Kripke's Account of the Argument Against Private Language
Crispin Wright
The Journal of Philosophy, Vol. 81, No. 12. (Dec., 1984), pp. 759-778.
Stable URL:
http://links.jstor.org/sici?sici=0022-362X%28198412%2981%3A12%3C759%3AKAOTAA%3E2.0.CO%3B2-D
The Journal of Philosophy is currently published by Journal of Philosophy, Inc..
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. 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/journals/jphil.html.
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.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.
http://www.jstor.org
Tue Aug 21 17:44:29 2007
KRIPKE ON PRIVATE LANGUAGE 759
KRIPKE'S ACCOUNT OF T H E ARGUMENT AGAINST 

PRIVATE LANGUAGE* 

S
AUL Kripke's Wittgenstein o n Rules and Private ~ a n ~ u a ~ e ?
suggests interpreting Wittgenstein's argument against private
language as a direct corollary of the considerations about
rule following which immediately precede those passages in the
Inuestigations (s243 and following) on which more traditional attempts
to understand Wittgenstein's thought on privacy have
tended to concentrate. For a long time I thought Kripke's interpretation
of these matters more or less coincident with that at which I
had arrived independently and which I had presented in my Wittgenstein
o n the Foundations of Mathematics and elsewhere.' A careful
reading of Kripke's book has convinced me both that this is not
the case and that the dominant impression given in my book of the
relation between the private-language argument (PLA)and the rulefollowing
considerations (RFC) is misleading.
The leading suggestibn about the PLA in that book was that it is
to be viewed as part of the considerations about rule following: an
argument, essentially, that the sort of objectivity of meaning necessary
if we are to think of the truth values of unconsidered, uninvestigated
statements as determinate independently of any investigation
we may carry out, can find no refuge in the situation of a single
speaker and his idiolect. The RFC were then depicted as taking
the argument outwards, as it were-arguing, first, that, within the
sphere of communal practice, concepts and distinctions can be
given currency, on the basis of which a "thinner" notion of correctness
and incorrectness in linguistic usage can be rehabilitated
than that sanctioned by objective meaning; but, second, that, as far
as the propriety of objective meaning is concerned, the community
at large ultimately fares no better than the would-be private linguist.
With none of this, at least as a potentially fruitful framework
for the investigation of Wittgenstein's later philosophies of mind
and mathematics, do I now disagree. But it does seem to me now
that the treatment in my book could be usefully supplemented in at
least two respects.
"This paper amplifies part of the discussion of my "Kripke's Wittgenstein," presented
at the 7th International Wittgenstein Symposium in Kirchberg-am-Wechsel,
Austria, in 1982, but omitted from the associated published volume.
?oxford: Basil Blackwell, 1981. All page references are to this text unless otherwise
stated.
'London: Duckworth, 1980. See also C. LRich and 0 . Holtzman, eds., Wittgenstetn:
To Follow a Rule (London: Routledge 6.Keagan Paul, 1981),pp. 99-106; and
"Strict Finitism," Synthese, Lr, 2 (May 1982): 203-282, pp. 248-252.
0022-362X/84/8112/0759$02.00 @ 1984 The Journal of Philosophy, Inc.
760 THE JOURNAL OF PHILOSOPHY
First, I think the involvement of "antirealist" premises in the arguments
against objective meaning was there overemphasized; it
seems to me that a more sensitive, sparing, and concept-specific use
of such premises may be possible without compromise of the
power of the argument. Second, more stress is wanted that Wittgenstein
has a differential claim about private language: that the
would-be private linguist and the community are not, in the end,
in the same predicament. I shall not, in this paper, attempt to enlarge
upon either of these claim^.^ The notable point is that an
analogue of each is a prominent feature of Kripke's interpretation:
the skeptical argument needs no anti-realist (verificationist) assistance;
and the bearing of the skeptical solution on private language
admits of no community-wide generalization.
Although I do not think that Kripke has Wittgenstein right, my
subject, except in the last section of the paper, is not the historical
Wittgenstein but Kripke's Wittgenstein. I shall argue that, even if
the main argument-the skeptical argument-which Kripke finds
in Wittgenstein, is sustained, there is strong prima facie reason to
doubt whether the accommodation with it-skeptical solutionwhich
Kripke represents Wittgenstein as commending can really be
lived with; whether, indeed, that accommodation is so much as coherent.
And I shall canvass ways, unconsidered (or only very cursorily
considered) by Kripke, for resisting the skeptical argument.
The upshot will be that the RFC, as interpreted by Kripke, are
flawed by a lacuna, and that, even if the lacuna were filled, the
PLA could nevertheless not emerge in the manner that Kripke
describes.
Because the gist of my remarks about Kripke's book is going to
be largely critical, it is perhaps worth emphasizing my admiration
of it. Whatever its relation to Wittgenstein's actual thought, and
whether or not ultimately cogent, Kripke's dialectic is tremendously
exciting. It will surely provide a great spur to improving our
understanding of Wittgenstein's philosophy.
I THE SKEPTICAL ARGUMENT
Fundamental to Hume's moral philosophy, as to his views about
causation, is a distinction between statements that are apt to express
real matters of fact and certain sentences that, although possessed
of standard features of the syntax of genuine statements-in
he^ are enlarged on in, respectively, "Rule-following and Constructivism," in
C. Travis and B. Gelder, eds., Inference and Understandrng, projected, and "Does
Wittgenstein Have a Cogent Argument against Private Language?" in J. McDowell
and P. Pettit, eds. Context, Content and Thought, forthcoming (New York: Oxford,
1985).
KRIPKE ON PRIVATE LANGUAGE 761
particular, the capacity to serve as arguments for various types of
statement-forming operator-are actually used not to state facts but
rather to project various aspects of speakers' attitudes and affective
responses. Moral discourse, and talk of causation, belong, for
Hume, in the latter category. Moral judgments, so viewed, do not
express our cognition of moral facts by which various moral sentiments
in us are generated; rather they serve to project those moral
sentiments upon the world. Likewise, those statements which the
non-Humean takes to aver the existence of causal relations, from
which certain observed regularities flow, serve for Hume to project
an attitude that we take up toward those observed regularities.
