bergmann’s terminology, it would make explicit the presence
of an individual actually perceived or imagined.
Bergmann suggests that if our natural expression was, e.g., “generalized for
this: if this is green then this is square,” rather than “all green things are square,”
 New Foundations, 23.

 New Foundations, 216 – 217.
 New Foundations, 235.
 New Foundations, 235.

Authenticated
then when we find the truth of “all f1’s are f2’s” obvious we would be presented with
the actuality of such complexes as \/ <a, f1(a) = \/ <b, f1(b), i.e., we would find the
truth of the statement “\/ <a, f1(a) = \/ <b, f1(b)” also obvious.²³⁶ In the Philosophical
Investigations Wittgenstein would not have agreed, but in the Tractatus he might
have been sympathetic. Surely, Bergmann’s view is plausible. Can one thoughtfully
assert that all green things are square without at least in principle being able to refer
to some particular thing, perceived or imagined, even if it happened to be one’s
hand or one’s computer, and say of it that if it were green then it would be square?
Bergmann of course held that one must actually, not just in principle, be able to,
make such reference, but the difference might be a matter of how we use the adverb
“thoughtfully,” not a matter of ontological import.
A detailed account of generality is needed, and Wittgenstein offered none of
his own, in the Tractatus or in his later works. Bergmann did. To appreciate it, we
should consider what alternatives to it there are. There is, first, the reductionist
account of universal statements as conjunctions, and of particular statements as
disjunctions, of their singular substitution instances. As we saw, Bergmann
found no merit in it, just as Frege and Russell did not. There is, second, Frege’s
account of generality as a second-level function “saturated” by first-level func-
tions. But this is very similar to Bergmann’s account, though it presupposed
Frege’s ontology, which Bergmann rejected for reasons independent of the
topic of generality.²³⁷ Third, there is Russell’s appeal to irreducibly general
facts. Bergmann’s view in “Generality and Existence” was similar to Russell’s,
and his view in New Foundations of Ontology may be described as a refinement
of it. But Bergmann provided an analysis of these facts, which Russell did not.
Indeed, Russell totally ignored the obvious and crucial question, In virtue of
what are general facts general? To have taken this question seriously was one
of Bergmann’s merits, though he had been anticipated by Frege. And, fourth,
there is the view, often but as we have seen wrongly attributed to Wittgenstein,
that all there is to generality is the general sentences. Bergmann probably
thought that this would be a “linguisticism” too crude to deserve discussion,
so he offered none, but he did hold that awareness of the referent of a general
statement includes perceptual or imaginative awareness of the sentence itself.
The merits of Bergmann’s position become evident when we contrast his tran-
scription of the sentence “all f1’s are f2’s” as “\/ <a, f1 (a) ⊃ f2 (a)>” with the standard
transcription of it as “(x) (f1x ⊃ f2x).” The latter includes the unrestricted individual
 New Foundations, 236.

 See Bergmann’s “Frege’s Hidden Nominalism,” Philosophical Review, 67 (1958). Included in
“Meaning and Existence,” (Madison: University of Wisconsin Press, 1959).

Authenticated
variable “x” and therefore can be read as saying something about all individuals,
as being about this computer, the page you are now reading, the moon, and so on.
Bergmann thought that if we had no particular individual in mind when we assert
the sentence we would have nothing relevant in mind, and would not be making a
genuine statement at all. Locke, Berkeley, and Hume would have agreed. The tra-
ditional empiricist tenet was that to understand what we say or hear we must have
an “idea” of what it is about. Bergmann seemed to accept this tenet in the case of
general statements, if by “idea” is meant an object perceived or imagined, rather
than a representation of it. He believed that whenever we make a genuine general
statement we must perceive or imagine a particular individual of which the state-
ment is true. But he made the empiricist tenet more plausible by allowing that we
can select, “alternatively and indifferently,” that individual.
The sentence “(x) (f1x ⊃ f2x)” does not mention this computer, the page you are
reading, the moon, or any other individual thing. In Wittgenstein’s Tractarian ter-
minology, it does not say that, e. g., if this computer is f1 then it is f2. Nonetheless,
presumably Wittgenstein thought that somehow it must show this. It must do so at
least in the sense that, if a thoughtful, circumspect, utterer of “(x) (f1x ⊃ f2x)” were
asked whether it entails that if this computer were f1 then it would be f2, the utterer
would agree. Bergmann did not use Wittgenstein’s terminology, but he might have
done so in order to explain the relevance of, say, this computer’s being a member
of one of the indefinite number of 2-tuples from which the quantifier “indifferently
builds” the general fact that all f1’s are f2’s. In his later works Wittgenstein rejected
the empiricist tenet about the indispensability of ideas, but he might have agreed
that for a statement about the sort of things that can be perceived or imagined to
make sense the speaker or hearer must in principle be able to, even if in fact does
not, perceive or at least imagine something of which the statement would be true.
Bergmann’s later and Wittgenstein’s Tractarian positions on generality
shared a negative but important feature – that, in Bergmann’s terminology, a
general statement does not mention the singular statement that provides it
with its target and scope, and in Wittgenstein’s, that it does not mention the el-
ementary statements that instantiate it. They also shared an important positive
feature. Bergmann argued that if one is aware of what is said by a general state-
ment, then one is aware also of the sentence used in making it, that in this sense
cognition depends on language in the case of generality (indeed, he held that it
does so in all cases except some perceivings, imaginings, and feelings). This de-
pendence, he held, is not causal or external; it is internal, constitutive.²³⁸
 New Foundations, 225.

Authenticated
“Thought is inseparably intertwined with language,” Bergmann wrote,²³⁹ thus

endorsing much of the linguisticism he had vehemently opposed earlier. Wittgen-
stein was the philosopher who in the Tractatus endorsed and in the Philosophical
Investigations celebrated that linguisticism.
It is simplistic to think that metaphysics provides descriptions of the world
that are additional to those provided by science and everyday thought, and to
view metaphysical disagreements as disagreements about the truth of such de-
scriptions. Metaphysicians do not discover entities hidden from the rest of us,
as chemists, physicists, and astronomers often do, nor do they have the sort
of training and instruments needed for making such discoveries. What they
can do, however, is to see, draw attention to, and emphasize similarities and dif-
ferences between fundamental kinds of items in the world that in everyday life,
and even in science, go unnoticed, not because they are hidden but precisely be-
cause they are fundamental and “there for everyone to see.” Bergmann wrote, “Is
there a felt difference between the external property, as some call it, of being
green and the internal one, as they also say, of being a property? Directly one
cannot argue on either side. That is one reason, though to be sure not the
only one, why at some place or places one must appeal to the phenomenological
basis. All I can say, therefore, is that this particular difference pierces my
eyes.”²⁴⁰ By “phenomenological basis” he meant what in the same work he
also called the phenomenological “rock bottom” and “the jumping-off place.”²⁴¹
Bergmann’s view that a general statement does not mention yet does involve
the singular statements that provide the quantifier with its “target” and “scope,”
and Wittgenstein’s that a general statement does not say that all elementary
propositions are given but this is shown, acknowledged, drew attention to,
and emphasized, in their own yet perhaps not incompatible terminologies, the
fundamental differences between general and singular statements. These differ-
ences are there for all to see, but they “pierce” few eyes. Even Aristotle, the
founder of logic, did not see them clearly when he counted both as subject-pred-
icate statements. Bergmann wrote of what is “presented,” and Wittgenstein of
what is only “shown.” These are metaphors and need not signify fundamental
disagreement. There is no established terminology for what Bergmann and Witt-
genstein wanted to say, because it was novel.
It would be a mistake to ask whether Wittgenstein’s semirealism is “true” or
“right,” and much the same can be said about Bergmann’s ontology. Neither is a
 New Foundations, 65.

 New Foundations, 59.
 New Foundations, e. g., 59, 21.

Authenticated
theory about some peculiar objects or facts. If they disagreed, it was not because
they saw different things, just as those who disagree with them do not. Both ex-
emplify a conception of philosophy more sophisticated than such a theory might
suggest. I shall say much more about this conception in chapter 12.
Wittgenstein and Bergmann were not realists regarding general statements. But
neither were they antirealists. They did not hold that general statements assert facts
in the way singular statements assert facts. But they also did not hold that all there
is to general statements is the general sentences and our use of them. Their views
regarding generality are paradigms of what I have called semirealism.

Authenticated
Chapter Eight: Generic Statements
1 The Ubiquity of Generic Statements
Universal and particular (“existential”) statements are not the only general state-
ments. There are also those that linguists call generic. In section 1 of this chapter
I shall argue that generic statements are ubiquitous in both everyday talk and
science, that they are far more important in cognition than universal statements,
and that universal statements themselves are commonly intended not as literally
universal but rather as generic. In section 2 I argue that, though realism about
the world presupposes the category of fact, in the robust Russellian sense of
“fact,” there are no generic facts even if there are universal facts. In section 3
I argue that generic statements are not reducible to any other kind of statement.
The ubiquity of generic statements, whether explicitly generic or disguised as
universal, provides the clearest and most eloquent argument for the antirealist
thesis. Such statements call for a straightforwardly antirealist interpretation, un-
like universal statements, which as we have seen admit of a semirealist interpre-
tation. Whatever the merits of semirealism regarding universal and particular
statements may be, regarding generic statements antirealism seems self-evident.
I have argued that the traditional arguments for antirealism have been exces-
sively abstract and based on obscure philosophical assumptions. A clear and plau-
sible defense of antirealism, I suggested, must bypass them and start afresh from
specific and readily understandable truths. One such truth is the implausibility of
a realist interpretation of universal, particular, and molecular statements, which
led us in the previous two chapters to endorse at most a semirealism regarding
such statements. The ubiquity of generic statements is another such specific
truth. It provides a noncontroversial reason for accepting not just semirealism but
antirealism about the world, cosmological antirealism, though not for antirealism
about things, ontological antirealism. Neither common sense nor philosophy
would accept the existence of facts that are just generic.
The 17th century French philosopher Antoine Arnauld found the statement
“Dutchmen are good sailors” puzzling..²⁴² It does not say that all Dutchmen
are good sailors. Some are not. But neither does it just say that some are.
Some Germans also are good sailors, but perhaps Germans are not good sailors.
What, then, does the statement say? We may be uncertain whether Dutchmen are
good sailors, but let us suppose it was common knowledge among those whose
 Antoine Arnauld and Pierre Nicole, Logic or the Art of Thinking, trans. and edited by Jill
Vance Buroker (Cambridge: Cambridge University Press, 1996), 116.

Authenticated
152 Chapter Eight: Generic Statements
judgment mattered when Arnauld wrote, presumably 17 century shipmasters. We

therefore also suppose that the statement was true. But if a fact is the sort of
brute extralinguistic entity that according to Russell makes a statement true
and which Wittgenstein had in mind when he declared in the Tractatus that
the world is the totality of facts, not of things, “Dutchmen are good sailors” cor-
responded to no such entity, for there was not such an entity in the world.
Perhaps there was a fact to which the particular statement “Some Dutchmen
are good sailors” corresponded. Perhaps there would have been a fact to which
the universal statement “All Dutchmen are good sailors” would have corre-
sponded had this statement been true. But there was no distinctive, third fact
to which “Dutchmen are good sailors” might correspond. Its truth did depend
on the truth of some statements of the form “x is Dutch and x is a good sailor,”
and perhaps these statements did correspond to Russellian facts, but “Dutchmen
are good sailors” is not the conjunction of these statements and thus did not cor-
respond to the fact, if there was one, that made the conjunction true. Neverthe-
less, the 17 century shipmasters had knowledge of its truth, and that truth mat-
tered greatly in their world. Yet what they knew was not in that world. This is
what puzzled Arnauld. There was no similar puzzle in the case of the other state-
ments mentioned here..
“Dutchmen are good sailors” is an example of a vast number of statements
of the form “Fs are G,” some of great practical and scientific importance. They
are general, not singular, but also not universal statements. Nor are they partic-
ular (“existential”) statements, which are much weaker. Generic statements are
usually made without intention to endorse the corresponding universal state-
ment and are so understood by the listener. Arnauld gave as examples also
“Frenchmen are brave,” “Italians are suspicious,” “Germans are large,” “Orien-
tals are sensuous,” and many others.²⁴³ In the recent literature of linguistics we
find “Birds fly” (penguins are birds but do not fly), “Frenchmen eat horse meat”
(most French people do not), and “John smokes a pipe” (sometimes he smokes
cigarettes).²⁴⁴ As the last example shows, a generic statement need not have the
grammatical form “Fs are G,” just as a universal statement need not have the
grammatical form “All Fs are G,” much less “For every value of x, if x is F
then x is G”: what matters is that the statement is intended and understood as
replaceable by a statement of that form.
 Logic or the Art of Thinking, 118.

 These and other generic statements are discussed in M. Krifka, F.J. Pelletier, G. Carlson, A.
ter Meulen, G. Chierchia & G. Link, “Genericity: An Introduction,” in Gregory N. Carlson and
Francis Jeffry Pelletier, eds., The Generic Book (Chicago: University of Chicago Press, 1995), 1–
124.

Authenticated
1 The Ubiquity of Generic Statements 153
Here are some other examples. After the German election in September 2005,
an observer wrote, “It is clear that Germans do not want to be governed by Angela
Merkel. There is no other way to explain the CDU’s collapse to a 35.2 % in the elec-
tion after reaching 49 % only a couple of months ago in opinion polls.” The author
obviously did not mean that all Germans were unwilling to be governed by Angela
Merkel. Yet, the statement is an example of coherent, perhaps astute political
thought, and it might have been true. The Encyclopedia Britannica informs us
that “The solubility of a gas in a liquid rises as the pressure of that gas increases,”
but it also says that “exceptions may occur at very high pressures.” Economists say
that reducing taxes leads to increased economic growth and therefore government
revenue, but they do not deny that sometimes it does not. No pharmaceutical com-
pany promotes its drugs as 100 % effective, and no responsible physician tells a pa-
tient that the recommended surgery would be 100 % safe. Parents, physicians, and
politicians insist that smoking causes lung cancer, but even politicians avoid saying
that it always does. Physicians do not even say that it is always bad for your health
– the Surgeon General only says that it may be. “Exercise prolongs life” is consid-
ered true but, notoriously, exercise often fails to prolong life. Abstention from uni-
versal statements is characteristic of serious thought and discourse. Discourse in
politics, diplomacy, and everyday business, even a family conversation, would be
impossible without generic statements.
Indeed, universal statements themselves are commonly intended and under-
stood as if they were only generic. Strawson noted that “there are many cases of
subject-predicate statements beginning with ‘all’ which it would be pedantry to
call ‘false’ on the strength of one exception or a set of exceptions.”²⁴⁵ It might not
be pedantry in the case of the universal statements in mathematics or highly the-
oretical areas in science. However, Strawson pointed out, these are also state-
ments philosophers often consider analytic – or disguised definitions, mean-
ing-postulates, reduction-sentences, inference-tickets, linguistic conventions –
not statements of fact.
In everyday discourse, we do make universal statements that allow for no ex-
ceptions, e. g., “All of Jack’s children attended the wedding,” but they are readily
replaceable with conjunctions of singular statements (the statement can be sup-
ported by a list of Jack’s children), which the typical universal statement is not.
We make universal statements commonly for rhetorical purposes, e. g., saying
“All politicians are crooked” when both speaker and listener know that some
are not, or “Everybody likes her” though obviously complete strangers do not.
In the interpretation and application of the law, universal statements are studi-
 P. F. Strawson, Introduction to Logical Theory (London: Methuen, 1952), 195.

Authenticated
ously avoided because the possibility of exceptions must be allowed – this is

why there are courts and lawyers. But in the areas of science where most scien-
tific endeavor takes place – geology, zoology, botany, medicine, anthropology,
psychology, linguistics, sociology, economics, even much of biology and chemis-
try – universal statements are scarce. It is generic statements that are common,
such as “Morotopithecus bishopi was a fruit-eater,” which does not mean that all
members of the species were fruit-eaters or that all they ate was fruit, or “Pa-
tients with prior strokes benefit from taking Lipitor,” which does not mean
that all do. Scientists shun universal statements perhaps because they believe
that such statements could be justified only by information about real causal
connections, which scientists seldom if ever have. They tend to rely, instead,
on statistical reports like “265 or 11.2 % of the patients who took Lipitor in a dou-
ble-blind, randomized, and placebo-controlled clinical trial suffered a stroke
over five years, while 795 or 37 % of those who took a placebo did.”
Aristotle noted that “it is the mark of an educated man to look for precision
in each class of things just so far as the nature of the subject admits.”²⁴⁶ Steven
Pinker said, regarding his theory of language acquisition, “I fully expect that [it]
will be met with some counterexamples. My defense is that an acquisition theory
that faces occasional counterexamples is better than no acquisition theory at
all.”²⁴⁷ The legal scholar Frederick Schauer remarks that “Universal generaliza-
tions, whether the source of the universality be definitional or empirical, tend
to interest philosophers, but most of the generalizations that the rest of us em-
ploy and encounter on a daily basis are not universal.”²⁴⁸
Indeed, “generalization invites exceptions,” “the exception proves the rule,”
and “rules are made to be broken” are sayings we hear often, the first implying
that all general statements are really generic, and the other two that even a rule
grounds only a generic, not a universal, statement. We resort to generic state-
ments not because of indifference to accuracy but because there is no acceptable
alternative. Usually, neither a universal statement understood strictly nor a par-
ticular statement or conjunction of singular statements would express what we
can legitimately say when making a generalization.
Arnauld contrasted universal truths about “the nature of things and their im-
mutable essences,” which admit of no exception, with universal truths about
“existing things, especially human and contingent events,” which “admit of
some exception” and, if we supposed that they did not, would be “judged falsely,
 Nichomachean Ethics, Book I, 3. W.D. Ross translation.

 Steven Pinker, Language Learnability and Language Development (Cambridge: Harvard Uni-
versity Press, 1984), 97.
 Frederick Schauer, Profiles, Probabilities and Stereotypes (Boston: Belknap/Harvard, 2003), 9.

Authenticated
2 Facts, Generic Facts, and Realism 155
except by chance.”²⁴⁹ The former are “metaphysically universal.” The latter are
only “morally universal,” like “the usual sayings ‘All women love to talk,’ ‘All
young people are inconstant,’ ‘All old people praise the past.’” But Arnauld cau-
tioned that “with respect to propositions having only moral universality” we
ought not to “reject them as false, even though we can find counterexamples
to them”²⁵⁰ I have suggested that such propositions are much more common
than those about “immutable essences.” But, pace Arnauld, even the latter, in-
cluding some that philosophers might call analytic, often admit of exceptions
and thus are in fact only “morally universal.”
Consider the venerable definition “Man is a rational animal,” meaning by
“man” human being and by “rational,” let us suppose, possessing intelligence
deserving to be called intellect. The definition states the “essence” of man,
what a man is, and logicians properly infer from it that all men are rational, in-
deed that this is necessarily so, “by definition,” “analytic.” But the logicians do
not mean that neonates display intelligence deserving to be called intellect. So,
metaphysicians revise the definition by inserting the adverb “potentially.” Some
neonates, however, are not even potentially rational – they are born with severe
and irremediable mental defects. The metaphysicians may revise the definition
further, perhaps by appealing (in the past) to Aristotle’s distinction between
first and second potentiality or (today) to the genetic roots of intellectual capaci-
ties. But, if they do, they are no longer interpreting the definition, they are trying
to rescue it. The original intention was just to say that men are rational animals,
and both the definition and the statement inferred from it should have been so
understood – and then left alone.
2 Facts, Generic Facts, and Realism
When Kant declared that things as they are known by us are not as they are in
themselves, he assumed that we can know things only if they are objects of sense
experience, “phenomena.” Our focus here, however, is not on things but on the
world, and not on sense experience but on description. There is a difference be-
tween things and the world: the world may contain all individual things but it is
the structure, not the mere collection, “totality,” of them. There is also a differ-
ence between sense experience and description. Sense experience is essentially
subjective and by itself barely qualifies, if at all, as knowledge. Neonates enjoy
 Logic or the Art of Thinking, 263.

 Logic or the Art of Thinking, 114– 115.

Authenticated
sense experience but know little if anything. Description, by contrast, almost cer-
tainly employs a public language and thus in principle is subject to the con-
straints of intersubjective agreement. It is taught, and when sophisticated it
may count as science. Now, the world is describable necessarily, though not sole-
ly, in general statements, whether universal or generic. But even if universal
statements correspond to items in the world, generic statements do not. Even
if there are universal facts, there are no generic facts. Yet almost all general state-
ments are generic, explicitly or tacitly. Cognition of the world, though perhaps
not cognition of individual things, requires the use of generic statements. Yet
the world contains nothing to which they correspond.
In chapter 6 I sketched the traditional argument for antirealism, in both the
Kantian and its more recent versions. I pointed out that, because of its forbidding
level of abstraction, the argument leaves unclear what it claims and what motivates
it. This is especially true of its extreme version, according to which what we think we
know corresponds to nothing in the world. I suggested that we should seek argu-
ments for a moderate version that claims only that certain essential, or especially
important, parts or aspects of what we think we know correspond to nothing in
the world. I suggested therefore that our focus should be on arguments from specific
and readily understandable truths. The form of such an argument would be as fol-
lows: (1) We cognize (perceive, understand, describe) the world as having a certain
uncontroversial and familiar particular feature F. But it is obvious (2) that the world
does not, perhaps cannot, have that feature. Therefore, (3) the world as we cognize
it, as it is “for us,” is not as it is “in itself.” The major defenders of antirealism, from
Kant to Goodman, did offer also arguments of this second sort. For example, Kant
pointed out that, though our knowledge of things rests on sense perception and we
necessarily conceive of them as causally related, causal relations are not themselves
perceived. In Ways of Worldmaking Goodman provided a rich display of concrete ex-
amples of features of the world – from apparent motion to the fairness of samples –
that are best understood as “made” by us. The argument for antirealism from the
ubiquity of generic statements is also of this sort: We think and talk about the
world as containing facts that are the object of the cognitive activity of generaliza-
tion, and generic statements are our chief vehicles of generalization, but there are
no generic facts in the world.
This is an argument for antirealism with respect to the logical structure of the
world, the structure that any world must have in order to count as a world. Indeed,
nothing would count as a world if it did not allow for true atomic statements such as
“This page is white,” compound statements such as “This page is not red” and “If
this page is white then so is the next page,” universal statements such as “All men
are mortal,” and particular statements such as “There is water on Mars.” But in ad-
dition to these there are also generic statements. A serious and knowledgeable Mis-

Authenticated
2 Facts, Generic Facts, and Realism 157
sourian might say that winters in Iowa are severe, but not that all are. Cardiologists
say that patients with prior strokes benefit from taking Lipitor, but not that all do.
Such statements are our chief vehicles of generalization, yet obviously they corre-
spond to nothing in the world. If there were no humans, all winters in Iowa
might still be severe, or some or none might be, but it would not be the case just
that winters in Iowa are severe.
This is an argument for antirealism with respect to a specific but essential
part of our cognition of the world. It resembles but is not the same as antirealism
with respect to the general statements that logic recognizes, which is familiar
and was defended by Wittgenstein in the Tractatus as part of his rejection of
Frege’s and Russell’s logical realism and defense of his thesis that “there are
no logical objects.” Wittgenstein’s logical antirealism or, as we have seen,
more precisely semirealism, may be plausible. But it is not nearly as plausible
as plain, straightforward, antirealism with respect to generic statements,
which he did not even consider.
Kant compared his transcendental idealism to the Copernican revolution in
astronomy. But despite our unquestioning acceptance of the heliocentric system,
we continue to see and think of the sun as rising in the east at dawn, then slowly
moving overhead, and setting in the west at dusk. This is no ordinary, transient
illusion. It is a phenomenological fact, in the sense of “phenomenology” ex-
pressed by Husserl’s slogan “We must go back to the ‘things themselves’ [zu
den Sachen selbst!].”²⁵¹ The case with our unswerving acceptance of realism is
similar. Putnam described the antirealist thesis as a virtual tautology. The helio-
centric system is not a tautology, but it is so amply confirmed as to function in
educated people’s thought as if it were. Nevertheless, the power of our belief that
the sun moves overhead has remained unaffected. The fate of transcendental
idealism is proof of the power of realism. Kant and philosophers like Goodman
continue to be met with incredulity, often plain incomprehension, even by most
of their philosophical readers. The reason is that Kant’s transcendental idealism
called for a radical change of perspective, as did Goodman’s “irrealism.”
But in the case of antirealism regarding generic statements there is no per-
spective to change. Unless we confuse generic statements with the corresponding
universal statements, we immediately see that there are no distinctive, generic,
facts that make them true. To acknowledge this, no revolutionary considerations,
Copernican or Kantian, would be needed. “Dutchmen are good sailors” does not
say what is said by “All Dutch people are good sailors,” or by “Some Dutchmen
are good sailors,” or by a conjunction such as “Lodewyk is a good sailor and Ry-
 Husserl, Logical Investigations, I, 25.

Authenticated
kaard is a good sailor and Hansel is a good sailor.” No subtle arguments are
needed, therefore, to acknowledge that there is no distinctive fact to which it
might correspond – the only relevant facts are those to which the latter state-
ments might correspond.
One who rejects generic facts need not also reject universal facts. Nor need
one who rejects both universal and generic facts also reject compound facts. And
the rejection of none of these implies rejection of atomic facts. But that the in-
tellectual activity of generalization is crucial to cognition and that the truth of
universal statements requires extralinguistic entities was taken for granted by
Frege and Russell, though they differed regarding what these entities might
be. Both argued vigorously against the reductionist view that universal state-
ments are just disguised conjunctions of their singular substitution-instances.
But generic statements, much less generic facts, were totally ignored by them,
as they have been by virtually all philosophers.
To be sure, Aristotle did note that the statement “Man is white,” or, as J. L.
Ackrill suggests, “Men are white,” allows both that some men are white and that
some men are not white, and acknowledged that such “indefinite” statements
have no place in the “syllogism.”²⁵² (Ackrill complains that they lack “an explicit
quantifier” and for this reason he says, somewhat presumptuously, “it is a pity
that Aristotle introduces [them] at all.”) Kant sharply distinguished what he
called strict universality from “assumed and comparative universality (through
induction),” which “is therefore only an arbitrary increase in validity from that
which holds in most cases to that which holds in all.”²⁵³ As we have seen, uni-
versal statements usually express only “assumed and comparative universality,
and perhaps Kant would have agreed that they should be understood as only ge-
neric. John Dewey did write about “generic” and “universal” propositions, but
explained that by the former he meant “propositions about kinds (general in
the sense of generic),” which have “existential import,” and by the latter “ab-
stract hypothetical propositions,” which are “nonexistential in import.”²⁵⁴
Quine in effect dismissed generic statements as involving “ambiguities of syn-
tax.” He wrote, “Sometimes the plural form of a general term does the work
merely of the singular form with ‘every’; thus ‘Lions eat red meat’…..Sometimes
it does the work rather of a singular with ‘an’ or ‘some’, but with an added im-
plication of plurality; thus ‘Lions are roaring.’”²⁵⁵ It was twentieth century lin-
 Aristotle, De Interpretatione 7, in J. L. Ackrill, Aristotle’s Categories and De Interpretatione

(Oxford: Oxford University Press, 1963), 129.
 Critique of Pure Reason, B 3 – 4.
 John Dewey, Logic: The Theory of Inquiry (New York: Henry Holt, 1938), 256.
 Willard V. N. Quine, Word and Object (Cambridge: MIT Press, 1960), 134.

Authenticated
3 The Irreducibility of Generic Statements 159
guists and some legal scholars, not philosophers, who explicitly and seriously
devoted attention to generic statements.
3 The Irreducibility of Generic Statements
Some logicians and philosophers of language have acknowledged the existence of

the nonstandard quantifiers “many,” “few,” and “most.”²⁵⁶ None is reducible to the
standard quantifiers “all” or “some.” Presumably, statements employing “many”
and “few” defy a realist interpretation – their truth value obviously depends, at
least in part, on our interests and attitudes, not on any distinctive facts about
what is many and what is few. Statements employing “most” would allow a realist
interpretation, if “most” is taken to mean “more than half.” But, unlike generic
statements, such nonstandard general statements, though useful and common,
are hardly indispensable for cognition.
Generic statements resemble statements employing “many” and “few” by
defying a realist interpretation. They also resemble them by defying reducibility
to statements employing “all” or “some.” Indeed, generic statements are not re-
ducible to any other kind of statement. Nicholas Asher and Jacques Morreau
have remarked that “the puzzling thing about generics [is that] their truth con-
ditions connect them at best only very loosely with particular facts about the
world,” and that they entail and are entailed only by other generic statements.²⁵⁷
The latter is not quite true. “Dutchmen are good sailors” does entail “Some
Dutchmen are good sailors,” and it is entailed by “All Dutchmen are good sailors
and there are Dutchmen.”
But “Dutchmen are good sailors” does not entail “All Dutchmen are good
sailors,” and is not entailed by “Some Dutchmen are good sailors.” It also nei-
ther entails nor is entailed by “All Dutchmen who are sailors are, always or usu-
ally, good sailors,” which was the analysis Arnauld seemed to favor.²⁵⁸ If only
two Dutchmen are sailors, their both being good sailors would not be enough
to make “Dutchmen are good sailors” true. The statement neither entails nor
 See, for example, J. Higginbotham and R. May, “Questions, Quantifiers, and Crossing,” Lin-
guistic Review, 1, 41– 79; J. Barwise and R. Cooper, “Generalized Quantifiers and Natural Lan-
guage,” Linguistics and Philosophy, 4, 159 – 219; Stephen L. Read, “Pluralitive Logic,” in Robert
Audi, ed., The Cambridge Dictionary of Philosophy.
 Nicholas Asher and Jacques Morreau, “What Some Generic Sentences Mean,” in The Gener-
ic Book.
 Regarding “The French are good soldiers,” Arnauld wrote: “[it] means that the French who
are soldiers are usually good soldiers” (Logic or the Art of Thinking, 116).

