You are on page 1of 22

On Performing and Explaining Linguistic Actions

Author(s): Quentin Skinner


Source: The Philosophical Quarterly, Vol. 21, No. 82 (Jan., 1971), pp. 1-21
Published by: Blackwell Publishing for The Philosophical Quarterly
Stable URL: http://www.jstor.org/stable/2217566
Accessed: 12/11/2008 20:31
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=black.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

The Philosophical Quarterly and Blackwell Publishing are collaborating with JSTOR to digitize, preserve and
extend access to The Philosophical Quarterly.

http://www.jstor.org

THE

PHILOSOPH

QUARTERL
VOL. 21

No. 82

JANUARY

1971

ON PERFORMING AND EXPLAINING


LINGUISTIC ACTIONS1
BY

QUENTIN

SKINNER

I
To utter any serious utterance is both to say something and to do someJ. L. Austin put it in his classic formulation of
thing. The speaker-as
this now familiar point2-can
thus be doing something in saying what he
not
as
a
and
says,
consequence of what is said. Any agent who utters
just
a serious utterance (in speaking or writing) can thus be said, in appropriate
circumstances, to perform a type of voluntary action. The question I wish
to raise is whether this fact can be shown to have any special bearing upon
the analysis of the concept of voluntary action itself. The answer I wish to
suggest is that if (as I shall assume) the things we do with words can genuinely
be regarded as forms of social behaviour, then the analysis of this class of
(linguistic) actions can in fact be used as a means of reassessing two prevailing theories in the philosophy of action : one concerned with the nature
and description of actions, the other with the appropriate method of explaining them.
Austin eventually isolated five types of so-called " illocutionary " act
which can be performed by a speaker in saying something. There are two
rather strictly convention-governed types
"verdictives " (concerned with
1I am very much indebted to Professor B. A. O. Williams for reading and commenting on successive drafts of this paper. He suggested the approach followed in
the first half, and saved me from some serious misunderstandings as well as overstatements, although he is not to be held responsible for any that remain. I am also
very grateful to Mr. John Dunn and Professor J. W. Yolton for their comments on
an earlier draft.
2J. L. Austin, How to do things with Words, ed. J. O. Urmson (Oxford, 1962), p. 94.
Hereafter cited by page only.

QUENTIN

SKINNER

giving verdicts) and " commissives " (concernedwith committing the speaker
to a course of action). But there are also three more conversational types,
and it will be these less strictly convention-governed cases on which I shall
concentrate : " exercitives " (such as entreating or warning) ; " behabitives "
(such as complimenting); and "expositives" (such as informing). The
relations between these classes of action and the concept of action as I wish
to discuss it can now be brought out in a single example. I choose to concentrate on the " exercitive " case of warning, since both Strawson (in his
extension of Austin's general theory)3 and Austin himself (in How to do
things with Words)have treated this as the paradigm case of an illocutionary
(as opposed to a " perlocutionary ") act. Consider then the following example : the case of a policeman who sees a skater on a dangerous pond and
utters to the skater the serious utterance 'The ice over there is very thin '.
(This is actually one of Strawson's examples.) Here the speaker says something and the words mean something. Austin's further point, however, is
of course that the speaker can also mean something in uttering an utterance
with such a meaning. The relations between the meaning of the utterance
and the question of how the speaker means the utterance to be taken are
not perhaps entirely clarified in Austin's account. But there can be no
doubt that two implications-both of them crucial to my own subsequent
argument-can now be drawn from this example.
The first is that if it is now said that what the policeman is doing in
saying " The ice over there is very thin " is warning the skater, this is to
provide, at least in the standard case of speaking with such illocutionary
force, an account of the speaker's intentions. This seems closely analogous,
moreover, to the account which H. P. Grice has given of the relations between intentions and meaning.4 An understanding, that is, of the nature
of the illocutionary act performed by the speaker in uttering his given
utterance seems analogous (perhaps equivalent) to an understanding of
what Grice originally called " non-natural " meaning-at least if we construe
that concept, as Grice himself now wishes to do,5 as referring to the meaning
of a speaker's utterance of a given utterance. And an understanding of
such non-natural meaning-that is, an " uptake " of what the speaker, in
uttering his given utterance, may have meant, or a specification of what
Grice now wishes to call the "utterer's occasion-meaning "-does seem
equivalent in turn (whatever may be true of " timeless meaning ") to an
3P. F. Strawson, "Intention and Convention in Speech Acts", The Philosophical
Review 73 (1964), pp. 439-51.
4H. P. Grice, "Meaning ", The Philosophical Review 66 (1957), pp. 377-88.
6H. P. Grice, "Utterer's Meaning and Intentions ", The Philosophical Review 78
(1969), pp. 147-77, greatly extends and refines Grice's 1957 discussion of " non-natural "
meaning, treating this in turn " as an attempt to define the notion of utterer's occasionmeaning " (p. 150). Grice's article has appeared since the completion of my own script.
I have thus retained his original term 'non-natural meaning', although I have of
course followed his new distinction (p. 149) between 'utterer's occasion-meaning' and
' utterance-type occasion-meaning' and his identification of 'non-natural meaning'
with the former category.

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

understanding of what the speaker intended to do in uttering his given


utterance.
The second implication concerns the exact sense-which
this example
now illustrates-in
which saying something can count as a case of performing
a voluntary (linguistic) action. For granted that the circumstances are
suitable, the policeman's utterance of his utterance actually constitutes the
action of warning the skater. And this applies equally to such cases as
complimenting, entreating, informing and so on, in a list of explicit performative utterances extending (Austin assures us) to the third power of ten.
To identify in such a case the nature of the illocutionary force co-ordinate
on a given occasion with the ordinary meaning of the given utterance is
equivalent to understanding the nature of the (linguistic) action performed
by the speaker in uttering his given utterance.