It is familiar that this sort of distinction is pivotal to a whole
class of important philosophical disputes. Realism not merely
about ethics and causation, but also about aesthetics, theoretical
science, pure mathematics, logical necessity, and Lockean secondary
qualities may each be opposed by appropriate versions of Humean
noncognitive "projectivism."
In this light, Kripke's Wittgenstein may be seen as first, by the
skeptical argument, confounding the ordinary idea that our talk of
meaning and understanding and cognate concepts has a genuinely
factual subject matter, and then, via the skeptical solution, recommending
an alternative projective view of its content. It is worth
emphasis that the skeptical solution is independent of the skeptical
argument: strictly, the option is open of simply accepting the latter
as demonstrating the vacuity of all our talk of meaning, etc., with
no prospect of its rehabilitation. (Similarly, a sympathizer might
regard Hume as having demonstrated that we should simply drop
all talk of causation.) Accordingly, it might seem as though one
way of resisting the PLA, as Kripke interprets it, would simply be
to take the skeptical medicine straight, eschewing the compensating
sweetmeat of the skeptical solution afterwards. I shall return to
that thought.
There are a variety of ways in which it might be argued that a
region of discourse apparently apt for the stating of facts does not
really perform that role. One way, the Humean strategy, would be
to argue that from within the framework of a certain preferred epistemology,
no reputable conception can be attained of the putative
species of fact in question. Another general strategy would be to let
the argument flow from a topic-neutral account of the ways in
which the distinction between fact-stating and non-fact-stating declarative
sentences comes out in their respective modes of employment
in the language. But the strategy of argument which Kripke
finds in Wittgenstein is different from both of these. Roughly, the
762 THE JOURNAL OF PHILOSOPHY
conclusion that there are no facts of a disputed species is to follow
from an argument to the effect that, even if we imagine our abilities
idealized to the point where, if there were such facts to be
known, we should certainly be in possession of them, we stzll
would not be in a position to justify any particular claim about
their character. So we first, as it were, plot the area in which the
facts in question would have to be found if they existed and then
imagine a suitable idealization, with respect to that area, of our
knowledge-acquiring powers; if it then transpires that any particular
claim about those facts still proves resistant to all justification,
there is no alternative to concluding that the "facts" never existed
in the first place.
The initial target class of putative facts comprises those which
you might try to express by claims of the form "By E, I formerly
meant so-and-so." The relevant idealization will involve your total
recall of all facts about your previous behavior and previous mental
history, it being assumed that facts about your former meanings
must be located in one of those two areas if they are located anywhere.
The argument will then be that, even in terms of the idealization, no
such claim is justifiable. It follows that your previous life in its entirety
is empty of such facts, and hence that there are none (cf.21 and
39).
I have sometimes encountered in discussion the complaint that,
whatever independent force Kripke's development of the argument
may have, its use of skepticism betrays its claim to represent Wittgenstein's
actual thought. In one way, this misconstrues the skeptical
argument; in another way, however, it may have a point. The
misunderstanding consists in a failure to see that Kripke's skeptic
is a mere device, annexed to the demonstration of a projectivist thesis
which might well be supported in other ways. (It is notable that
the historical Wittgenstein, though undoubtedly hostile to classical
forms of skepticism, unmistakably displays projectivist leanings in
ceitain of his remarks, e.g., on first-person ascriptions of sensation
and when he compares mathematical statements to rules.) Classical
forms of skepticism purport to discover inadequacies in our actual
cognitive powers: the skeptic about induction, or other minds, or
memory, holds that the best we can do, in attempting to arrive at
justified opinions concerning statements in the relevant classes,
always falls short of anything that ought really to be counted as
justification. The skepticwhom Kripke finds in Wittgenstein, in contrast,
is concerned to teach us something about the range of items that
exzst to be known. That said, there will still be a point to the complaint
if it turns out that, despite these differences, the technzques
763KRIPKE ON PRIVATE LANGUAGE
utilized by Kripke's skeptic are importantly similar to those which
feature in traditional skeptical arguments. Wittgenstein undoubtedly
thought those arguments mistaken; it is hardly likely that he would
have allowed himself to succumb to an argument which, even if tending
toward a conclusion congenial to him, needed to rely upon epistemological
principles that, if granted, would enormously strengthen
the traditional skeptic's case. We shall consider the matter in due
course.
Suppose it granted that there are indeed no facts that we can express
by statements of the form, "By E, I formerly meant so-andso."
How exactly do destructive consequences follow about the notions
of meaning and understanding in general? Kripke himself is
fairly brief on the point (13), but it is not difficult to see. Remember
that the argument will have involved an extensive idealization
of your knowledge of your previous behavior and mental history:
you will have been granted perfect recall of all such facts. If it
turned out that you still could not justify any preferred claim of the
form "By E, I formerly meant so-and-so," then how can you be better
placed to justify a claim of the form "By E, I presently mean soand-so"?
For anything true of your mental life and behavior up to
and including the present will be known to you tomorrow, in accordance
with the idealization. And the argument will have shown
that tomorrow you won't be able to justify any claim of the form
"By E, I yesterday meant so-and-so." Hence you cannot be in a position
to justify the present-tense counterpart of that claim today.
The idealization also entails that nobody else is better placed than
you to justify any such claim. It follows that nobody can justify
any claim about what they, or anybody else, formerly meant or
means. Hence, in the presence of the idealization, there can be no
facts about what anybody means by any expression. And it is impossible
to see how, consistently with that admission, there might
yet be facts about what expressions, as it were impersonally, mean.
The strategy of the skeptical argument thus appears sound and ingenious.