Authenticated
is entailed by “Most Dutchmen are good sailors.” Most Dutchmen are not even
sailors, good or bad. And, if they were, but only 52 % of them, while 70 % of Ital-
ians, 80 % of Germans, and 90 % of Norwegians are good sailors, this might not
be enough to make “Dutchmen are good sailors” true. 52 % of Americans are
women, but it is not true that Americans are women. However, even if only
10 % of Dutchmen are good sailors, this might be enough, as long as 2 % of Ital-
ians, 3 % of Germans, and 4 % of Norwegians are good sailors. That the word
“enough” is needed here indicates that we take generic statements to be true
not because we find generic facts in the world that make them true but partly
because of our interests and attitudes. In the 21st century “Dutchmen can read
and write” would not be true if only 45 % could read and write, but in the 17th
century perhaps it was.
“Dutchmen are good sailors” does not entail that more Dutchmen than peo-
ple of any other nationality are good sailors, absolutely or proportionally. We do
not and need not compare Dutchmen with all other nationalities in order to
make or accept the statement. If comparison does take place (usually implicitly),
it is largely, though not wholly, up to us with whom to compare them. Instead of
Norwegians and Italians, we might pick Germans and Spaniards. But perhaps we
would not pick Hungarians or Mongolians, because Hungary and Mongolia are
landlocked and we might think the comparison would be “unfair.” At any
rate, if only four Dutchmen and only two persons of any other nationality are
good sailors, we are not likely to say that Dutchmen are good sailors. If only
four Dutchmen and only two persons of any other nationality are graduates of
the Dubuque College of Cosmetology, we would not say that Dutchmen are grad-
uates of the Dubuque College of Cosmetology.
It has been suggested that “adverbs such as usually, typically, and in general
are closest in meaning to the generic operator.”²⁵⁹ This would be trivially true of
“in general” if inserting it in “Dutchmen are good sailors” merely makes explicit
that the statement is generic, and perhaps of “typically” if it is used as a syno-
nym of “stereotypically” (see below). Not so of “usually.” How usual must it be
for a Dutchman to be a good sailor if the statement “Dutchmen are good sailors”
is to be true? It might be true even if 10 % are good sailors, as long as only 6 % of
Italians, 7 % of Germans, and 8 % of Norwegians are.
Nor, contrary to another suggestion, need the statement be saying that all
Dutchmen are normally good sailors. What being a good sailor involves, say,
standing firmly at the wheel in raging seas, might be abnormal for all people,
 M. Krifka, F.J. Pelletier, G. Carlson, A. ter Meulen, G. Chierchia & G. Link, “Genericity: An
Introduction,” in The Generic Book, 25.

Authenticated
Dutch or not, though admirable. Even becoming a sailor might be abnormal, in

some legitimate sense of this vague word. It might conflict with emotions that
are normal, such as fear of drowning. At any rate, as Gregory Carlson has con-
clusively pointed out, generic statements can also be made about normal kittens
and abnormal drunk physicians. Are we to take such statements to be about nor-
mal normal kittens and normal abnormal drunk physicians?²⁶⁰
Shall we say, instead, that “Dutchmen are good sailors” means that all
Dutchmen are good sailors in normal circumstances? But what are these circum-
stances? Sailing on merchantmen or sailing on warships? Serving under de-
manding or serving under easy-going shipmasters? On short or long voyages?
Perhaps people even become sailors mainly when the economic circumstances
are abnormal. Being a sailor might be attractive mainly in such circumstances.
But, again, what are these circumstances? High unemployment in the Nether-
lands, or high unemployment just in its coastal areas?
What such examples show is that no brute fact makes “Dutchmen are good
sailors” true. But this does not mean that the statement is “subjective.” Although
it was of special interest to shipmasters, its truth was not dependent on their per-
sonal wishes. It might have been accepted by all other people in a position to
know, including first mates and ship owners. Its truth was objective in the proper
sense of being intersubjective, agreed to by competent judges, people with
knowledge and open mind about navigation. It was not what Kant called
“mere fancy.” As Carlson says, we know that not all dogs bark, but also we
know that “Dogs bark” is true. He adds, “the knowledge that there are three-leg-
ged rabbits does not falsify the statement that rabbits have four legs.”²⁶¹ If we
say that nonetheless “Dutchmen are good sailors” is not really objective, we
must mean that it does not admit of a realist interpretation, that it does not cor-
respond to a fact. But this is exactly what I have argued.
Generic statements have been called vague, but their vagueness is unlike that of
predicates. According to Peirce, “A proposition is vague when there are possible
states of things concerning which it is intrinsically uncertain whether, had they
been contemplated by the speaker, he would have regarded them as excluded or
allowed by the proposition. By intrinsically uncertain we mean not uncertain in con-
sequence of any ignorance of the interpreter, but because the speaker’s habits of
language are indeterminate.”²⁶² Perhaps generic statements are indeed vague in
this sense. But what Peirce had in mind was vagueness of propositions due to
 Gregory N. Carlson, Reference to Kinds in English (New York & London: Garland, 1980), 38.
 Reference to Kinds in English, 30, 36.
 C.S. Peirce, “Vague,” in Dictionary of Philosophy and Psychology, ed. J.M. Baldwin (New
York: MacMillan, 1902), 748.

Authenticated
the presence of vague predicates like “bald” – the quantity of hair on a person’s
head may be such that it is intrinsically uncertain whether a speaker would
apply the predicate to it. Generic statements are not vague for such a reason.
Their vagueness is not due to the presence of a vague predicate. “Good sailor”
may be a vague predicate, but this would not be the main reason “Dutchmen are
good sailors” is vague. It is vague because of its logical form. It would be vague
even if we replaced the predicate “good sailor” with a predicate that is not
vague. This is why generic statements are useful, indeed indispensable. Predicates
such as “bald” are also useful and perhaps indispensable because they are vague.
But their vagueness is different from that of generic statements.
Generic statements have been said also to be inexact, imprecise. Again, this
is true, but how we understand it calls for caution. The inexactness of a generic
statement is not due to the presence in it of an inexact word. The statement “Jack
is there” is inexact, but if we wished we could state Jack’s location with reason-
able precision by saying, e. g., “Jack is in the kitchen,” and might be happy to
replace the former statement with the latter. In the case of “Dutchmen are
good sailors,” however, any attempt at precision is likely to yield a statement
that, whatever its merits, we would not put in place of the original. Either it
would significantly differ in truth value, as “All Dutchmen are good sailors”
would, or it would not be even a general statement, as a conjunction of state-
ments of the form “x is Dutch and x is a good sailor” would not.
Carlson distinguishes inductively established correlations from “real rules or
regulations,” associating generic statements with the former and universal state-
ments with the latter.²⁶³ It may be unclear what he means by “real rules or regula-
tions.” But his phrase “inductively established correlations” is reasonably clear. Its
use implies that, as Kant might have put it, generic statements possess at most as-
sumed universality. Of course, Kant had in mind universal, not generic, statements,
and, as we saw earlier, contrasted those possessing only “assumed and comparative
universality, through induction,” with statements possessing “strict universality,”
meaning by “strict” that they are necessary and a priori. But generic statements
lack even assumed universality – this is why they are generic.
Arnauld would have said that universal statements established inductively
are only “morally universal.” Russell and most other epistemologists in effect
have agreed: they are only “probable.” According to Russell, even if the sun
rose every day in the past, it is only probable that it will rise tomorrow. (He wise-
ly avoided assigning a numerical value to that “probability.”) This was “the prob-
 Gregory N. Carlson, “Truth Conditions of Generic Sentences: Two Contrasting Views,” in
The Generic Book.

Authenticated
lem of induction.” Indeed, reasonable people seldom expect inductive reasoning

to yield more than a generic statement, unless it involves also causal informa-
tion. We appeal to what “history teaches” – e. g., in predicting election results,
hurricanes, and the gyrations of the stock market – precisely when we lack
such information. In both everyday and scientific reasoning, induction unsup-
ported by causal information, which may be called primitive induction, is usual-
ly taken to justify only generic statements, as the frequent occurrence of the
phrase “ceteris paribus” shows. This is why scientific writing routinely includes
caveats such as “The precise mechanism through which fluticasone propionate
affects allergic rhinitis symptoms is not known.” The closer scientists are to
the field or lab, the less willing they are to venture universal statements.
Nevertheless, though based only on induction, “Dutchmen are good sailors”
does not mean that all Dutchmen are probably good sailors, as Russell might
have said. If only 10 % of Dutchmen are sailors, it would be false that all Dutch-
men are probably good sailors, whether in the statistical or the epistemic sense
of “probably.” But it might still be true that Dutchmen are good sailors if, say,
only 6 % of Italians, 7 % of Germans, and 8 % of Norwegians are good sailors.
Nor does the statement mean that all Dutch sailors are probably good sailors.
It might be true even if only 40 % of Dutch sailors are good sailors, as long
as, say, only 15 % of Norwegian, 14 % of German, and 13 % of Italian sailors
are good sailors.
Asher and Morreau say that it is “reasonable” to infer from “Fs are G” that
something is G given that it is F. But inferring from “Dutchmen are good sailors”
that Maarten is a good sailor given that Maarten is Dutch would not be reason-
able if, as surely is the case, less than 50 % of Dutchmen are sailors.
I noted earlier that the suggestion that “Fs are G” means “All Fs are typically
G” might be acceptable if “typically” is understood as a synonym of “stereotypi-
cally.” Indeed, a common complaint about generic statements is that they in-
volve stereotyping, misrepresentation, or at least exaggeration of the facts. The
complaint targets mainly generic statements that, like Arnauld’s examples, con-
cern nationality, gender, race, age, or religion. These are sensitive matters, and
people care deeply how statements about them might be intended or under-
stood. Many resent, even find insulting, that such statements are made at all.
For this reason, “stereotyping” is a pejorative today, and it does apply to ge-
neric statements involving abuse of conceptualization or classification. But it is
misplaced if applied to all generic statements. Like most conceptualization and
classification, most generic statements are innocent. New York Times columnist
David Brooks writes, “I … believe most of human thought consists of stereotypes.
I’m not against stereotypes; I’m against crude stereotypes.” If we say that all ge-
neric statements involve stereotyping, then we must say that so do almost all

Authenticated
universal statements, since almost all are intended and understood as though
they are only generic. That the conceptualization generic statements involve is
sometimes abused counts against them no more than the frequent abuse of in-
ductive reasoning counts against induction. Indeed, abuses of generic state-
ments, including those charged with stereotyping, are usually just abuses of in-
duction. One is more likely to be struck by lightning than to be attacked by a
shark, resort managers in the Bahamas say, and this is true, but it does matter
whether one is swimming in the ocean or sleeping in a hotel bed. The type of
inductive “reasoning” exemplified in “We don’t need fire insurance because
we’ve never had a fire” is unfortunately familiar.
But deductive reasoning, too, is frequently abused. The validity of a deductive
argument is not a matter of choice, but its premises are. One can “prove” deductively
any proposition if free to choose the premises. It is not the validity of the standard
proofs of the existence of God that usually has been questioned – if not already
valid, they can be made valid by adding suitable premises – but the truth of
their premises, e.g., that existence is a property (“perfection,” “real predicate”),
that the universe has a cause, or that the human eye manifests intelligent design.
The variety of antirealism defended here is modest and measured. It does
not deny the reality of individual things or even the reality of atomic, compound,
and universal facts. It denies only the reality of generic facts. This would hardly
cause common sense to rebel. Even philosophers have not claimed that there are
such entities. Yet, if cognition of the world requires the intellectual activity of
generalization, and generic statements are the chief vehicles of generalization,
then this modest and measured variety of antirealism is not much weaker
than traditional antirealism. Moreover, since it is based solely on the ubiquity
and irreducibility of generic statements, it is more plausible than standard anti-
realism. It is also less obscure.
Nevertheless, two important caveats are needed. The first is that while gener-
ic statements cannot be given a realist interpretation, they resist an unqualifiedly
antirealist interpretation. There is no fact to which “Winters in Iowa are severe”
corresponds, but the statement is ordinarily taken to be true while “Winters in
Florida are severe” is not. The reason, of course, is that, though equivalent nei-
ther to universal statements nor to singular statements or conjunctions of singu-
lar statements, generic statements would not be regarded as true were it not for
the truth of some singular statements. This does not mean that they are equiva-
lent to the corresponding particular statements. What “Winters in Iowa are se-
vere” says is quite different from what “Some winters in Iowa are severe”
says. Yet it does entail the truth of the latter. Of course, the converse does not
hold: “Some winters in Iowa are severe” does not entail “Winters in Iowa are se-
vere.” “Some winters in Missouri are severe” is true, but “Winters in Missouri are

Authenticated
4 Logical experiences 165
severe” is not. Generic statements have no relationship to generic facts because

there are no such facts, but they do have a relationship to relevant singular facts,
even though they are not reducible to statements asserting those singular facts.
Nonetheless, this relationship is not like that of universal statements to their sin-
gular instances. As Wittgenstein pointed out, while the former do not say what
the latter assert, they show it. This is not true in the case of generic statements.
Even a semirealist interpretation of them cannot be defended. “All winters in
Iowa are severe” entails “The 1903 winter in Iowa was severe,” but “Winters
in Iowa are severe” does not. Yet, though not unqualified, the antirealist interpre-
tation of generic statements is not like that of statements that are just false. Most
generic statements are true, or false, for obvious empirical reasons. But these
reasons do not include knowledge of generic facts.
The second caveat is that while generic statements are the common means of
generalization, we nonetheless understand them as approximations of universal
statements. The word “all” remains the paradigmatic expression for generality.
This is why the primacy of generic statements may be said to be “pragmatic,”
not logical or semantical. “Patients with prior strokes benefit from taking Lipi-
tor” is indeed different from “All patients with prior strokes benefit from taking
Lipitor,” but the latter remains our guide to understanding the former. The same
can be said, respectively, of “Patients with prior strokes do not benefit from tak-
ing Lipitor” and “No patients with prior strokes benefit from taking Lipitor,” even
though both are false.
4 Logical experiences
In the case of universal, particular, and molecular statements we have found

that, at most, semirealism is warranted. In the case of generic statements, how-
ever, full-fledged, though not unqualified, antirealism seems inescapable. Both
reject logical realism. But many philosophers have drawn attention to what
may be called logical experiences. William James acknowledged the occurrence
of “a feeling of and, a feeling of if, a feeling of but, and a feeling of by,” and
pointed out the dependence of the thought of something as “existent extra men-
tem” on “repeated experiences of the same.”²⁶⁴ H. H. Price also wrote of “a feel-
ing of if,” which “arises through the experience of questioning or doubting,”²⁶⁵
 See William James, The Principles of Psychology (Cambridge, MA: Harvard University Press,
1983), especially Chapter IX.
 Thinking and Experience, 134.

Authenticated
and suggested that “Disappointed expectation is what brings NOT into our
lives.”²⁶⁶ According to Sartre, questioning is what “introduces nothingness in
the world,” for example when looking in a café for someone you eagerly expect
but fail to see.²⁶⁷ We may have that experience when looking for the Times but
failing to find it, a distinctive experience of its absence from the rack. Since
we would be failing it to find it among all the newspapers on the rack, it
would be also an experience associated with generality, whether that expressed
in universal or in generic statements. Gustav Bergmann drew attention to an ex-
perience associated with conjunction when he remarked that “there is an obvi-
ous distinction between two simultaneous awarenesses, one of this, the other of
that, on the one hand, and a single awareness of this-and-that, on the other.”²⁶⁸
The occurrence of such experiences and their association with the respective
logical expressions cannot be denied. Nor are they philosophically unimportant.
Perhaps they even anchor the use of the expressions, consciously or uncon-
sciously. Of course, they are not experiences of logical objects, nor need they
occur whenever the expressions are used. They may play a role analogous to
that of Kant’s “schemata” and Wittgenstein’s “criteria.” If so, the analogy
would be illuminating. Logical semirealism differs from both logical antirealism
and logical realism much as Kant’s position on causality differed from both anti-
realism and realism regarding causality, and Wittgenstein’s position on other
people’s sensations differed from both antirealism and realism regarding
“other minds.” Both could be described as semirealist.
There seems to be no connection between what Kant called the pure concepts of
the understanding, such as the concept of causality, which according to him are a
priori and necessary for cognition of the world of experience but stand for nothing
in experience, and the world of experience to which nonetheless they apply. To deal
with this problem, Kant proposed his doctrine of the schematism of the pure under-
standing. He wrote: “[P]ure concepts of the understanding being quite heterogene-
ous from empirical intuitions, and indeed from all sensible intuitions, can never be
met with in any intuition. For no one will say that a category, such as that of cau-
sality, can be intuited through sense and is itself contained in appearance. How,
then, is the subsumption of intuitions under pure concepts, the application of a cat-
egory to appearances, possible…? We must be able to show how pure concepts can
be applicable to appearances…Obviously there must be some third thing…”²⁶⁹ This
“third thing” was the schema. A few pages later Kant explained that “The schema of
 Thinking and Experience, 124, upper case in original.

 Jean-Paul Sartre, Being and Nothingness, 21– 45
 New Foundations of Ontology, 223.
 Critique of Pure Reason, A 137/B 176–A 138/B 177.

Authenticated
4 Logical experiences 167
cause, and of the causality of a thing in general, is the real upon which, whenever
posited, something else always follows.”²⁷⁰ This, of course, is what Hume had called
constant conjunction. The pure concept of causality must be understood as associ-
ated with the concept of constant conjunction if it is to have application to experi-
ence. Thus Kant agreed with Hume that constant conjunction is what we depend on
in applying the concept of causality, but of course he sharply denied that it is the
same as causality.
In the Philosophical Investigations Wittgenstein denied that the use of words
for sensations depends on anything they refer to. Such words, he claimed, refer
neither to anything “outer” such as behavior, nor to anything “inner.”²⁷¹ But he
was aware that so far he had told us only what their use does not depend on.
What does it depend on, then? The answer that it depends on nothing is deeply
unsatisfactory, even if true, as most readers of Wittgenstein would testify. Witt-
genstein offered an answer that appealed to his notion of a (nondefining) “crite-
rion,” as contrasted with a mere “symptom.” It replaced the notion of reference
in the case of words for sensations, yet it did stand for a connection between
their use and something else.
A person’s holding one’s cheek in a certain distinctive way, Wittgenstein
wrote, may be a “criterion” for saying that the person has a toothache, in the
sense of grounding the learning and then governing the use of the expression.²⁷²
On the other hand, the person’s facial expression may be just a “symptom” of
having a toothache, an event that only “happens” to occur when the person
has a toothache. Holding one’s cheek and having a toothache, of course, are
not the same. They are only “connected” in some way. But we do not infer
this connection from our own past experience – such inductive reasoning
would not be serious, since it would be based on only one instance.²⁷³ Nor do
we infer the connection from other persons’ past or present experience – we
are not aware of other persons’ toothaches. Perhaps holding one’s cheek is
not a criterion for the use of “toothache” (Wittgenstein mentioned it only as a
possible example), but surely some patterns of behavior, however complex,
are such criteria, in the sense that, though logically neither sufficient nor neces-
sary for the correctness of the use of “toothache,” if they were not at least occa-
sionally satisfied that use would be bewildering and presumably never mastered.
A person can have a toothache without holding a cheek or doing anything else
 Critique of Pure Reason, A 144/B 183.

 Philosophical Investigations, 243 – 248.
 Ludwig Wittgenstein, The Blue and the Brown Books (Oxford: Oxford University Press, 1958),
24– 25.
 The Blue and the Brown Book, 24.

Authenticated
connected with having a toothache, and a person can smile, sing, and dance de-
spite an aching tooth, but if this were usually or even often the case we would
have doubts about not only about the truth of the assertion that the person
has a toothache but about its being a proper use of language.
Logical antirealism faces a difficulty analogous to those Kant and Wittgen-
stein faced: there seems to be no connection between the use of logical expres-
sions and what we perceive, think, or talk about. We cannot deal with this diffi-
culty by following Wittgenstein or Kant, for they were concerned with very
different topics and relied on psychological and linguistic assumptions we
need not make. But the logical experiences mentioned earlier may serve as a
“third thing” that at least sometimes mediates between the use of logical expres-
sions and what we perceive, think, or talk about. There is little doubt that the use
of “if” bears some relation to the experience of questioning or doubting, and of
“not” to disappointed expectation.
This is not the place to attempt a psychological or phenomenological anal-
ysis of logical experiences. Their occurrence may not be essential to the uses of
logical expressions. They are not what these expressions stand for, nor are they
experiences of anything else in the world. But they do seem essential to human
life and thus to the world as it is for us. It is a world of familiarity and strange-
ness, expectation and disappointment, presence and absence. It is a world of ac-
tion. We are immersed in it. Our world is not the earth viewed from the moon.

Authenticated
Chapter Nine: Facts and Truth
1 Realism and antirealism regarding facts
The position we have reached so far does not question realism regarding things
(ontological realism), and it accepts nothing stronger than semirealism, e. g.,
Wittgenstein’s and Bergmann’s positions, regarding the sentential connectives
and the quantifiers. It is antirealist only regarding generic statements. It says
nothing about the physical structure of the world, or about space, time, and
causal order. It acknowledges even the occurrence certain experiences associat-
ed with logical expressions. It is thus moderate and inoffensive to common
sense, the mature and thoughtful judgment we all share and must respect
even if not accept. Neither semirealism regarding the sentential connectives
and the quantifiers nor antirealism regarding generic facts denies anything
dear to common sense.
Nonetheless, as I noted in the Introduction, sentences themselves, not just
the sentential connectives and the quantifiers, are logical expressions. This is
often unclear because of the frequent use of “proposition,” instead of “sen-
tence.” The use is legitimate when what is said is intended to apply also to
the synonymous sentences in other languages, but highly controversial if
taken to refer to an entity that is neither a sentence nor a fact nor a judgment.
The logical “objects” for which sentences supposedly stand are facts. In addition
to the sorts of logical antirealism we have considered so far, there is antirealism
with respect to facts as such. It questions the legitimacy of the very category of
fact – atomic, molecular, universal, particular, or generic – and thus of any theo-
ry of truth as correspondence to fact. It questions the reality of the world even if
the world is the totality only of atomic facts.
The case for the reality of facts – robust Russellian facts – is straightforward,
though seldom properly made by realists. It is essential to cosmological realism,
even if not to ontological realism. My example of four worlds containing Jack,
Jill, and the relation of admiring, and differing only in respect to facts about
Jack, Jill, and the relation of admiring, should have made this evident. If Jack
admires Jill but Jill does not admire Jack, only the fact that this is so, not the
mere presence of Jack, Jill, and admiration in the real (“actual”) world, would
distinguish that world from a world in which Jill admires Jack but Jack does
not admire Jill, a world in which they admire each other, and a world in
which neither admires the other, if these worlds differ in no other respect. We
could not make such distinctions if there were no such entities as facts. There
would also be no world, if the world is the totality of facts. There might still

Authenticated
170 Chapter Nine: Facts and Truth
be things, but realism regarding things, ontological realism, does not entail real-
ism regarding the world, cosmological realism.
Facts are thus indispensable for our very understanding of the dispute be-
tween realism and antirealism, as Richard Fumerton has forcefully argued.²⁷⁴ De-
fenders of realism often rest their case on the “obvious” reality of material ob-
jects: of one’s hands, a tree in the park when no one sees it, the sun and the
earth before there were human beings, and so on. But the realism they defend
and the antirealism they oppose are ontological, not cosmological, they are real-
ism and antirealism regarding things, not the world. Israel Scheffler objected to
Goodman’s Ways of Worldmaking by saying that surely we didn’t make the
stars.²⁷⁵ Scheffler’s objection was simplistic, and he failed to notice the very
title of Goodman’s book. To acknowledge the mind-independence of stars is
not to acknowledge the mind-independence of the world. We may remain realists
regarding things, including stars, even if we abandon realism regarding the
world. We may be both ontological realists and cosmological antirealists.
Ontological antirealism is certainly repugnant to common sense. It questions
even the reality of the child’s animals, vegetables, and minerals. But cosmological
antirealism questions only the reality of the world conceived as the totality of
facts, a conception purely philosophical and too recherché for common sense,
which seldom contemplates “the world” and when it does conceives it as the totality
of things. Had cosmological antirealism questioned the reality of the world as the
totality of things, it would have been no less repugnant to common sense than on-
tological antirealism. It would have questioned the reality of animals, vegetables,
and minerals. Ontological antirealism is a position, as Hume saw, doomed to pro-
duce no conviction. It is so implausible that the few philosophers, perhaps Berkeley,
who seem to have held it are viewed by nonphilosophers with astonishment.
Are there facts? Is there a strong case for realism regarding facts? In defense
of his ontology of facts, Reinhardt Grossmann wrote, “Among the furniture of the
world, there are quite obviously facts. It is a fact that the earth is round, a fact
that whales are mammals, and a fact that 2 plus 3 equals 5.”²⁷⁶ But Grossmann
 Richard Fumerton, Realism and the Correspondence Theory of Truth (Lanham, MD: Rowman
& Littlefield, 2002). Fumerton focuses squarely on the correspondence theory of truth, calling it
“alethic realism,” and is skeptical about the possibility of achieving an independent and inter-
esting conception of metaphysical realism. The term “alethic realism” is also used by William
Alston in A Realist Conception of Truth (Ithaca, N.Y.: Cornell University Press, 1996). Paul Hor-
wich has argued for a “minimalist” theory of truth, but it is unclear that his view of reality is
minimalist. See his Truth (Oxford: Clarendon, 1998), second edition.
 Israel Scheffler, “The Wonderful Worlds of Goodman,” Synthese (v. 45, 1980), 204.
 Reinhardt Grossmann, Phenomenology and Existentialism (London: Routledge, 1984), 227.

Authenticated
1 Realism and antirealism regarding facts 171
also held that facts are abstract entities, like properties and relations, which he
regarded as universals: While an apple may be red at time t, the fact that this is
so is not itself located at t or any other time, nor is it located in space where the
apple is or anywhere else.²⁷⁷ Grossmann did add, “There is only one world, a
world in which universals and particulars dwell side by side in harmony” and
what “keeps this world together…is the unity of the fact,” but he also cautioned:
“The problem is not how universals can be kept tied to particulars, but how facts
are to be analyzed into their constituents.”²⁷⁸ David Armstrong has claimed that
Grossmann’s position involves “absurd lack of economy.”²⁷⁹ I suggest, however,
that it involves subtlety and sophistication. It bears important, though unintend-
ed, similarities to Wittgenstein’s ontology of facts in the Tractatus. Like Wittgen-
stein, Grossmann was sensitive both to the special need for the category of facts
in ontology and to the special nature of facts. In describing facts as abstract en-
tities, he came close to recognizing that they are logical objects.
Metaphysical realists’ grasp and defense of their position are seldom clear or
unambiguous. For example, in his spirited defense of realism, Michael Devitt writes
that “the sentence ‘a is F’ is true because it has a predicational structure containing
words standing in certain referential relations to parts of reality and because of the
way that reality is.”²⁸⁰ Yet, in the same paragraph, he denies that truth requires
“mysterious entities” such as facts. One wonders what Devitt could mean, if not
a fact, by a “way that reality is,” or by the word “situation,” which he uses several
pages later in speaking of “pairing of sentences with situations.”²⁸¹
Devitt is not alone in taking such a puzzling stand on facts. Hilary Putnam
writes that a state of affairs (he could have said “possible fact”) is “a kind of
ghostly double of the grammarian’s sentence.” But he then says, “Whether a de-
scriptive sentence is true or false depends on whether certain things or events
satisfy the conditions for being described by that sentence.”²⁸² Presumably,
this page satisfies the conditions for being described by the sentence “This
page is white.” How do those “conditions” differ from the state of affairs or
fact that this page is white?
 Reinhardt Grossmann, The Existence of the World (London: Routledge, 1992), 10.
 Phenomenology and Existentialism, 114.
 David Armstrong, A World of States of Affairs (Cambridge: Cambridge University Press,
1997), 136.
 Michael Devitt, Realism and Truth, Second Edition (Princeton: Princeton University Press,
1997), 28.
 Realism and Truth, 32.
 Hilary Putnam, Words and Life (Cambridge: Harvard University Press, 1994), 301.

Authenticated
Paul Horwich defends a “minimal theory” of truth, which he thinks avoids

commitment to Russellian facts. He claims that his theory is “perfectly consistent”
with the “intuitions” that “whenever a proposition or an utterance is true, it is true
because something in the world is a certain way…For example, <Snow is white>’s
being true is explained by snow’s being white.”²⁸³ Horwich, too, appeals to “ways”
something is without telling us how these “ways” might differ from Russellian
facts. Moreover, snow’s being white would not explain <Snow is white>’s being
true unless “snow’s being white” was a synonym of “that snow is white,”
which it is not. The genitive expression “snow’s being white” obviously refers to
being white, a color, and would refer to a way snow is only if by “way” we
mean a property, in this case the property of being white. And this property is
only part of what explains the truth of “Snow is white.” It is that snow has (exem-
plifies, instantiates) that property, not the property by itself, which explains the
truth of “Snow is white.” Speaking of snow’s being white is quite different from
saying that snow is white. Consider the analogy with the genitive expression
“Jack’s coat.” It is not a synonym of “that Jack has a coat,” and so it does not
fully explain the truth of “Jack has a coat.” “Jack’s coat” refers to an individual
thing, a coat, just as “snow’s being white” refers to a property, and that thing is
only part of what explains the truth of “Jack has a coat.” It is that Jack has a
coat, not the coat itself, that explains the truth of “Jack has a coat.”
William P. Alston calls his theory of truth “minimalist realism,” and in a circum-
spect defense of it he writes, “I see no reason to suppose that facts are not objec-
tively real, and as such capable of rendering true propositions true in a nontrivial
sense.” He adds, however, that the “mode of reality” of facts is “quite different
from that of substances, states, properties of substances, and events, as Strawson
and others have been at pains to point out.”²⁸⁴ Alston does not tell us what the dif-
ference is, or what he means by “mode of reality.” However, William Vallicella, also
a defender of realism, does. He argues that true propositions require “truth-making
facts.” And he astutely points out that facts could be truth-making only if they are
“proposition-like,” “structured in a proposition-like way” – only if “a fact has a
structure that can mirror the structure of a proposition.”²⁸⁵ Though his formulation
of it may invite deflationist attacks, Vallicella’s view is in the spirit of Wittgenstein’s
position in the Tractatus, where we find subtle and deep, even if brief, accounts of
the notions of fact and correspondence to fact. But Wittgenstein applied these no-
 Truth, 104– 105.