II
I now turn to consider the possible impact of these considerations upon
the current discussion of two general issues in the philosophy of action. My
first concern is with the descriptive element in the concept of action. Here
I wish to re-examine two theses which both seem to have become generally
accepted in recent discussions. The first states that the locution primitive
for the concept of action is that of something being caused to happen. If
a given episode in which an agent is doing something is to be classified as
a case of action, it is said, then it must be possible to redescribe the episode
in the form of the locution 'A bringing it about that p ', where A stands
for the agent and p the result of 9ing, where 9 in turn signifies a verb of
action, such verbs being defined in turn as those which are applicable as
answers to questions of the form 'What is A doing ? '.
I shall assume that what the proponents of this claim wish to assert is
that it states a necessary condition for applying the concept of action. It
is true that the condition has sometimes been treated as "the essential
feature " of action, and thus as the means of distinguisihing the things we
do from the things that happen to us. It is also true that the condition has
sometimes been represented by critics as an attempt to give the " definition
of an act ", as a claim that if the primitive locution can be applied this
must mean that the episode in question is an action.6 But these appear to
be misunderstandings.
They certainly represent an unjustifiably strong
way of putting the point. Consider for example the case of a skater on thin
ice who brings it about in skating that he falls into the water. The episode
can certainly be described as a case of A bringing it about that p, but we
should hesitate to call it the description even of an unintended action on
the part of the skater. I think it is usually clear, however, that it is only
6For these claims see respectively Alan R. White, ed., The Philosophy of Action
(London, 1968), p. 11 ; John Ladd, " The Ethical Dimension of the Concept of Action ",
The Journal of Philosophy 62 (1965), p. 640; A. I. Melden, Free Action (London, 1961),
p. 23.

QUENTIN

SKINNER

the converse claim which the proponents of this thesis wish to make on
behalf of the primitive locution-the
claim that if A is ying, A is performing
a voluntary action only if A in ying is thereby bringing it about that p.
And it is of course in this form that the thesis has become generally accepted.
The other thesis I wish to re-examine states the set of conditions which
must be met by any verb of action q such that if it is true that A is ping
it will equivalently be possible to redescribe the episode in the form of the
locution agreed to be primitive for action. It has of course been a weakness
of much recent discussion about the nature of action that this further question has not always been raised, so that the primitive locution has often
been introduced-as
it is, for example, by Bennett, Chisholm, White and
others7-without
any explicit justification, and without any account of the
conditions for applying it. It has also been a feature, however, of several
recent studies both that this lacuna has been recognized, and that a particular
method has been developed specifically to elucidate the conditions such
that, when they are met, it will follow from the fact that A is ying that A
is thereby bringing it about that p.
These conditions, it is said, can (and can only) be derived from a study
of the tense logic of verbs of action. The progressive refinement of this
essentially Aristotelian insight has served to yield a number of distinct
categories into which verbs of action can be classified, all with an evident
purchase upon the philosophical analysis of voluntary action. There is
Ryle's distinction between "task " and "achievement " verbs; Vendler's
distinction between "states" and "achievements"
on the one hand and
' activities " and " accomplishments " on the other; and Kenny's isolation
of the concept of a "performance ", which he distinguishes both from
" states " and " activities ". It has further been claimed that the conditions under which this basically syntactical concept of a " performance "
can be applied are equivalent to the conditions such that if it is true that A
is (ing, then it follows that the episode can be redescribed in the form of
the locution agreed to be primitive for action. Kenny's most essential
claim, in fact, is that for any case in which it is true to say that A is (ing,
and in which the verb of action 9 passes the tests for being classified as a
" performance " verb, it will equivalently be true to say that A is thereby
bringing it about that p.8 Again, this thesis has been extensively discussed
and refined (by Potts, Taylor, Evans and others9), and appears to have
become quite generally accepted.10
7See Daniel Bennett, "Action, Reason and Purpose ", The Journal of Philosophy 62
(1965), p. 85; Roderick M. Chisholm, "The Descriptive Element in the Concept of
Action ", The Journal of Philosophy 61 (1964), p. 615; White, op. cit., p. 2.
8Anthony Kenny, Action, Emotion and Will (London, 1963), pp. 177, 182. Hereafter cited by page only.
9See Timothy C. Potts, " States, Activities and Performances ", Proceedings of the
Aristotelian Society, Supplement 39 (1965), pp. 65-84; C. C. W. Taylor, " States,
Activities and Performances ", loc. cit., pp. 85-102; C. O. Evans, " States, Activities
and Performances ", The Australasian Journal of Philosophy 45 (1967), pp. 293-308.
l?See, for example, the citation of these conditions in White, op. cit., pp. 3-4 and

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

All these philosophers are in essential agreement, moreover, about the


nature of the conditions which must be met if an episode in which A is
ying is to be classed as a "performance ". (I shall follow this neologism,
rather than Vendler's proposal to speak of " accomplishments ", since it
has gained wider currency.) First, the sentence reporting that A is ying
must take a singular or distributed plural form of grammatical object.11
(For example, 'The policeman is warning the skater' rather than 'The
policeman is warning skaters', which would describe an " activity " rather
than a "performance ".) Next, the verb of action ( must pass a basic
grammatical test which isolates rather than merely applies to "performance " verbs: the test that if it is said that A is (ing B (for example,
warning B), this must imply (as it does in this case) that A has not yed B
(has not warned B) rather than (as with "activity" verbs) that A has yed
B (Kenny, pp. 172-3). Finally, the report of the episode must pass five
tests (Kenny, pp. 176-8 and 183-4). The first two can be
"criterial"
represented by saying that whereas " activities " can go on for a time, it
is characteristic of " performances " that they take time, and can be done
either quickly or slowly. (For example, it takes time to issue a warning,
and we can imagine an emergency in which A might not have enough time
to warn B.) The next criterial test states that "performances ", unlike
" activities ", can always be either complete or incomplete. (For example,
A might start to warn B and be cut off, in which case A may not succeed
in performing the given action. But this situation cannot arise if A in ying
is merely engaged in an " activity " such as running.) Finally, any " performance" can both have a purpose and be commanded. (For example, A
can without absurdity and for a purpose be ordered: " Warn B ! ".)
To summarize and apply these purely exegetical remarks: I have now
said that if A is warning B (or complimenting, entreating, informing and
so on), then A is performing a voluntary (linguistic) action; and I have now
shown that the action which A takes in warning B meets all the conditions
for being classified as a "performance ". (This will equally be found to
apply, in suitable sentences, to the case of A complimenting or entreating
or informing B, and so on.) So it follows, according to the theory I am considering, that any case in which A is ying B, and ying is warning (or complimenting and so on), will be a case in which it will equivalently be true to
say that A is thereby bringing it about that p.
The only truth, however, which this form of analysis seems to yield so
far about warning, complimenting and so pn is that these verbs of action
are capable of being put through a passive transformation. If A is ying B
and ying is warning, then it is undoubtedly but also trivially true to say
their deployment in Richard Taylor, Action and Purpose (Englewood Cliffs, 1966),
pp. 104, 107, 111-2.
l1Here and in stating the " criterial " tests I follow Evans's refinements of the
tests for " performances ".