Everything depends upon the details of its execution.
Simplifying the details somewhat, the execution runs essentially
like this. Suppose you claim, on December 21 1984, that by 'green'
you meant, on December 20 1984, green. The skeptic challenges
you to justify your claim. You are idealized to have perfect recall of
all your previous linguistic and nonlinguistic behavior, together
with your entire mental life-the whole pageant of your thoughts,
sensations, imaginings, dreams, moods, etc. (At this point in the
argument there is, of course, no doubt entertained as yet about the
supposition that you presently know what you mean by 'green1-
764 THE JOURNAL OF PHILOSOPHY
the skeptic gets you to stand on the rug before he pulls it away; cf.
1112.) Now no doubt you can cite a lot of behavior that is broadly
consistent with your preferred account of your former understanding
of 'green'. The skeptic will point out, however, that there are
no end of alternative interpretations of your former meaning, all of
which rationalize your behavior equally well. Perhaps, for example,
by 'green' you formerly mean grue1984; where an object is grue,
at time k just in case k is earlier than the inception of January 1 in
year t and the object is green, or k is some later time and the object
is blue. Seemingly there are indefinitely many grue-interpretations
that can be used to make sense of your previous applications of
'green', all of which are incompatible both with the supposition
that by 'green' you formerly meant green and with each other.
It might be objected that this is to consider only one kind of use
of 'green', viz., simple ascriptions and withholdings, and that account
will need to be taken of all sorts of other more sophisticated
uses, including embeddings in descriptions of your own and others'
propositional attitudes, which you previously will likely have
made. But, of course, the point of the example is that 'green' will
be assigned the same extension, u p to and including the time of
your dialogue with the skeptic, under indefinitely many grue-interpretations;
so the skeptic should have no difficulty in rationalizing
your previous uses both in extensional contexts and in all attributions
of propositional attitudes whose possession, by yourself
and others, can be explained in terms of the extension of the property
of being green.' And even if grue-interpretations don't work out
in general, the decisive consideration is surely that your previous
behavior with 'green' is finite; hence, it must be possible, it appears,
with sufficient ingenuity, to come u p with some interpretation
of your previous understanding of 'green' which will be as
unwelcome to you as the grue-interpretations. Finite behavior cannot
constrain its interpretation to within uniqueness.
The arena of battle now shifts to the mental. Perhaps considerations
uniquely determining your previous understanding of 'green'
can be recovered from your previous thoughts, imaginings, etc. It is
evident, however, that, if the search is to succeed, the relevant mental
items must have a certain generality. It is no good remembering
imagining certain green things, or green after-images that you may
have experienced, or thoughts about what you would have said if
asked to describe the color of that liqueur we had on such-and-such
an occasion. For the constraints imposed by introducing such considerations
cannot be stronger than if the images had been public
objects or if the imaginary and hypothetical situations had actually
765KRIPKE ON PRIVATE LANGUAGE
taken place; and the effect of those transformations would merely
be finitely to enlarge an inadequately finite pool of actual data.
You have to come up with some mental episode that somehow has
sufficient content to exclude all the unwanted interpretations of
your former understanding of 'green', including all the grue-interpretations,
at one go.
The only candidate, it appears, is some sort of general thought:
you need to have entertained a thought that has something to say
about each of the situations in which the difference between the
true interpretation of your former understanding of 'green' and
each of the successive grue-impostors successively comes to light.
On the face of it, though, it is not far-fetched to suppose that you
might very well have entertained such a thought. What if you remember
having thought, say, "By 'green' I certainly don't mean
any concept which, at some particular time, will continue to apply
to an object only if that object changes color at that time." Does
not that at least force the skeptic to work a bit harder at the concoction
of unwelcome interpretations?
It wouldn't be all that satisfactory if this were the best you could
do. After all, it is pretty much fortuitous whether any such thought
ever occurred to you; and your knowledge about your former understanding
of 'green' will not seem to you contingent on such an
occurrence. What you will want to be able to say is that you know
what you formerly meant by 'green' whether or not you happen to
have had a convenient thought that can be used to scotch a particular
line of unwelcome interpretation. But the skeptic, in any case,
has a stronger, indeed a seemingly decisive reply. His challenge,
after all, was general. No special interest attaches to the justifiability
of claims about your former understanding of color predicates:
the question was, can a n y claim of the form "By E, I formerly
meant so-and-so" be justified? Clearly the challenge is not met if,
in the attempt to justify one such claim, you presuppose your right
to be sure of another. But just such a presupposition is made by the
attempted play with the general thought above. If, for example, by
'color' you had previously meant schmolor (20),where the concept
of schmolor stands to all grue-type concepts exactly as color stands
to blue, green, etc., then your having entertained that general
thought is quite consistent with the skeptic's being correct in interpreting
you as having meant gruels84 by 'green'.
The point is perfectly general. Thoughts you may have had
about how, quite generally, you should be prepared to use an expression
will suffice to meet the skeptic's challenge only if you presuppose
their proper interpretation. But that is just to presuppose
766 THE JOURNAL OF PHILOSOPHY
that the skeptic's challenge can be met with respect to the expressions
that figure in those thoughts. Yet no category of mental item
can be appropriate to the challenge except a general thought; only
such a thought can have enough to say, can cover the indefinitely
many potential situations in which you should wish to regard
some determinate use of 'green' as mandated by the understanding
that you believe you have long possessed of that expression.
It, therefore, appears that the only ploy that has any chance of
accrediting your understanding of 'green' with an appropriately
general normative role (11, 23, 24) totally fails to meet the skeptic's
challenge. And now "it seems the whole idea of meaning vanishes
into thin air" (22).
11. THE SKEPTICAL SOLI'TION
Suppose that the skeptical argument is sound. Could we simply accept
its conclusion and abandon all talk of meaning and understanding
as founded upon error? Or must we seek some sort of rehabilitation
of the concept of meaning, a skeptical solution?