 William Alson, A Realist Conception of Truth (Ithaca & London: Cornell University Press,
1996), 41.
 William F. Vallicella, A Paradigm Theory of Existence: Onto-Theology Vindicated (Dordrecht,
Boston, London: Kluwer Academic Publishers, 2002), 13, 166 – 167, 192– 193.

Authenticated
tions, respectively, to atomic facts (Sachverhalte, states of affairs), which are “config-
urations of simple objects,” and atomic propositions, which consist of names of sim-
ple objects. Not surprisingly, he could give no examples of simple objects and there-
fore no examples of atomic facts or of atomic propositions. Wittgenstein famously
argued that his notion of correspondence to fact has no application to the state-
ments of mathematics, logic, ethics, and religion, which, according to him, say noth-
ing even though some of them show much, including “the higher.” But he saw that
it also had no application to the compound and general statements of everyday talk.
The importance of the category of fact (or state of affairs) is seen clearly and
discussed at some length by Jan Westerhoff in his recent book Ontological Cate-
gories. He relies on two arguments. The first he calls “semantic.” Westerhoff
writes, “Sentences are indispensable for a comprehensive description of the
world…therefore states of affairs must be parts of the world.”²⁸⁶ The premise,
of course, is true, but the conclusion does not follow. Compare “Numerals are
indispensable for a comprehensive description of the world…therefore numbers
must be parts of the world.” Westerhoff calls his second argument for states of
affairs “cognitional.” Gestalt, not associationist, psychology provides the more
plausible account of perception, he says. We always perceive wholes (“com-
plexes”); we do not first perceive and then put together their parts.²⁸⁷ But to
admit that this is so is not to admit that what we perceive is states of affairs.
For this to be true, propositions of the form “S perceives x” must be reducible
to propositions of the form “S perceives that p.” Clearly, few if any are. A neonate
may see a hand, but probably not that it is a hand. This is true also of adults.
When I see my hand, usually I do not also see that it is my hand or even that
it is a hand. Westerhoff says that “states of affairs provide the primary epistemic
point of contact between us and the world.”²⁸⁸ Indeed, one of Hegel’s most influ-
ential contributions was to argue that there is no direct consciousness of objects
that is unmediated by judgment. In American philosophy, Wilfrid Sellars held
much the same view. But infants do seem to begin by perceiving things, e. g.,
their mothers, long before they make any judgments. Perceiving that what
they see is Mother is at first too intellectual for them. Both complex individual
things and states of affairs have constituents, but it does not follow that when
we perceive an individual thing we perceive a state of affairs, that individual
 Jan Westerhoff, Ontological Categories: Their Nature and Significance (Oxford: Clarendon
Press, 2005), 74.
 Ontological Categories, §28.
 Ontological Categories, 71.

Authenticated
things are states of affairs. Perhaps, as Wittgenstein remarked in his Notebooks,

all complexity is propositional, but this is not obvious.²⁸⁹
The ontology of facts is usually considered in relation to the theory of truth as
correspondence to fact. But Strawson wrote, “The only plausible candidate for what
(in the world) makes a statement true is the fact it states; but the fact it states is not
something in the world….Of course, statements and facts fit. They were made for
each other. If you prise the statements off the world you prise the facts off it too;
but the world would be none the poorer…. Facts are not broken or overturned, in-
terrupted or prolonged, kicked, destroyed, mended, or noisy.”²⁹⁰ Strawson took the
latter point to be relevant because he believed that “the whole charm of talking of
situations, states of affairs or facts as included in, or parts of, the world, consists in
thinking of them as things, and groups of things”²⁹¹
Donald Davidson concurred: “The realist view of truth, if it has any content,
must be based on the idea of correspondence, correspondence as applied to sen-
tences or beliefs or utterances, to entities that are propositional in character; and
such correspondence cannot be made intelligible.” He went on to add: “The real
objection [to correspondence theories] is that such theories fail to provide enti-
ties to which truth vehicles (whether we take these to be statements, sentences,
or utterances) can be said to correspond.”²⁹² Quine agreed: “Objects in abun-
dance, concrete and abstract, are indeed needed for an account of the world;
but facts contribute nothing beyond their specious support of a correspondence
theory.”²⁹³ I have already mentioned Hilary Putnam’s remark that a state of af-
fairs is “a kind of ghostly double of the grammarian’s sentence.”
Devitt was right that, if understood as Russell and Wittgenstein did, facts are
mysterious entities. There do seem to be no such entities. Alan White pointed out
that, contrary to standard philosophical practice, facts are not even properly de-
scribable as states of affairs, for “[F]acts, but not events, situations, or states of af-
fairs, can be disputed, challenged, assumed, or proved.” ²⁹⁴ Indeed, facts do seem to
be man-made fictions. The general argument against them is that to serve the pur-
 In Tractatus Wittgenstein wrote: “To perceive a complex means to perceive that its constit-
uents are related to one another in such and such a way” (5.5423). Gustav Bergmann agreed, es-
pecially in New Foundations of Ontology, 199.
 F. Strawson, “Truth,” Aristotelian Society Supplementary Volume 24 (1950), 136 – 137. See
also his “Reply to John R. Searle” in Lewis Edwin Hahn, The Philosophy of F. Strawson (Chicago
& LaSalle: Open Court, 1998), 402– 04.
 Aristotelian Society Supplementary Volume 24 (1950), 139.
 Donald Davidson, “The Structure and Content of Truth,” Journal of Philosophy LXXXVII, 6
(June 1990), 304.
 W.V.O. Quine, The Pursuit of Truth (Cambridge: Harvard University Press, 1992), 77– 88.
 Alan R. White, Truth (New York: Anchor, 1970), 80 – 81.

Authenticated
pose for which they are introduced they must be a sort of items that the world can-
not be plausibly supposed to contain. They cannot be said to be entities in the sense
in which animals, vegetables, and minerals, or even colors and spatial relations, are
said to be entities. They bear uncanny resemblance to sentences, in particular the
sentences they supposedly make true, and thus appear to be no more than reified,
hypostatized sentences. The fact that Dick likes Jane much too conveniently has ex-
actly the distinctive features of the sentence “Dick likes Jane”: three constituents,
configured in the same order as the words, one of them (the relation of liking) cat-
egorially different from the other two (the individuals Dick and Jane), much as the
verb in the sentence is syntactically different from the names. The fact seems tailor-
made for the sentence.
That there are such entities as facts is especially implausible in the case of
the subcategory of relational facts. Indeed, there is a close connection between
the category of fact and the category of relations, since a fact is conceived not as
the mere collection of its constituents but as a complex entity with constituents
related to one another. Now, if there are facts, at least some relational facts about
observable things presumably would themselves be observable. But, clearly,
none are. The reason is straightforward: relations are not observable. You may
see the left and the top edge of this page, but surely you do not also see an
item, which might be called the relation longer-than, that holds between them.
If you think that you do, what is its location in your visual field and what are
its color and shape? Surely one cannot see an item that is not in one’s visual
field, and which has neither shape nor color. But if you do not see the relation
longer-than, surely you also do not see the relational fact to which the true state-
ment “The left edge of this page is longer than the top edge” is supposed to cor-
respond. Of course, you see the page, as well as its edges, and you also see that
the left edge is longer than the top edge. But the question is whether when you
see all this you see an item that is the relation between the two edges of being
longer than, and therefore whether you also see a relational fact of which that
relation is a constituent. If the relational fact is visible while the relation is
not, is this so because the relation is somehow hidden? Or too small to see?
One could ask such questions also about the location in the visual field, and
the color and shape, of the relational fact itself, not just of the relation, and sure-
ly the answer would be that it has none.
To ask these questions is not to display sectarian empiricism. Perhaps there
are many unseen and unseeable, unobserved and unobservable, entities. But the
relational fact that the left edge of this page is longer than the top edge should
not be one of them, since the two edges of this page – the page you see in front of
you at this very moment – are quite visible, as is the page itself. If there were

Authenticated
such entities as facts, surely facts about observable things should themselves be
observable.
Bertrand Russell was perhaps the most astute champion of relations. What
he eventually said about them is therefore instructive: “For my part, I think it
is as certain as anything that there are relational facts such as ‘A is earlier
than B.’ But does it follow that there is an object of which the name is ‘earlier’?
It is very difficult to make out what can be meant by such a question, and still
more difficult to see how an answer can be found. There certainly are complex
wholes which have a structure, and we cannot describe the structure without re-
lation-words. But if we try to descry some entity denoted by these relation-words
and capable of some shadowy kind of subsistence outside the complex in which
it is embodied, it is not at all clear that we can succeed.”²⁹⁵ Russell did not ex-
plain how he could be certain that there are relational facts such as ‘A is earlier
than B’ but not at all certain that there is an object of which the name is ‘earlier.’
The case for the “complex wholes” that are relational facts is no better than the
case for their constituents, including relations. The fact that a is earlier than b is
no easier to “descry” than is the relation earlier. How could the former be des-
cried if the latter could not? Pace Russell, there is no special difficulty in making
out what is meant by the question whether “there is an object of which the name
is ‘earlier.’” What he found difficult was how to avoid a negative answer to it.
Russell in effect admitted that there is no object of awareness, whether per-
ceptual or intellectual, to which the relational expression in a relational state-
ment might correspond. Exactly the same can be said about the expressions
for predication, usually a form of the verb “to be,” in subject-predicate state-
ments. To say that this sheet of paper is white presupposes mastery of the con-
cept expressed by the word “is,” but surely there is no relevant object of aware-
ness, in addition to this sheet of paper and its color, to which “is” corresponds. It
would be absurd to suggest that there really is such an object but it is not per-
ceived. Is it not perceived for the reasons that the atoms in the paper are not?
The interpretation of relational expressions in the history of philosophy has
usually been antirealist. Hume and Leibniz held that relations are just “the prod-
uct of the comparison of ideas,” not independent denizens of the world. Accord-
ing to Kant, our concepts of relations such as causality and inherence are a pri-
ori, nonempirical, imposed on what we perceive but representing nothing. The
nonempirical status of relations was vigorously defended, and celebrated, by
the British idealists, especially T. H. Green and F. H. Bradley. Frege, who resolute-
ly rejected the category of fact, did assume that relational expressions stand for
 Bertrand Russell, My Philosophical Development (London: Allen and Unwin, 1959), 172– 173.

Authenticated
what he called concepts, but he held that all expressions standing for concepts
are incomplete, unsaturated, fundamentally different from names, and thus in-
capable of serving as logical subjects of sentences.
The metaphysical notion of fact is grounded in our use of declarative sentences,
and the supposition that there are facts in the world depends at least in part on the
assumption that true sentences must correspond to something in the world, that,
though not names, they somehow must be like names. But sentences cannot
serve as grammatical subjects or objects of verbs, which is the mark of names.
Despite appearances, logic does not need them for such a service. Regarding
the “paradox” of material implication (according to the truth-functional defini-
tion of implication, “p ⊃ q” would be true if q is true or if p is false, regardless
of what p and q are about), Quine pointed out that “This controversy would not
have arisen if the notion of statements as naming had been carefully avoided,
and the variables ‘p,’ ‘q,’ etc., had been treated explicitly as standing in posi-
tions appropriate to statements rather than names.” The natural reading of “p
⊃ q” is “p only if q,” rather than the misleading “p implies q.”²⁹⁶ Indeed, “p is
true,” if taken literally, is gibberish. “Snow is white is true” is grammatically
ill-formed. “‘Snow is white’ is true” is not, but its subject-term is not a sentence
– it is the name of a sentence.
The category of fact is of little use in contemporary logical theory, which usu-
ally adopts a Fregean approach to the semantics of sentences by taking their “de-
notations” to be their truth-values, though it does not hypostatize truth and fal-
sity, as Frege did. The paradigms of expressions that stand for entities, whether
individuals or predicates, are words assigned to them by convention, explicit or
implicit. This is why the number of names and predicates is finite, in principle
determinable for any language. Not so with sentences. They are not introduced
in language by convention. And their (potential) infinity is one of the most im-
portant data in linguistics.
Of course, there are nominalizations of sentences, such as “Snow’s being
white.” They indeed are nouns and often do serve as subjects or objects of
verbs. But they are not sentences and could not be taken to stand for facts even
if there were facts. Consider the phrase “John’s whistling” as it might occur in
the sentence “John’s whistling awoke her.” It stands for a whistling, which is an
action, a doing, perhaps just a noise, not for a fact. Hence the tendency to refer
to facts instead with what Fowler called fused participles, like “Snow being
white” or “John whistling awoke her.” (Fowler’s example was “Women having
 W. V. Quine, “Whitehead and Modern Philosophy,” originally published in 1941 and reprint-
ed in Selected Logic Papers, enlarged edition (Cambridge: Harvard University Press, 1996), 16.

Authenticated
the vote reduces men’s political power.”) One of the two words in “John whistling”
is presumably intended to modify the other; otherwise, they would be just a pair of
unrelated words. But if “whistling” is intended to modify “John,” then the gram-
matically proper sentence would be “Whistling John awoke her” or, better, “John,
who was whistling, awoke her.” Both are obviously different from “John’s whis-
tling awoke her” – they could be true when it is false and false when it is true.
And if “John” is intended to modify “whistling,” then the grammatically proper
sentence would be the original “John’s whistling awoke her,” the subject-term
of which refers to something John does or makes, an action or a noise, not a
fact. Contrary to Roderick Chisholm, Fowler’s objection to fused participles was
not a mere “stricture.”²⁹⁷ They are bad logic, not just bad grammar.
I ended Chapter Eight drawing attention to certain experiences associated
with our use of some logical expressions. Since I have counted sentences also
as logical expressions, one may ask, are there distinctive logical experiences as-
sociated with sentences as such? Indeed, there are, the obvious example being
beliefs, in the sense of experiences or feelings of conviction, confidence, perhaps
faith. But assertions of sentences, i. e., statements, usually are not accompanied
by such feelings – they are just made, correctly or incorrectly. Contrary to wide-
spread philosophical opinion, they usually express no beliefs if beliefs are occur-
rent states or events. It is common in current philosophy to speak of people’s
“systems of beliefs,” and even to say that theories are such systems. But people
are at most just confident in some judgments they make. Theories are systems of
propositions, which in the present context means no more than that they are sys-
tems of sentences. It is unlikely that anyone’s totality of beliefs deserves to be
called a system – many are haphazard, unrelated to the person’s other beliefs
2 Semirealism regarding facts
In the previous section I considered the case for realism and then the case for
antirealism regarding facts. But there is a third way of understanding facts,
which is neither realist nor antirealist. It is semirealist. In general, if a proposi-
tion is in dispute between realism and antirealism, with the realist asserting and
the antirealist denying it, the semirealist would differ from both by holding that
it is an improper proposition, perhaps even that there is no such proposition,
and thus that both asserting and denying it are improper. There is an analogy
 Chisholm suggested that the objection was a mere stricture in his reply to my “States of Af-
fairs,” in Radu J. Bogdan, ed., Roderick M. Chisholm (Dordrecht: Reidel, 1986).

Authenticated
2 Semirealism regarding facts 179
here with sophisticated agnosticism. The theist asserts the proposition “God ex-
ists” and the atheist denies it, but the sophisticated agnostic questions, for vary-
ing reasons we need not consider here, its propriety or sense and neither asserts
nor denies it.
The semirealist way of understanding facts relies on Wittgenstein’s distinc-
tion between saying and showing, and applies to the puzzle regarding facts
the kind of resolution he proposed of the puzzle regarding universal statements.
Indeed, “fact” was on Wittgenstein’s list of formal concepts, together with “ob-
ject,” “complex,” “function,” and “number.” And according to him, “when
something falls under a formal concept as one of its objects, this cannot be ex-
pressed by means of a proposition” but “instead is shown in the very sign for this
object” (4.126). This is why “it is nonsensical to speak of the total number of ob-
jects. The same applies to the words ‘complex’, ‘fact’, ‘function’, ‘number’, etc.
They all signify formal concepts, and are represented in conceptual notation
by variables, not by functions or classes (as Frege and Russell believed)”
(4.1272). Wherever “fact” is used as a proper concept-word, nonsensical pseu-
do-propositions would result, just as wherever “object” is used as a proper con-
cept-word. It cannot be “said” that what the sentence “This page is white” as-
serts is a fact, for the same reason it cannot be said that this page is an
object. Indeed, as we have seen, to say “p is a fact” would be worse, because
if we replaced the sentential variable in the phrase with a sentence, as we should
be able to do, the result, e. g., “This page is white is a fact,” would be grammati-
cally ill-formed, which “This page is an object” is not. Of course, we may say
“That this page is white is a fact,” but “That this page is white” is not a sentence
and thus not a value that p can take.
We can no more intelligibly talk about facts than we can talk about objects.
But this does not mean that Wittgenstein was an antirealist regarding facts or
objects. When an item, say, what the true sentence “This page is white” allegedly
corresponds to, falls under the formal concept “fact,” this cannot be expressed
by means of a proposition but neither is it nothing. For it is “shown” in the sign
for the item. Just as “a name shows that it signifies an object, a sign for a number
that it signifies a number” (4.126), a sentence shows that it signifies a fact. We
cannot properly say “there are facts” but also cannot properly say “there are
no facts,” just as we cannot properly say “there are objects” but also cannot
properly say “there are no objects.” The antirealist holds that it is false that
there are facts. The antirealist does not.
If it cannot be said that there are facts, then also cannot be said that a true
sentence (or statement, proposition, judgment, thought, belief, etc.) corresponds
to a fact. To this extent, antirealism regarding facts appears right. Yet, as we have
seen, the presence in the world of the individuals, properties, or relations men-

Authenticated
tioned in a true sentence does not suffice for its truth. “Jack admires Jill” men-
tions Jack, Jill, and the relation of admiration. But their presence in the world
does not suffice for the truth of the sentence. It does not distinguish the world
in which the sentence is true from worlds in which the sentence is false but oth-
erwise are just like it. Surely, however, there is such a distinction to be made. To
this extent, realism regarding facts appears right.
Semirealism regarding facts differs from realism regarding facts by denying
that true sentences stand for special entities, additional to and categorially dif-
ferent from the entities mentioned in the sentences, that can be referred to, de-
scribed, and analyzed independently of the sentences. This page is rectangular,
white, made of paper, and held in our hands, but none of these can be said of
the fact that this page is white. But semirealism regarding facts differs also from
antirealism regarding facts by acknowledging that there is more to truth than the
sentences – and the statements, propositions, judgments, thoughts, beliefs, etc.
– that are true. It may even label this additional feature “correspondence to
fact.” The label would be reasonable since in everyday talk we do say that a
true sentence “fits the facts.” The terminology of facts thus need not be shunned
by the semirealist. There is more to the truth of a sentence than the sentence that
is true, there is also what “makes” the sentence true, but we cannot say what it is
except by repeating the sentence or uttering a synonym of it.
Antirealism regarding facts is plausible insofar as facts cannot be referred to
or described independently of the sentences supposed to stand for them. But the
theory of truth as correspondence to fact is also plausible. Sentences reporting
direct observation, like “That’s a cat, not a dog,” obviously are not true just in
virtue of the words they contain, or some relation they bear to other sentences,
the speaker, a culture, or an institution. This is why realism regarding facts is
plausible. If the correspondence theory is defective, the reason is not that it dis-
tinguishes the sentence from the fact that makes it true but that it fails to ac-
knowledge the special nature of facts, the special status of the concept of “fact.”
3 Truth
Facts are not entities that true sentences name, and if truth were a relation be-
tween sentences and facts, as the correspondence theory of truth holds, it would
not be an ordinary relation, like next-to or earlier-than. This does not mean, how-
ever, that “there are no facts,” or that “truth is not correspondence to fact.” It
means that we must free ourselves from the confines imposed by the usual the-
ories of truth as correspondence, and see that the question about what makes

Authenticated
3 Truth 181
true sentences true ignores the complexity, subtlety, and wide range of the use of
the word “true.”
Only a poor philosophy of language would hold that to have meaning an ex-
pression must stand for an entity, but it is sound metaphysics to identify what
entities there are by first asking what entities, if any, various expressions appear
to stand for. Realism about Jack and Jill need not infer Jack’s and Jill’s reality
from the existence of their names, but it might identify what it is about as the
people those names name. Realism about properties (“universals”) need not
hold that there are such entities on the grounds that there are predicates, but
it usually identifies what it is about as what predicates seem to stand for. Real-
ism about facts identifies what it is about as what true sentences seem to stand
for, though it need not hold that sentences would have no meaning or sense if
there were no such entities as facts.
In Chapter Eight I argued that generic statements are essential to cognition.
Many generic statements, in everyday life as well as in science, are considered
obviously, unquestionably, true. Yet I also argued for antirealism regarding ge-
neric facts. It might have been asked, if there are no generic facts, what makes
generic statements true? But the use of the word “make” in such a context,
though common in philosophy, is metaphorical and misleading. We should rath-
er speak of the features, relational or nonrelational, in virtue of which sentences
are taken to be true, and acknowledge that these features are many and diverse.
Frege held that truth is the object that is the reference (Bedeutung) of all true
sentences. He did not claim to have discovered a new object, as an astronomer
might claim to have discovered a new planet. He explained that the reasons for
his position were, first, his general distinction between the sense and reference
of expressions – names as well as predicates – other than sentences and, sec-
ond, the obvious connection between the reference of those expressions and
the truth-value of the sentences in which they occur. (Whether “She is young”
is true depends on who she is.) Frege mentioned two alternatives to his view, ar-
guing that neither is credible. Regarding the alternative that truth is the relation-
al property of correspondence to fact, he remarked, as we saw in chapter Six,
“What is a fact? A fact is a thought that is true.”²⁹⁸ And regarding the alternative
that truth is some nonrelational property of sentences or thoughts, he wrote,
“One can, indeed, say: ‘The thought that 5 is a prime number is true.’ But closer
examination shows that nothing more has been said than in the simple sentence
‘5 is a prime number.’” Unlike recent deflationist theorists of truth that also
would say this, however, Frege explained why it is so: “By combining subject
 “Thought,” in The Frege Reader, 34

Authenticated
and predicate, one reaches only a thought, never passes from a sense to its Be-
deutung, never from a thought to its truth-value.”²⁹⁹ According to Frege, the pri-
mary bearers of truth-value are thoughts, not sentences, but he did not mean by
“thought” a subjective, psychological state or event: “By a thought I understand
not the subjective performance of thinking but its objective content, which is ca-
pable of being the common property of several thinkers.”³⁰⁰ Frege did not say
that one never passes from a thought to the world. He said that one never passes
from a thought to its truth-value. Russell and Wittgenstein needed a world of
facts. Frege needed just plain Truth.
Anyone who thinks that Frege’s focus on truth, rather than on facts or the
world, is obscurantist may be reminded that the notion of a fact and the notion
of a world, whether understood as the totality of facts or the totality of things, are
hardly clearer. It would be absurd to identify the world with the totality of Fre-
gean thoughts, since according to him thoughts belong in a “realm” that is nei-
ther physical nor mental, but not more absurd than to identify the world with the
totality of true sentences. What might not be absurd, however, is to identify the
world with truth – it would only be uncommon. The actual world is what we seek
to know, and knowledge requires truth.
Frege’s view of truth resembled the medieval view that it is one of the tran-
scendentalia., together with Being, One, Good, and Beautiful. Both views do jus-
tice to the wide diversity in the application of the word “true.” Neither allows for
a simplistic definition of truth. Frege, of course, did not attempt to provide one.
And according to the medieval view, since the transcendentalia range across the
highest genera, i. e., the ontological categories, they cannot be defined per genus
et differentiam.. The placement of both Truth and Good among them accords with
the suggestion made in our time by Putnam, Goodman, and Dummett that truth
is a sort of goodness (Putnam), rightness (Goodman), or correctness (Dum-
mett).³⁰¹ Putnam’s suggestion is preferable, since rightness and correctness are
plausibly themselves regarded as kinds of goodness. It accords with the impor-
tant use of “true” in such phrases as “true friend” and “true art,” where “true”
has the meaning of “real” or “genuine.” And it accords with the diversity in the
uses of “true,” which may be no greater than the diversity in the uses of “good.”
 “On Sinn and Bedeutung,” in The Frege Reader, 158.

 The Frege Reader, 156.
 Cf. Hilary Putnam, “The Face of Cognition,” Journal of Philosophy 91, no. 9, 1994, Dewey
Lectures, Lecture 3, and most of his other writings since then; Nelson Goodman, Problems
and Projects (New York: Bobbs-Merrill, 1972); Michael Dummett, Truth and Other Enigmas (Cam-
bridge: Harvard University Press, 1978).

Authenticated
3 Truth 183
The several “theories” of truth emphasize different uses of “true,” and each
is plausible in its own way. Some sentences, most notably those reporting obser-
vation, are taken to be true in virtue of surviving what Quine called confronta-
tion with sense experience. Their case lends support to the standard correspond-
ence theory of truth. Some simple mathematical sentences perhaps are taken to
be true in virtue of surviving confrontation with intellectual intuition, a priori in-
sight, reason. But most mathematical sentences and all theoretical sentences in
science are taken to be true mainly, if not solely, in virtue of their membership in
systems that survive confrontation with experience or with reason. Their case
lends support to the theories of truth as “coherence” and as “idealized warrant-
ed assertability.” Sometimes sentences are accepted only because they belong to
theories judged more “beautiful” or “elegant” than their alternatives. Their case
could have given rise to an aesthetic theory of truth. And sentences like “I will be
alive tomorrow” are accepted, at least tacitly, by the speaker as true mainly, if
not solely, for practical reasons. Their case perhaps lends support to the “prag-
matic” theories of truth. It calls for explanation.
In ordinary, not all, circumstances, I do not consider whether the sentence “I
will be alive tomorrow” is true, but of course “take” it, however implicitly, to be
unquestionably true. I do so neither on the basis of experience or reason, nor
because it coheres with other sentences. If I appealed to experience, reason,
or coherence, I might not take it to be true, at least not unquestionably true –
and as a result perhaps suffer disastrous consequences today, or at least post-
pone paying the life insurance premium until tomorrow. Rather, I take it to be
true insofar as its truth is presupposed by virtually everything I do and plan
today. My life would be radically different if I did not have unquestioning faith
that it will continue for at least one more day. Acceptance of the sentence “I
will be alive tomorrow” is thus practically necessary, while the acceptance of
some sentences of mathematics and science may be said to be theoretically nec-
essary, and the acceptance of some observation sentences (“It’s hot!”) perhaps
palpably necessary.
The word “true” is versatile enough to allow without equivocation for such
diversity in what “makes” sentences true. It does resemble “good” in this re-
spect. Gustatory pleasure, knowledge, compassion, right conduct, and justice
are all standard examples of good things, but they seem to have little else in
common. Yet there is no equivocation in calling all of them good.
The ways of knowledge and truth are not neat and tidy. This is especially evi-
dent in ethics, as we saw in Chapters Three and Four. The moral there, however,
was not that we should accept noncognitivism regarding goodness and rightness.
A doctor’s orders are imperatives, but usually their legitimacy and authority are un-
questionably cognitive because their ground is usually cognitive. Theology often

Authenticated
grounds the authority of God in his omniscience, not his status as our creator, but
ordinary religious thought usually holds the latter to be a sufficient ground. It is in
this second way, presumably, that the authority of what Kant called practical reason
(Vernunft) and his description of ethical judgments as both imperatives (Imperative)
and cognitions (Erkenntnisse) should be understood.³⁰² If so, perhaps we should call
ethical judgments “valid” rather than “true,” as Kant indeed often did. They would
be valid in the sense in which we call valid both a doctor’s orders and the propo-
sitions grounding them. But ethical judgments can be valid also in the sense in
which a traffic policeman’s telling us to move to the other lane of the street is (some-
times) valid. We may follow Nelson Goodman and just use “right” for all four: eth-
ical judgments, a doctor’s orders, the propositions grounding those orders, and the
traffic policeman’s orders.
Similarly, the moral to be drawn from the diversity of the uses of “true” is
not that we should stampede into a theory of truth that renounces correspond-
ence to fact altogether. Even in the case of generic statements, where neither cor-
respondence to fact nor coherence or practical utility suffices for truth, all three
are relevant. “Dutchmen are good sailors” is taken to be true because a “suffi-
cient” number of singular statements of the form “x is Dutch and x is a good sai-
lor” are taken to be true, usually on the basis of direct observation. But how
many such statements suffice would depend on how the generic statement co-
heres with various other statements – about sailors, ships, shipping lanes, ma-
rine weather, piracy, the presence of men-of-war protecting merchantmen. And
its truth would depend also on practical considerations, best known to shipmas-
ters and ship owners, such as the purpose of most sailings, the value of cargoes,
the availability and cost of labor at the docks.
This is why the various theories of truth are all legitimate. The truth of “I
have a headache” is plausibly viewed as correspondence, if not to a “fact”
then to a “thing”: the ache felt in the head. But even though many sentences
can be paired off with bits of the world in this way, most cannot, including
some that especially interest us. Obvious examples are counterfactuals like “If
Hitler had not invaded the Soviet Union, Germany would have won the Second
World War,” and complex, largely dispositional sentences like “She liked him,
admired his intelligence, was attracted to him, but did not love him because
he reminded her of her father.” The truth of such statements is better viewed
as their coherence with a vast number of other sentences. And many sentences
about the future, like “I shall be alive tomorrow,” are surely taken as true be-
 Immanuel Kant, Critique of Practical Reason, trans. Lewis White Beck (Indianapolis: Bobbs-
Merrill, 1956), 68.