QUENTIN

SKINNER

that A in ying B (warning B) will thereby be bringing it about that p (that


B is in fact warned). If the analysis is to be rescued from this triviality it
is obvious that some independent value must be assigned to the 'p ' which
is said to be brought about whenever a "performance " takes place. It is
perhaps a methodological weakness of certain existing discussions about
" performances " that they have scarcely emphasized the crucial importance
of this point or treated it in a very systematic way. It is certainly possible,
however, to gain from an analysis such as Kenny's an account of the features
of the situation which it is said must necessarily be brought about as the
result of a "performance ".
The essence of Kenny's claim is in fact that ' p' must designate some
new and discernibly changed state of affairs which is the result of A's action
(pp. 177-8). In the case of "intensional" (psychological) "performances"
the change which must take place is in A, the subject of the action. To say
that a " performance " is " intensional " is thus to say that when the agent
acts and brings about a change, " the change is in the subject and not, save
per accidens,in the object " of the action (p. 196). In the case of any " performance" which is not just psychological, however, the change which it is
said must be brought about is not in A (the subject of the action) but in B
(the patient or object of the action). In any such " performance " it is
said to be essential that " after the action is over, the patient must have
changed " (p. 181). There must be some state " newly true of the patient "
which "must be different from its state before the event " (p. 181). And
the change which is thus brought about will be a difference which we can
" inspect " and submit to " examination " (p. 180), a difference for example
in the properties, the position, or the nature of B, the object of A's action
(p. 178). The value being assigned to 'p' in the case of these ordinary,
non-psychological " performances " is thus that when A ps B, the state of
B which A is thereby said to bring about must be such that " after the event
B should have changed " in some evident and corresponding way (p. 180).
The thesis I now wish to argue is that once these accounts of " performances " are rescued from tautology in this way, they can be shown to be
founded upon a mistake. Once the value being assigned to 'p ' is clarified,
it can be shown that no necessary equivalence exists at all between reports
that A is ping B and reports that A is bringing it about that p. I shall
continue, in trying to make good this claim, to concentrate on the somewhat complex but central case in which the (linguistic) action which A performs in ping B is in fact that A warns B. I believe the same case could
equally well be argued, however, for any of the other cases of linguistic action
I have mentioned.
I turn first to consider the situation in which A might truly be said to
have brought it about that p (that B has been warned), where 'p ' duly has
the value required by Kenny's type of analysis. A might for example have
brought it about that B was caused to act in such a way that B successfully

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

avoided some potential danger. We can certainly say of such a situation


that B has been warned, and we may expect this to be reflected in some
state newly true of B which represents the result of a warning having been
issued to B. Now according to the theory I am considering, any situation
of which it can be said that A is (ing B, where (ing is a " performance ",
will be a situation of which it can equivalently be said that A is bringing
it about that p. So it must follow, according to these accounts, that if A
has in fact brought it about that p-as in the present example-then A
must have (ed B (warned B) and therebybrought it about that p (that B
was in fact caused to act in such a way as to avoid some potential danger).
But this does not necessarily follow at all. It is possible for A to bring
about such a result without issuing any warning himself or indeed performing
any linguistic action at all. The policeman might for example succeed in
bringing it about that the skater on the dangerous pond is duly warned
merely by signalling to another policeman to warn the skater of the potential
danger. It will certainly be true to say of this situation that the skater
has been warned, and that A (the first policeman) has brought it about
that p (that the skater has duly been warned). But it is not equivalently
true, and in this case it is not true at all, to say that A (the first policeman)
(ed B (warned the skater) in any way required by the existing accounts of
( performances ".

This failure of equivalence can of course be rectified by an expansion of


'.
If the "p " which the first policeman has to bring about in (ing B
'p
(warning the skater) is that the skater is warned by himself, the policeman,
then the loss of equivalence in this case between qing and bringing it about
that p is certainly restored. But this does not really meet the difficulty. It
merely seems to remove the alleged point of converting such a report to
the effect that A has done something into a report that A has brought about
a result. The alleged point of such a conversion is that the latter form of
description should yield what Kenny (p. 185) calls a more " fundamental "
description of the action. But the simplest version of this allegedly more
fundamental description in the present case is that "The policeman is
bringing it about that he, the policeman, is warning the skater ". This
report, however, of what the policeman brings about not only fails to add
anything to the original description of his action. It actually represents a
repetition, in a syntactically clumsier form, of exactly the original report of
what the policeman was doing. The price in this case of restoring the
equivalence between ying and bringing it about that p is thus that the
analysis of the action simply moves in a circle-a result which might be
thought to cast a quite general doubt on the status being claimed for the
locution said to be primitive for voluntary human action.
The main point I wish to make in criticism of the existing attempts to
apply the concept of a " performance " to elucidate the concept of voluntary
action is the converse of the claim I have now argued. I have argued that

QUENTIN

SKINNER

there are cases of " performances " in which it can be true that A is bringing
it about that p without its being equivalently true that A is (ing B. I now
wish to argue that there can be cases in which A is genuinely (ing B, and
in which the verb of action q passes all the tests for being a " performance "
verb in a suitable sentence, but in which A in (ing B is not thereby bringing
it about that p, according to any value of 'p' assigned in the existing
discussions about "performances ".
I begin by reverting to the dichotomy laid down in Kenny's account
between intensional and non-intensional "performances ". It is clear that
the linguistic actions with which I am concerned cannot be said to fall on
the intensional side of this dichotomy, if intensionality is heuristically
defined (as in Kenny's account) as that property peculiar to psychological
events. If A is warning B, for example, this involves something more than
a change merely in A's affective or cognitive state, as when A is hating or
remembering B. And if A in warning B does something more than bring
about a change merely in his own mental condition, this can scarcely be
said to happen merely per accidens. The attempt to do something more
seems on the contrary to reflect the standard intention of any agent who
performs such a linguistic action.
The peculiarity, however, of the linguistic " performances " with which
I am concerned is that they not only fail in this way to assimilate to the
class of intensional " performances ". They also fail to meet the conditions
for being classed as ordinary non-intensional " performances " as well. They
pass all the tests, as I have shown, which have been laid down for the isolation of " performances ". And yet it does not follow that the result of A
(ing in these types of case can be represented either as a change essentially
in the agent A or as a change in B, the object of A's action, in any sense
required by the existing discussions of " performances ".
I shall consider six different ways in which it might be argued that a
genuine linguistic " performance " can take place, in which A is authentically
(ing B, but in which no corresponding change is thereby brought about by
A in B's state or activity. The speaker might first of all fail at a purely
locutionary level to bring about any change in the object of his warning.
It might happen (1) that A is warning B but that B does not hear the words
of warning; or (2) that B hears the words but does not understand their
meaning. Next, the speaker might fail at the illocutionary level and in
consequence fail to bring about any change in B's state or behaviour. It
might happen (3) that A is warning B, that B hears the words and understands their meaning, but that B misunderstands their intended illocutionary
force. An important example of this type of failure, which Austin did not
discuss but which can certainly arise, would be the situation in which the
conventions (social as well as linguistic) are such that A's intention (to
warn or compliment B and so on) is incapable of being assessed by B as
being a case of that intention.l2 If A then compliments B, for example, in
12For a discussion of this and other cases of oblique and non-avowable illocutionary