Kripke himself writes:
. . . I choose to be so bold as to say: Wittgenstein holds, with the skeptic,
that there is no fact as to whether I mean [green or gruel]. But if
this is to be conceded to the skeptic, is this not the end of the matter?
What can be said on behalf of our ordinary attributions of meaningful
language to ourselves and to others? Has not the incredible and selfdefeating
conclusion, that all language is meaningless, already been
drawn? (70/1)
There is, however, a certain awkwardness here. Suppose someone
runs a similar skeptical argument about moral obligation, concluding
that statements about what people morally ought, or
ought not, to do, lack a factual subject matter. It would be, to say
the least, an infelicitous expression of this result to say, "So undertaking,
and refraining from, any particular projected course of action
are always both morally permissible." For the conclusion of
the argument would apply equally to judgments of moral permissibility:
to claim that a course of action is morally permissible is
just to say that it is not the case that it ought to be refrained from.
It is natural to wonder, correspondingly, whether the conclusion of
Kripke's skeptic is indeed "incredible and self-defeating" only if
the notion of meaninglessness which Kripke uses in its formulation
presupposes the notion of meaning as moral permissibility
presupposes the notion of obligation. If that is so, the right conclusion
is surely that such is not the way to formulate the conclusion
of the skeptical argument. Once it is better formulated, such un-
KRIPKE ON PRIVATE LANGUAGE 767
happy claims as that all language is meaningless, that nobody ever
succeeds in understanding anybody else, etc., will presumably not
be entailed.
If somebody wishes to reject the suitability of a certain class of
concepts to figure in statements apt to be genuinely true or false,
this rejection cannot coherently take the form, it appears, of any
kind of denial of statements in which those concepts figure. What
then is the proper way of formulating the conclusions of Kripke's
skeptic? One influential view of the concepts of meaning and understanding,
associated chiefly with the writings of W. V. Quine, is
that their paramount function for us is as theoretical terms in a
deeply entrenched but philosophically suspect scheme of explanation
of human linguistic behavior and of nonlinguistic but language-related
patterns of social activity. If we think of this scheme
as issuing in a large class of only semi-articulated theories about
particular individuals and groups of individuals, then one way of
expressing the conclusion of the skeptical argument is that scientific
realism about these theories is not an option; there are no
facts, describable only by recourse to the concepts of meaning and
understanding, which such theories might succeed in codifying. An
immediate consequence of this perspective is that two quite different
lines of response to the skeptical argument are apparently
open. One is a kind of instrumentalism: a view which tries to retain
the propriety of theorizing of the sort in question while granting
its nonfactual status. The skeptical solution attempts just this.
The other response is to regard theorizing of this sort as discredited,
and to seek better approaches involving quite alternative systems
of concepts. That, in general, is Quine's own response to the
difficultiesthat he finds in meaning and other intentional notions.
If it admits satisfactory development, then the skeptical solutionand
with it Kripke's reading of the PLA-would seem to be de trop.
This picture of the role of the concepts of meaning and understanding
in our ordinary thinking is, however, an oversimplification.
It ignores the larger class of self-directed statements concerning
meaning and understanding which we make-the class that
Wittgenstein himself gives special attention to-and, still more
importantly, it makes nothing of various platitudes that articulate
our conception of the connections between meaning and truth.
One such platitude is that the truth value of a statement depends
only upon its meaning and the state of the world in relevant respects.
Equivalently:
An utterance of S expresses a truth in a particular context if and only
if what, in that context, S says is so, is so.
768 THE JOURNALOF PHILOSOPHY
The obvious corollary is that, if we take the view that the skeptical
argument discredits all talk of meaning, understanding, and cognate
concepts-like the concept of what a sentence is used to sayit
is not clear how much purchase we can retain on our ordinary
notion of a statemenťs being true. A proponent of the Quinean
view has the choice either to abandon the notion of truth altogether
or to reconstruct it in a fashion that liberates it from conceptual
ties with the discredited notion of meaning. The former
course, however, is hardly an option unless we are prepared to
abandon the idea that it is ever the case that language has a factstating
function. (And if that were our view, why see the conclusion
of the skeptical argument as calling into question the propriety
of talk involving the concepts of meaning and understanding?)
The reconstructive project, on the other hand, looks to be utterly
daunting. (Indeed it is doubtful whether, in the present context, it
is coherent to suppose that there can be such a project; for whatever
reconstruction of truth, free of all play with meaning and cognate
notions, were proposed, it is not clear why an analogue of the skeptical
argument would not be available to rob any particular assignment
of truth conditions to a sentence of all possible behavioral
or psychological corroboration.)
So the strategy incorporated in the skeptical solution may seem
more attractive. It is in any case more comfortable to think of any
errors involved in our talk of meaning, or our moral language, as
philosophical: as belonging to our picture of what is going on in
those areas of linguistic practice, rather than as undermining the
practices themselves. Thus, Kripke suggests (73 ff) that statements
involving the notion of meaning have no truth conditions, properly
so described, but only conditions of justified or warranted use:
All that is needed to legitimize assertions that someone means something
is that there be roughly specified circumstances under which
they are legitimately assertable, and that the game of asserting them
under such conditions has a role in our lives (77/8).
Kripke's interpretation of the PLA now follows elegantly from this
reorientation. Without attempting to do justice to the detail of his
exposition (81 ff, summarized 107/8), we find that the most natural
account of the justification conditions of statement forms like:
(i) Jones means addition by '+'
and
(ii) If Jones means addition by '+', then he will answer '125' when
asked. "What is 47 +78?"
769KRIPKE ON PRIVATE LANGUAGE
involves essential reference to a community of practitioners with
the symbols they mention. Very roughly, (i)will be considered justified
if Jones performs satisfactorily often enough with '+' and
marks his acceptance into the community of '+' users-those
whose uses of '+' can generally be depended upon. And (ii) expresses
a test for membership in that community, ratified by the responses
of those already accredited with membership. Accordingly,
such statement forms simply have no legitimate application to
symbols whose use is essentially "private" and which cannot, in
the nature of the case, competently be taken up by a community.