Authenticated
3 Truth 185
cause the speaker must accept them in order to engage in normal activities today.
No account of truth should fail to note such differences between true sentences,
such kinds of truth.

Authenticated
Authenticated
Chapter Ten: I and the World
1 The paradox of antirealism
In the preface to Ways of Worldmaking, Nelson Goodman wrote, “I think of this book
as belonging to that mainstream of modern philosophy that began when Kant ex-
changed the structure of the world for the structure of the mind….”³⁰³ Indeed, con-
temporary antirealism is best understood as the heir of Kant’s transcendental ideal-
ism. Kant held that although there is reality as it is in itself (“things-in-themselves”),
we can know the world only as it is for us (“things-for-us”). Not only our knowledge
but all our judgments, whether true or false, are shaped by our cognitive faculties,
our senses and our concepts. We can no more “get at” reality as it is in itself, apart
from us, than we can get outside our skins.
Perhaps it is false that nothing unperceivable can be known by us – such
radical empiricism is seldom defended today. But the proposition that nothing
unconceptualizable can enter in epistemic relations with our judgments is, in
Putnam’s words if not sense, a “virtual tautology.” Only judgments can enter
in epistemic relations with other judgments, and judgments necessarily involve
concepts. To state the proposition as an explicit tautology would require exten-
sive accounts of the notions it involves. But there can be little doubt that the
proposition is true. It should not be confused with the thesis of idealism. Kant
did not hold, as Berkeley did, that everything is mental. Nor should it be con-
fused with the antirealist thesis that all reality is dependent for its existence
or at least nature on us, which is hardly a tautology, not even a virtual one.
For there is no contradiction in thinking or speaking of “things in themselves.”
This is why Kant described his view as both transcendental idealism and empiri-
cal realism. Contemporary antirealists are seldom sensitive to Kant’s distinction
between the two.
My concern in this book, however, has not been to engage in the general and
rather amorphous dispute between realism and the many varieties of antireal-
ism. It has been to appraise the much better defined version of metaphysical an-
tirealism I called logical antirealism. Logical antirealism resembles Kant’s tran-
scendental idealism but places the dependence of the cognized world on our
language rather than on our mental faculties, and even then it does so only
with respect to the logical expressions in language.
In the Introduction, after noting the implausibility of the traditional argu-
ments for antirealism, I promised to bypass them and start afresh, from specific
 Ways of Worldmaking, x.

Authenticated
190 Chapter Ten: I and the World
and readily understandable truths, not abstract and obscure philosophical as-
sumptions. I hoped to arrive in this way at an antirealism that is modest and
credible. The antirealist position explained in Part Two does not deny the reality
of things. It is a cosmological, not ontological, antirealism. It denies only the re-
ality of generic facts, and regarding other general facts it accepts, at most, semi-
realism, like that defended, though in different ways, by Wittgenstein and Berg-
mann. But it retains the metaphysical bite of traditional antirealism. For generic
statements are essential to cognition of the world, even if not to cognition of
things. Such an antirealist position would hardly be opposed by common
sense. Even philosophers have not claimed that there are generic facts. As to
semirealism regarding other general facts, common sense would find the distinc-
tion between it and realism too technical to worry about. Our position may seem
excessively complex but the realism/antirealism issue is complex. It does not
admit of simplistic answers.
The thesis of antirealism is that the world, at least insofar as it is cognizable,
depends on our cognition of it. But any cognition that is at all advanced seems im-
possible without statements. Even a mere conception of the world requires state-
ments, be they true or false, if it is to count as conception of a world, rather than
an assemblage of inventoried things. And statements are uses of language. Insofar
as cognition of the world, whether knowledge or mere conception, requires state-
ments, it depends on language and therefore so does the world insofar as it is cog-
nizable. But the only language we know or can even conceive is our language, a
human language, or at least one translatable into ours. Could there be languages
that are nonhuman and in principle untranslatable into a human language?
There is no clear sense in which they would count as languages. As Quine re-
marked, “illogical cultures are indistinguishable from ill-translated ones.”³⁰⁴
It appears, therefore, that the world, insofar as it is knowable, still depends
on humans, though now more specifically on their language. The position ex-
plained in Part Two, however moderate and plausible its antirealist and its semi-
realist parts may seem to be, still appears to imply that the world is dependent
on a tiny part of itself – if not just on me, P.B., then on us, the members of the
human species. In either case, we seem to face cosmological humanism, even
human creationism. Both, of course, would be absurd. They would be anthropo-
centrism at its worst. The absurdity of supposing that the world depends on me
is obvious. The absurdity of supposing that it depends on us may seem less glar-
ing, but the supposition remains absurd.
 W. V. Quine, The Ways of Paradox (New York: Random House, 1966), 105.

Authenticated
1 The paradox of antirealism 191
The task of this Part is to free antirealism of anthropocentrism. This will re-
quire five steps. The first is to show that there is an impersonal use of the first-
person personal pronoun “I.” The second step, which is not the same as the first,
is to show that there is no such entity such as the philosophical, metaphysical,
self or ego. The third, closely related to the second, step is to argue that there is
also no such entity as consciousness if conceived as something the philosophical
self has or engages in. It will be taken in the next chapter. The fourth step, also to
be taken in the next chapter, is to argue that the notion of oneself presupposes
the notion of others, at least in advanced cognition, and thus that there is also
an impersonal use of the plural personal pronoun “we.” The fifth step is to ex-
plain the thesis that the world depends on our cognition of it as really asserting,
not that the world and our cognition of it enter in some logical or a causal rela-
tion, but that they are identical. They are identical not because, as idealism
claims, the world is mental but because there is neither a self nor a conscious-
ness from which it might be distinguished. This fifth step will be taken in Chap-
ter Twelve.
Despite appearances, neither antirealism nor semirealism need have the ab-
surd implications mentioned earlier. They need not imply that the world depends
on me or on us. They need not imply a mad solipsistic creationism that says “I
made the stars,” or a bizarre human creationism that says “We made the
stars.” As we saw in Chapter Two, the first-person pronouns “I” (“me,” “my”)
and “we” (“us,” “our”) can and must be understood as impersonal when used
in Cartesian epistemology and the realism/antirealism debate, indeed in most
if not all philosophical contexts. If the pronoun “I” could be so understood,
then the consequence would be the demise of the narrowest variety of anthropo-
centrism, subjectivism, in all three of its forms: solipsism, skepticism, and ego-
ism. If the pronoun “we” could be understood as impersonal, then the conse-
quence would be the demise of the broader variety of anthropocentrism
exemplified by traditional antirealism. And, as we shall see in the next chapter,
the question whether antirealism is a form of idealism – subjective, objective,
transcendental, or absolute – would also have been answered. In this chapter
our concern will be with the first-person singular pronoun “I.”
Philosophical attention since Descartes has focused on what Hume called
the self. The validity of Descartes’ argument “I think, therefore I am” has often
been questioned, but its premise has remained, even if only tacitly, a central as-
sumption of modern philosophy. “I” ordinarily refers to oneself, the speaker
using it. (Continental philosophers prefer its Latin synonym, ego.) The focus
on the self is not only anthropocentric, it leads anthropocentrism to its logical
extreme by concerning itself with just one human being. In ethics this extreme
is egoism, in epistemology it is skepticism about the “external world” (including

Authenticated
“other minds”), and in metaphysics it is solipsism. Indeed, in the history of phi-

losophy, ethics has been largely preoccupied with battling egoism, and episte-
mology with battling skepticism. There has not been a battle against solipsism
in metaphysics, perhaps because no one has seemed to hold it.³⁰⁵
In Chapter Two I considered how subjective epistemology would face the
challenge of anthropocentrism. To avoid begging the question against the skep-
tic, it must be concerned, not with entities, whether all humans or just oneself,
but rather with the necessary conditions of thought and talk. In this and the fol-
lowing two chapters I shall consider how the challenge of anthropocentrism
would be faced by metaphysics.
In nonphilosophical contexts “I” refers to, say, P.B., and “we” refers to, say, you
and P.B. Both you and P.B. are located on the planet earth. The world, however, is
not. I shall argue that in the relevant philosophical contexts first-person pronouns,
both singular and plural, refer to cognition of the world and thus to the world itself,
not the human beings to which they refer in nonphilosophical contexts. That there
is such an impersonal use of “I” will be argued in the next section. Of course, the
argument would not show that there is no such entity as what Hume called the self.
To show this will be the task of section 3 of this chapter.
2 First-person singular pronouns
In Chapter Two I argued that when engaged in Cartesian doubt I must use the
pronoun “I,” not the noun “P.B.” or any definite description such as “the speaker
of this sentence.” That pronoun must be used without an antecedent noun, it
must be a dangling pronoun. Descartes could not have offered “Descartes ex-
ists,” instead of “I exist,” as the first truth he discovered. The reason is that “Des-
 Anthropocentrism may be considered discrimination in favor of humans, what defenders of

animal rights call speciesism. Its narrowest form, egoism, may be considered discrimination in
favor of just one human being, oneself, what Kant called “the dear self,” and is commonly re-
garded as a paradigm of immorality. Egoism has conceptual descendants, ordered according
to their distance from it. Nepotism – discrimination in favor of one’s relatives – seems to
come first. Next, perhaps, is discrimination in favor of one’s friends, hence Polemarchus’s def-
inition of “justice” in Plato’s Republic. Then might come chauvinism, discrimination in favor of
one’s country. Farther, we would find racism: discrimination in favor of one’s race. Sexism, dis-
crimination in favor of one’s gender, also belongs among the conceptual descendants of egoism.
So does classism (Thrasymachus’s “advantage of the stronger,” Marx’s “dictatorship of the pro-
letariat”), though, unlike nepotism, chauvinism, and sexism, it has been endorsed by important
theories, namely, Platonism and Marxism, which rest on independent grounds.

Authenticated
2 First-person singular pronouns 193
cartes” was the name of a 17th century Frenchman, part of the external world the
existence of which Descartes had not yet proved.
This is why philosophers have usually held that the primary reference of the
indexical “I” is the philosophical self, the metaphysical subject, the ego, das Ich.
Presumably, it was such an entity that Descartes meant when he asserted later
that he was “a thinking thing.” For there is a difference between using “I”
and using one’s name or a definite description of oneself. John Searle offers
the following example. If I make a mess in a supermarket by spilling a bag of
sugar on the floor, I may be ashamed, look to see if anyone saw me, and
worry about what to do. Whether I use “I” or my name does not affect the
truth-value of saying that I made a mess, yet there is an important difference.
As Searle says, what is essential to the case is that “it is me that is making a
mess.”³⁰⁶ There is a difference between my making a mess and P.B.’s making a
mess. The connection with the experience of shame is evident and direct only
in the case the former. For less Anglo-Saxon examples of this sort, such as hear-
ing steps behind me when peeping through a keyhole, we may go to Sartre.
If I use “I” in an autobiography, “P.B.” could replace “I” throughout without
change in truth-value – there would be a change merely in literary genre, from
autobiography to biography. There would be no difference in reference, though
there might be subtle differences of the kind Searle noted. If we say that there
would be a difference in meaning or sense, our use of these terms would be tech-
nical and therefore requiring extensive and inevitably controversial explanation
of how they differ from “reference,” the sort of explanation I have tried to avoid
in this book and is notably absent from Searle’s and Sartre’s examples. But in a
Cartesian context none of the relevant statements in the biography would be al-
lowable because all would beg the question against the skeptic, who questions
the existence of a physical world and therefore the existence of its inhabitants,
including human beings such as P.B. – or Descartes.
I pointed out in Chapter Two that it would have been futile for Descartes to
say at the initial stage of his reasoning, as he did later after he thought he had
proved his existence, that in his argument “I think, therefore I am” “I” referred
only to a thinking thing, not to a 17th century Frenchman. Presumably, Louis XIII
also was or had a thinking thing, but Descartes did not hold that the argument
proved the existence of that thinking thing. It would have been futile for him to
say that “I” referred to this thinking thing. He might have been using the phrase
“this thinking thing” to refer to Louis XIII’s thinking thing. Russell held that the
 John Searle, Intentionality (Cambridge: Cambridge University Press, 1991, 218. But Searle’s
conclusions are different from mine.

Authenticated
demonstrative pronoun “this” can refer only to objects of “acquaintance,” mean-

ing direct awareness, but surely he was mistaken. I can speak about a man men-
tioned in the newspaper story I just read by saying “This man is horrible,”
though I am not acquainted with the man, in Russell’s or the ordinary sense
of “acquainted.” At any rate, even if Descartes had held Russell’s view, he
would still have needed to say who was acquainted with this thinking thing, Des-
cartes or Louis XIII? Of whose direct awareness was the thinking thing an object,
Descartes’ or Louis XIII’s?
It would not have helped if Descartes had replaced “I” with “TT,” a name he
might have given to the thinking thing the existence of which he thought he had
proved with his argument. The reason is not that there was no such thing as TT,
or that the premise of the argument was false or the argument invalid, but that
referring to TT would also beg the question against a sophisticated skeptic.
Which thinking thing would Descartes have named “TT”? Surely, not Louis
XIII’s, for Louis XIII, a human though royal being, would not exist if the physical
world did not. But for the same reason Descartes could not have named Des-
cartes’ thinking thing “TT.”
Could Descartes have just said, “It is my thinking thing”? No, because he
would be using the first-person singular pronoun “my.” Its use presupposes
that of “I.” It is the possessive adjective corresponding to “I.” Moreover, contrary
to Descartes’ intentions, to speak of my thinking thing would imply that I am not
that thinking thing, that I am only the one who owns it, a relation I bear also to
my car and my nose, I being the owner and the thinking thing the owned. My
bearing that relation to TT would be just as questionable as my bearing a rela-
tion, say, that of admiration or distaste, to Louis XIII’s thinking thing. Any skep-
tic worth his salt would forthwith ask, How do you know that there is this think-
ing thing to which you say you bear the relation? The skeptic would ask the
question just by being true to skepticism, not because of metaphysical prejudice
against thinking things or subtle semantic opinions about pronouns. It would be
useless to say that TT is the thinking thing to which only I have access. Even if
this were so, and we ignored this unacceptably metaphorical sense of “access”
as well as the implied crude distinction between “inner” and “outer,” saying it
would presuppose an independent answer to the earlier question about the ref-
erence of “I.” Who is the one who alone has access to that thinking thing?
It is not a principle of logic but surely true that “Someone is F” readily fol-
lows from “I am F” only if the indexical term “I” could be replaced with a name
or a nonindexical definite description, even if it is never so replaced and even if
we do not, perhaps cannot, know a name or description with which to replace it.
We readily accept statements employing indexicals because we take for granted
that in principle they can be replaced with nonindexicals, even if sometimes we

Authenticated
find ourselves unable to do so. We accept statements employing “here” or “now”

since usually, though not always, we are able to replace them with names or def-
inite descriptions, however rough, of places or times. This is true also of “I” in
ordinary contexts. “P.B.” could replace “I” in “I am writing this book,” even if
we agree that the replacement might be more than just a change of words, as
in Searle’s and Sartre’s examples. In some legal documents it may even be re-
quired. But in a Cartesian context it is prohibited.³⁰⁷
In the argument “I think, therefore I am,” the pronoun “I” is profligately
used twice – in the premise and the conclusion. Does it have the same reference,
be it to a man, a thinking thing, or a mere thinking? If it does not, the argument
is invalid. Does it have reference in the premise at all? If it does not, the argu-
ment lacks a premise. But if it does, the question against the skeptic is begged.
When the reality of the whole world is at issue, whether in epistemology or
in metaphysics, the reality of none of its parts, whether Descartes, P.B., or all hu-
mans, may be presupposed. Descartes could not have claimed that “I exist” was
the first truth he had discovered if by “I” he referred to Descartes. Nor could he
have begun his reasoning, as he did, by writing “Let us suppose, then, that we
are dreaming” if by “we” he was referring to his readers. Like Descartes, his
readers are inhabitants of the external world the existence of which Descartes
had proposed to doubt. What was distinctive of Descartes’ epistemology was
not his initial commitment to the existence of only one entity (himself), or the
nature of that entity (a thinking thing), but his decision to presuppose reference
to nothing that might be doubted, and therefore the requirement, which he in
fact met, that initially he use only indexical expressions in the relevant contexts.
This decision, however, has far-reaching implications of which Descartes
seemed totally unaware. When I say that now it is cold here, I am not saying
that on November 7, 2011, it is cold in Woodbury, even if on November 7, 2011,
I am in Woodbury. Yet neither am I explicitly referring to a time other than No-
vember 7, 2011, or a place other than Woodbury. This causes no historical or geo-
graphical problems as long as I can replace “now” with “November 7, 2011” and
“here” with “Woodbury,” or some other date and name – there would still be a
time and place I am talking about. But if I cannot do so, then as suggested in
chapter 2, I would be like a historian only of now or a geographer only of here.
There is a difference between indexical and nonindexical expressions. There
is a difference between saying “It is cold here” and saying “It is cold in Wood-
 Bertrand Russell used the suggestive phrase “egocentric particulars,” instead of “indexi-
cals.” He wrote: “One of the aims of both science and common sense is to replace the shifting
subjectivity of egocentric particulars by neutral public terms” (Human Knowledge [New York:
Simon & Schuster, 1962, 85]).

Authenticated
bury,” though it is neither the sort of difference that there is between “It is cold
here” and Il fait froid ici, nor the sort of difference that there is between “It is cold
in Woodbury” and “It is cold in the capital of Washington County,” even though
all four seem to be saying the same thing. It certainly is not just a difference of
words. Perhaps this is why some have thought that there is an entity such as here
or hereness, which is the primary reference of the indexical “here,” Perhaps the
difference between “It is cold now” and “It is cold on November 7, 2011” is why
some have thought that there is an entity such as the present or presentness,
which is the primary reference of the indexical “now.”³⁰⁸ The latter view is
more plausible because of the seeming connection between the notion of time
and the notion of existence. Augustine noted that what does not exist now
seems to not exist at all, which is not the case with what does not exist here.
Kant held that while space is the pure form of outer sense, time is the pure
form of both outer and inner sense. And Heidegger named his classic work
Being and Time, not Being and Space.
To be consistent, a subjective epistemology such as Descartes’s would be
like a geography of here or a history of now that is in principle unable to say
where is here and when is now. It would be an epistemology without nouns
and with dangling pronouns. Of course, our interest here is not in exegesis
and criticism of Descartes. But considerations similar to those that apply to Des-
cartes’ epistemological project apply also to the metaphysical dispute between
realism and antirealism. The distinction between the world and me must not
be a distinction between the world and P. B. or anyone else. There is no place
for the ordinary use of the first-person singular pronoun when debating the de-
pendence or independence of the world on our cognitive faculties, just as there is
no place for it when engaged in Cartesian doubt. The pronoun does have a use in
both, indeed an essential one. But it cannot be their ordinary use. It must be im-
personal. For the reasons already given, in such contexts “I” cannot refer, as it
ordinarily does, to the speaker. For the same reasons, it cannot refer to any
other part of the world. Though grammatically a personal pronoun, the use of
“I” in such contexts must be understood as impersonal. That there is such a
use is evident from the intelligibility of Cartesian doubt and of the realism/anti-
realism debate even if they are misconceived or defective in some other way. How
is this use to be understood, then?
Like all indexicals, “I” serves to indicate, refer. Unlike a name or a definite de-
scription, however, what an indexical indicates depends on the circumstances of its
 For a detailed argument, see Quentin Smith, Language and Time (New York: Oxford Univer-
sity Press, 1993).

Authenticated
utterance, in particular, on the speaker’s spatial, temporal, perceptual, conceptual,

and linguistic “perspective” in those circumstances. As etymology suggests, a per-
spective is a view, a view from a standpoint, from a point of view. But, despite ety-
mology, a view need not be optical, or even perceptual. It could be conceptual – an
understanding or a thought – as in “political view.” And it could be “linguistic” in
the straightforward sense that it involves language necessarily. A normal adult’s
view of almost anything differs from a neonate’s view largely because it is insepa-
rable from how the thing is, would, or could be described. In this broad sense, a
view is the same as a cognition. It need not be a particular person’s cognition. If
a travel agent praises the panoramic view from the observation deck of the Empire
State Building, it would be absurd to ask “Whose view is it?” – as absurd as asking
“Whose truth is it?” when a proposition is described as true, or “Whose knowledge
is it?” when physics is described as a body of knowledge.
In ordinary contexts, “I” indicates or refers to the speaker, and the view on
which its use depends is the speaker’s view. In philosophical contexts such as Des-
cartes’ epistemological project and the metaphysical dispute between realism and
antirealism, however, that reference, as well as reference to anything identified as
associated with the speaker, indeed to any inhabitant of the world, would be ques-
tion-begging. In such contexts “I” could indicate only the view itself that is essential
to its use. Of course, it would not be a view of Manhattan or of any other particular
object. It would be a view of the world, a “worldview.”
Even in everyday speech “I” sometimes refers to a view, perspective, or cog-
nition, not the speaker. One may say, for example, “I can’t believe you left your
children in the car unattended!” in order to provide information not about one’s
beliefs or anything else about oneself but to indicate the view that leaving chil-
dren unattended in a car is grossly imprudent. “I” occurs rarely in scientific dis-
course. “We” is more common, but usually in impersonal declarative sentences.
In science, “We know that…” is likely to serve the same role as “It is known
that…,” the point of which is precisely to avoid reference to a person. If it in-
volves reference at all, it is likely to be the reference to the current state of knowl-
edge, whether generally or in the specific discipline. A state of knowledge, of
course, is a cognition, a “view.”
The use of “I” coheres with and usually presupposes the use of five other
indexicals: “here,” “now,” “thus perceived,” “thus understood,” and “thus de-
scribed.” (“Thus” is etymologically related to the indexicals “this” and
“that.”)³⁰⁹ The view that “I” indicates is not defined by just a geometrical
 David Kaplan has distinguished between “pure indexicals,” such as “I,” which involve no
genuine demonstration and do not admit attachment of a noun, and demonstratives such as

Authenticated
point, a position in space (here). It is defined also by position in time (now), per-
ceptual modality (thus seen, heard, felt), conceptualization (thus understood),
and linguistic expression (thus described).
But a view is necessarily a view of something, real or unreal. To indicate a
view therefore is also to indicate its object. To indicate a view of Manhattan is
to indicate also Manhattan. In the philosophical contexts that would render ref-
erence to the speaker or any other inhabitant of the world question-begging, “I”
indicates a worldview and thus also the world.
In both the narrow optical sense and the wider sense of “cognition,” a view
is a state of consciousness, but we need not presuppose a particular theory of
consciousness, whether Kant’s, Hegel’s, or Husserl’s. Nevertheless, in this sec-
tion we have moved closer to understanding assertions such as Hegel’s “the Ab-
solute [i. e., ultimate reality] is Thought,” which otherwise are baffling. By insist-
ing that a state of consciousness must receive linguistic expression if it is to
count as advanced cognition we have moved closer to understanding also Witt-
genstein’s no less baffling assertion “I am my world.”
3 The Self
I argued in the previous section that there is an impersonal use of the first-per-
son personal pronoun “I,” not that there is no entity such as the philosophical,
metaphysical, self or ego. To argue the latter is the task of the present section.
Showing that the first-person singular pronoun has an impersonal use was the
first step in the project of freeing antirealism from anthropocentrism. The second
step, which is not the same as the first, is to show that there is no such entity
such as the philosophical, metaphysical, self or ego.
A glance at the history of the topic of a philosophical self is instructive. In
the Treatise of Human Nature Hume described the self as “that to which our sev-
eral impressions and ideas are suppos’d to have a reference,”³¹⁰ Much later, Witt-
genstein called it “the philosophical self” or “the metaphysical subject,” which
“thinks or entertains Ideas” (Tractatus 5.631). John Locke had written that “Expe-
rience convinces us that we have an intuitive knowledge of our own existen-
ce….If I know I feel pain, it is evident that I have as certain perception of my
own existence as of the pain I feel.”³¹¹ But Hume vigorously disagreed. In “look-
“this,” which do, e. g., in “this man.” See his “Demonstratives,” in J. Almog et al., eds, Themes
from Kaplan, (Oxford: Oxford University Press, 1977), 565 – 614.
 David Hume, A Treatise of Human Nature (Oxford: Clarendon Press, 1951), •.
 An Essay Concerning Human Understanding, Book IV, chapter ix, section 3.

Authenticated
3 The Self 199
ing within” himself he found no such entity: “when I enter most intimately into
what I call myself, I always stumble on some particular perception or other, of
heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch
myself at any time without a perception, and never can observe any thing but
the perception.”³¹² Hume then proposed that the self is merely a “collection”
of perceptions, perhaps the collection of all of “his” perceptions, which in effect
would be the whole world that he thought was knowable by him. He agreed that
there are ideas in Locke’s sense, though he called them “perceptions,” but unlike
Locke he held they were the only knowable or even conceivable entities.
Kant had not read Hume’s Treatise, relying for his knowledge of Hume on
Hume’s later work An Inquiry concerning Human Understanding, which contains
no discussion of the self. But he reached a similar conclusion. The Self was one
of the three topics of the transcendental dialectic in the Critique of Pure Reason,
the other two being the World-as-a-Whole and God. None is an object of experi-
ence, whether of sense perception (“outer sense”) or of introspection (“inner
sense”), and thus none can be an object of knowledge. Kant proceeded to replace
the notion of the self with the notion of what he called the transcendental unity
of self-consciousness, or apperception. But the term “self-consciousness” (Selbst-
bewusstein) is ambiguous. It may mean (1) consciousness of a self, (2) conscious-
ness of a state of consciousness by another state of consciousness, or (3) con-
scious mental states, as contrasted, for example, with the unconscious desires
that some take Freud to have postulated.
A century and a half later Sartre argued that there is no self-consciousness in
sense (1). He called consciousness in sense (2) positional, thetic, reflective, and
pointed out that only occasionally is consciousness self-conscious in this sense.
But a major tenet of his view was that all consciousness is self-consciousness in
sense (3). Only in that third sense is self-consciousness essential to conscious-
ness. The French language does not have one word for self-consciousness, as
German and English do. Conscience de soi would be the natural translation of
the English “self-consciousness” or the German Selbstbewusstein. But conscience
de soi seems to express sense (1) or (2). Sartre was compelled therefore to invent
the phrase conscience (de) soi in order to express sense (3).
Kant was explicit that in his phrase “unity of self-consciousness” he was
using “self-consciousness” in sense (3). “The I think must be able to accompany
all my representations,” he wrote, but went on to explain that this is so because
“otherwise something would be represented in me that could not be thought at
all, which is as much as to say that the representation would either be impossible
 An Essay Concerning Human Understanding, 25

Authenticated
or else at least would be nothing for me.”³¹³ One would have no access to such a
representation, it “could not be thought at all.” This indeed was supposed to be
the case with Freudian unconscious desires. It is because of the requirement of
accessibility that Kant classified the unity of self-consciousness as transcenden-
tal, a fact about consciousness that is required for the very possibility of con-
sciousness, rather than an empirical and therefore contingent fact that might
be due to the presence in all consciousness of an intuition, outer or inner, of
a thing that thinks, for which “I” might stand.
In the first edition of the Critique Kant declared that “we do not and cannot
have the least acquaintance” with “the constant logical subject of thinking,”³¹⁴
and he did not change his mind in the second edition, where he wrote that
the representation “I” is “wholly empty,” “a bare consciousness which accompa-
nies all concepts.”³¹⁵ It is needed only “insofar as [the manifold representations
of intuition] must be capable of being combined in one consciousness.”³¹⁶ Rep-
resentations must be so combined if they are to be cognitively relevant, to be
“thought at all.” They are combined, however, not as representations for the
same subject, whether a phenomenal or a noumenal ego, but as belonging in
the same consciousness. As Kant remarked in the first edition of the Critique,
“no cognitions can take place in us, no connection and unity among them, with-
out that unity of consciousness which precedes all data of intuition and in rela-
tion to which alone all representation of objects is possible.”³¹⁷ In the second ed-
ition he pointed out that “inferences from [‘I think’] can contain a merely
transcendental use of the understanding, excluding every admixture of experi-
ence.”³¹⁸ Descartes’ inference of “I exist,” meaning by “I” a thinking thing,
which its owner intuits or is directly aware of, was therefore illegitimate.
Kant did appeal to a noumenal self in his ethics, because he thought that the
“commands of practical reason” required a free agent, in violation of the princi-
ple of causality. But he made it clear that the existence of that self, like the ex-
istence of God and the freedom of the will, which he thought were also required,
was a matter of faith, not knowledge: “How a law can be by itself and immedi-
ately a ground of determination of the will (which is, after all, the essential fea-
 Critique of Pure Reason, B 131– 13

 Critique of Pure Reason, A 350.
 Critique of Pure Reason, B 404.
 Critique of Pure Reason, B 136 – 137.
 Critique of Pure Reason, A 107.
 Critique of Pure Reason 406.