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

such a situation, he may thereby bring it about only that B takes him to
have had the intention to perform some wholly different action-the
intention, perhaps, to flatter or defer to B, possibly even to ridicule him. A
second type of failure at this illocutionary level, which Austin does discuss,
might happen (4) when B hears, understands and correctly grasps the
intended illocutionary force of A's utterance, but (incorrectly) believes that
A's warning was not seriously intended. The utterance might be taken,
that is, as a mock performance-a
case of parodying, joking or otherwise
only pretending to warn, compliment and so on. Finally, the speaker may
still fail in 9ing to bring it about that p even when his locution and its
intended illocutionary force are both fully understood. There could be a
situation (5) in which A believes there is a danger to B and points it out,
but fails to persuade B of the danger and so fails to bring about any change
in B's state or alteration of his behaviour. There could also be a situation
(6) in which A believes there is a danger to B and points it out, in which
B also believes there is a danger to himself, but in which A still fails thereby
to bring about any change in B's state or behaviour, simply because B is
reckless, so that A's warning has no effect upon B's will.
I shall now consider and try to meet two objections which might possibly
be levelled against the claim that these are all cases of genuine " performances" which nevertheless do not involve the bringing about of any
change in the object of the action. It might first of all be pointed out (quite
fairly) that the lack of equivalence between ying and bringing it about that
p in these cases rests very much on the fact that (to put it colloquially) it
is not altogether up to A, the agent in these cases of linguistic action, whether
or not in ping he does bring about a change in the state of B, the object of
his action. It might thus be objected that all these cases show is that
linguistic or other inter-personal actions are very peculiar cases of action,
in which the power of agent causality is unusually restricted or diminished.
This objection depends in turn, however, on assuming that any voluntary
action must be a case in which the successful bringing about of an intended
result is entirely up to the agent, and depends merely upon his will. But
it is not even characteristic of the most basic " performances " that the
changes capable of being brought about by the agent are in fact entirely
up to him in the sort of way that would be required to sustain this objection.
Consider for example the following reported episode : " Mary is cutting the
cake ". This is an unambiguous case of a simple performance. It is indeed
offered as such in Kenny's account (p. 172). If it is true that Mary is qping
B (cutting the cake), then it will be equivalently true that Mary is bringing
it about that p (that the cake is in a cut state). Yet it is clear even in this
case that there could be restrictions upon the power of the agent, such that
Mary fails in trying to q to bring it about that p. It does not seem, thereacts, see my own article, " Conventions and the Understanding of Speech Acts ", The
Philosophical Quarterly 20 (1970), pp. 118-38.

10

QUENTIN

SKINNER

fore, that there is any important distinction to be drawn between this type
of simple " performance " and the type I am considering, over the question
of whether the achievement of the result said to be intrinsic to the description
of the action as a " performance " is in fact entirely up to the agent.
The second and more serious objection, however, might consist of questioning whether the six episodes I have described really constitute cases in
which the " performance " of A warning B can be said to have taken place
at all. It might be claimed that these are not really cases of A warning B,
but are merely cases of A trying to warn B. And if this objection is now
tested against the six types of asymmetry I have cited it certainly appears
to have some purchase. It seems more or less certain, at least with cases
(1) and (2), that the correct description of the episode must after all be
that A is trying (but failing) to warn B. There still seem good grounds for
doubting, however, whether it would be true to say of any of the other cases
that they should be described as cases of A trying to warn B rather than as
cases of A warning B. It seems unquestionable, at least in cases (5) and
(6), that the objection cannot be sustained. Here the speaker A utters an
utterance with the form and intended force of a warning to B, which B
correctly understands and of which B gains " uptake " as a warning. If B
disbelieves the warning, or else believes but fails to heed it, then A undoubtedly fails to bring about that there is any state of activity "newly
true " of B as a result of the warning having been issued to him. But there
can be no doubt in both these cases that A still did warn B. It would
undoubtedly be correct for A to report of episode (5) "I warned him but he
took no notice ", or of episode (6) "I warned him but he refused to take
any notice ". It is at least not clear, moreover, that it would be mistaken
in cases (3) and (4) to insist on a similar point. In case (3) the failure to
bring about a change in B's state might be due to nothing more than B's
" non-playing " (in Austinian parlance) of A's warning (or compliment and
so on) as a means of registering his disapproval of it. It might thus seem
quite appropriate to say of such a case that A did warn B (or compliment
him and so on) but that B refused to " take " (but did not fail to grasp)
the warning. Similarly in case (4) the failure is solely due to B's misunderstanding of how seriously the warning was intended. Again, it might thus
seem quite appropriate to say that A did warn B (since B did gain "uptake " of the warning as a warning) but that this had no effect upon B
because-as we do in fact say-he treated the warning as a joke.
The objection, therefore, that these asymmetric cases can be dismissed
as cases of A merely trying to q B and failing to bring it about that p does
not hold generally. The break between attempting to perform and successfully performing the relevant action seems to come at some point between
case (1) and case (6). I am not concerned with the question of exactly where
this break occurs. All I am concerned to claim is that it is unquestionable
with cases (5) and (6), and at least arguable with cases (3) and (4), that A