So the concepts of meaning and understanding have no proper
place in the description of an apparent linguistic practice of an individual,
if that practice is one in which others could not competently
share.
The elegance of Kripke's interpretation does not, however, long
conceal its difficulties. One immediate difficulty is presented by the
meaning-truth platitude. If the truth value of S is determined by its
meaning and the state of the world in relevant respects, then nonfactuality
in one of the determinants can be expected to induce nonfactuality
in the outcome. (A rough parallel: If among the determinants
of whether it is worth while going to see a certain exhibition
is how well presented the leading exhibits are, then, if questions of
good presentation are not considered to be entirely factual, neither
is the matter of whether it is worth while going to see the exhibition.)
A projectivist view of meaning is thus, it appears, going to
enjoin a projectivist view of what is for a statement to be true.
Whence, unless it is, mysteriously, possible for a projective statement
to sustain a biconditional with a genuinely factual statement,
the disquotational schema "P1'is true if and only if P' will churn
out the result that all statements are projective.
Kripke's own remarks are confusing in this regard. He quotes
with approval (73) Michael Dummetťs suggestion that the central
contrast between the picture of language and meaning proposed in
the Tractatus and that of the Investigations resides in a shift from a
conception of statement-meaning as truth-conditional to the view
that the meaning of each statement is fixed by its association with
conditions of justified assertion. But Dummett, at least as I read
him, never intended that reorientation to involve a total rejection
of the category of fact-stating discourse. It could not be so intended
with any plausibility, since, as we have noted, the historical Wittgenstein
thought that we are apt to be misled by the form of our
discourse in certain selected areas into thinking that its role has to
be that of stating facts. He could hardly have considered that we
770 THE JOURNAL OF PHILOSOPHY
were likely to be so misled unless he thought that form of discourse
to be very often associated with the activity of fact-stating. In any
case, whatever intention Dummett, or Wittgenstein, may have had,
it is doubtful that it is coherent to suppose that projectivist views
could be appropriate quite globally. For, however exactly the distinction
be drawn between fact-stating and non-fact-stating discourse,
the projectivist will presumably want it to come by way of a
discouery that certain statements fail to quality for the former class;
a statement of the conclusion of the skeptical argument, for instance,
is not itself to be projective. But can Kripke's exposition
make space for this admission? According to Kripke, what is distinctive
of fact-stating is the possession by one's statements of "real
truth conditions" (whatever that may mean). And how can the
judgment, "S has (real)truth conditions," be genuinely factual ifin
accordance with the platitude and the considerations of a moment
ago-"S is true" is not?
Another way of seeing that the situation cannot really be satisfactory
is to inquire what status, once the skeptical argument is accepted,
is supposed to be possessed by the sort of account adumbrated
by Kripke of the assertion conditions of statements about
meaning and understanding. Could it yesterday have been true of a
single individual that he associated with the sentence "Jones means
addition by '+' " the sort of assertion conditions Kripke sketches?
Well, if so, that truth did not consist in any aspect of his finite use
of that sentence or of its constituents; and, just as before, it would
seem that his previous thoughts about that sentence and its use will
suffice to constrain to within uniqueness the proper interpretation
of the assertion conditions he associated with it only if he is
granted correct recall of the content of those thoughts-exactly
what the skeptical argument does not grant. But would not any
truths concerning the assertion conditions previously associated by
somebody with a particular sentence have to be constituted by aspects
of his erstwhile behavior and mental life? So the case appears
no weaker than in the skeptical argument proper for the conclusion
that there are no such truths; whence, following the same routine,
it speedily follows that there are no truths about the assertion
conditions that any of us presently associates with a particular
sentence, nor, a fortiori, any truths about a communal association.
It follows that the premises, requisite for Kripke's version of the
PLA, about the community-oriented character of the assertion conditions
of statements concerning meaning and understanding are
not genuinely factual, and the same must presumably be said of the
conclusion, that the concepts of meaning and understanding have
no proper application to a private linguist.
KRIPKE O N PRIVATE LANGUAGE 771
The skeptical solution seems to me, therefore, to be a failure.
More: to sustain the skeptical argument is to uncage a tiger whose
depredations there is then no hope of containing.
I11 RESISTING THE SKEPTICAL ARGUMENT
Kripke himself considers two possible sources of error in the argument.
The first is the assumption that facts about my former understanding
of E must be constituted by aspects of my former behavior
and mental life. Is not a more plausible candidate, a certain
former dzsposztzon, the disposition to use E in certain sorts of way?
Against this suggestion Kripke brings (26-37) two prima facie very
telling sets of considerations to bear. First, the relevant sorts of disposition
are, with respect to any particular expression, presumabiy
finite, since all my capacities are finite; whereas, intuitively, we
want the meaning of E to contribute toward the determination of
its correct use in literally no end of potential cases. Second, meanings
are, whereas dispositions are not, normative: I may, in certain
circumstances, be disposed to use an expression in a way which is
out of accord with my understanding of it and which, therefore,
constitutes wrongful use of that expression; whereas I can scarcely
be said to have a disposition to use an expression in a way out of
accord with the way in which I am disposed to use it. Now there is,
no doubt, scope for discussion about how decisive these two rejoinders
are.3In particular, it need not be contradictory to suppose
that someone may be disposed to act in a way in which he is not
disposed to act-provided his dispositions are appropriately stratifzed.