Authenticated
3 The Self 201
ture of all morality), that is for human reason an insoluble problem and the same
as how a free will can be possible.”³¹⁹
In the closing years of the 18th century, Fichte rejected Kant’s distinction be-
tween noumena and phenomena and declared that the self does exist but “only
insofar as it is conscious of itself.”³²⁰ But about a century later Nietzsche com-
mented that “a thought comes when ‘it’ wishes, and not when ‘I’ wish; so that
it is a perversion of the facts of the case to say that the subject ‘I’ is the condition
of the predicate ‘thinks’.”³²¹ It was Hegel, however, who, in the intervening years,
refined Fichte’s position and prepared the ground for Nietzsche’s.
In the Phenomenology of Spirit Hegel rejected the assumptions “that there is a
difference between ourselves and … cognition,” and that “the Absolute [reality]
stands on one side and cognition on the other, independent and separated from
it, and yet is something real.”³²² Later, in the Encyclopedia of the Philosophical Sci-
ences, congratulated Kant on “emancipating” philosophy “from the ‘soul-thing [See-
lending].’”³²³ He remarked that “‘I’ is the vacuum or receptacle for anything and ev-
erything: for which everything is and which stores up everything in itself.”³²⁴ A
couple of pages earlier he had written, “in point of contents, thought is only true
in proportion as it sinks itself in the facts; and in point of form it is no private or
particular state or act of the subject, but rather that attitude of consciousness
where the abstract self, freed from all the special limitations to which its ordinary
states or qualities are liable, restricts itself to that universal action in which it is
identical in all individuals.”³²⁵ This sentence in effect encapsulated the five steps
mentioned earlier as needed if antirealism is to be freed from anthropocentrism.
Cognition of the world is not private and particular but rather identical in all indi-
viduals and thus abstract. And it does not correspond to the facts, it is in the facts,
“sunk” in them. I shall return to this view in chapter 12.
Early in the 20th century, G.E. Moore expressed doubt about the existence of
a self: “It is quite possible, I think, that there is no entity whatever that deserves
to be called ‘I’ or ‘me.’”³²⁶ And, elsewhere, he agreed with William James that
 Critique of Pure Reason, Part I, Book I, Chapter 2.

 Fichte, The Science of Knowledge (Cambridge: Cambridge University Press, 2008, edited and
translated by Peter Heath and John Lachs), 98.
 Beyond Good and Evil, 17
 Phenomenology of Spirit, 47.
 Hegel’s Logic, § 47.
 Hegel’s Logic, § 24.
 Hegel’s Logic, § 23.
 “The Status of Sense-Data,” in Philosophical Studies, 174. See also “A Defence of Common
Sense” in J. H. Muirhead (ed.) Contemporary British Philosophy, 2nd series (Allen and Unwin, Lon-
don: 1925), 211– 212, 223.

Authenticated
“The present thought is the only thinker.”³²⁷ At roughly the same time, in Logical
Investigations, Husserl explicitly denied the existence of the ego, though later, in
Ideas, he accepted it. Further thought on the topic had to await Wittgenstein’s
writings, early and later, as well as Sartre’s. Both endorsed Hume’s rejection
of the self, and each offered a novel account of the role of the first-person sin-
gular pronoun.
In Tractatus Logico-Philosophicus Wittgenstein asked sarcastically, much as
Hume had done, “Where in the world is a metaphysical subject to be found?”
(5.633). His answer was that “There is no such thing as the subject that thinks
or entertains ideas” (5.631) and that “I am my world” (5.63).³²⁸ The answer was
not as idiosyncratic as it may seem. It was at least partly phenomenological,
like Hume’s and, later, Sartre’s. The metaphysical subject is not to be “found.”
A simple experiment by the reader might confirm the claim. Let the reader
focus, not on this page or something remembered or imagined, nor on the read-
er’s nose or eyeglasses or anything else that might be seen, but rather on “what
sees this page.” Few have claimed success in such an experiment. Few have
claimed to find, become aware of, something that does the seeing, the subject
of seeing. Nor was the positive view of the self that Wittgenstein offered, “I
am my world,” idiosyncratic. As we saw earlier, Hume described the self as “a
heap or collection of different perceptions,” perceptions being all that “is ever
really present to the mind” and thus the world insofar as it was knowable. Sartre
appealed to an “absolute, impersonal consciousness,” which “constitutes” the
world, the ego or the subject being just one of the objects in that world.
If one thinks of the self as oneself, say, P.B., it would indeed be absurd to say
that one is the world. But it is not one’s existence that is in question here. Hume
did not deny Hume’s existence, Wittgenstein did not deny Wittgenstein’s exis-
tence, Sartre did not deny Sartre’s existence. What they denied was the existence
of a philosophical self, a metaphysical subject. They thought it was phenomeno-
logically evident that there is no such entity. It might be evident also from the
grammatical monstrosity of expressions like “the I” (das Ich), “I-hood” (die Ich-
heit), and “I-like” (ichlich), which have marred many philosophical writings.
 G. E. Moore, Lectures on Philosophy (London: Routledge, 1966), 163.

 Following certain remarks by Wittgenstein in The Blue and the Brown Book (Oxford: Oxford
University Press, 1958, 66 – 67), G. E. M. Anscombe has argued that “I” is not a referring expres-
sion. In using it, she says, “getting hold of the wrong object is excluded, and that makes us think
that getting hold of the right object is guaranteed. But the reason is that there is no getting hold
of an object at all” (“The First Person,” in Samuel Guttenplan, ed., Mind and Language, Oxford:
Oxford University Press, 1975, 45 – 65). Her point is not Hume’s or Wittgenstein’s. It has nothing
to do with what one could or could not “find” in one’s mind.

Authenticated
Chapter Eleven: We and the World
1 Consciousness
The ubiquity of the pronoun “I,” which Kant noted, encourages the philosophical
thought that there is a self, an ego, a special entity that engages in activities
such as thinking, imagining, and perceiving. The ubiquity of the verbs “think,”
“imagine,” and “perceive” (as well as the more specific “see,” “hear,” “feel,”
“smell,” and “taste”) encourages the philosophical thought that there is conscious-
ness, awareness, a relation of which thinking, imagining and perceiving are species
that consists in the self’s “intending,” being “directed upon,” objects. Taken togeth-
er, these two thoughts encourage the philosophical picture of consciousness as an
arrow shot by the self and aimed at objects “external” to the self. (Ancient physiol-
ogy indeed held that seeing involves light coming out of the eye, hitting the object
seen, and bouncing back.) The project of freeing antirealism from anthropocentrism
rejects this picture. Of course, it does not deny that there are people or that people
think, imagine, and perceive. It rejects the conception of thinking, imagination, and
perception that rests on that philosophical picture.
The rejection of the philosophical self was the second of the five steps, listed
in the previous chapter, that the project of freeing antirealism from anthropocen-
trism takes. The rejection of consciousness understood as an entity, a relation be-
tween a self and its objects, is the third step. Hegel and, much later, Moore, Witt-
genstein, and especially Sartre, in effect combined it with the second. The
second step implies rejection of subject-object dualism. The third implies rejec-
tion of act-object dualism. If consciousness is understood as a relation between
subject and object but there is no subject, then there is no such relation. But, like
the second step, the third step can be defended also phenomenologically. The
reader may attempt now to find, become aware of, not the entity that sees this
page, but rather the reader’s seeing it – the seeing itself, the visual conscious-
ness. I believe few would claim success.
In “The Refutation of Idealism,” Moore argued that “[T]he moment we try to
fix our attention on consciousness and to see what, distinctly, it is, it seems to
vanish: it seems as if we had before us a mere emptiness. When we try to intro-
spect the sensation of blue, all we can see is the blue: the other element is as if it
were diaphanous.”³²⁹ “In general,” he wrote, “that which makes the sensation of
 “The Refutation of Idealism,” 25.

Authenticated
204 Chapter Eleven: We and the World
blue a mental fact seems to escape us: it seems, if I may use a metaphor, to be
transparent – we look through it and see nothing but the blue.”³³⁰
Sartre described the role of consciousness in “constituting” the world as the
“revelation” of the objects conceptualized as a world. It “exhausts” itself in its
objects, he wrote, precisely because it is nothing but the revelation of them:
“Consciousness is outside; there is no ‘within’ of consciousness.”³³¹ It has no in-
habitants. Whether perceptual or conceptual, consciousness is not a “thing.” One
may even go so far as saying that it is nothing. To use a word Heidegger had ap-
plied, consciousness is only the “lightening” of its objects, like the coming of
dawn, which lightens, reveals, the rocks, bushes, and hills that had been invis-
ible in the darkness of the night, but is not itself an object of sight.³³²
If Hume’s and Wittgenstein’s denial of the existence of a self, and then
Moore’s and Sartre’s conception of consciousness, are accepted, we may come
closer to grasping what conception of ourselves would be needed for under-
standing the thesis of antirealism, that, as Nelson Goodman put it, we make
the world, and even Hegel’s signature assertion that that the Absolute is
Thought. Goodman’s impish “we make the world,” of course, was a metaphor.
It meant that what the world is “for us” depends, at least in part, on our con-
sciousness, cognition, of it, on how we perceive, conceptualize, and describe
it. He did not mean that it was made by Goodman, alone or in the company
of others, at Harvard or somewhere else. And Hegel used “thought” roughly in
the broad sense of “cognition” in which I have used it. But it was not Hegel’s
or any other human being’s thought. Nor could the Absolute be a thought of any-
thing other than itself, since the Absolute is all-encompassing reality. There is
nothing to which it could be relative, whether as subject or as object.
As I pointed out in the Introduction, the thesis of antirealism is not entirely
foreign to common sense. We easily understand what would be meant by saying
that the world of the fly is quite different from our world, and that of the octopus
even more so, as long as we are aware that the sense organs of a fly or an octo-
pus differ radically from ours and therefore that they perceive the world not at all
as we do. We easily understand what is meant by saying that people in very dif-
ferent cultures live in different worlds because the concepts through which they
understand their surroundings are different. We take for granted that the world
of an intelligent extraterrestrial life-form would be fundamentally different from
 “The Refutation of Idealism,” 20.

 Jean-Paul Sartre, “Interview,” The Philosophy of Jean-Paul Sartre, ed. by Paul Arthur
Schilpp (LaSalle, IL: Open Court, 1981), 11.
 Poetry, Language, and Thought, 121.

Authenticated
1 Consciousness 205
our world, perhaps indescribably so, since almost certainly it would be both per-
ceived and conceptualized very differently.
If we say, however roughly, that according to Kant the world is dependent on
our consciousness of it, we can say, also roughly, that according to Hegel the
world is that consciousness. Both positions become more plausible if we do not
limit consciousness to perception and allow also for conception (understanding)
as a mode of cognition, especially if we suppose that perception necessarily involves
conception. The defect of empiricism, Hegel wrote, is that “it makes sense-percep-
tion [Wahrnehmung] the form in which fact is apprehended,” but “the process of
knowledge…proceeds to find out the universal and permanent element in the indi-
vidual apprehended by sense. This is the process leading from simple perception to
experience [Erfahrung].”³³³ Hegel’s distinction between “simple perception” and ex-
perience parallels Sherlock Holmes’s distinction between seeing and observing
(heeding, watching, noticing). He told Dr. Watson (in “A Scandal in Bohemia”),
“You see, but you do not observe.” Empirical knowledge rests on observation,
not on mere perception. This is why, in recent philosophy, Wilfrid Sellars argued
against “the myth of the given,” holding that all cognition presupposes conceptual-
ization, that sense perception as such grasps no facts.³³⁴
Even if perception involves conception, this need not mean that it is propo-
sitional, that statements of the form “S perceives x” entail statements of the form
“S perceives that p.” It need only mean that statements of the form “S perceives
x” entail statements of the form “S perceives x as (an) F.” This, of course, is not
true of a neonate’s perception. A neonate first just perceives (sees, feels) Mother,
and neither perceives Mother as the neonate’s mother nor that Mother is the neo-
nate’s mother. The latter are later stages of the child’s cognitive development.
Presumably perceiving Mother is first followed by perceiving Mother as the
child’s mother, and perceiving that Mother is the child’s mother comes later, per-
haps never if it presupposes the acquisition of language.
I mentioned earlier Sartre’s view of consciousness as like the coming of
dawn, the “lightening,” revealing, of its objects. Had those objects been
“there,” had they existed, before the “coming of dawn”? Or do they exist only
when revealed? In other words, are they mind-dependent? If by “mind” we
mean, as Moore and Sartre did, a consciousness that is not a thing but merely
the revelation of things, then the mind cannot be meaningfully said to enter
in causal, logical, or any other relations, and thus to depend on anything or
have anything depend on it.
 Hegel’s Logic, § 38.

 Empiricism and the Philosophy of Mind, 176.

Authenticated
The image of the coming of dawn has a clear, natural, application to percep-
tion. At dawn things become visible, knowable. Hence the initial plausibility of
an antirealism such as Berkeley’s: to be is to be perceived. The image does not
have a clear or natural application to conception. For such an application, we
need to go beyond Berkeley – to Kant and especially Hegel.
The distinction between perception (Kant’s “sensibility”) and conception
(Kant’s “understanding”) gives rise to important questions, many dealt with by
Kant in detail. One, already mentioned, is whether objects can be perceived with-
out being conceptualized. Clearly, even if they can, such perception would not be
a case of advanced cognition. But another, arguably more fundamental, question
is whether existence and nonexistence are themselves outcomes of conceptual-
ization, of the application of the concepts of existence and nonexistence. If
they are, then the answer to the question whether the objects revealed by the
coming of dawn existed before they were revealed would be that they neither
did nor did not.
2 The I that is We and the We that is I
Antirealism faces the paradox of seeming to say that the world is dependent, at least
for its nature if not also existence, on some of its zoological parts, at least on me. I
have argued that there is no special, philosophically relevant entity – self, ego, sub-
ject – denoted by the first-person singular pronoun. I have also argued that to avoid
the paradox we must understand first-person singular pronouns (“I, “me,” “mine”)
when used in philosophical contexts such as Cartesian doubt and the realism/anti-
realism debate as impersonal. I shall argue now that in such contexts the first-per-
son plural pronouns should also be understood as impersonal.
Indeed, the choice between “I” and “we” is often stylistic. A politician may use
the latter in order to avoid displaying conceit. It is often stylistic even in philosophy.
In epistemology we speak of skepticism regarding the limits and extent of our
knowledge, though what we actually consider, as Descartes did, is the extent and
limits of my knowledge. The choice is not stylistic in ethics, however, which from
Plato to Hobbes to Sidgwick to Rawls has been preoccupied with egoism, the posi-
tion that advocates the pursuit of “my own good” rather than of “our good.” Nor is
it stylistic in stating the metaphysical doctrine of antirealism.
I have argued that in philosophical contexts such as Cartesian epistemology
and the realism/antirealism debate the first-person singular pronoun should be
understood as indicating a cognition, a perspective, a view, not a person. In such
contexts, indeed in any advanced context, cognition is inherently social. My
worldview is the same as our worldview, just as my view of Manhattan is the

Authenticated
2 The I that is We and the We that is I 207
same as our view of Manhattan when both are the view of Manhattan from the
Empire State Building, the one that, say, a tourist guide describes. In those con-
texts there is no difference between the role of first-person singular pronouns
and the role of first-person plural pronouns, the distinction between me and
us lacks significance. If my skepticism regarding the existence of an external
world is justified, so is our skepticism regarding it, and if the latter is justified
then so is my skepticism.
Needless to say, this is not true in most nonphilosophical contexts. In every-
day life as well as in science, “I” usually indicates, say P.B., and “we” indicates
P.B. and you. The difference is obvious. P.B. weighs 160 pounds, but P.B. and you
together weigh 300 pounds. P.B. visited Charleston last year, we did not. But
when doubting the existence of the world or considering the question of its
mind-independence, there is no relevant difference between me and us, between
my doubt and our doubt, between my stand and our stand. There are differences
only when we assume that P.B. and you are parts of the world, a natural assump-
tion in nonphilosophical contexts but one that cannot be made when doubting
the existence of that world or considering whether it is mind-independent.
In the philosophical contexts in which the use of the first-person singular
pronoun “I” is impersonal, the use of the first-person plural pronoun “we” is
also impersonal, and for the same reason. This impersonal use of “we” is not
limited to philosophy. It is common in science, as in “We know that the speed
of light cannot be exceeded,” the impersonal sense of which becomes explicit
when restated as “It is known that the speed of light cannot be exceeded,”
“Physics has found (established, discovered, confirmed) that the speed of light
cannot be exceeded,” or “According to physics, the speed of light cannot be ex-
ceeded.” An impersonal use of “we” is also common in moral contexts, as in
some occurrences of “We don’t torture prisoners,” the impersonal sense of
which becomes explicit when restated as “Torturing prisoners is unacceptable”
or “Morality prohibits the torture of prisoners.” It is common also in etiquette, as
in “We don’t talk with the mouth full” and “We don’t wear flip-flops at wed-
dings,” which mean that talk with the mouth full and wearing flip-flops at a
wedding are violations of table manners, custom, proper dress rules, or expect-
ation. Clearly, in these cases, “we” does not refer to any particular persons’
knowledge, moral attitude, table manners, or dress rules. This is made explicit
when the speaker is asked “Who knows that the speed of light cannot be exceed-
ed?”, “Who doesn’t torture prisoners?”, “Who doesn’t talk with the mouth full?,
or “Who doesn’t wear flip-flops at weddings?” and replies, “I didn’t mean any-
one in particular.”
The deeper explanation of the occasional interchangeability of “I” and “we”
even in nonphilosophical contexts is that there cannot be a drastic divergence

Authenticated
between my sense perception or concepts and those of other persons. The “other”
is taken by me to be capable of judgments that I must consider only if also taken
to perceive and understand the world at least roughly as I perceive and under-
stand it. Otherwise, not only I would not understand what the other says, I
may not recognize it as language. Verbal communication with humans presup-
poses that we are not speaking about different things or in a different lan-
guage.³³⁵ This is why Wittgenstein held that for language to be possible there
must be agreement in judgment. Even nonverbal communication with nonhu-
man animals presupposes that they perceive, however differently, the same
things that we perceive. We place the food tray where both we and the cat can
see it, and we expect the dog to come through the door that we leave open.
The distinction between one’s own and others’ cognition of the world has little
application in the case of advanced cognition. Its application to perception (includ-
ing introspection) may be clear, but beyond early childhood one’s cognition almost
always includes conception, understanding. And then it depends essentially on
what one learns from others, including the common, public language normally mas-
tered in childhood. It depends not on the aggregate of others’ cognitions but rather
on a systematic impersonal whole that seems to have life and properties of its own.
This is familiar in the case of scientific cognition, but it is true also of everyday cog-
nition, say, of a country, a town, or an automobile.
We usually acknowledge that our beliefs and opinions are at least partly due
to our parents and teachers, the books and articles in magazines and newspa-
pers we read, the people we talk with at home and in the street or have heard
in the classroom, on radio, or on television. If one has knowledge that is strictly
“one’s own,” underived from and unaffected by others, it is of the sort enjoyed
by neonates and qualifies as knowledge only barely. Educated persons explicitly
or implicitly take physical things to be as physicists say they are, the past as his-
torians say it was, and mathematical truths as those accepted by mathemati-
cians. And physics, history, and mathematics are inherently social. The require-
ment in all sciences that experiments be repeatable – by others, not just by
oneself – is sacrosanct. The 2009 status report by the Large Hadron Collider list-
ed 2900 authors. The Wikipedia on the Internet is committed to impersonality.
No single person can master today the whole or even a major part of a discipline.
A law is often so complex that no single attorney knows everything in it or has
even read it. Orthopedists who do knee surgery usually decline to do hip or ankle
 Cf. Donald Davidson, “The Very Idea of a Conceptual Scheme,” Proceedings and Addresses
of the American Philosophical Association 47 (1974), 5 – 20.

Authenticated
surgery. Even in everyday life, individual claims to knowledge count for little if
they do not survive scrutiny by others.
The cognitive dependence of oneself on others in such cases is not just caus-
al. It is logical. This was an essential thesis of Wittgenstein’s later philosophy as
well as of Sartre’s. Much of the history of 20th century Anglo-American philoso-
phy was epitomized by Wittgenstein’s move from the thesis in the Tractatus that
the world is “my” world to the thesis in the Philosophical Investigations that a
private language is logically impossible and that understanding a public lan-
guage presupposes “agreement in judgment.” Many consider the highlight of
20th century continental philosophy to be Sartre’s argument that only through
the Other’s “look” can I see myself as an object, as the entity in the world
that I am – that the Other “defines” me. Much of contemporary thought, e. g.,
feminism, depends on it, sometimes explicitly.
But it is to Hegel that we must go for the original account of the logical de-
pendence of oneself on others. Hegel insisted on the necessity of a move from
“the I (das Ich) of the colourful show of the sensuous here-and-now” to the “I
that is We and We that is I” (Ich, das Wir, und Wir, das Ich ist).³³⁶ It was a
move from individual cognition to collective cognition. As Terry Pinkard has
put it, Hegel reconceptualized “the unity of thought and being” as an “intersub-
jective unity.”³³⁷ J. N. Findlay explains: “Hegel holds that the understanding of
other minds, far from being more obscure than the understanding of things, is
the model and paradigm in terms of which intercourse with things can assume
a limited clarity.”³³⁸ But Pinkard and Findlay are downplaying the metaphysical
significance Hegel attached to this view. Recognition of the primacy of society
over the individual is only a step, however necessary, to grasping the Absolute,
he held. Human cognition may be social, but the Absolute transcends society.
According to Hegel, Spirit (mind, Geist) develops from its primitive stage of
subjective spirit – sensuous cognition – to its second stage of objective spirit, by
which he meant the normative customs and traditions of everyday life [Sittlich-
keit], the family, the state, and institutions such as corporations and professional
guilds (which would include what today we call the academic disciplines), and
culminates in its third stage of absolute spirit, which includes art, religion, and
the complete, perfect, knowledge that Hegel called philosophy. If by “mind” is
 Phenomenology of Spirit, 177.

 Terry Pinkard, “Hegel’s Phenomenology and Logic: an overview,” in Karl Ameriks, ed., The
Cambridge Companion to German Idealism (Cambridge: Cambridge University Press, 2000), 164.
See also Pinkard’s Hegel’s Phenomenology: The Sociality of Reason (Cambridge: Cambridge Uni-
versity Press, 1996).
 Phenomenology of Spirit, analysis of 177.

Authenticated
meant cognition, then at the stage of sensation the mind is primitive, undevel-
oped, and comes into its own only at the social stage.
Nevertheless, we seem still left with anthropocentrism. Wittgenstein’s public
language surely is a human language, Sartre’s Other is a human being, and He-
gel’s objective spirit is a society of humans, not of ants or Martians. (Hegel ex-
plicitly held even that the Absolute achieves self-knowledge through human be-
ings, those he regarded as philosophers, his position thus appearing to be not
just anthropocentric but anthropomorphic.) However, though all this is true, it
is not the whole truth. Statements about a public language, Wittgenstein’s agree-
ment in judgment, and Hegel’s social institutions are not reducible to statements
about human beings, even though the language, agreement, or institutions
would not exist if human beings did not. The public that speaks the public lan-
guage, the parties to the agreement in judgment, and the institutions of society
are not mere collections of human beings. Nor are they, of course, themselves
human beings. In an important sense, they are impersonal. Hegel and Wittgen-
stein could still be charged with anthropocentrism in an extended and rarified
sense, but their views also suggest how anthropocentrism can be avoided, the
topic of the next chapter.
Hegel famously thought that the discipline of philosophy itself was an exam-
ple of the priority of society over the individual. For philosophical views are in-
separable from the history of philosophy. Earlier philosophical systems, despite
their diversity, are preserved in those developed later.³³⁹ Hegel was fully aware
that a philosopher’s thought, however original, is rooted, not only in the culture
to which it belongs and the language it employs, but also in the thought of other
philosophers, past and present, read or heard. The content of a philosophical
work is incomprehensible in abstraction from that of previous philosophical
works. Instruction in philosophy has always been mainly instruction in the his-
tory of philosophy, even if only the history of its latest period and in a single
country. The same is true, of course, of the other disciplines.
Commentators sometimes say that when Hegel described philosophy as the
most advanced stage in the development of spirit he meant his own philosophy.
Indeed, he did hold that the Absolute achieves self-knowledge through humans.
But Hegel would have insisted that no individual human being’s knowledge
could be identified with the Absolute’s self-knowledge. Perhaps he thought, or
at least hoped, that his system was the closest a human being could come,
though I doubt that he did. If he did, he would not have been unique among phi-
losophers in this respect.
 Hegel’s Logic, § 86.

Authenticated
That advanced cognition is social is obvious in the case of cognition that

necessarily involves the use of language. Some cognition, of course, is possible
without language. A neonate’s and a dog’s sense perception are examples. Dis-
tinctively human cognition, however, like that exemplified by the sciences and
mathematics, obviously is not. Perhaps neonates and dogs have an innate “lan-
guage of thought” that is needed for sense perception, and perhaps even God
could not know the world without employing a divine language, but neither pos-
sibility need be discussed here. Our concern is with distinctively human cogni-
tion, the sort of cognition neonates and dogs clearly lack and God is believed
to surpass. Wittgenstein’s perhaps most persuasive reason for holding that lan-
guage is necessarily public, “social,” was his description of an imaginary private
language for keeping track of one’s sensations. Even if possible, such a language
would be so impoverished as to be pitiful. A public language is indispensable to
any cognition deserving to be called human. And it does presuppose agreement
in judgment – if in nothing else, at least in which sounds or marks to count as
words and which words those sounds or marks happen to be.
Until Hegel, epistemologists – from Plato and Aristotle to Descartes, Hume,
and Kant – took for granted that cognition is individual, personal. In the Intro-
duction I called this view cognitive individualism, contrasting it with what I
called cognitive collectivism. Hegel saw that it is true only on the primitive, sub-
jective, level, and that advanced, truly objective, cognition is necessarily social.
“It is the nature of humanity,” he wrote, “to press onward to agreement with oth-
ers; human nature only really exists in an achieved community with others.”³⁴⁰
Hegel became famous partly because of the moral and political implications his
views were taken to have. Cognitive collectivism, however, has no such implica-
tions. One of its elementary tenets is that advanced cognition requires a public
language. This is hardly a moral or political tenet.
Initially, as in Kant’s version, antirealism insists on the dependence of the
world on my cognition. In Hegel’s version, it insists on its dependence on our
cognition. It does so also in its recent versions, for example, Goodman’s, Dum-
mett’s, and Putnam’s, which focus on the dependence of the world on our lan-
guage – public, not private, language.
To say that the world depends on me seems absurd. To say that it depends
on us may appear to just multiply the absurdity. But what is at issue is depend-
ence on cognition, and the dependence of the world on our cognition is less par-
adoxical. Its dependence on me would be abhorrent to common sense because it
 Hegel’s Phenomenology of Spirit, trans. A. V. Miller (Oxford: Oxford University Press, 1977),
43.

Authenticated
makes the world, the paradigm of what is objective, appear subjective. But its
dependence on us may seem less alarming when understood as dependence
on advanced cognition. Not only is the latter not dependence on a solitary per-
son, it is not dependence even on a collection of persons. It is a dependence on a
cognition that is irreducibly institutional. There is no mystery about what such a
cognition might be. The cognition, knowledge, embodied in any developed dis-
cipline would be an example. As cognition, a body of presumed knowledge,
physics is not reducible to the collection of individual physicists’ cognitions.
And while it would be glaringly absurd to hold that the physical world is what
I say it is, it is not absurd to hold that it is what physics says it is. The move
from “I” to “we” does not signify total abandonment of anthropocentrism. But
it does signify abandonment of subjective anthropocentrism.
For Berkeley, the paradigm of cognition was perception, not conception. (He
argued vigorously against what Locke had called “abstract ideas.”) Indeed, this
is the immediate, natural, way to think of cognition. It explains why we also
think of our cognitive faculties as personal, not social, and of knowledge as a
personal, not social, achievement. It explains why we find cognitive individual-
ism more plausible than cognitive collectivism and why antirealism usually has
been understood as subjective. Cognitive collectivism comes into its own mainly
when we recognize the role in cognition of conception, the understanding, and
especially of description, language. While perception is inherently individual,
personal, and subjective, conception, and especially description, is inherently
social, public, and objective. A neonate’s cognition of Mother is entirely percep-
tual and dependent on no one else’s perception of Mother. But the child’s later
conception of Mother as mother is dependent on others, obviously so when ex-
pressed in language.
Antirealism is far more plausible in the case of conception than in the case
of perception. Moreover, if perception presupposes conception, as held by both
Kant and Hegel, and in recent philosophy by Wilfrid Sellars and many others,
then conception is primary. Even if Berkeley’s idealism could be defended on
the level of perception, it was Kant’s transcendental idealism and Hegel’s abso-
lute idealism that would be plausible on the level of conception. These “ideal-
isms” were the forms antirealism took in traditional philosophy. It cannot be un-
derstood in abstraction from them.
3 Idealism
Berkeley made idealism a technical philosophical position, Kant’s transcenden-

tal idealism inaugurated modern antirealism, and Hegel’s absolute idealism

Authenticated
3 Idealism 213
seemed to free it from anthropocentrism. Berkeley’s idealism rested on the asser-

tion that to be is to be perceived, a version of the thesis of antirealism limited to
our perceptual cognitive powers. This is why, though it seems incredible, there is
no puzzle about what it says – similar thoughts have occurred to many teen-
agers. Kant’s transcendental idealism acknowledged the essential place among
our cognitive powers of conception, the “understanding,” but it is not likely to
occur to teenagers. Hegel’s absolute idealism seemed to avoid the anthropocen-
trism latent in Berkeley’s as well as Kant’s versions, and as a consequence it has
baffled even philosophers.
According to the idealist, everything is mental. According to the antirealist,
everything is dependent on our cognitive faculties, at least insofar as it is know-
able. But cognitive faculties are paradigms of what is mental. The idealist and
the antirealist agree that insofar as the world is knowable, it is mind-dependent.
Moreover, the idealist is likely to support idealism with many of the same argu-
ments that the antirealist employs. But, despite their obvious similarities, ideal-
ism and antirealism differ in important ways. Idealism avoids a major difficulty
faced by antirealism: the putative difference between a cognition and its object.
The idealist denies that there is such difference, claiming for example that the
perception of a tree and the tree perceived are no more distinguishable than
are a feeling of pain and the pain felt.
Understood as holding that everything is mental, idealism is often rejected
as obviously false, but sometimes also greeted as a pleasingly warm view of
the universe, which its chief 20th century critic G. E. Moore said he wished
were true.³⁴¹ Antirealism, by contrast, stays in the bloodless realm of philosoph-
ical technicality and arouses neither enthusiasm nor alarm, even though it may
seem to have absurd implications, for example, that there was no earth before
there were humans.
Though they often employ the same arguments, idealism does not entail an-
tirealism and antirealism does not entail idealism. The idealist says that all real-
ity is mental, the antirealist usually does not. What is mental need not be de-
pendent on our cognitive faculties, and what is dependent on our cognitive
faculties need not be mental. Idealism is a theory about the nature of reality. An-
tirealism is not. It says nothing about the constituents of reality. Its thesis that
knowable reality depends on our cognitive faculties says nothing about the na-
ture of that reality. Whether reality is mental remains for the antirealist a further
question. How it is answered, indeed how idealism is ultimately understood,
 G. E. Moore, “The Refutation of Idealism,” included in Philosophical Studies (New York:
Harcourt Brace, 1922).