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

11

is definitely ping B (warning B), even though A is definitely failing thereby


to bring it about that p (that any corresponding change takes place in B's
state or behaviour).
If these are genuine asymmetries, two implications can now be said to
follow from the argument I have presented. The first is that the value of
'p' seems to have been incorrectly stated in the existing discussions of
"performances ". It is of course true that if A succeeds in bringing it about
by issuing a warning to B that B is thereby put on his guard, then B has
certainly been warned by A. If any of the cases (3) to (6) are genuine cases
of A warning B, however, it follows that this condition of bringing about a
change in B does not have to be met for the concept of warning to have
application. It is only necessary, for A to have warned B, that A should
have given notice of the fact to B that B is in some potential danger. The
same applies to the other cases of linguistic action I have mentioned. If A
succeeds in bringing it about in complimenting B that B is in fact flattered,
then B has certainly been complimented by A. As the O.E.D. puts it,
however, it is only necessary, if it is to be said that A is complimenting B,
that A should be addressing B " with formal expressions of civility ". So
too with A entreating B. If A brings it about in entreating B that B is in
fact prevailed upon to act in some way, then B has certainly been entreated
by A. But it is only necessary (as the O.E.D. again puts it), if it is to be
said that A is entreating B, that A should be " pleading " with B. And so
too with A informing B. If A brings it about in informing B that some item
of knowledge is in fact gained by B, then B has certainly been informed by
A. But it is only necessary (to appeal to the authority of the O.E.D. once
more), if it is to be said that A is informing B, that A should be " issuing
instruction " to B.
The value which I have now stated for ' p ', however, renders the formula
'A bringing it about that p' unnecessary in these cases. If "A warning
B " (or complimenting and so on) can simply be a matter of A pointing out
to B that B is in potential danger, it becomes merely otiose, and nothing is
gained at all, if we insist on saying that A is thereby " bringing it about
that he is pointing out to B that B is in potential danger ". Once this is
granted, however, a second and more general implication can be said to
follow from the argument I have presented. It is that some doubt must
attach to the claim that the formula 'A bringing it about that p' can really
be regarded after all as " palmary for the description of voluntary action "
(Kenny, p. 152). This is not to deny, of course, that there are voluntary
actions which fit the description of being episodes in which an agent brings
about a result. The point is rather that even when we are armed with this
locution, we are left confronting something elusive about the concept of
action itself. It can readily be shown that the locution does not state a
sufficient condition for applying the concept of action: for an agent, as I
have shown, can also be causally connected in the required sense with the

12

QUENTIN

SKINNER

things that happen to him. I have now sought to argue that the locution
also fails to state even a necessary condition for applying the concept of
action: for the situation which A brings about in (ing B, in the various
cases of linguistic action I have considered, is simply the situation in which
A is indeed ping B, and is not necessarily to be represented as some further
change in the object of A's action which can be represented as the result
of the action.
III
I now turn to consider the possible significance, for the question of how
voluntary actions are to be explained, of the account of linguistic action
which I have now tried to set out and defend. I wish in particular to consider this account in the light of the traditional thesis-the thesis both of
Aristotle and of Hobbes, which has again been re-stated in the course of
the very recent reaction against Wittgensteinian philosophical psychologyto the effect that " the relation between a reason and an action when the
reason explains the action by giving the agent's reason for what he did "
is " a species of ordinary causal explanation ". This is Davidson's introduction to his classic reformulation of the traditional position.l3 The same
general commitment has also been announced in a large number of other
recent studies (by Ayer, Berofsky, Hamlyn, Maclntyre, Yolton and others),14
all of whom have sought to establish the same general conclusion by showing
that when either a motive or an intention is cited (as they are " in the normal
way ", as Ayer puts it, p. 9) as a means of explaining an action, then " the
relations between intentions, volitions and emotions on the one hand and
action on the other, is causal, not logical " (Yolton, p. 26). The way is thus
left open to insisting, as many of these philosophers want to do, that (as
MacIntyre puts it, p. 224) " the dependence of the concept of human action
upon the concept of intention does not exclude the possibility of causal
explanation from the realm of human action ". The general significance of
this line of argument is of course that it appears to hold out some hope for
a fully causal and predictive science of human action. It has already been
concluded, indeed, by some of the less cautious of these philosophers, that
there is " nothing about the concept of action which lifts human behaviour
out of the realm of causal explanation ".15
13Donald Davidson, " Actions, Reasons and Causes ", The Journal of Philosophy 60
(1963), p. 685.
14A. J. Ayer, " Man as a Subject for Science ", in Philosophy, Politics and Society,
Third Series, ed. Peter Laslett and W. G. Ruiciman (Oxford, 1967), pp. 6-24; B.
Berofsky, " Determinism and the Concept of a Person ", The Journal of Philosophy 61
(1964), pp. 461-75; D. W. Hamlyn, " Causality and Human Behaviour ", Proceedings
of the Aristotelian Society, Supplement 38 (1964), pp. 125-42 ; Alasdair MacIntyre, " The
Antecedents of Action ", in British Analytical Philosophy, ed. Bernard Williams and
Alan Montefiore (London, 1966), pp. 205-25; J. W. Yolton, " Agent Causality ", The
American Philosophical Quarterly 3 (1966), pp. 14-26. All citations hereafter by page
only.
15G. Madell, " Action and Causal Explanation ", Mind 76 (1967), p. 34. My italics.

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

13

I now wish to argue that in the face of such claims the special features
of linguistic action take on a further significance. I begin by reverting to
the brute facts about my basic example: the policeman uttering to the
skater the utterance 'The ice over there is very thin '. So far I have defended two theses about this situation. First, the episode thus described
might in suitable circumstances be correctly redescribed by saying " The
policeman is warning the skater ". And secondly, what this unique (illocutionary) redescription of the episode further reveals is what the policeman
in uttering his utterance intended or (in Grice's sense) non-naturally meant.
I now wish to illustrate and defend two further theses about such illocutionary
redescriptions. First, that when the facts about such an episode are capable
of being redescribed in the unique form in which the intended illocutionary
force of the utterance is revealed, this provides not merely a redescription
but also a mode of explanation of the given episode. And secondly, that
although such explanations function by citing reasons, these cannot in this
case be construed as causes of which the corresponding actions are effects.
Consider the situation in which the policeman's utterance is heard by a
puzzled bystander who for some reason fails to grasp the policeman's intention in uttering this utterance. The bystander might of course be puzzled
by various features of the episode. But one request for an explanation might
undoubtedly take the form of the question 'Why did he say that? ' (or
more exactly, 'Why did the policeman utter that given utterance ? '). And
one reply, providing an explanation of the policeman's action, might be
' He said it to warn the skater' (or more exactly, 'The policeman's reason
for uttering that given utterance was to point out the potential danger to
the skater of skating where the ice is very thin '). The illocutionary
redescription serves as an explanation of the action.
There seems no doubt, moreover, that there must at least be a standard
case in which it is correct to speak of such illocutionary redescriptions as
supplying explanations of at least some puzzles about such types of action.
For it is one thing if the bystander understands what A's utterance to B
itself means, so that he may be capable of giving an account of what A said.
But it is quite another and further thing if the bystander understands what
A's utterance of an utterance with that meaning was itself intended to
mean on the given occasion, so that he may be able to give an account of
why A said what he said. Colloquially, what this further information explains about A's action is its point.
There also seems no doubt about the manner in which valid explanations
can be supplied by means of such illocutionary redescriptions : they cite a
reason for A's performance of his action. To know that A in a given instance
meant to warn B is to know that if A was acting intentionally in the ordinary
way, then one at least of his reasons for uttering his given utterance must
have been to point out to B that he regarded B's situation as being one of
potential danger. But it can scarcely be said that to cite such a reason is