So much, at any rate, is certainly part of our ordinary concept
of a disposition: almost all the dispositional properties about
which we ordinarily speak are such that their display is conditional
on the absence of certain interfering factors, and there is no contradiction
in the idea that such interference might be widespread and
even usual. The matter is obviously one of some subtlety. Here I
car, do little more than record my own view that Kripke is ultimately
right, at least as far as our intuitive conceptions of meaning
and understanding are concerned. Understanding an expression is,
intuitively, more like an ability than a d i ~ ~ o s i t i o n . ~Roughly, it is
the (fallible) ability to suit one's employment of the expression to
certain constraints. Even at the most fundamental level, then, and
when nothing interferes with the exercise of a disposition, there
ought to be a distinction between what somebody's understanding
'See, e.g., Simon Blackburn, "The Individual Strikes Back", Synthese, Lvrrr, 3
(March 1984):281-301, and Graeme Forbes, "Skepticism and Semantic Knowledge",
Proceedings of the Aristotelian Society, ~xxxrv(1983/4): 221-237.
Cf. G. P. Baker and P. M. S. Hacker, Wittgenstein: Meaning and Understanding,
vol. r of Essays on the Philosophical lnuestigations (Oxford:Blackwell, 1983), ch. xvr.
772 THE JOURNAL OF PHILOSOPHY
of E requires of him and the use of that expression which he actually
makes; it is just that, if nothing interferes with the exercise of
the disposition, the use he makes will be the use required of him.
You could put the point by saying that, intuitively, understanding
generates rule-governed behavior; to suppose that it is at some fundamental
level simply a matter of a disposition is to ignore the distinction
between suiting one's behavior to a rule and merely behaving
in such a way that, when the rule is construed as a descriptive
hypothesis, it fits what one does.
A second response to the skeptical argument which Kripke discusses
(41-50) is the idea that meaning green by 'green' "denotes an
irreducible experience, with its own special quale, known directly
to each of us by introspection." If there were such an experience
"as unique and irreducible as that of seeing yellow or feeling a
headache," then-in the presence of the relevant idealizations-it
could simply be recalled in response to the skeptic's challenge and
that would be that. Kripke's response to this proposal, drawing extensively
on themes explicit in the Investigations, is surely decisive.
Quite apart from the introspective implausibility of the suggestion,
it is impossible to see how such an experience could have the content
that understanding is conceived as having, could have, as it
were, something to say about the correct use of E in indefinitely
many situations. There might, indeed, be a distinctive experience
associated with meaning so-and-so by E; but then, in order for recall
of the experience to meet the skeptic's challenge, it would be
necessary additionally to recall the association-and that would
presuppose recall of one's former understanding of E, the possibility
of which is exactly what is at issue.
There are, however, a number of other ways in which the skeptic's
routine, seductive though Kripke's presentation makes it seem,
is open to serious question. One concerns the play the skeptic
makes with the finitude of previous linguistic behavior. There is
no question, of course, but that the relevant behavior is finite and
that it is thereby debarred from supplying a conclusive ground for
affirmingthat your former understanding of E was, indeed, so-andso.
But to suppose that it follows that there is no rational basis for
preference among indefinitely many competing hypotheses, all of
which are consistent with your previous linguistic behavior, is tantamount
to the supposition that Goodman's "new riddle of induction"
admits of no solution. This point does not, of course, depend
on the fact that we actually used a Goodman-type example in the
development of the skeptical argument. Rather, Goodman's riddle
is exactly the challenge to explain in what, if any, sense it is ra-
773KRIPKE ON PRIVATE LANGUAGE
tional to prefer, on the basis of finite evidence, the sorts of general
hypotheses which we invariably do prefer to any of the other indefinitely
many alternatives whose formulation he illustrates. Of
course, this assimilation is not in itself a satisfactory rejoinder to
Kripke's skeptic. But it does at least show him u p for a fairly familiar
animal. knd it teaches us that we ought notto regard the skeptical
argument as, so to speak, establishing a theorem unless we
think it right to despair of a solution to Goodman's riddle.'
What is unsatisfactory about the suggestion is that it gets the intuitive
epistemology of understanding wrong. Recognition that a
certain use of an expression fits one's former (and current) understanding
of it would not, it seems, except in the most extraordinary
circumstances, have to proceed by inference to the best semantic
explanation of one's previous uses of that expression. The kind of
fact-if, against the skeptical argument, there can indeed be such a
fact-which having formerly had a particular understanding of an
expression is, is misrepresented by this response.
There is, however, a further response, focusing on the second
stage of the skeptical argument, at the point where it is argued
that, no matter how rich a battery of explicit thoughts you may
'Afficionados of Kripke's text might feel that he, in effect, answers this point (38).
"Let no one-under thr influence of too much philosophy of scienre-suggest that
thr hypothesis that I meant plus is to be preferred as the simplest hypothesis. I will
not herr argue that simplirity is relativ? or that it is hard to define, or that a Martian
might find thr quus function simpler than the plus function. Surh replies may
have considerablr merit, but the real trouble with thr appeal to simplicity is morr
basic. Such an appral must br based on a misundrrstanding of the skrptiral probIrm,
or of thr role of simplicity ronsiderations, or both. Recall of the skeptical problem
was not merely epistemic. T h e skeptic argues that there is no fact [my italics] as
to what I meant, whether plus or quus. Now simplicity considerations can help us
decide betwren competing hypotheses, but thry obviously can never tell us what the
competing hyopthrsrs are. If we do not understand what two hypothrsrs state, what
does it mean to s,iy that one is 'more probable' brcause it is 'simpler'? If the two
comprting hypothrses arr not grnuinr hypothrsrs, not assertions of grnuinr mattrrs
of fact, no 'simplirity' considerations will makr them so."
I do not wish to suggrst that canons of simplicity provide an appropriate responsr
to Goodman's riddle. However, liripkr's point is grnrral. Whatever criterion
of prrfrrability among comprting hypotheses we come u p with, its application can
br appropriate only if we do genuinely have rompeting hypotheses, only if there is
some "fart of the matter" about which we are trying to arrive at a rational view.