Authenticated
would depend on what meaning is attached to “mental.” Presumably the idealist

does not claim that rocks, firewood, and puddles of water are mental in any fa-
miliar sense of “mental.” In everyday discourse they are paradigms of what is
not mental, just as perceiving, thinking, and feeling are paradigms of what is
mental. If we use “mental” in accord with those paradigms, we must say that
if the idealist is right, then there are no such things as rocks, firewood, or pud-
dles of water.
In his classic paper “The Refutation of Idealism,” which marked the end of
the dominance of idealism in British philosophy and inaugurated 20th century re-
alism, G. E. Moore defined idealism as the view that “the Universe is spiritual.”³⁴²
This in effect means that reality is either a spirit or consists of states of one or
several spirits. Moore used “spirit” as a synonym of “mind,” a spiritual state
being what today would be called an irreducibly mental state, in particular, a
state of consciousness. According to Moore, the idealist’s thesis (he obviously
had Berkeley’s in mind) that to be is be perceived seems plausible only because
it confuses the object of (perceptual) consciousness with its content, a confusion
made possible by the fact that consciousness itself has no content, that it is di-
aphanous, transparent. Moore thus anticipated Sartre, who decades later wrote
that consciousness has no inhabitants and exhausts itself in its objects. But,
as we saw earlier, Sartre also proposed that the activities traditionally attributed
to the ego are performed by an impersonal consciousness that “constitutes” the
world, and that the ego is no more than one of the objects in the so-constituted
world, not the subject that constitutes it. He thus combined the rejection of Ber-
keley’s idealism with a quasi-Kantian transcendental idealism. Moore would not
have endorsed the latter, but he did agree with Kant that the idealist, again pre-
sumably meaning Berkeley, employs the verb “perceive” in a sense that excludes
conception, thus impoverishing idealism.
When asserting that to be is to be perceived, Berkeley could not have consistent-
ly regarded the subject-term of “perceive” to be the name “Berkeley,” for Berkeley,
the 18th century bishop, had a material body even if also an immaterial soul. The
subject-term therefore could only be “I,” understood as standing, not for Berkeley,
but for a self or ego, which may belong to but is not identical with a human being.
Hume, Wittgenstein, and Sartre looked for such an entity but failed to find it. With-
out it, however, there would not be the perceiving to which Berkeley appealed as the
measure of what there is. For there would be nothing that would engage in that per-
ceiving, even if there were something perceived.
 Philosophical Studies, 1.

Authenticated
3 Idealism 215
Idealism became dominant in philosophy after philosophy took the “new

way of ideas” in the 17th century, and its development was dependent on the
sense attached to “idea.” Descartes and Locke proposed that the primary and
perhaps only objects of knowledge are our ideas, though neither was an idealist.
Descartes wrote, “Of my thoughts some are, as it were, images of things, and to
these alone properly belongs the name IDEA.”³⁴³ Locke used the term in a similar
sense: “the term [‘idea’] …stand[s] for whatsoever is the Object of the Under-
standing, when a man thinks.”³⁴⁴ This sense was too broad, however, because
it implied that if material objects were directly perceived, which both Descartes
and Locke denied, they would count as ideas, which both would have found ab-
surd. Locke explained further that by “idea” he meant, not only concepts (“ab-
stract ideas”) and mental images, but also sensations and feelings (“passions”).
This use of “idea” was entirely different from Plato’s use of ιδɛα, to which none-
theless it must be traced.
Berkeley used “idea” in Locke’s sense, but earned his classification as an
idealist by holding that minds and their ideas are all that there is, a view Des-
cartes and Locke would have found appalling. Hume used “perception” in
Locke’s sense of “idea,” but distinguished within perceptions between “impres-
sions” and “ideas”: “Those perceptions, which enter with most force and vio-
lence, we may name impressions….By ideas I mean the faint images of these
in thinking and reasoning.”³⁴⁵ Both Descartes and Locke distinguished sharply
between ideas and their owner, Descartes’ “thinking thing” or Locke’s “self.”
German philosophers called it “ego,” or just “the I” (das Ich). But by denying
that he could find such an entity within himself, Hume originated the totally dif-
ferent line of thought that culminated in Wittgenstein and Sartre.
Kant used Vorstellung, commonly translated as “representation,” roughly in
Locke’s broad sense of “idea,” but he made a sharp distinction between “intu-
itions (Anschauungen),” which are “immediately related to the object” and are
“singular,” and concepts (Begriffe), which are “mediate,” related to objects by
means of “marks” that “can be common to several things.” Such common
“marks” are what philosophers call universals. A concept, Kant held, is either
empirical or pure: “Pure concepts have their origin solely in the understanding,
not in images of sensibility, and are called ‘notions.’” Kant used also “idea”
(Idee), and like Descartes’s and Locke’s his use was rooted in Plato’s ιδɛα. But
he described ideas very differently: “A concept made up of notions, which
 Meditations, III.

 Essay Concerning the Human Understanding, I, 1, 8.
 Treatise of Human Nature, Book I, Part I, Sect. I.

Authenticated
goes beyond the possibility of experience, is an idea or a concept of reason.”³⁴⁶

Unlike the concepts of the understanding, the ideas of reason have no applica-
tion to experience. We have no knowledge of God, freedom (of the will), and im-
mortality, but we do have ideas of them, he famously held. We have no experi-
ence of God, freedom, or immortality, or of anything from which they can be
validly inferred, but God, freedom, and immortality are unavoidable “objects”
of reason, and as such are essential to morality.
Hegel distinguished between the “several modes of feeling, perception, de-
sire, and will,” which “are in general called ideas (mental representations [Vor-
stellungen]),” and “thoughts [Gedanken], categories, or, in more precise lan-
guage, adequate Notions [Begriffe],” which philosophy puts “in the place of
the generalized images [Vorstellungen] we ordinarily call ideas.”³⁴⁷ But he ac-
knowledged that “the mind [consciousness, das Bewusstsein] makes general im-
ages of objects, long before it makes Notions of them”³⁴⁸ While in the case of
Kant the usual English translation of Begriff has been “concept,” in the case
of Hegel it has been “Notion,” in order to mark Hegel’s distinctively metaphysical
understanding of the nature of Begriffe. The notion of something is often said to
express its essence or nature. According to Hegel, however, it does not merely
express but is that essence: “[I]n point of contents, thought is only true in pro-
portion as it sinks itself in the facts.”³⁴⁹ This assertion may be puzzling, but it
would be less so if understood as a statement of the thesis of antirealism.
Though Hegel described notions as thoughts, he so described also judg-
ments and syllogisms. Gedanke in German is as ambiguous as “thought” is in
English. It can refer to (1) a thinking, a subjective mental action or event, whether
a judging, a supposing, or a reasoning, (2) the objective content of a judgment
(as it does in Frege’s writings), or (3) a component of that content. For Hegel,
it referred to a notion only when used in the third sense. This is also the main
sense of “concept” in contemporary Anglo-American philosophy. But by “the no-
tion of x” Hegel meant the comprehension, grasp [Griff] of x, which is a sense
much stricter than the usual sense of “concept.” Hegel called the notion “the
soul [Seele] of objective reality.”³⁵⁰ It is in this sense that he held that the Abso-
lute is Thought, and thus earned his own classification as an idealist. Hegel did
not mean by “the Absolute” some mysterious object or part of reality. He meant
reality itself. And reality of course must be understood as relative to nothing, as
 Critique of Pure Reason, B 376 – 377.

 Hegel’s Logic, § 3.
 Hegel’s Logic, § 1.
 Hegel’s Logic, § 23.
 Science of Logic, 597.

Authenticated
3 Idealism 217
unqualified, unconditional, all-encompassing, self-contained, complete, and in-

finite in the literal sense of unlimited, “limited by nothing.”
Finite things are limited, incomplete, parts of reality. They are dependent on
the Absolute for both their nature and their existence, but the dependence is nei-
ther causal nor logical. It is better understood as analogous to the dependence
on a biological organism of its parts. The Absolute is itself dependent on its finite
parts, as an organism is dependent on its parts. Hence Hegel’s characterization
of the Absolute as concrete, not a mere abstraction. Knowledge of it must be an
organic whole, in the sense of being systematic. Hegel famously demanded that
philosophy be systematic. But, we may note, this is true of any advanced cogni-
tion, e. g., a theory in a developed science. The statements in it depend on each
other for both their content and truth. Its concepts presuppose each other.
Although Hegel did call his system “absolute idealism,” it would be super-
ficial to interpret him as holding that everything is mental. His idealism bore lit-
tle resemblance to Berkeley’s. It was much closer to what we have meant by “an-
tirealism.” If antirealism did not allow for Kantian things-in-themselves, as
Hegel did not, then the antirealist thesis that the world depends on our cognition
of it need not be different from Hegel’s thesis that they are the same. They are the
same in the sense in which a headache and the feeling of that headache are com-
monly believed to be the same. Berkeley thought that what would be true of un-
felt pains is true also of unperceived trees. If he was wrong, the reason was not
his belief that trees are dependent on cognition but his refusal to acknowledge
modes of cognition other than perception that provide cognitive access to trees.
The difference between Berkeley and Hegel is commonly said to be that
while Berkeley’s idealism was subjective, Hegel’s was objective. These character-
izations are not wrong, but they require deeper explanation and major qualifica-
tions. We can say that according to the subjective idealist the world depends on
me, while according to the objective idealist it depends on us. But this would not
allow us to describe Berkeley’s idealism as subjective, because he insisted that
something would exist even if perceived only by God. And it may not do justice
to Hegel’s idealism. Hegel did not hold that the world depends on human beings.
He insisted on the social nature of developed human cognition, and insofar as
he also held that the world depends on such cognition his idealism was objec-
tive, in the straightforward sense in which, say, physics is objective while person-
al fancy is subjective. But Hegel described his idealism as absolute, as holding
that the Absolute is Thought. And this “thought” is neither subjective, if this
means that it belongs in subjective spirit, nor objective, if this means that it be-
longs in objective spirit. It belongs nowhere.

Authenticated
Chapter Twelve: Mind and the World
1 Concepts, Properties, and Universals
The relativity of an object to cognitive perspective in the case of visual perception is

familiar because in optics, painting, photography, and everyday life we must view
and represent the object from an “angle,” and what is seen depends in part on
that angle. Much the same is true of the other species of perception. To feel an object
one must touch it with part of one’s body, and what is felt depends in part on the
characteristics of that part of body. What one hears notoriously depend on the con-
dition of one’s hearing. The relativity of an object to cognitive perspective in the case
of understanding, conception, is different, however. It requires explanation.
The most original and extensive explanation was proposed by Hegel. It was,
in effect, the final of the five steps listed in Chapter Ten as needed to free anti-
realism from anthropocentrism. Put briefly and somewhat crudely, it is that (1)
understanding involves thoughts, (2) thoughts are concepts, (3) concepts are
properties of things, (4) properties are universals, (5) there is nothing but their
properties in things, therefore (6) there is nothing in things but thought. There
are no Lockean substrata or Bergmannian bare particulars. “[T]he universal of
the notion is not a mere sum of features common to several things, confronted
by a particular which enjoys an existence of its own. It is, on the contrary,
self-particularizing or self-specifying.”³⁵¹
Whether the world is the totality of facts or the totality of things, it may thus
be described as conceptual. But the concepts the world involves are no more per-
sonal than are the properties of the things in the world. They are in the world, in
the things that have them. No anthropocentrism is implied.
The noun “thought” may mean thinking, something one does, but more
often it is used as a synonym of “concept” and “idea.” “Concept” became com-
mon in 20th century Anglophone philosophy as replacement of “idea,” which in
17th and 18th century British philosophy had explicitly stood for a subjective state,
often a mental image. It is the usual translation of the German Begriff. The latter
was a key term in Kant’s and Hegel’s works, which profoundly influenced British
philosophy in the 19th century, as well as in Frege’s, which influenced it no less
profoundly in the 20th century. As I noted earlier, in the case of Hegel Begriff is
usually translated as “Notion,” because of his distinctive metaphysical view of
the nature of Begriffe.
 Hegel’s Logic, § 163.

Authenticated
1 Concepts, Properties, and Universals 219
Understood as subjective states of the mind, ideas are incapable of serving

as principles of objective classification. But, except perhaps at its most rudimen-
tary stages, cognition necessarily involves objective classification, whether ex-
plicitly, as in botany and zoology, or implicitly, as in ordinary subject-predicate
statements. To say that this page is white is to assign it to the class of white
things. If we take classification to be something that we do, we may say that it
consists in applying a concept to what is classified, whether by using a general
word (adjective or common noun) or by thinking of it in a certain way. The aim,
of course, is objective classification. We seek truth, not fancy.
Locke had stressed the importance of classification, but he claimed that its
principles are “abstract ideas,” obtained from particular ideas by leaving out the
characteristics that distinguish them from one another. The existence of such
ideas was vigorously denied by Berkeley and Hume, on the reasonable grounds
that there are no abstract mental images. The principles of objective classifica-
tion were called by Kant and Hegel concepts (Begriffe). This is also how the
word “concept” is understood in current Anglophone philosophy, even when
limited, as it often is, to the meanings or uses of words – what a word means
or how it is used is of interest in philosophy only if it is an objective fact, not
personal whim. A classification, of course, could be subjective, fanciful, but
then it would be unsuitable for cognition.
If a classification is objective, the concept that serves as its principle is usually
taken to stand for a property common to the things classified, for what in metaphy-
sics is often called a “universal.” The metaphysical realist holds that the things clas-
sified have (exemplify, instantiate) the property for which the concept stands inde-
pendently of the application of the concept or any other mode of cognition. The
concept and the property are taken to be distinct entities, the former consigned to
the mind or language of the classifier, and the latter to the object that exemplifies
it. The metaphysical antirealist rejects this distinction between properties and con-
cepts. A subjective antirealist, perhaps Berkeley, does so by regarding concepts as
subjective states (“ideas in the mind”), and by denying that there are things external
to the mind in effect denies that concepts stand for properties of things. Hegel re-
jected the distinction between properties and concepts by taking concepts to be
the same as the properties of things. They are neither subjective states of human
minds nor, of course, objective states of human brains.
Hegel was explicit about the intimate connection between the category of
concepts, which initially seems psychological or epistemological, and the meta-
physical category of properties. There is no thought without concepts, there are
no things without properties, and concepts are the same as the properties of
things. This is why the world and cognition of it are one and the same. The
world is not just dependent on mind, it is mind: “thoughts, far from being merely

Authenticated
220 Chapter Twelve: Mind and the World
ours, must at the same time be the real essence of the things.”³⁵² Hegel insisted
that “It is not we who frame the concepts….Rather the concept is the genuine
first; and things are what they are through the action of the concept, immanent
in them, and revealing itself in them.”³⁵³ This is why the Absolute, i. e., ultimate
reality, is identical with Thought.
Hegel’s rejection of the distinction between concepts and properties was the
essential tenet of his version of antirealism. It was an “objective” antirealism,
free of anthropocentrism. The properties of things “make,” constitute, the things
in the world, they are the principles of the objective classification of them and so
may be called concepts, but they are not states of human or any other minds,
they are in the things themselves.
It is usually admitted today that in the case of understanding cognition of
things involves concepts of their properties; this is why instead of understanding
we often speak of conceptualization. Few would agree with Hegel that the con-
cepts of the properties of things are the properties themselves. They probably
would agree, however, that concepts are neither subjective mental states nor
just the words expressing them. They would agree with Sartre that consciousness
has no inhabitants, and with H. H. Price that cognition can occur without the use
of words, as it often does in the case of recognition. They would agree that to
understand something is neither to imagine it nor to utter its name, but they
would not deny that we can think of, even know, some properties of some things,
whether perceived or not.
Hegel’s assertions that the absolute is thought and that thought is “sunk” in
things are often misunderstood because of failing to grasp his position regarding
universals on which they rest. “The form or character peculiar to thought is the
universal, or, in general, the abstract,” he wrote.³⁵⁴ Hegel’s position was radically
different from Locke’s, Berkeley’s, or Hume’s. It was closer to Aristotle’s view
that universals are in things, but it also shared important elements with Plato’s
view that they are not parts of things. The position becomes clearer when under-
stood in the context of the traditional debate about universals and particulars.
By a “universal” is usually meant a property that can be common to, exem-
plified by, two or more individual things, e. g., the white color of this and the next
page. Common properties are “universals” in the straightforward and innocuous
sense of “Applicable to or involving the whole of a class or genus, or all the in-
dividuals or species comprising it… Opposed to particular” (Oxford English Dic-
 Hegel’s Logic, § 41.

 Hegel’s Logic, § 163.
 Hegel’s Logic, § 20.

Authenticated
tionary). Universals so understood are what in everyday life we take general

terms, i. e., adjectives and common nouns, to stand for. In logic properties are
the values of predicate variables, say, Φ, individual things (“particulars”)
being the values of individual variables, say, x, and the logical form of a nonre-
lational atomic sentence usually represented as Φx. So understood, there is no
“problem” of universals, for there is no temptation to think that there cannot
be common properties, that, e. g., this page and the next page cannot be both
white, any more than there is for thinking that two persons cannot have the
same father or that two ships cannot be headed in the same direction. Realism
regarding universals thus appears true.
The “problem of universals” arises when we assume that a property of things
must be somewhere and then ask, “Where is it?” Plato’s answer seemed to be
that it is nowhere, not in space, but then he added to the puzzle by saying
that it is in a special, separate, realm of being, the “world of Forms.” More plau-
sible seemed to be the answer that a common property is “present” in the things
that have it as a part of them, that, e. g., the white color of this and the next page
is a part of both pages even though they have different locations in space. But
then one may ask, how can something be in two places at the same time? The
antirealist regarding universals, the “nominalist,” would say that it cannot,
and thus deny that there are universals.
There is little, however, to recommend the assumption that the properties of
things are “somewhere” and therefore perhaps parts of them. If I am 6 feet tall, it
would be nonsense to suppose that being 6 feet tall is somewhere, let alone a
part of me or indeed of anything. A position on universals that might be called
“semirealist” appears preferable: there are properties but they are not parts of
the things whose properties they are and thus they are not where the things
are or anywhere else. Philosophers often employ, instead of “part,” the term
“constituent,” saying that a property is a constituent of what has the property,
but this merely disguises the nonsense.
All along, the dispute between realism and antirealism regarding universals
has been stated as if it concerned a matter a fact. But, surely, it does not. No em-
pirical facts about this page and the next are in question when we ask whether
they have the same color, and if any were we would settle them not by philo-
sophical discussion, contemplation, intuition, or analysis but by just looking
at the pages, perhaps also employing a scientific instrument. And surely it is
not nonempirical facts that are in question. Even if they were, it would be un-
clear why philosophers should enjoy special knowledge of them. Those who sup-
pose that the truths of arithmetic correspond to nonempirical facts about num-
bers leave the study of them to mathematicians.

Authenticated
The “puzzle” about where properties might be was felt mostly under the influ-
ence of the British empiricists, who held that all things must be actual or possible
objects of perception and therefore found themselves unable to imagine an individ-
ual object as being anything other than the collection of its perceived properties. The
properties must, therefore, be where the thing is. But the idea that a property must
be perceivable in the way the things that have it are perceivable is like the idea that
the 20th century must be perceivable (seen? touched? heard? smelled?), and the idea
that a property must be “somewhere” is like the idea that the 20th century must be
somewhere (in Europe? Asia? Africa? South America?) Both confuse conception
with perception, and properties with the things that have properties. On these mat-
ters, Plato and Frege were closer to the truth. The color of an apple is not in the
apple in the way a worm might be.
When discussing the problem of universals it is convenient to employ the
terminology proposed by H.H. Price in his classic work Thinking and Experience.
He used the phrase universalia in rebus for properties as understood by Aristotle:
universals are “in” things as their properties. Universalia post rem are not prop-
erties but the subjective states, “ideas,” perhaps mental images, that are taken to
correspond to general terms, or those terms themselves. Universalia ante rem are
universals as understood by Plato, entities that exist independently of both spa-
tiotemporal objects and any subjective states. Berkeley and Hume, of course, de-
nied that there are universalia ante rem. But they also denied that there are uni-
versalia in rebus, on the grounds that the properties of particulars are themselves
particulars. This position was developed and defended two centuries later by D.
C. Williams.”³⁵⁵ It is known today as “trope” theory.
The view Price preferred was “the philosophy of universals,” essentially Ar-
istotle’s view of universals as properties of things, universalia in rebus. He was
especially clear about the intimate connection between the category of concepts
and the category of properties. Concepts are not words or mental images, but
rather “recognitional capacities.” They are principles of classification. As
such, concepts are not clearly distinguishable from the corresponding classes
of individual things and thus from the corresponding properties. Indeed, the
very distinction between class and property becomes murky. Goodman and
Quine argued at one time for “nominalism” regarding both properties and
classes, on the grounds that both are abstract entities: “We do not believe in ab-
stract entities. No one supposes that abstract entities – classes, relations, proper-
 D. C. Williams., “The Elements of Being,” Review of Metaphysics 7: 3 – 18, 171– 192. I have
discussed the view in detail in Resemblance and Identity: An Examination of the Problem of Uni-
versals (Bloomington & London: Indiana University Press, 1996).

Authenticated
ties, etc. – exist in space-time; but we mean more than this. We renounce them
altogether.”³⁵⁶ Later, Quine abandoned this stand because he believed that
classes (sets) are required in mathematics.
Price was not an antirealist, but he might have agreed that as a matter of
fact, not philosophical argument, we cannot distinguish between what the
world really is and how we understand it. It is commonplace to speak of the
child’s world, the soldier’s world, the scientist’s world, and so forth. Such de-
scriptions do not refer to different portions of the world. They refer to the
world itself, but as understood or conceptualized in a certain way: the child’s,
the adult’s, or the scientist’s.
Hegel rejected Plato’s universalia ante rem, but he also had no patience with the
British empiricists’ universalia post rem, and probably would have dismissed trope
theory as failing to “see the universal in the particular.” At first glance, therefore, he
may seem to have accepted Aristotle’s view of universalia in rebus. Indeed, we find
him writing, “Animal, qua animal, does not exist: it is merely the universal nature of
the individual animals…But to be an animal … is the property of the particular an-
imal, and constitutes its definite essence. Take away from the dog its animality, and
it becomes impossible to say what it is.”³⁵⁷ This remark is best understood in the
context of Aristotle’s Categories. It concerns what Aristotle called “secondary sub-
stances,” e.g., animal and dog. In chapter 5 I called substance universals. They
are the genera and species of “primary substances,” that is, of individual things,
e.g., this animal, this dog. Primary substances, according to Aristotle, are neither
“present in” nor “said of” (i.e., predicable of) anything. But he made clear that
by “being present in I do not mean present as parts are present in a whole, but
being incapable of existence apart from the said subject.” Secondary substances
are said of but are not present in individual things. For centuries, animality (“the
Animal”) served as a standard example of a secondary substance, but so would
have caninity (“the Dog). The latter is a species of the former, which is its genus.
Hegel’s remark “Take away from the dog its animality, and it becomes impos-
sible to say what it is” applies also to those of Aristotle’s “accidents” that are pres-
ent in and also said of things. We may call them universal properties, e.g., round
shape and white color. But “a certain whiteness may be present in the body…. yet
it is never predicable of anything.” Aristotle wrote. Take away the whiteness of
this page the color white, we may say, and it becomes impossible to say what it
is. The whiteness of this page is not predicable of the page – the page is not its
 N. Goodman and W. V. Quine, “Steps Toward a Constructive Nominalism,” Journal of Sym-
bolic Logic, 12 (1947).
 Hegel’s Logic, § 24.

Authenticated
whiteness, it is a page. The color white is predicable of the page, though only be-
cause it is predicable of its whiteness: the page is white because its color is
white. We may call Aristotelian accidents such as the whiteness of this page partic-
ular properties, thus granting antirealists regarding universals their due. But the
whiteness of this page must not be confused with the color white, a universal prop-
erty, which is said directly of that particular property and indirectly of the page. We
must also grant realists regarding universals their due.
We perceive this page as well as its whiteness, and we can be said to perceive
also the universals of which they are instances, namely, being a page and the color
white. When we perceive the whiteness of the page, we may be aware, know, also
that it is the color white, and thus aware of the species, as well as that it is a color,
the genus, though not by perceiving them – we do not perceive three properties:
the whiteness of the page, the color white, and color. We are aware of both the spe-
cific property and the generic property in the sense that we know that one of the
properties present in the page is the color white and that the color white is a color.
We do not perceive them in addition to perceiving the whiteness of the page. But
we can think of them. And we can think of them even when we do not see a par-
ticular property falling under them. We can think of them without seeing anything.
We can think of them also without employing a suitable mental image. This is evi-
dent when we try to recognize a color in its instances. (“What would you say is the
color of that shirt?”) Such thinking may be described as employing the relevant
concept. But this need not mean more than that we are capable of thinking of
the universal and recognizing it in its instances. We do not “have” the concept
in the way we might have a mental image, let alone a pencil. And we would
have the concept, but not the image, even when asleep.
“It is not we who frame the concepts [die Begriffe],” Hegel wrote. “Rather the
concept is the genuine first; and things are what they are through the action of
the concept, immanent in them, and revealing itself in them.”³⁵⁸ Cognition, at
least in the case of understanding, necessarily involves concepts. But concepts
are not mere abstractions. Though universals, they are “sunk” in things, they
are properties of things. They are not denizens of a Platonic realm, separate
from the world, which individual things only “imitate.” They are “concrete,”
not “abstract,” universals. But even when they are properties of spatiotemporal
things, they are not spatiotemporal parts of those things, they are not spatiotem-
poral parts of the world. They are “concrete,” not “abstract,” universals: “[T]he
universal of the notion is not a mere sum of features common to several things,
 Hegel’s Logic, § 163.

Authenticated
2 Solipsism and Pure Realism 225
confronted by a particular which enjoys an existence of its own. It is, on the con-
trary, self-particularizing or self-specifying”³⁵⁹
What Hegel meant by saying that universals particularize themselves, that
they self-particularize, need mean no more than that there is nothing in individ-
ual things that “connects” them to their properties, that there are no Lockean
substrata or Bergmannian bare particulars. Hegel’s famous view that the Abso-
lute “develops” from the Logical Idea to Nature need mean no more than that
concepts have no reality apart from individual things, that universals must be ex-
emplified. And when Hegel says that from Nature the Absolute develops to Spi-
rit, this need mean no more than that individual things have no reality apart
from concepts.
Hegel’s view of concepts accords with the standard use of the word in the
cognitive disciplines. An account of the fundamental concepts of physics refers
to Newton or Einstein no more than it refers to their states of mind. It refers to
how certain properties, e. g., gravitation and light, are understood in physics.
They are understood as universals, though of course not so described. Physics
is no more interested in particular properties, or tropes, than it is interested in
particular objects. The whiteness of this page is of no greater interest to it
than this page. Just as geometry is concerned not with the rectangular shape
of this page but with rectangular shape as such, physics is concerned not with
the mass of this page but with mass.
2 Solipsism and Pure Realism
The thesis of this book has been that there is no place for anthropocentrism in
philosophy, that philosophy ought to be “dehumanized.” In the case of episte-
mology and ethics, I argued in Part One, this can be done directly, without qual-
ification, by resolutely shifting their focus. In metaphysics, however, anthropo-
centrism has been present as antirealism, most notably Kant’s transcendental
idealism. And we cannot just return to a pre-Kantian metaphysics. Part Two
was devoted to explaining why this is so, in effect, to showing the virtues of an-
tirealism. For our goal has been not to reject antirealism but to free it from an-
thropocentrism. This means that we must understand the central claim of anti-
realism as nonanthropocentric, as making no reference to humans. This can
be done only by radical rethinking of the role of personal pronouns in philo-
sophical contexts, which was attempted in the preceding two chapters.
 Hegel’s Logic, § 163, 227.