14

QTUENTIN SKINNER

equivalent to citing a cause of A's action. The explanation is supplied from


a grasp of the conventions governing the illocutionary force attaching to
the utterance itself. It is supplied, that is, simply from an understanding of
what A meant, in the non-natural sense of understanding not just what the
utterance meant (for it is not even essential that the utterance itself should
in that sense have any meaning at all) but of understanding what it conventionally means to utter an utterance with such a meaning. To know the
answer to this further question is to know how the speaker intended his
utterance to be taken. And this in turn is to know the intentions of the
speaker in uttering that given utterance. And this in turn is to know why
he uttered that utterance. And this is to have an explanation of the
(linguistic) action.
The independence of such explanations from causality, however, has
now been overstated. The explanation of why A uttered his given utterance
is not merely a matter of citing the convention that to utter a particular
utterance can count in suitable circumstances as a case of warning someone.
It is a matter of citing this convention and of assuming that the speaker
both knew of this convention and intended to follow it in the given case.
This is to concede, however, that in order to complete even these conventional types of explanation a further condition which is at least arguably
causal needs to be added-the condition that the agent should first have
attained a knowledge of the relevant conventions and should then have
formed the intention to act according to these intentions in the future.
It still seems possible, however, to distinguish the type of reasonexplanation with which I am concerned, which consists of focusing primarily
on the agent's intentions in his actions, from the type of reason-explanation
which focuses primarily on the agent's cognitive and affective states prior
to action, linking these by inductive generalization to the performance of
certain sorts of action. The type of conventional explanations with which I
am concerned precisely functions by appealing not to inductive generalizations but rather by appealing to the conventions which may seem to apply
in the situation to be explained. It is true that the convention which thus
serves to explain the action must still be psychologically linked to the agent
who performs the action. It remains true, however, that the primary focus
in this type of explanation lies in the elucidation of conventions rather than
in the elucidation of the agent's mental states. It is not essential to allude
directly to these states at all in order to provide a conventional explanation
of an action which, while partial, inferential and so highly fallible, may
nevertheless be illuminating and correct. It still seems possible, therefore,
to vindicate the sole claim I am concerned to make : that the recovery of
a conventional (illocutionary) redescription, by pointing to the intentions
of an agent in the performance of a given action, still possesses in itself a
measure of explanatory force.
It must be admitted, however, that this primary role which I have

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

15

assigned to recovering the agent's intentions in order to explain his (linguistic)


actions has usually been quite explicitly rejected by the theorists and practitioners of precisely those disciplines in which the argument might be
thought to apply most significantly. I have in mind those disciplinessuch as literary criticism and the various forms of intellectual historywhich are essentially concerned with the analysis of complex utterances and
the explanation of their special features. It happens that one of the pieces
of conceptual bric-a-brac which the practitioners of these disciplines have
been taught to handle with reverence has been the argument exposing the
alleged "intentionalist fallacy ". And so it has frequently been insisted
that it is possible to establish the " general irrelevance of intentions "16 in
any attempt to interpret and explain the complex utterances (such as
literary or philosophical works) to which such critics and historians characteristically devote their attention.
This argument, however, appears to have been mounted in a somewhat
confused way. Those who have argued against the relevance of intentions
have tended to assume that to ask about an author's intentions must
necessarily be equivalent to asking about " what made him write ",17 and
that an author's intentions in his works must thus be ranked " amongst the
factors that caused those works to come into being ".18 And those who
have sought to defend the need to grasp an author's intentions in his works
have tended to assume that this task must consist of showing the relevance,
in explaining the special features of his works, of " the artist's personal
interests, problems, life-history and psychic peculiarities ".19 It is not surprising, therefore, that it has come to seem a fairly decisive objection against
this latter view to point out that it is surely " completely irrelevant " to
the understanding of a work to know, say, that it may have been composed
"primarily in order to make money ".20
This alleged exposure of the intentionalist fallacy, however, scarcely
seems to constitute a very relevant threat to the argument I have attempted
to present. Despite the claims being made for them, neither of these arguments really bears at all on the question of an author's intentions in writing
his given works. Both the arguments are in fact addressed to the question
of whether an author's motivesin writing (such as the desire for gain or other
' psychic peculiarities ") can be said to be relevant to the understanding of
his works. Once this essential distinction has been drawn, moreover, between the study of an author's motives in writing and the study of his
intentions in writing, my general argument can then be applied to show
16John Kemp, " The Work of Art and the Artist's Intentions ", The British Journal
of Aesthetics 4 (1964), p. 151.
7W. K. Wimsatt, Jr., The Verbal Icon (New York, 1958), p. 4.
18Monroe C. Beardsley, Aesthetics (New York, 1958), p. 17.
19Richard Kuhns, " Criticism and the Problem of Intentions ", The Journal of
Philosophy 57 (1960), p. 14.
20Kemp, loc. cit., p. 150.

16

QUENTIN

SKINNER

that (whatever may be true of motives) the recovery of an author's intentions


in writing is not merely relevant but essential in any attempt to explain
the special features of his works.
Consider for example the case (the rare case) of the critic who has in
fact asked, as the central question of a study about Milton, " What did
Milton mean in Paradise Lost ? ".21 It is evident that this is not (or not
merely) a question about what the words of the poem may mean. It is
also (or rather) posed as a question about what Milton, in writing such a
poem and so in using words with such meanings, may (in Grice's sense)
have non-naturally meant. But this is a question which asks in effect for
the (illocutionary) redescription of the complex of (linguistic) actions which
Milton, in writing the poem, may be said to have performed. I have already
shown, however, that to ask such a question is precisely to ask about the
agent's intentions in uttering his given utterance. And I have sought to
argue that to answer such a question will be to provide a valid and indispensable (not of course complete) guide to assessing why the agent said what
he said or wrote what he wrote. It follows, therefore, that if my general
argument is correct, it contains an implication of some practical importance :
that it must actually be an exegetical duty, and not a fallacy at all, for
critics and historians to concentrate on attempting to recover the intentions
of speakers or writers in the performanceof these complex types of (linguistic)
action.
I shall now consider and try to deal with two possible objections which
might be raised against the entire line of argument I have sought to present.
It might be objected (to revert to my basic example) that the bystander
could still be wholly unsatisfied with the claim that the policeman's action
has been explained, since he may in asking for the explanation already have
known that the policeman's utterance had the form and the intended force
of a warning. This suggests, however, a corroborationrather than an objection to my basic argument. It suggests that there may be two valid forms
of reply which may be given in answer to a simple request for the explanation
of such an action. One such reply clearly does consist of treating the question
as a request to be told why the policeman should have wanted (at all) to
utter his given utterance, granted that the utterance itself is correctly
assessed as having the intended force of a warning. The other consists,
however, of treating the question not as a request to be told why the policeman might have wantedto, but merely why he did, utter the given utterance.
The first type of answer provides an explanation by turning to the affective
states of the agent and seeking to discover what might have prompted the
policeman to want to warn the skater. It may well be, moreover, that the
desire (whatever it was) that functioned as the motive for acting should be
treated as a cause of the policeman's resulting action. My essential contention, however, is that whatever may be said about the concepts involved
21A. J. A. Waldock, Paradise Lost and its Critics (London, 1961), p. 1.