Therefore-or so Kripke's thought presumably runs-we beg the question against
the skeptic in appealing to any such criteria at this stage. But this surely gets everything
back to front. It is only after the skeptiral argument has come to its conclusion
that the skeptic is entitled to the supposition that there is indeed no such fact of
the matter. In the rourse of the argument, he rannot assume as much without begging
the question. At the stage at which we might appeal to the sort of refined methodology
which could be used to answer Goodman's riddle, there simply is not yet
any basis for thinking that talk of meaning and understanding is not factual. For
what is, I think, a better response to the point, see the sequel in the text.
774 THE JOURNAL OF PHILOSOPHY
formerly have entertained concerning your understanding of E,
these thoughts will not turn the trick. Kripke's skeptic discounted
the attempt to bring your previous general thoughts against unwelcome
interpretations of your previous use of E, on the grounds
that you thereby presuppose knowledge of the proper interpretation
of those thoughts-which is, in detail, knowledge of the very
putative species currently under suspicion. This can seem reasonable.
On inspection, however, it cannot always be possible to justify
a presumed genre of knowledge "from without" in the way the
skeptic is here demanding. At any rate, it is obvious enough that, if
we were to allow the propriety quite generally of this skeptical
move, the results would be calamitous. Imagine, for example, a
skeptic who questions a claim about my former perceptions, say,
"Yesterday, I saw it raining." And suppose the ground rules are as
for the dialogue with Kripke's skeptic; that is, I am to be permitted
to adduce any relevant fact so long as I do not thereby presuppose
that there is such a thing as knowledge of what I formerly perceived-since
it is of belief in the very existence of the genre of
knowledge that the skeptic is demanding justification. So I cannot
simply claim to remember what I perceived: my ammunition will
be restricted to my present seeming-memories, the presently available
testimony of others, presently accessible putative traces, like
damp ground, etc. and meteorological office and newspaper records.
It ought to be a straightforward, if tedious, exercise for the
skeptic to accommodate all that without granting me the truth of
my claim about my perception of yesterday's weather. So I can
know "all relevant facts" without knowing anything about what I
formerly perceived. So there is no fact of the matter about what I
formerly perceived. So, since the arguments will work just as well
in the future when now is "then," there is no fact of the matter
about what I presently perceive. So, since the argument applies to
all of us, there is no such thing as perceptual knowledge. "There's
glory for you!"
The trouble, evidently, is the assumption that knowledge of a
former perception has to be inferential, that the ultimate grounds
for such knowledge must reside in knowledge of a different sort.
That is true only if knowledge of what I am presently perceiving is
inferential; otherwise, the skeptic may satisfactorily be answered
simply by recalling what one formerly perceived. So, too, Kripke's
skeptic persuades his victim to search for recalled facts from which
the character of his former understanding of E may be derived. And
that is fair play only if knowledge of a #resent meaning has to be
inferential; otherwise the skeptic is satisfactorily answered simply
by recalling what one formerly meant.
775KRIPKE O N PRIVATE LANGUAGE
The claim, then, is that the methodology of the skeptical argument
is appropriate, if ever, only to cases where it is right to view
the putative species of knowledge in question as essentially inferential.
And no ground for that supposition in the present case has
so far been produced. But if it is to be possible simply to recall the
character of former meanings, can the requisite presupposition,
that knowledge of present meanings may be noninferential, really
be made good? Kripke, in effect, confronts this suggestion when he
considers the possibility (50/1) that meaning so-and-so by E might
simply be an irreducible, sui generis state, a state "not to be assimilated
to sensations or headaches or any 'qualitative' states, nor to be
assimilated to dispositions, but a state of a unique kind of its
own." His reply, only very briefly developed (52/3), is that it is utterly
mysterious how such a state could have the requisite properties,
in particular how, although a finite state realized in a finite
mind, it could nevertheless have the potential infinity of content
that the normativity of meaning requires. How can there be a state
which each of us knows about, in his own case at least, noninferentially
and yet which is infinitely fecund, possessing specific directive
content for no end of distinct situations?
This may be a good question. But Kripke's discussion contrives
to leave the impression that it is rhetorical, that we have not the
slightest idea what such a state might be. Whereas a little reflection
shows that both these features-noninferentiality and indefinite
"fecundity"-are simply characteristic of our standard intuitive notion
of intention. Normally, we are credited with a special authority
for the character of our own intentions; asked about them, it is
considered that we ought to know the answer, and, saving lying
and slips of the tongue, etc., that our answers should be given a
special weight. Admittedly, this authority does not have to be taken
to suggest noninferential knowledge; it might be, for example, that
it derived from authority for the premises of an inference-say, certain
occurrent thoughts. But to think of self-knowledge of intention,
in any case where the subject would be credited with authority,
as invariably based on inference from associated occurrent
thoughts is to caricature the ordinary notion. For one thing, each
of us regularly carries out intentional acts without necessarily
thinking about what we are doing at all. Usually these are routine
activities in which we are expert. It is perfectly proper to say of
such activities that they are knowingly and intentionally performed
and, indeed, that they are preceded by the appropriate intention. (If
you were asked, in advance, whether you had the appropriate intention,
you would unhesitatingly confirm that you did.) Notice
also that we can in general make no ready sense of the question,
776 THE JOURNAL OF PHILOSOPHY
"How do you know?" directed at an avowal of intention; if there
were an inference in the offing, the question ought to admit of a
straightforward answer. But the decisive consideration is this. Even
when an intention is accompanied by certain occurrent thoughts
relating to its content or the circumstances of the (envisaged)
course of action, one's knowledge of the character of the intention
is not to be thought of as achieved via reflection on the content of
those thoughts. If it were, by what principle could I assure myself
that those were the thoughts on which I should be concentrating,
rather than some other recent (or, if I am clever enough, simultaneous)
train? To come to know that you have a certain intention
is not to have it dawn on you that you have an intention of some
sort and then to recover an account of what the intention is by reflecting
upon recent or accompanying thoughts. It is the other way
round: you recognize thoughts as specifying the content of an intention
that you have because you know what the intention is an
intention to do.