Authenticated
Hume in effect denied that there is what Wittgenstein was to call “the philo-
sophical self” or “the metaphysical subject.” If he was right, there would not be
the entity with which the world might be contrasted and on which it might depend.
Antirealism would become what Wittgenstein called pure realism: “solipsism, when
its implications are followed out strictly, coincides with pure realism. The self of sol-
ipsism shrinks to a point without extension, and there remains the reality co-ordi-
nated with it” (5.64). The world becomes “my world” because, with respect to the
content of any cognition, primitive or advanced, perceptual or conceptual, no dis-
tinction between me and the world that would be relevant to the realism/antirealism
issue can be made. Hegel would have agreed, though he would have added that no
individual, personal cognition is independent of societal, public cognition. If so,
then there is also no distinction relevant to the realism/antirealism issue between
us and the world. Wittgenstein himself held, though only in his later works, that in-
dividual cognition is dependent on public cognition, insofar as it requires the use of
language, by denying that a private language is possible and insisting that a public
language presupposes agreement in judgment.
When by “I” or “we” is meant mind in the sense of cognition, and by
“world” the object of that cognition, the philosophical distinction between the
world and cognition of the world does begin to resemble the distinction between
a headache and the feeling of that headache. We attach little sense to specula-
tion about unfelt headaches or about feelings of a headache that are feelings
of nothing. Indeed, Berkeley noted that his thesis “to be is to be perceived”
was most plausible in the case of pains: “nobody will pretend that real pain ei-
ther is, or can possibly be, in an unperceiving thing, or without the mind, any
more than its idea.”³⁶⁰ Headaches are feelings of a headache, and feelings of
a headache are headaches. A headache is identical with the feeling of that head-
ache, though of course in the sense in which the Evening Star is identical with
the Morning Star, not the sense in which the Evening Star is identical with the
Evening Star. Hegel explained the two senses by saying that he was concerned
only with identity “in unity with difference,” not the “abstract identity” asserted
in “silly” statements like “mind is mind” or “planet is planet.”
The relation between a headache and the feeling of that headache is a famil-
iar and reasonably uncontroversial instance of the relation between the world
and cognition. Feeling a headache is an experience, a cognition, and the felt
headache is the “thing,” res, reality, that is cognized. The realist, if there is
one in this case, would say that there are, or at least could be, unfelt headaches.
The skeptic would say that one cannot know whether there are unfelt headaches.
 Berkeley, Principles of Human Knowledge, Part I, 41.

Authenticated
The antirealist would say that there is no difference between the feeling of a
headache and the felt headache. But the instances of the relation between the
world and cognition that realists, skeptics, and antirealists usually consider –
such as the relation between material things and sense perception, causality
and knowledge of it, or an esoteric elementary particle in physics and the theory
that is the only reason for accepting its reality – are unfamiliar and seldom un-
controversial. In chapter 8 we found an instance that though unfamiliar is also
uncontroversial. It is the relation between the world and the cognition of it ex-
pressed in generic statements. No one thinks that there are generic facts. In
their case antirealism seems unassailable.
The antirealist thesis has seemed paradoxical, absurd, because it is inter-
preted as saying that the whole world – from the page you are reading now to
the remotest known galaxies, and since the Big Bang to the farthest conceivable
future – depends for its existence and nature on the minds, the cognitive capaci-
ties, of humans, a species in one of its planets’ fauna. This would be a zoological
thesis. It would also be absurd zoology, as well as absurd physics and astronomy.
But antirealism is not a theory in zoology, or in physics or astronomy. The mind
relevant to the philosophical topic of the dependence of the world on the mind
cannot be a thing in that world, nor can it be a part or property of such a thing.
This is why the antirealist thesis is not that the world is dependent on the human
mind. There is no mind on which the world might depend. There is only the
world. To be sure, it is the cognized or at least cognizable – perceivable, conceiv-
able, describable – world, but no other world, if there is one, is relevant in phi-
losophy, science, or everyday life. There are humans in the cognized world, of
course, just as there are also whales and chimpanzees. But there is nothing in
or about humans, just as there is nothing in or about whales or chimpanzees,
on which the world might depend.
The idea of mind usually employed in careless talk about mind-dependence
or independence rests on a powerful but misleading picture of the mind as some-
thing “in” us, a self or a soul, often fancied as located somewhere, somehow,
behind the eyes. More than half a century ago, in his attack on Descartes’s dual-
ism, Gilbert Ryle called this picture the dogma of the ghost in the machine. But
the word “dogma” suggests that the picture was a philosophical invention, which
it was not. We do sometimes point to our heads when speaking of our thoughts,
and the belief that they are located within the skull is ancient, not acquired from
philosophy or science. But sometimes we also point to our hearts when speaking
of our feelings, especially love, and the belief that feelings are located some-
where within the chest is also ancient, though few if any educated people
share it today. To be sure, there is a brain behind the eyes, but only a poor neuro-

Authenticated
science or philosophy would identify it with the mind at issue in the realism/an-
tirealism debate.
At any rate, even if thoughts were in the head, pointing to our heads in phil-
osophical contexts like Cartesian doubt or the realism/antirealism issue would
be inadmissible. Thought can be no more in the head when considered in
those contexts than love can be in the heart when considered in biology. To
say this is not to question anything that neuroscience and cardiology say. It is
to acknowledge that, unless we flagrantly beg the question against the skeptic
and the antirealist, we cannot take the mind relevant in such philosophical con-
texts to be a part or feature, mental or physical, of a thing, human or nonhuman,
in the world. Neuroscience says nothing about this.
Although there is an obvious nonphilosophical distinction between P.B. and
the world, there is no philosophical distinction between me and the world. The
reason is that in the latter case each must be understood in terms of the
other. “The point without extension” to which the self “shrinks” determines
“the limit of the world,” of the “reality co-ordinated” with that point. To say
this indeed is to express a sort of solipsism, Wittgenstein’s “I am the world,”
but it is a solipsism that “coincides with pure realism.” Such solipsism is not
alarming, because the “I” in “I am the world” refers, not to Wittgenstein, P.B.,
or any other person or thing in the world, physical or mental, but to the world
itself, even if we add that it refers to the world as cognized, viewed, in a certain
way. The impression of absurdity plaguing antirealism vanishes when the impli-
cations of its most extreme variety, solipsism, are “followed out strictly.”
Indeed, there would be no room left for any sort of subjectivism, be it solip-
sism in metaphysics, skepticism in epistemology, or egoism in ethics. There is no
room for solipsism because without the contrast between me and the world, the
solipsist’s assertion “Only I exist” becomes empty. There is no room for skepti-
cism about the external world: “Scepticism is not irrefutable, but obviously non-
sensical, when it tries to raise doubts where no questions can be asked” (6.51). It
is nonsensical because there would be nothing with which the external world
could be contrasted, nothing to which it would be “external.” And no room is
left for egoism in ethics, insofar as rejecting “the philosophical self” implies
that “The sense of the world must lie outside the world” (6.41), not in the
world. The fatal weakness of all subjectivism – solipsism, skepticism, or egoism
– is its implicit commitment to a pre-Humean view of the self. Without a subject,
there cannot be a subjectivism. Hume, Wittgenstein, and Sartre saw this clearly.
Their rejection of subjectivism, though highly technical, would please common
sense, which firmly disapproves of egoism, rejects skepticism without hesitation,
and dismisses solipsism as madness.

Authenticated
Berkeley claimed that the world is dependent on the mind insofar as the
world is perceived. Kant pointed out that it is dependent on the mind also insofar
as it is understood. Contemporary antirealism has added that the world depends
on the mind insofar as it is described. Berkeley could not consistently mean by
perceiving a relation that a human bears to an object, and he did not. Nor could
Kant consistently mean by concepts inhabitants of human skulls, and he did not.
And contemporary antirealists who appeal to meanings or uses of words cannot
consistently appeal to human activities such as speaking and writing. But few
philosophers who took the linguistic turn saw that a nonzoological view of
mind demands a nonzoological view not only of perception and conception
but also of language.
Gilbert Ryle was an exception, as I noted in the Introduction. He made a sharp
distinction between the use and the usage of a word: “Hume’s question was not
about the word ‘cause’; it was about the use of ‘cause’. It was just as much
about the use of ‘Ursache’. For the use of ‘cause’ is the same as the use of ‘Ursache’,
though ‘cause’ is not the same word as ‘Ursache’. Hume’s question was not a ques-
tion about a bit of the German language. The job done with the English word ‘cause’
is not an English job, or a continental job.”³⁶¹ In a symposium with J. N. Findlay,
Ryle remarked, “The famous saying: ‘Don’t ask for the meaning; ask for the use’,
might have been and I hope was a piece of advice to philosophers, and not to lex-
icographers or translators.” Findlay agreed, but he also added: “ [We] cannot fully
say, in a great many cases, how an expression is used, without saying what sort of
things it is intended to refer to, or to bring to mind…”³⁶² If Ryle meant by “use of a
word” roughly what Kant meant by “concept,” then Findlay’s important addition
that the use of a word involves the things to which it is applied was closer to Hegel’s
view of concepts as “sunk” in the things to which they apply, or, as Strawson put it,
“permeating,” “soaking” them.
Incompatibility between conceptual schemes, Strawson held, need not be
formal inconsistency, the clashes between them need not call for choice between
contradictory propositions. For example, he argued, there is no genuine contra-
diction between science and common sense regarding the reality of colors. The
seeming difference between what they say is due to the different concepts
they employ. But we cannot separate the concepts from their subject matter,
and then judge independently which are faithful to it.³⁶³ For the subject matter
is “permeated” by the concepts. To question this would be “an invitation to
 “Ordinary Language.”

 Gilbert Ryle and J. N. Findlay, “Symposium: Use, Usage and Meaning,” Proceedings of the
Aristotelian Society, Supplementary Volumes 35 (1961), 233.
 P. F. Strawson, Scepticism and Naturalism (New York: Columbia University Press, 1985), 45.

Authenticated
step outside the conceptual scheme which we actually have – and then to justify
it from some extraneous point of vantage. But there is nowhere to step; there is
no such extraneous point of vantage.”³⁶⁴ There is no “metaphysically absolute
standpoint from which we can judge between the two standpoints.”³⁶⁵ “The pic-
ture of a concept-free access to facts, to reality,” Strawson wrote, “is confused
and ultimately self-contradictory.”³⁶⁶
There can be no genuine contradiction between different conceptual practi-
ces. They are just different. There is no arbiter of the propriety of the application
of any concept which is external to and independent of our actual conceptual
practice. We cannot exit the practice so that we can “see” what it is about and
then judge whether it is the “right” one. “[Y]ou can have no cognitive contact
with, hence no knowledge of, Reality which does not involve the forming a be-
lief, making a judgment, deploying concepts.”³⁶⁷ This, of course, is the thesis of
what I have called conceptual antirealism. Speaking of Wittgenstein’s remarks
on aspect-seeing, in the case of a picture that may be seen equally legitimately
as a duck or a rabbit, Strawson wrote, “the visual experience is irradiated by, or
infused with, the concept; or it becomes soaked with the concept.”³⁶⁸ Hegel would
have agreed: “in point of contents, thought is only true in proportion as it sinks
itself in the facts.”
If concepts permeate, soak, the things conceived, and if to speak of the use
of a word is to speak of the things to which it refers, then the mind itself, insofar
as it requires concepts and words, not only cannot be in the world, it must be the
world. This, of course, was Wittgenstein’s pure realism. It was also Hegel’s “ab-
solute idealism,” which he proposed after rejecting Kant’s “transcendental ideal-
ism” as unacceptably psychological and thus subjective. Hegel would have glad-
ly agreed that absolute idealism could be called pure idealism. He might have
also agreed that it is pure realism, had he anticipated Wittgenstein’s reasoning.
Berkeley began with the simplistic notions of a self (“mind”) confronted with
an object (material or mental) and of consciousness as a relation between the
two (perceiving, imagining). Kant reduced the self to a unified consciousness.
Hegel identified that consciousness with its objects. Wittgenstein rejected the
self altogether, thereby removing also the motivation for consciousness, the “in-
tentional’ relation of the self to objects, and allowed only for the world. Hegel
 P. F. Strawson, Analysis and Metaphysics (Oxford: Oxford University Press, 1992), 64.
 Scepticism and Naturalism, 38.
 Analysis and Metaphysics, 86.
 Analysis and Metaphysics, 86.
 P. F. Strawson, “Imagination and Perception,” in Freedom and Resentment and Other Essays
(London: Methuen, 1974), 57.

Authenticated
was philosophically far apart from Berkeley, and Wittgenstein was even more so.
But it is unclear that on our topic Hegel and Wittgenstein were far apart from
each other. It is unclear that absolute idealism and pure realism differ in more
than literary and philosophical style or approach, striking though these differen-
ces may be. Wittgenstein’s pure realism is a realism that has earned its keep, un-
like the pre-Kantian realism (“naïve realism”) that, as Descartes, Berkeley, and
Hume saw, lay helpless before the skeptic. Hegel’s absolute idealism held that
reality is thought, but it did not hold that vegetables, animals, and minerals
are sensations, feelings, or mental images. It held only that there is no “casha-
ble” difference between vegetables, animals, or minerals and how we perceive
and think of them. There is a world, but its nature and contents can be known
only as what perception reveals, conception allows, and language expresses. Jus-
tice is thus done to both realism and antirealism.
The cognizer on whose cognition the world has been held to depend must be
the philosophical self, the metaphysical subject, not Descartes, Hegel, or P.B. But
there is no such cognizer. There is only the world. Solipsism, if properly under-
stood coincides with pure realism, the rejection of all subjectivity. The most ex-
treme form of antirealism thus leads to the most extreme form of realism, realism
without a self, a pure realism. The most extreme form of subjectivism gives way
to the most extreme form of objectivism. The seeming paradox of the antirealist
thesis that the world depends on humans’ cognition of it is resolved: the world
does not depend on its human inhabitants, or on anything associated with them.
Of course, this is only a reason for rejecting the most serious objection to the the-
sis. It is not a reason for accepting it. The reason for accepting the thesis was ex-
plained in Part Two. It rested mainly on the role of logical cognition, especially
in generic statements.
Claims like Kant’s that the objects of knowledge conform to knowledge, not
knowledge to its objects, Goodman’s that we make the world, and Hegel’s that
Reality is Thought are at first glance incredible. But they can be seen now as
rhetorical flourishes of what follows from a proposition that is a tautology and
a proposition that, though not a tautology, is self-evident.
The tautology is that the only world we perceive, understand, and describe is
the world perceived, understood, and described by us. The self-evident proposi-
tion is that, in philosophical contexts like Cartesian doubt and the realism/anti-
realism issue, we cannot coherently regard ourselves as a part, mental (an ego, a
colony of egos) or material (a brain, a collection of brains), of that world. Togeth-
er with the tautology, this proposition yields a metaphysics that is antirealist but
not anthropocentric.

Authenticated
3 Philosophical method
The reader may ask, how can a substantive view rest on so little? The question
betrays commitment to a conception of philosophy I have warned against repeat-
edly. Philosophers can claim no expertise on any things or facts, empirical or
nonempirical. They have neither the training nor the means to make empirical
discoveries, nor do they have the training or the means to make nonempirical
discoveries, for example like those in mathematics. Philosophy is neither ama-
teur psychology, devoted to the study of “the ideas in the mind,” nor amateur
lexicology, devoted to the study of “the workings of our language.”
Appeals to a tautology are often the best way to break the sway of a picture
that we find misleading and dissolve an illusion. “You can’t both spend and save
it” exposes an all too frequent illusion in financial planning. When made in phi-
losophy, such an appeal can be conducive to understanding by exposing a phil-
osophical picture as misleading and thus dispelling the illusion the picture gen-
erates. The tautology that the only world we perceive, understand, and describe
is the world perceived, understood, and described by us can dispel the illusion
generated by the philosophical picture of ourselves as spectators of an external
world, like astronauts observing the earth from the moon.
Appeals to self-evident propositions are often properly acknowledged to be
decisive. “Well, doctor, say what you will, it still hurts” sometimes silences the
intelligent physician, and “You can’t live forever” sometimes ends agonizing in-
decision. Reductio ad absurdum, in effect an appeal to the self-evidence of the
principle of noncontradiction, is often the only available form of argument in
mathematics. Appeals to self-evident propositions are sometimes decisive also
in philosophy. “We cannot both regard ourselves as inhabitants of the world
and deny its reality” is surely an example.
The goal of philosophy is not to seek information but to achieve understanding.
And understanding, in philosophy as well as in everyday life and science, rests on
seeing similarities and differences – in looks, structure, function, causal role – not
seen before. This is why the word “model,” originally applied, as it still is, to a piece
of wood for making moulds in metal casting, has increasingly replaced “theory,”
which in Greek meant a view, contemplation. It is a mistake to think that saying
what something is like is at most second best, and that we should aim at saying
what it is. When we say what something is we are really saying what it is like, name-
ly, that it is like the paradigms for the application of the relevant concept. Ordinarily
the likeness is so close that its presence and our reliance on it escape our attention.
To say that a certain animal before us is a dog is, by implication, to say that it is like
the paradigms for the application of “dog.” But we are seldom conscious of the im-
plication because the likeness is ordinarily so close that the animal can be taken to

Authenticated
3 Philosophical method 233
be such a paradigm itself, even though we did not learn the conventions regarding
the meaning or use of “dog” by reference to it.
Plato and Aristotle, or Hume and Kant, Husserl and Russell, did not disagree
because they were “presented,” in sensory or intellectual intuition, with different
things or facts, or because some of them offered false descriptions of those
things and facts or made formal errors in their inferences from the true descrip-
tions. In chapter 7 I remarked that if Wittgenstein and Bergmann disagreed about
generality, the reason was not that they “saw” different things, just as those who
disagreed with both of them did not. At most, all three saw things differently.
In the present chapter I suggested regarding the “problem” of universals that
what philosophers discussing it in fact do is to propose, mostly by means of met-
aphor and analogy, certain descriptions, verbal “pictures,” of the world, ways of
understanding it on the most abstract level, hoping usually to free us from the
illusions produced by what they consider misleading other metaphors, analo-
gies, descriptions, or pictures.³⁶⁹ The defender of universalia in rebus does not
happen to see the identity of the color of this page and the color of the next
page, the trope theorist does not just fail to see it, and Plato did not enjoy an
extrasensory awareness of his Forms that most of us do not. To think otherwise
would be to misunderstand what philosophers do.
The central question, according to trope theorists, is whether, e. g., the color
of this page and the color of the next page are identical, one and the same entity,
or only resemble each other, exactly or inexactly. Thus the appeal to tropes is an
attempt to avoid commitment to universals by appealing instead to the presence
of a relation of resemblance. But there is no fact of the matter here. If there were,
we should be able to tell which is the right view in the case of the two pages by
just looking at them – after all, we can see both pages and we can also see their
color. We would expect to see whether the relation between the color of this page
and the color of that page is identity, i. e., that a certain color is a property com-
mon to both, or resemblance, i. e., that they are particular properties related by
resemblance. But how would such identity and resemblance differ? The truth is
that neither the philosophers who say that properties are universals nor those
who say that they are particulars are reporting anything to be settled by or infer-
red from experience, they are not reporting a discovery of facts about the world.
The idea that there is a “fact of the matter” they disagree about is surely absurd.
Are the disagreeing parties each claiming better powers of vision than the other?
 The view that making analogies is “the driving force behind all thought” was common in
ancient and medieval philosophy. For a recent defense of it by cognitive scientists, see Douglas
Hofstadter and Emmanuel Sander, Surfaces and Essences: Analogy as the Fuel and Fire of Think-
ing (New York: Basic Books, 2013).

Authenticated
Surely, they can only claim to be offering a more illuminating description of a

familiar and uncontroversial fact.
The problem of universals is whether this fact is more like the paradigms of
identity, say, the identity of the Morning Star and the Evening Star, or like an or-
dinary relation, say, the spatial relation between the two pages.³⁷⁰ Science and
common sense seem to hold that it is the former. This is why they choose
“Both are white,” or “This page is white and the next page is white,” to describe
the fact, rather than a relational statement of the form xRy. Indeed, there are no
relevant statements of the latter sort because there are no names of particular
properties, of tropes. In aRb, where R stands for resemblance, a and b would
not be names, respectively, of the color of this page and the color of the next
page since there are no such names. “The color of this page” and “the color
of the next page” are not names. And it is unclear that names could be coherent-
ly introduced. Could they be reusable? Would they be public?
Of course, an antirealist may say that even genuine discoveries, say, in sci-
ence, are no more than proposed pictures, since there is no such thing as getting
at the things themselves, at reality. Such an antirealist should review Kant’s dis-
tinction between transcendental idealism and empirical realism. But in the case
of philosophical theories what the antirealist says would be true even if the anti-
realism were false. Once the illusion is dispelled that the task of philosophy is
“conceptual” analysis or describing the “workings” of language, as was the ear-
lier illusion that it was to inquire into “the mind and its ideas,” philosophical
theories can no longer be seen as companions, much less competitors, of scien-
tific theories. Even if science could “get at” reality through observation, experi-
ment, and inference, surely, for the reasons repeatedly explained in this book,
philosophy could not. What it can do, however, is to draw attention to, and em-
phasize, similarities and differences between fundamental kinds of entities in
the world that in everyday life, and even in science, may go unnoticed – some-
times precisely because they are fundamental.
The answer then to the question, how can a substantive view rest on so little
as a tautology and a self-evident proposition, is that we do not want our view in
this book to be substantive, if this means description of some things or facts. It is
intended not as report of discovery but as proposal of a way of understanding
the relationship between us and the world. When discussing Wittgenstein’s semi-
realism, I avoided judgments about its “truth,” stressing instead its importance
as an alternative to the stark positions of metaphysical realism and antirealism.
 I develop this point in detail in Resemblance and Identity.

Authenticated
4 Philosophy without anthropocentrism 235
4 Philosophy without anthropocentrism
A philosophical picture is not a philosophical system. Our project has been to

free philosophy – first epistemology and ethics, then metaphysics – from anthro-
pocentrism, not to construct epistemological, ethical, and metaphysical systems.
But even if we distrust the claims to philosophical knowledge that such systems
make, and content ourselves with drawing verbal pictures, we have left room for
them. The metaphysical picture of the world we have drawn shares with Hegel’s
the absence of the usual subject-object dualism. Is it then an idealist picture? If it
is, the idealism certainly is not Berkeley’s. Is it Hegel’s? It depicts no “Absolute”
that develops from Logical Idea to Nature and then to Spirit, nor does it employ a
“dialectical” method.
Is our metaphysical picture Sartre’s? It does shares with it the explicit rejec-
tion of both subject-object dualism and act-object dualism. But it completely by-
passes Sartre’s existentialist concerns with freedom, bad faith, sadism, and mas-
ochism, which surely are irremediably anthropocentric. Is our metaphysical
picture Wittgenstein’s, in which the subject-object and act-object dualisms are
also absent? If Wittgenstein had a metaphysical picture in the Tractatus, it
was so sketchy that the question is unanswerable. Had he had one, what our pic-
ture and his would share is the rejection of the metaphysical self and thus of
subjectivism in all its forms. In this respect both would be descendants of He-
gel’s metaphysical picture. Hegel’s assertion that thoughts are the essences of
things and Wittgenstein’s that the self of solipsism shrinks to a point without ex-
tension need not convey different metaphysical pictures just because they are
very different in terminology, style, and historical context.
Our metaphysical picture is compatible with other philosophical pictures as
long as they avoid divorcing the world from cognition. I have held that con-
sciousness has no intrinsic nature and no “inhabitants,” not even an ego, that
there is just the world. Hence, there is some plausibility of the physicalist picture
of the world as matter. But, unlike it, ours does not exclude consciousness – it
merely does not include it, much as a group portrait of a family usually does not
include the photographer. Our metaphysical picture is also compatible with that
of a phenomenology faithful to Husserl’s motto “We must go back to the things
themselves.” Consciousness is not in our picture because it is not among “the
things themselves,” an additional item in the world competing with them for a
place in it. It also ought not to have been in Husserl’s.
Our metaphysical picture is compatible with that of pure idealism, of the
world as mind. It does justice to what makes idealism plausible. Consciousness
has no qualitative nature and no inhabitants. There are only its objects. In itself
it is nothing. Nonetheless, it is the revelation of things and the regulator of which

Authenticated
are admissible into the world. It is a revelation on the level of perception (sen-
sibility, outer or inner; intuition, sensory of intellectual) and a regulator on
the level of conception (understanding, thought, reason, language). Perception
opens the gate to the world, but conception is the gate-keeper. So understood,
consciousness is the same as cognition, sometimes even knowledge.. We can re-
sponsibly regard as objects in the world only those revealed and then allowed by
it. But though our picture contains only what consciousness reveals and accepts,
it does not exclude Kantian things-in themselves. It merely does not include
them, much as the group portrait of a family might not include the grandparents.
In The Vocation of Man, Fichte offered a characteristically dramatic, and un-
happy, view of consciousness: “All that I know is my consciousness itself…..I
know of no being, not even of my own. There is no being….Pictures [Bilder]
are: – they are the only things which exist.”³⁷¹ Fichte thought the moral to be
drawn is that we must rely on faith, rather than knowledge. The views of con-
sciousness we find in Hegel, Wittgenstein, and Sartre, however, suggest that
we should rather say: “All that I know is the world itself. The things in it are:
– they are the only things which exist.” But we might be wiser if we refused
to choose between the metaphysical pictures of Wittgenstein’s pure realism
and Fichte’s pure idealism, and welcomed both. Our philosophical understand-
ing would then be richer. Of course, the welcome need not be unreserved. Pure
realism does not entail physicalism, and thus is free from the latter’s rashness.
Pure idealism does not entail a doctrine about an Absolute and its dialectical de-
velopment, and it need not be expressed in the jargon of German idealism. As to
which is preferable, suffice it to note the advantage pure realism enjoys of being
close to both common sense and science. Pure idealism lacks that advantage.
Our rejection of anthropocentrism in philosophy allows for the investigation
of the numerous traditional topics in philosophy to which the realism/antireal-
ism debate is irrelevant. As I pointed out in the Introduction, though much of
traditional ethics and epistemology consists of inquiries that properly belong
in the empirical sciences, not philosophy, both also contain much that does
not, for example, what I called epistemology-as-logic in the case of epistemology
and theories of goodness like Plato’s, Aquinas’s, and Moore’s in the case of eth-
ics And while much of post-Kantian metaphysics has seemed committed to the
prima facie absurd view that the world depends on humans, there are many met-
aphysical inquiries of which this is not true, especially in pre-Kantian philoso-
phy. The “dehumanization” of metaphysics proposed in this book is compatible
with most theories to which those inquiries led.
 Fichte, The Vocation of Man, trans. William Smith (Chicago: Open Court, 1910), 88 – 90.

Authenticated
4 Philosophy without anthropocentrism 237
Our picture leaves room for the project of ontology: listing and describing
the most general kinds (“categories”) of entities and the relations among
them. One example is the question whether individual things are merely bundles
of their properties. There is also, regarding each of the metaphysical pictures
mentioned earlier, the question whether what the picture does not include none-
theless ought to be included, for example, consciousness in the case of the
physicalist picture and Kantian things-in-themselves in the case of the idealist
picture. No science can answer these questions. If they can be answered at
all, only philosophy can do it. If they cannot be answered, only philosophy
can explain why they cannot.

Authenticated
Bibliography
Ackrill, J. L. Aristotle’s Categories and De Interpretatione (Oxford: Oxford University Press,
1963).
Anscombe, G. E. M. “The First Person,” in Samuel Guttenplan, ed., Mind and Language
(Oxford: Oxford University Press, 1975).
Anscombe, G. E. M. Introduction to Wittgenstein’s Tractatus (London: Hutchinson’s University
Library, 1959).
Alston, William. A Realist Conception of Truth (Ithaca, NY: Cornell University Press, 1996).
Aristotle. Nichomachean Ethics, trans. W.D. Ross (Oxford: Clarendon Press, 1908).
Aristotle. Categories, in J. L. Ackrill, Aristotle’s Categories and De Interpretatione (Oxford:
Oxford University Press, 1963).
Aristotle. De Interpretatione 7, in J. L. Ackrill, Aristotle’s Categories and De Interpretatione
(Oxford: Oxford University Press, 1963).
Aristotle. Metaphysics, trans. Hugh Lawson-Tancred (London: Penguin Books, 2004).
Armstrong, David M. Universals and Scientific Realism (Cambridge: Cambridge University
Press, 1978).
Armstrong, David M. A World of States of Affairs (Cambridge: Cambridge University Press,
1997).
Arnauld, Antoine and Nicole, Pierre. Logic or the Art of Thinking , trans. and edited by Jill
Vance Buroker (Cambridge: Cambridge University Press, 1996).
Audi, Robert. “Intrinsic Values and Reasons for Action,” in Terry Horgan and Mark Timmons,
eds., Metaethics after Moore (Oxford: Clarendon Press, 2006).
Austin, J. L. How to Do Things with Words (New York: Oxford University Press,1976).
Austin, J. L. Sense and Sensibilia (Oxford: Clarendon Press, 1962),
Bambrough, J. R. “Universals and Family Resemblances,” Proceedings of the Aristotelian
Society, LXI (1960 – 61).
Barwise, J. and Cooper, R. “Generalized Quantifiers and Natural Language,” Linguistics and
Philosophy, 4.
Bergmann, Gustav. Meaning and Existence (Madison, University of Wisconsin Press, 1959).
Bergmann, Gustav. New Foundations of Ontology (Madison: University of Wisconsin Press,
1992).
Bergmann, Gustav. Logic and Reality (Madison: University of Wisconsin Press, 1964).
Blackburn, Simon. Essays in Quasi-Realism (Oxford: Oxford University Press, 1993).
Bloom, A. H. The Linguistic Shaping of Thought: A Study in the Impact of language on
Thinking in China and the West (New York: Laurence Erbaum, 1981).
Bloomfield, Leonard. Language (Chicago: University of Chicago Press, 1933).
BonJour, Laurence. In Defense of Pure Reason. (London: Cambridge University Press, 1998).
Brentano, Franz. Psychology from an Empirical Standpoint, trans. by A.C. Rancurello, D.B.
Terrell, and L. McAlister. 2nd ed., intr. by Peter Simons, 1995 (London: Routledge, 1973).
Butchvarov, Panayot. “Epistemology Dehumanized,” in Quentin Smith, ed., Epistemology:
New Essays (New York: Oxford University Press, 2008).
Butchvarov, Panayot. “Ethics Dehumanized,” in Mark Timmons and Terry Horgan, eds.,
Metaethics After Moore (Oxford University Press, 2006).
Butchvarov, Panayot. “Metaphysical Realism and Logical Antirealism,” in Richard Gale, ed.,
Guide to Metaphysics (Oxford: Blackwell, 2002).