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

17

in this type of explanation, the other type may already have served to provide some explanation of the policeman's given action simply by recovering
the illocutionary redescription of the action itself. This objection, in short,
might be answered by claiming that there can be a case in which to ask in
this way for the explanation of the action may be to ask a fundamentally
ambiguous question.
The second objection, however, might be that even if such linguistic
actions are open to this mode of explanation, this again suggests only that
they must constitute a very special case, and should not be used to illustrate
any more general thesis about the explanation of action. But this objection
For the fact is that the illocutionary
would, I think, be misconceived.
features of linguistic actions, on which I have so far concentrated, seem by
no means to be confined to this class of actions. It can readily be shown
that a range of ordinary physical actions exhibit similar illocutionary features.
(There is no doubt, incidentally, that Austin would have sanctioned this
extension of his concept of " illocutionary " force. While his own main
concern was to contrast illocutions with locutions, he explicitly mentions
(p. 118) that certain illocutions are invariably performed non-verbally.)
As an illustration of this extension of my argument, consider the list of
possible explanations for a given action which Ayer draws up in conceding
that a given set of bodily movements can according to context represent a
variety of actions. Ayer takes the example (pp. 10-11) of someone drinking
a glass of wine, which can in different circumstances represent " (1) an act
of self-indulgence, (2) an expression of politeness, (3) a proof of alcoholism,
(4) a manifestation of loyalty, (5) a gesture of despair, (6) an attempt at
suicide, (7) the performance of a social rite, (8) a religious communication,
(9) an attempt to summon up one's courage, (10) an attempt to seduce or
corrupt another person, (11) the sealing of a bargain, (12) a display of professional expertise, (13) a piece of inadvertence, (14) an act of expiation,
(15) the response to a challenge ". Now it has sometimes been argued that
the possibility of such radical lack of correlation between physical movements and actions already rules out any possibility that such episodes can
be brought under causal forms of explanation. But while such disjunctions
may certainly suggest that the criteria for applying certain action concepts
may be in dispute, they scarcely seem to bear upon the question of whether
the corresponding episodes may be causally explained. And if my present
thesis is valid, this way of putting the point may in any case be beside the
point. The point-if the aim is to set limits to the application of causalityis rather that the list itself continues to leave unresolved a confusion between two types of situation : those in which the performance of the given
action does not, and those in which it does, bear some conventional (nonnatural) meaning or (illocutionary) force.
It is true that this response to the objection is rather hard to make
good in terms of Ayer's particularly elaborate and eccentric list. It is not

18

QUENTIN

SKINNER

clear that the explicans in either (3), (5) or (13) even yields an explanation
of anything that could be called a voluntary action. And it is not clear how
the explicans in either (6), (10) or (12) is even to be understood. It is hard
to see, that is, how any of these answers could be offered as possible explanations for the action simply of drinking a glass of wine. It must also be conceded that it does seem possible to insist more or less certainly with (1),(2),
(7) and (9) that it would not make much sense to ask what the agent in
performing his given action of drinking the glass of wine might have meant.
Or if such a question is asked it seems to resolve into a question about the
agent's motives. And Ayer is then able to mount a strong case for saying
that such motives function as the causes of actions. (I do not think it has
yet been made clear how this form of causality is supposed to operate. But
I do not wish or need for my present purposes to engage with that argument.)
This still seems to leave us, however, with (4), (8), (11), (14) and perhaps
(15). The explanation in all these cases, pace Ayer's apparent confusion
between motives and intentions, does seem to take the form of a redescription
which directs us not primarily to the agent's motives but rather to his
intentions in performing the given action of drinking the glass of wine. It
does seem possible, that is, at least in these cases, to arrive at a reason for
the agent's action by recovering the unique illocutionary redescription of
the action itself, in terms of which the agent's performance of it can be
shown non-naturally to mean something. It follows, even in these nonlinguistic types of case, that an understanding of what the agent may have
intended in performing his given action can still serve as a mode of explaining why that particular action was performed, even when the action
itself consists only of going through certain appropriate physical movements.
As a final caution, it is perhaps worth distinguishing the claim I have
now sought to argue from two further claims which are current in the
philosophy of action. First, it does not follow from the claim I have made
about the meaning of actions that what " distinguishes a mere bodily movement from an action " is " the meaning of that movement ".22 I have

rather sought to argue that it is in part because, in the required sense, it


often does not make sense to ask for the meaning of an action, that, when it
does, the recovery of this meaning can sometimes constitute a mode of
explanation of the action. And secondly, it does not follow from my emphasis on recovering the intentions of individual agents that the explanation
of action should essentially be treated as an ad hoc enterprise in contrast
with law-like causal explanations.23 If there were any interesting contrast
to be drawn at this point (I do not think there is) it might rather be argued
that it is the nomological mode of explanation which contains something
more like an ad hoc element. For it is undoubtedly possible, and often
22May Brodbeck, " Meaning and Action ", Philosophy of Science 30 (1963), p. 309.
23This contrast is urged in A. R. Louch, Explanation and Human Action (Oxford,
1966), pp. 3-5 and passim.