What of the mysterious fecundity? Well, suppose I intend, for example,
to prosecute at the earliest possible date anyone who trespasses
on my land. Then there can indeed be no end of distinct responses,
in distinct situations, which I must make if I remember
this intention, continue to wish to fulfill it, and correctly apprehend
the prevailing circumstances. But if we are at ease with the
idea that my intention has a general content, noninferentially
known to me, then there is no more a puzzle about the "infinity"
of this content than there is a puzzle about the capacity of any universally
quantified conditional, (x)(Fx-+Gx), to yield indefinitely
many consequences of the form, Ga,Gb, . . . . , when conjoined
with corresponding premises of the form, Fa,Fb, . . . .
I want to stress that this is merely to describe what seem to be
features of our intuitive notion of intention. The notion is not unproblematic.
It could be that it is radically incoherent. The fact
remains that it is available to confront Kripke's skeptic, and that,
so far as I can see, the skeptical argument is powerless against it.
The ordinary notion of intention has it that it is a characteristic of
mind-alongside thought, mood, desire, and sensation-that a
subject has, in general, authoritative and noninferential access to
the content of his own intentions, and that this content may be
open-ended and general, may relate to all situations of a certain
kind. In order, then, to rebut the skeptical argument, it would have
sufficed, at the point where the skeptic challenged you to adduce
some recalled mental fact in order to discount the grue-interpretations,
to recall precisely your former intention with respect to the
777KRIPKE O N PRIVATE LANGUAGE
use of 'green'. To be sure, any specification that you might give of
the content of that intention would be open to unwelcome interpretation.
But, if you are granted the intuitive notion of intention,
you can reply that you do not in any case know of the content of an
intention via a specification of it; rather, to repeat, you recognize
the adequacy of the specification because you know of the content
of the intention.
The point, in summary, is not that it is particularly comfortable
to think of your former meaning of 'green' as consisting in your
having had a certain general intention, construed along the lines of
the intuitive conception, but rather that the skeptical argument has
absolutely no destructive force against that proposal.
IV KRIPKE'S H'ITTGENSTEIN
I conclude with but the briefest indication of the most important
difference, as I see it, between Kripke's Wittgenstein and Witt-
genstein.
There is an evident concern in the Investzgations with a large
class of psychological predicates which, like intention intuitively
conceived, seem to have a content that can somehow transcend that
of any accompanying thoughts in the subjecťs mind. Examples
are: recalling how a piece of music goes (without hearing it right
through "in one's head"); deciding to have a game of bridge (without
thinking through all the rules); realizing how to continue a
series (without per zmposszble, thinking through the entire infinite
expansion); grasping the meaning of an expression "in a flash"
(without having all its possible uses run before one's mind); and so
on. Each of these predicates, it seems, can come to be true of a subject
quite abruptly, yet involves some sort of reference to things he
need not, on that occasion, think about explicitly. Wittgenstein
thought that we were greatly prone to misunderstand the "grammar"
of these notions and to form quite false pictures of the nature
of the connection that obtains between the psychological state of
someone of whom such a predicate comes to be true and the "absent
aspects" noted in the parentheses above. In particular, there
need be n o connection between the subjective content, properly so
regarded, of such states and the detail of the "absent aspects".6 Accordingly,
the normatzve power of intention-the determinacy,
when it is determinate, in the matter of whether a particular course
of conduct fulfills a prior intention-cannot always be accounted
for by reference only to the previous subjective content of the subjecťs
psychological states. Wittgenstein's conclusion, however, is
For an excellent discussion of these examples, see Malcolm Budd, "Wittgenstein
on Meaning, Interpretation and Rules," Synthese, r.vIII, 3 (March 1984): 303-323.
778 THE JOURNALOF PHILOSOPHY
emphatically not that there is no such thing as the fulfillment of a
prior intention-the conclusion, in effect, of Kripke's skeptic.
Is it correct for someone to say: "When I gave you this rule, I meant
you to . . . in this case"? Even if he did not think of this case at all as
he gave the rule? Of course it is correct. For "to mean it" did not
mean: to think of it (Investigations s692; my italics).
Rather, a satisfactory philosophy of intention has to validate our
claim to noninferential authority for our present (and previous) intentions
without succumbing to the mythology of infinite, explicit
introspectible content. The intuitive conception of intention utilized
against Kripke's skeptic above perennially tempts us toward
this mythology. But there has to be something right about it if@acethose
who would wish to reanimate a dispositional account of
meaning and intention-Kripke's skeptic is riot to win the day.
It is this dilemma which is prominent in the last sections of part
I of the Znuestigations (§ 591 to the conclusion), sections about
whose evaluation there is so far little consensus. The insight that
there is a problem here, of the most profound importance for the
philosophies both of language and of mind-whether or not Wittgenstein
solved it-is one of the principal lessons of the Znuestigations,
and one which Kripke's book ought to make it easier to learn.
CRISPIN WRIGHT
University of St. Andrews
BOOK REVIEWS
Frege's Conception of Numbers as Objects. CRISPIN WRIGHT. The
Scots Philosophical Monograph Series. Aberdeen: Aberdeen University
Press, 1983. xxi, 193 p. E8.50.
Despite its title, this book is no scholarly study of Frege on the
numbers. Rather it is an original contribution to the philosophy of
mathematics, an attempt to develop and defend the conception of
numbers expressed in Frege's Foundations of Arithmetic. Wright
takes this to consist of three independent elements: (1) numbertheoretic
realism-statements about the numbers are determinately
true or false in virtue of states of the world irrespective of our ability
to ascertain them; (2)number-theoretic platonism-the numbers
"constitute a unique domain of genuine objects" (xx);(3) numbertheoretic
logicism-logic is the epistemological foundation for
0022-362X/84/8112/0778$00.50 01984 The Journal of Philosophy, Inc.