Authenticated
Bibliography 239
Butchvarov, Panayot. Resemblance and Identity: An Examination of the Problem of Universals

(Bloomington & London: Indiana University Press, 1996).
Butchvarov, Panayot. Skepticism about the External World (New York: Oxford University Press,
1998).
Butchvarov, Panayot. Skepticism in Ethics (Bloomington and London: Indiana University Press,
1989).
Butchvarov, Panayot. “States of Affairs,” in Radu J. Bogdan, ed., Roderick M. Chisholm
(Dordrecht: Reidel, 1986).
Carlson, Gregory N. Reference to Kinds in English (New York & London: Garland, 1980).
Carlson, Gregory N. and Pelletier, Francis Jeffry, eds., The Generic Book (Chicago: University
of Chicago Press, 1995).
Casullo, Albert. A Priori Justification (New York: Oxford University Press, 2004).
Chisholm, Roderick M. A Realistic Theory of Categories (Oxford: Cambridge University Press,
1996).
Chomsky, Noam. “Comments: Galen Strawson, Mental Reality”, Philosophy and
Phenomenological Research, vol. LVIII, no. 2 (June 1998).
Chomsky, Noam. “Language and Nature,” Mind, 104: 1 – 60.
Chomsky, Noam. Language and Thought (Wakefield, Rhode Island & London: Moyer Bell,
1993).
Chomsky, Noam. Powers and Prospects (London: Pluto, 1996).
Chomsky, Noam. Rules and Representations (New York: Columbia University Press, 1980).
Davidson, Donald. “Rational Animals,” in Subjective, Intersubjective, Objective (Oxford:
Davidson, Donald. “The Structure and Content of Truth,” Journal of Philosophy LXXXVII, 6
(June 1990).
Davidson, Donald. “The Very Idea of a Conceptual Scheme,” Proceedings and Addresses of
the American Philosophical Association 47 (1974).
Devitt, Michael. Realism and Truth, Second Edition (Princeton: Princeton University Press,
1997).
Dewey, John. Logic: The Theory of Inquiry (New York: Henry Holt, 1938).
Diamond, Cora. The Realistic Spirit: Wittgenstein, Philosophy and the Mind (Cambridge, MA:
MIT Press, 1991).
Diamond, Cora. “Realism and Resolution: Reply to Warren Goldfarb and Sabina Lovibond,” in
Journal of Philosophical Research XXII (1997).
Dummett, Michael. The Logical Basis of Metaphysics (London: Duckworth, 1991).
Dummett, Michael. Truth and Other Enigmas (Cambridge: Harvard University Press, 1978).
Fales, Evan. Causality and Universals (London: Routledge, 1990).
Fichte, J.G. The Science of Knowledge, edited and trans. by Peter Heath and JohnLachs
(Cambridge: Cambridge University Press, 2008)
Fichte, J.G. The Vocation of Man, trans. William Smith (Chicago: Open Court, 1910).
Fodor, Jerry. The Language of Thought (Cambridge: Harvard, 1975).
Fogelin, Robert. Wittgenstein (London: Routledge, 1987, second edition).
Frege, Gottlob. The Basic Laws of Arithmetic (Los Angeles: University of California Press,
1987).
Frege, Gottlob. The Frege Reader, ed. M. Beaney (Oxford: Blackwell, 1997).
Frege, Gottlob. Translations from the Philosophical Works of Gottlob Frege (Oxford: Blackwell,
1970).

Authenticated
240 Bibliography
Fumerton, Richard. Realism and the Correspondence Theory of Truth (Lanham, MD:Rowman &
Littlefield, 2002).
Geach, P.T. “Saying and Showing in Frege and Wittgenstein,” in Essays on Wittgenstein in
Honor of G.H. von Wright, ed. Jakko Hintikka, Acta Philosophica Fennica 28 (Amsterdam:
North Holland, 1976).
Gelder, Tim van. “What Might Cognition Be, If Not Computation?” The Journal of Philosophy,
Volume XII, number 7 (July 1995).
Goldfarb, Warren. “Metaphysics and Nonsense,” Journal of Philosophical Research XXII
(1997).
Goodman, Nelson. Problems and Projects (New York: Bobbs-Merrill: 1972).
Goodman, Nelson. Ways of Worldmaking (Indianapolis: Hackett, 1978).
Goodman, Nelson and Quine, W. V. “Steps Toward a Constructive Nominalism,” Journal of
Symbolic Logic, 12 (1947).
Goodman, Nelson and Elgin, Catherine Z. Reconceptions in Philosophy (Indianapolis &
London: Hackett, 1988).
Grossmann, Reinhardt. Phenomenology and Existentialism (London: Routledge, 1984).
Grossmann, Reinhardt. The Existence of the World (London: Routledge, 1992).
Hacker, M.S. Insight and Illusion (Oxford: Clarendon, 1972).
Harris, Zellig. “Distributed Structure”, in The Structure of Language: Readings in the
Philosophy of Language, eds. Jerry A. Fodor and Jerrold J. Katz (1965).
Hegel, G.W.F. Hegel’s Logic, trans. William Wallace (Oxford: Clarendon Press, 1975).
Hegel, G.W.F. Hegel’s Lectures on the History of Philosophy, (London & New York: Routledge
and Kegan Paul, 1974).
Hegel, G.W.F. Hegel’s Phenomenology of Spirit, trans. A. V. Miller (Oxford: Oxford University
Press, 1977).
Hegel, G.W.F. Science of Logic, trans. A. V. Miller (London: George Allen, 1969).
Heidegger, Martin. Basic Writings, ed. D. F. Krell (New York: Harper & Row, 1977).
Heidegger, Martin. Being and Time (New York: HarperOne; Revised edition, 1962).
Heidegger, Martin. Poetry, Language, Thought, trans. Albert Hofstadter (New York: Harper &
Row, 1971).
Higginbotham, J. and May, R. “Questions, Quantifiers, and Crossing,” Linguistic Review 1.
Hofstadter, Douglas and Sander, Emmanuel. Surfaces and Essences: Analogy as the Fuel and
Fire of Thinking (New York: Basic Books, 2013).
Horn, Gabriel. “Thought without Language in Birds,” in L. Weiskrantz, Thought without
Language (Oxford: Clarendon Press, 1988).
Horwich, Paul. Truth (Oxford: Clarendon, 1998).
Humboldt, Wilhelm von. On Language (Cambridge: Cambridge University Press, 1988).
Hume, David. A Treatise of Human Nature (Oxford: Clarendon Press, 1951).
Husserl, Edmund. Logical Investigations (Dordrecht: Kluwer, 2010).
Hutchinson, Brian. G.E. Moore’s Ethical Theory: Resistance and Reconciliation (New York:
Inhelder, Bärbel and Piaget, Jean. The Growth of Logical Thinking (New York: Basic Books,
1958).
Kant, Immanuel. Critique of Practical Reason, trans. Lewis White Beck (Indianapolis:
Bobbs-Merrill, 1956).
Kant, Immanuel. Critique of Pure Reason, trans. Paul Guyer and Allen W. Wood (Cambridge:

Authenticated
Bibliography 241
Kant, Immanuel. Groundwork of the Metaphysics of Morals, trans. H. J. Paton (New York:
Harper and Row, 1964).
Kant, Immanuel. Logic, trans. Robert S. Hartman and Wolfgang Schwartz (New York: Dover,
1974).
Kaplan, David. “Demonstratives,” in J. Almog et al., eds, Themes from Kaplan, (Oxford:
James, William J. The Principles of Psychology (Cambridge, MA: Harvard University Press,
1983).
Jansons, Kalvis M. “A personal view of dyslexis and of thought without language,” in
Thought without Language, ed. L. Weiskrantz (Oxford: Clarendon Press, 1988).
Lemos, Noah. Common Sense (Cambridge: Cambridge University Press, 2004).
Locke, John. An Essay Concerning Human Understanding (London: Dent, 1960).
Malcolm, Norman. Knowledge and Certainty (Englewood Cliffs, NJ: Prentice-Hall, 1963).
Marcus, Ruth Barcan. “Interpreting Quantification,” Inquiry 5, 1962, 252 – 259.
Moore, G. E. “A Defense of Common Sense,” in Philosophical Papers (London: Allen &
Unwin, 1959).
Moore, G. E. Ethics (Oxford: Oxford University Press, 1912).
Moore, G. E. “Identity,” Proceedings of the Aristotelian Society, n.s., v. I, 1900 – 1901.
Moore, G. E. Lectures on Philosophy (London: Routledge, 1966).
Moore, G. E. Principia Ethica (Cambridge: Cambridge University Press, 1903).
Moore, G. E. “The Refutation of Idealism,” included in Philosophical Studies (New York:
Harcourt Brace, 1922).
Nietzsche, Friedrich. Beyond Good and Evil (New York: Millenium, 2014).
Nozick, Robert. Anarchy, State, and Utopia (New York: Basic Books, 2013).
Pears, David. The False Prison (Oxford: Oxford University Press, 1987).
Pinkard, Terry. “Hegel’s Phenomenology and Logic: an overview,” in Karl Ameriks, ed., The
Cambridge Companion to German Idealism (Cambridge: Cambridge University Press,
2000).
Pinkard, Terry. Hegel’s Phenomenology: The Sociality of Reason (Cambridge: Cambridge
University Press, 1996).
Pinker, Steven. Language Learnability and Language Development (Cambridge: Harvard
Pinker, Steven. The Language Instinct (Harvard University Press, 2007).
Pinker, Steven. The Stuff of Thought (New York: Viking, 2007).
Price, H. H. Perception (London: Methuen, 2nd ed., 1950).
Price, H. H. Thinking and Experience (London: Hutchison’s University Library, 1953).
Putnam, Hilary. Ethics without Ontology (Cambridge: Harvard University Press, 2004).
Putnam, Hilary. Realism and Reason. Philosophical Papers. Volume 3 (Cambridge: Cambridge
Putnam, Hilary. “Sense, Nonsense, and the Senses: An Inquiry into the Powers of the Human
Mind” (The Dewey Lectures), Journal of Philosophy, XCI (1994).
Putnam, Hilary. Words and Life (Cambridge: Harvard University Press, 1994).
Quine, W.V.O. From Stimulus to Science (Cambridge, MA: Harvard University Press, 1995).
Quine, W.V.O. The Pursuit of Truth (Cambridge, MA: Harvard University Press, 1992).
Quine, W.V.O. Theories and Things (Cambridge, MA: Harvard University Press, 1981).
Quine, W.V.O. “Two Dogmas of Empiricism,” in From a Logical Point of View (Cambridge, MA:
Harvard University Press, 1994).

Authenticated
242 Bibliography
Quine, W.V.O. The Ways of Paradox (New York: Random House, 1966).
Quine, W.V.O. “Whitehead and Modern Philosophy,” in Selected Logic Papers, enlarged
edition (Cambridge, MA: Harvard University Press, 1996).
Quine, W.V.O. Word and Object (Cambridge, MA: MIT Press, 1960).
Reagan, Tom. Bloomsbury’s Prophet (Philadelphia: Temple University Press, 1986).
Russell, Bertrand. A History of Western Philosophy (New York: Simon and Schuster, 1972).
Russell, Bertrand. The Philosophy of Logical Atomism (Chicago & La Salle: Open Court, 1985).
Russell, Bertrand. Human Knowledge (New York: Simon & Schuster, 1962).
Russell, Bertrand. Introduction to Mathematical Philosophy (New York: Simon and Schuster,
1971).
Russell, Bertrand. Mysticism and Logic (London: Allen & Unwin, 1917).
Russell, Bertrand. My Philosophical Development (London: Allen and Unwin, 1959).
Russell, Bertrand. The Problems of Philosophy (Oxford: Oxford University Press, 1953).
Russell, Bertrand. Theory of Knowledge: The 1913 Manuscript (London & New York:
Routledge, 1984).
Russell, Bertrand and Whitehead, Alfred North. Principia Mathematica to *5 (Cambridge:
Ryle, Gilbert; Findlay, J. N. “Symposium: Use, Usage and Meaning,” Proceedings of the
Aristotelian Society, Supplementary Volumes 35 (1961).
Sapir, Edward. Language (New York: Harcourt, Brace & World, 1921).
Sapir, Edward. Language, Culture and Personality, ed. David G. Mandelbaum (Berkeley & Los
Angeles: University of California Press, 1963).
Sartre, Jean-Paul. Being and Nothingness, trans. Hazel E. Barnes (New York: Philosophical
Library, 1956).
Sartre, Jean-Paul. “Interview,” in The Philosophy of Jean-Paul Sartre, ed. by Paul Arthur
Schilpp (LaSalle, IL: Open Court, 1981).
Saussure, Ferdinand de. Course in General Linguistics (LaSalle, IL: Open Court, 1986).
Schauer, Frederick. Profiles, Probabilities and Stereotypes (Boston: Belknap & Harvard,
2003).
Scheffler, Israel. “The Wonderful Worlds of Goodman,” Synthese (v. 45, 1980).
Schilpp, Paul Arthur. The Philosophy of Bertrand Russell (La Salle, IL: Open Court, 1989).
Searle, John. Intentionality (Cambridge: Cambridge University Press, 1991).
Skinner, B.F. Verbal behavior (New York: Appleton-Century-Crofts, 1957).
Smith, Quentin. Language and Time (New York: Oxford University Press, 1993).
Spinoza. Ethics (New York: Hafner, 1963).
Ryle, Gilbert. “Ordinary Language,” Philosophical Review LXII (1951).
Ryle, Gilbert. “Use, Usage, and Meaning,” Proceedings of the Aristotelian Society
Supplementary Volume XXXV (1961).
Strawson, P. F. Analysis and Metaphysics (Oxford: Oxford University Press, 1992).
Strawson, P. F. “Imagination and Perception,” in Freedom and Resentment and Other Essays
(London: Methuen, 1974).
Strawson, P. F. Introduction to Logical Theory (London: Methuen, 1952).
Strawson, P. F. “On Referring” (Mind, 1950).
Strawson, P. F. “Truth,” Aristotelian Society Supplementary Volume 24 (1950).
Strawson, P. F. “Reply to John R. Searle” in Lewis Edwin Hahn, The Philosophy of F. Strawson
(Chicago & LaSalle: Open Court, 1998).
Strawson, P. F. Scepticism and Naturalism (New York: Columbia University Press, 1985).

Authenticated
Bibliography 243
Vallicella, William F. A Paradigm Theory of Existence: Onto-Theology Vindicated (Dordrecht,

Boston & London: Kluwer Academic Publishers, 2002).
Watson, J. B. and McDougal, W. The Battle of Behavior (New York: Norton, 1929).
Westerhoff, Jan. Ontological Categories: Their Nature and Significance (Oxford: Clarendon
Press, 2005).
White, Alan R. Truth (New York: Anchor, 1970).
Whorf, Benjamin Lee. Language, Thought and Reality: Selected writings of Benjamin Lee
Whorf, ed. John B. Carroll, (New York: John Wiley and Sons, 1956).
Williams, D. C. “The Elements of Being,” Review of Metaphysics 7: 3 – 18, 171 – 192.
Wittgenstein, Ludwig. The Blue and the Brown Books (Oxford: Oxford University Press, 1958).
Wittgenstein, Ludwig. Culture and Value, ed. G.H. von Wright, trans. Peter Winch (Chicago:
University of Chicago Press, 1980).
Wittgenstein, Ludwig. Letters to Russell, Keynes and Moore, ed. G.H. von Wright (Oxford:
Blackwell, 1974).
Wittgenstein, Ludwig. MS 302, “Diktat für Schlick” 1931 – 33. (Quoted by David Stern,
“Sociology of Science, Rule Following and Forms of Life,” in M. Heidelberger and F.
Stadler, eds., History of Philosophy and Science, 347).
Wittgenstein, Ludwig. Notebooks, 1914 – 1916, trans. G. E. M. Anscombe (New York: Harper &
Row, 1961).
Wittgenstein, Ludwig. Philosophical Grammar (Berkeley & Los Angeles: University of
California Press, 1974).
Wittgenstein, Ludwig. Philosophical Investigations, trans. G.E.M. Anscombe (Oxford:
Blackwell, 1991).
Wittgenstein, Ludwig. Philosophical Remarks, ed. Rush Rhees, trans. Raymond Hargreaves
and Roger White (Chicago: University of Chicago Press, 1975).
Wittgenstein, Ludwig. Philosophical Occasions 1912 – 1951 (Indianapolis, IN: Hackett, 1993).
Wittgenstein, Ludwig. Remarks on the Foundations of Mathematics (Cambridge: MIT Press,
1994).
Wittgenstein, Ludwig. Tractatus Logico-Philosophicus, trans. D. F. Pears and B. F. McGuinness
(London: Routledge, 1972).

Authenticated
Index
Ackrill, J. L. 158 Davidson, Donald 92, 174, 208
Alston, William P. 170, 172 definite descriptions 50 – 51
analytic ethics 52 – 56, 58, 59, deontic way 13 – 14, 42
animal rights 1 Descartes 9 – 10, 32, 38 – 40, 44, 47 – 48,
Anscombe, G.E.M. 136, 203 87 – 88, 191 – 196, 215
anthropomorphism 9 Devitt, Michael 171
Arnauld, Antoine 151 – 152, 154 – 155, 159, Diamond, Cora 5, 109 – 110, 110, 133
162 direct awareness 39
Aquinas, Thomas 8, 30, 56, 62, 116, 237 disjunctive facts 108
Aristotle 9, 30, 43, 46, 98, 104, 108, 154, Dummett, Michael 6, 16, 86, 92 – 93, 112,
158, 223 – 224 128, 182
Armstrong, David 64, 85, 171
Audi, Robert 55, 60 egocentric particulars 195
Austin, J. L. 13, 35 egoism 192
Ayer, A. J. 31 epistemic justification 36 – 38
epistemic probability 36 – 38
Bacon, Roger 91 epistemology naturalized 29
Bambrough, Renford 92 epistemology-as-logic 22, 29, 42 – 51
belief 24, 33 – 34, 178 evidence 34 – 36
Bergmann, Gustav 20 – 21, 86, 89, 104, 105, existence, concept of 15, 49 – 51, 76 – 77, 79,
106 – 107, 108, 124 – 125, 130 – 133,141 – 141 – 142, 206
150, 166, 169 extraterrestrial cognizers 24, 29, 90
Berkeley, George 3, 10, 15, 21, 49, 51, 112,
212 – 213, 215 facts 18, 20, 21, 106, 107, 108, 118 – 119,
Blackburn, Simon 22 124, 130, 151, 155 – 159, 169 – 178
Bloomfeld, Leonard 95 Fales, Evan 64
Brentano, Franz 87, 88 Fichte 11, 201, 236
Brooks, David 163 Findlay, J. N. 209, 229
first-person perspective 40, 197
Carnap, Rudolf 71 first-person pronouns 32 – 42, 38, 191 – 198,
Carlson, Gregory N. 152, 162 206 – 207
Cartesian, epistemology 29, 33 – 42, 38, Fodor, Jerry 93
206 Fogelin, Robert 136
Casullo, Albert 102 Frege, Gottlob 22, 43, 89, 104 – 105, 108,
chauvinism 192 111, 113 – 114, 118 – 119, 122 – 128, 131,
Chisholm, R. M. 31, 32, 178 181 – 182
Chomsky, Noam 4, 97 – 98 Fumerton, Richard 170
cognitive collectivism 12, 23, 211 – 212 fused participles 178
cognitive individualism 12, 211
common sense 7, 115, 130, 170 Geach, P.T. 111
conceptual analysis 52 – 59 Gelder, Tim van 95
conceptual antirealism 21 general facts 115 – 116, 122 – 129, 155 – 159
cosmological realism 113, 130, 170, 190 general statements 20, 117, 130 – 150, 152,
165

Authenticated
Index 245
generic statements 19, 151 – 164, 181 Lemos, Noah 7

Genesis 9 Leslie, Alan M. 100
Gettier, Edmund 31 linguistic antirealism 22, 85, 104, 120
Goldfarb, Warren 74, 109 – 110, 133 linguistic turn 14, 118
Goodman, Nelson 3, 16, 17, 50, 93, 98 – 99, Locke, John 10, 88, 91, 198, 215, 219
105, 116, 156, 170, 182, 189 logic 20, 42 – 43, 117, 121
Grossmann, Reinhardt 170 – 171 logical antirealism 19, 104, 121
logical cognition 20, 101
Hacker, P. M. S. 109 logical constants 72, 101, 103, 108, 122, 126
Heald, William 21 logical experiences 165 – 168
Hegel 3, 11 – 12, 24, 198, 201, 205, 209, logical expressions 20, 22, 101 – 102, 113,
210 – 211, 216 – 223, 224 – 225 117 – 118, 122, 169
Heidegger 3, 12, 14, 85, 112 logical form 71 – 74
Hobbes, Thomas 91 logical objects 19, 21, 22, 101, 113 – 114, 169
Hofstadter, Douglas 233 logical pictures 68 – 69
Horn, Gabriel 100 logical semirealism 19
Horwich, Paul 172 logical statements 101, 121, 125
Human creationism 17, 83 logical structure 20, 21, 84, 98, 99, 102,
Humboldt, Wilhelm von 94, 97 – 98 113, 114 – 115, 116, 117, 119 – 121, 156,
Hume, David 10, 42, 63, 116, 176, 191, 198 –
199, 226 Malcolm, Norman 31
Husserl 3, 87, 202, 235 – 236 Marcus, Ruth Barcan 135
Hutchinson, Brian 56 mathematics 4
Marxism 192
ideal utilitarianism 56, 65 metaethics 55 – 56, 58 – 59
identity, concept of 22, 206 Mill, John Stewart 49, 50, 115
immaterialism 3, 6 Moore, G..E. 8, 24, 47, 50, 52 – 65, 66 – 67,
immediate experience 39 74, 75, 201, 203 – 204, 213 – 214
Inhelder, Bärbel 97
intuitions 14, 39, 54, 64 Nagel, Thomas 14
naturalism 5
James, William 3, 52, 94 – 95, 165, naturalistic epistemology 29 – 31
Jansons, Kalvis M. 95 naturalistic ethics 56 – 58
justified belief 33, 34 negative facts 116, 122 – 129
nepotism 192
Kant, Immanuel 1, 6 – 8, 10 – 11, 13, 15, 16, nonrealism 6
17, 41, 42, 45, 46, 49 – 50, 51, 67, 83, noumena 13
88, 112, 155,156, 157, 162, 166 – 168,
189, 199 – 201, 212, 215 ontological realism 18, 113, 130, 170, 190
Kaplan, David 197 ontology 7
Keller, Helen 95 other minds 48 – 49
knowledge 36 – 38
Koro 96 Pears, David 109
Kuhn 23 Peirce, C. S. 161
Pelletier, Francis Jeffry 152
language of thought 93 perceptual antirealism 21
Leibniz 10, 91, 176 phenomenology 3, 235 – 236

Authenticated
246 Index
physicalism 235 Skinner, B. F. 96

Piaget, Jean 97 Smith, Quentin 196
Pinkard, Terry 209 Socrates 9
Pinker, Steven 16, 99 – 101, 154 speciesism 1, 192
Plato 8, 9, 30, 52, 91, 106 Spinoza, Benedict de 1, 10
Platonism 116, 192 statements of logic 101, 121
practical anthropology 52 states of affairs 18, 108
presupposition 44 – 51 stereotyping 163 – 164
Price, H. H. 88, 91 – 92, 115, 165, 222 – 223 Strawson, P. F. 45 – 46, 153, 174, 229 – 230
Prichard, H. A. 56, 64 subjective epistemology 29, 31 – 42, 43 – 44
private language 48 – 49 systems of beliefs 24
Putnam, Hilary 16, 18, 97, 105, 157, 171, 182
things-in-themselves 10, 112
quasi-realism 22 Titchener, Edward Bradford 3
Quine, W.V.O. 3, 5, 30, 32, 53, 92 – 93, 158, transcendental arguments 44
174, 177, 190 transcendental idealism 8, 10, 57, 83, 189,
212
racism 192 transcendentalia 43, 62, 182
relations 175 – 177 tropes 64, 222 – 225, 233 – 234
relativism 23 truth 20, 36 – 37, 43, 109, 174, 180 – 185
religion 1
Ross, W. D. 56, 60 universals 64, 104, 105, 181, 218, 220 – 226,
Russell, Bertrand 1, 15, 20, 41, 46, 50 – 51, 234
61 – 62, 103, 104, 105, 108, 122 – 129,
131, 134, 137, 151, 162, 176, 195 Vallicella, William 172
Ryle, Gilbert 4, 229
Watson, J. B. 95, 120
Sander, Emmanuel 233 Williams, D. C. 222
Sapir, Edward 96 way of ideas 13
Sartre, Jean-Paul 3, 12, 41, 48 – 49, 50, 166, way of words 13
199, 204, 205, 214, 235 Westerhoff, Jan 173
Saussure, Ferdinand de 94 White, Alan 174
Schauer, Frederick 154 Whorf, Benjamin Lee 96 – 97
Scheffler, Israel 170 Wittgenstein, Ludwig 3, 12, 14, 15, 18, 21,
scientific revolution 9 22, 24, 30, 31, 53, 58, 61, 62, 66 – 85,
Searle, John 94, 193 91, 92, 95, 101, 108 – 111, 112, 120, 130 –
semirealism 22, 68, 72, 76, 108, 112, 114, 141, 144 – 150, 166 – 168, 169, 174, 179,
115, 122, 128, 131, 132, 150, 151, 157, 226, 235
165, 166, 169, 178 – 180, 190, 191, 235 words 3, 13, 22, 30, 53, 59, 69, 90, 91, 94,
sexism 192 95, 99, 100, 108, 110, 119, 142, 167, 177,
Sextus Empiricus 33 211, 219, 220, 222, 230
skepticism 22 – 23, 32 – 33, 35 – 38, 42 – 44, Wundt, Wilhelm Maximilian 3
45, 47, 48, 49, 50 – 51, 65, 112, 191 – 192, Wynn, Karen 99 – 100
194, 207, 228

Authenticated

KR 6 KJ 6 RDTK

Hochgeladen von

Dokumentinformationen

Originaltitel

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

KR 6 KJ 6 RDTK

Hochgeladen von

Copyright:

Verfügbare Formate

Panayot Butchvarov

Brought to you by | Michigan State University

Brought to you by | Michigan State University

Realism, Antirealism, Semirealism

Brought to you by | Michigan State University

Library of Congress Cataloging-in-Publication Data

Bibliographic information published by the Deutsche Nationalbibliothek

© 2015 Walter de Gruyter Inc., Boston/Berlin

Brought to you by | Michigan State University

Brought to you by | Michigan State University

Part One: Epistemology and Ethics Dehumanized

Chapter Two: Three Varieties of Epistemology 29

Chapter Three: The Property Good 52

Chapter Four: Saying and Showing The Good 66

Part Two: Metaphysics Humanized

Chapter Five: The Role of Language in Cognition 85

Chapter Six: Metaphysical Realism and Logical Antirealism 104

Brought to you by | Michigan State University

Chapter Seven: Logical Semirealism 130

Chapter Eight: Generic Statements 151

Chapter Nine: Facts and Truth 169

Part Three: Metaphysics Dehumanized

Chapter Ten: I and the World 189

Chapter Eleven: We and the World 203

Chapter Twelve: Mind and the World 218

Brought to you by | Michigan State University

Brought to you by | University of Michigan

Brought to you by | University of Michigan

Brought to you by | University of Michigan

Brought to you by | University of Michigan

A familiar thesis in ethics is that one ought to treat others “humanely,” as

Brought to you by | University of Michigan

ing concepts” or describing the “workings of our language,” as if concepts and

 Nelson Goodman, Ways of Worldmaking (Indianapolis: Hackett, 1978), 6.

Brought to you by | University of Michigan

Brought to you by | University of Michigan

issue is irrelevant to many metaphysical inquiries, especially those in pre-Kant-

Brought to you by | University of Michigan

The history of anthropocentrism in philosophy is illuminating, though we can

Brought to you by | University of Michigan

Brought to you by | University of Michigan

 Critique of Pure Reason, B 69.

Brought to you by | University of Michigan

 Ludwig Wittgenstein, Tractatus Logico-Philosophicus, trans. D. F. Pears and B. F. McGuinness

Brought to you by | University of Michigan

Brought to you by | University of Michigan

ics, of course, may be essentially an elaborate proposal of ways of acting, espe-

Brought to you by | University of Michigan

3 Antirealism and its Varieties

Antirealism is not limited to metaphysics. There is also antirealism in ethics, aes-

Brought to you by | University of Michigan

spectacles, you could be sure of seeing everything blue.”²⁶ Neuroscience unambig-

Brought to you by | University of Michigan

Brought to you by | University of Michigan

 Ways of Worldmaking, especially chapters I and V.

Brought to you by | University of Michigan

Brought to you by | University of Michigan

4 Logical Antirealism and Semirealism

Brought to you by | University of Michigan

 See Gustav Bergmann, New Foundations of Ontology (Madison: University of Wisconsin

Brought to you by | University of Michigan

Brought to you by | University of Michigan

Brought to you by | University of Michigan