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

19

necessary, to provide causal explanations without being in a position to


state the relevant covering laws. But it is hard to see how the citation of
an agent's intentions in acting can be explanatory unless the agent's behaviour appears to be of a recognizable, classifiable character: not necessarily rule-governed, but necessarily conventional at least in the sense of
suggesting comparisons and possible inferences about the way in which
the agent might himself be able to characterize and account for why he
was doing what he was doing.
I am not concerned here with the further question of the determinants
of the distinction I have now sought to draw, within the class of voluntary
actions, between those which do and those which do not possess this element
of what might be called institutional significance. My concern so far has
only been to establish the significance of the claim that there is such a
distinction to be drawn. And if this claim has by now been sufficiently made
out, and the possible objections to it sufficiently met, my final concern will
now be to trace out two ways in which I believe my thesis has application
in the current debate about the explanation of action.
The first application arises in connection with certain current doctrines
about the relations between redescribing and explaining actions. It is sometimes claimed (as Ayer puts it, p. 16) that " merely to redescribe a phenomenon is in no way to account for it ". Against this type of claim, however,
I have now sought to argue that for at least one class of actions there can
be a unique form of redescription which, by recovering the illocutionary
force of the action, is thereby capable of serving as a mode of explanation.
It is also claimed that even if redescriptions can be explanatory, this can
only be because they enter into causal explanations. Thus it is said (for
example by Davidson, p. 692) that "we can't infer, from the fact that
giving reasons merely redescribes the action and that causes are separate
from effects, that therefore reasons are not causes ". It is even said that
the relation between a convention and an action must be regarded as causal
when the one is cited to explain the other. " The only way " (as Ayer for
example puts it, pp. 22-3) " in which the social context enters the reckoning " is as part of the agent's motivation. There is thus said to be nothing
in the claim that " actions need to be understood in terms of their social
contexts " from which it " follows that the performance of the action is not
causally explicable ". Against these types of claim, however, I have now
sought to argue that there can be a unique form of illocutionary redescription
which recovers what an agent in acting may non-naturally have meant, and
which is thus capable of giving an explanation by giving the agent's reason
for acting, but which does not consist of recovering anything that could be
regarded as a cause of which the action was an effect.
This first application of my thesis, moreover, can readily be corroborated
for the case of ordinary physical actions as well as linguistic actions. Consider, for example, item (4) on Ayer's list of possible explanations for the

20

QUENTIN

SKINNER

agent drinking the glass of wine : that it could be a manifestation of loyalty.


On the one hand, it may be that an understanding of the desire or other
affective state which may have prompted the agent to want to swear loyalty
is assimilable to an understanding of the cause of which his drinking the
glass of wine was the effect. On the other hand, such an understanding
seems quite separate from an understanding of the fact that the agent may
already have meant to swear loyalty (rather, say, than to seal a bargain)
simply in drinking the glass of wine, just as the policeman in my basic
example meant to warn the skater simply in uttering his given utterance.
And while the first of these forms of explanation may be, the second cannot
be, assimilable to a causal model. For while the citation of the agent's
desire or motive may plausibly be regarded as equivalent to citing a causal
condition necessary and sufficient to bring about the resulting action, the
citation of the agent's intention in acting is equivalent to citing the nonnatural meaning of the action itself, and is thus a matter of citing a feature
-not a condition-of the given action.
The second application of my thesis concerns the current doctrines about
the relations between motives and intentions in the explanation of actions.
If an illocutionary redescription can serve as a mode of explanation, but
merely by citing the agent's intention (not his motive) in performing his
given action, then it is arguable that the relations between motives, intentions and actions must have been mis-stated both by those philosophers
who have wished to assert and by those who have wished to deny that
intentions must function as causes of actions. Those who have sought to
insist on the " internal " or logical relations between all affective states and
actions have sometimes based their case, as for example Melden seems to
do, on a misdescription of the agent's motives for action. The citation of a
motive is treated as the means of " giving a fuller characterization of the
action ".24 Here the concept of a motive seems to be mistakenly assimilated
to the agent's intentions in acting. Those who have sought, on the other
hand, to insist on the contingent character of all such relationships have
sometimes based their case, as for example both Davidson and MacIntyre
seem to do, on a misdescription of the agent's intentions in action. Here
the concept of an intention in acting seems to be mistakenly assimilated to
the contingent antecedent conditions for the occurrence of the given action.
This conflation appears particularly clearly in Davidson's account, which
first cites the claim that "a person knows his own intentions in acting
infallibly ", and then argues that even if " your knowledge of your own
reasons for your actions is not generally inductive ", this cannot be held to

" show the knowledge is not causal ", since " you may be wrong about which
motive made you do it ".25 Both sides in the debate thus seem to be com24A. I. Melden, op. cit., p. 88.
25Davidson, loc. cit., p. 699. My italics. For similar assimilations of intentions to
the contingent antecedents of action, see MacIntyre in op. cit., pp. 224-5 ; B. N. Flem" On Intention
", The Philosophical Review 73 (1964), p. 320.
ing,

ON PERFORMING

AND

EXPLAINING

LINGUISTIC

ACTIONS

21

mitting the same mistake. The source of the confusion might perhaps be
characterized as a failure to recognize that motives in acting, but not intentions in acting, imply wants. The confusion itself seems to consist, on both
sides, of confusing the sense of intentionality which serves to characterize
actions and their illocutionary force with such affective states of the agent
as his motives, desires and generically the wants that prompt (and perhaps
cause) his actions. It certainly seems strange that the same concept, that
of intention, should be capable of being applied in these two quite different
ways. My sole concern, however, has been to try to consider the significance
rather than the undoubted curiosity of this fact.
It will be clear by now that my thesis occupies a middle ground which
has, I think, been generally overlooked in current discussions about the
explanation of action. Those philosophers who have (correctly, I believe)
emphasized the importance of intentions and conventions in the study of
voluntary human action often write as though it follows that the attempt
to apply causal models to such actions must be a confusion, even a "pernicious confusion "; that it must in any case be " wholly irrelevant "; and
that the whole vocabulary of causality ought accordingly to be " expunged "
from the discussion of voluntary action, on the grounds that " no nonteleological account is valid " in " the account of our behaviour implicit in
our ordinary language ".26 Conversely, those philosophers who have insisted
(again correctly, I believe) on the impossibility of such an exclusion often
write (as I have indicated in the case of Ayer, Davidson and others) as
though it follows that intentions and conventions must necessarily be treated
as causes of action. I have sought to argue that neither of these implications
follows, and that both are in fact mistaken. To accept that actions may
be causally explained, which I have not sought to deny, does not entail
accepting that every successful explanation of an action must be causal
in form; but to insist that there can be valid non-causal explanations of
actions, in the way I have sought to argue the case, does not entail saying
anything to conflict with an ultimate belief in a neuro-physiologically based
determinism.
Christ's College,Cambridge.

26For these four claims, see respectively Louch, op. cit., p. 238; A. I. Melden, op.
cit., p. 184; Raziel Abelson, " Because I Want To ", Mind 74 (1965), p. 541; Charles
Taylor, The Explanation of Behaviour (London, 1964), p. 54.