PHILOSOPHICAL TOPICS

VOL. 34, NOS. 1 & 2, SPRING AND FALL 2006

The Bounds of Sense

A. W. Moore
St Hugh’s College

I.

My title is, of course, a direct echo of the title that P. F. Strawson gave to his famous
study of Kant’s Critique of Pure Reason.1 Strawson’s title, as he pointed out in his
preface, was in turn a partial echo of a title that Kant himself had considered for the
Critique, or at any rate for an embryonic version of the Critique.2 In a letter to
Marcus Herz, written in 1771, Kant had told Herz that he was busy on a work
which he called The Bounds of Sensibility and Reason.3 Not that this was Strawson’s
only reason for his choice of title. The title also, he claimed, “[alluded] compen-
diously to the three main strands in [Kant’s] thought.”4 As he went on to amplify
this claim, he indicated how, among other things, his title played on the ambiguity
of the word “sense”, with its connotations of both experience and meaning. He
wrote:
In two ways [Kant] draws the bounds of sense, and in a third he trav-
erses them. He argues, on the one hand, that a certain minimal structure
is essential to any conception of experience which we can make truly
intelligible to ourselves; on the other, that the attempt to extend beyond
the limits of experience the use of structural concepts, or of any other
concepts, leads only to claims empty of meaning . . . [But Kant] seeks to
draw the bounds of sense from a point outside them, a point which, if
they are rightly drawn, cannot exist.5
That Strawson managed to accomplish so much in the mere naming of his book
prompted one reviewer to write, “The title itself is a roguish stroke of genius.”6

327
 My aim in this essay is to explore some of the resonances of that book’s won-
derful Kantian title, and to relate these to some key moments in twentieth-century
analytic philosophy, one of whose defining features has been the aspiration to draw
the bounds of sense. Both experience and meaning have been of fundamental con-
cern to twentieth-century analytic philosophy, but the latter quintessentially so, and
my chief concern in this essay will be with the bounds, in particular, of meaning.
Henceforth I shall accordingly use “sense” only in that sense.
This aspiration of twentieth-century philosophy to draw the bounds of sense
has in turn been dogged by the very threat of self-stultification to which Strawson
saw Kant fall prey. That threat, at a highly schematic level, is clear. Any attempt
to draw the bounds of sense by dividing some metaphorical space into two, that
whereof one can make sense and that whereof one cannot, looks as if it must fall
foul of the fact that one cannot make sense of the divide itself unless one can make
sense of both sides. As Wittgenstein famously says, in his preface to the book whose
contribution to this dialectic has arguably been more significant than any other and
whose own echoes will have been clear in what I have said so far, “in order to be
able to draw a limit to thought, we should have to be able to find both sides of the
limit thinkable (i.e., we should have to be able to think what cannot be thought).”7
There are various episodes in the history of analytic philosophy in the last century
that have been, in effect, more or less self-conscious attempts to work around this
aporia, either by drawing the bounds of sense in some entirely different way, or by
drawing them in this way and denying that doing so is self-stultifying, or by draw-
ing them in this way, acknowledging that doing so is self-stultifying, and learning
to live with the self-stultification.

II.

Let us begin with the Tractatus. Among the countless exegetical problems posed by
that extraordinary and perplexing book, perhaps the deepest are problems about
where it stands in relation to this aporia. All three of the ways just mentioned of
trying to work around the aporia have been read into the book. The two extremes,
whereby on the one hand Wittgenstein is read as attempting some quite different
way of demarcating sense, and whereby on the other hand he is read as attempting
to live with self-stultification, are emblematic of what I shall call “new” readings of
the book and “traditional” readings of it.8
New readings of the Tractatus hold the following:
The limit to be drawn is the straightforward limit that separates those
signs to which, as a matter of brute historical fact, meanings have been
assigned from those signs to which, as a matter of brute historical fact, no
meanings have been assigned; and there is not the least self-stultification
in characterizing what lies on the latter side of the limit in precisely that
way. What gives the project the interest it has, as a philosophical rather

328
 than a merely lexicographical exercise, is in part its generality. But its
chief interest lies in the fact that there are temptations of a distinctively
“philosophical” kind to see meaning where it is lacking. Wittgenstein’s
aim in the Tractatus is to eliminate such temptations. And the way in
which he tries to achieve this aim is by producing signs to which such
temptations attach; then indulging the temptations to such an extent
that they eventually become unsustainable and disappear. Again, there
is not the least self-stultification in this.
Traditional readings of the Tractatus, by contrast, hold the following:
There are, in Wittgenstein’s view, things that cannot be put into words.
And what divides these things from the things that can be put into
words is something that itself cannot be put into words. Any attempt to
say what divides them must therefore issue in nonsense. If this means
that any attempt to say what divides them must be self-stultifying, so be
it. It does not follow that any such attempt must be a failure. For there
may be a special kind of nonsense that is able to serve the very function
required here. And indeed Wittgenstein thinks there is. Such nonsense
is precisely what he takes the bulk of the Tractatus to consist in. He thinks
that the nonsense that he himself produces can help us to apprehend
some of the things that cannot be put into words. In particular, it can
help us to apprehend what divides the things that cannot be put into
words from the things that can. The self-stultification is benign.

It is not my purpose here to arbitrate in this exegetical debate. But I do want

to highlight one advantage that new readings certainly have over their more tradi-
tional rivals, namely that they chime well with that section in the preface from
which I quoted earlier. Significantly, my quotation, though relevant to the structure
of the aporia that I was using it to illustrate, was concerned with drawing a limit to
thought, not with drawing a limit to sense (= meaning). But it is the latter that is
currently of concern to us. And precisely Wittgenstein’s point, in the larger context
from which I quoted, is that, although the former is impossible—for the very rea-
son given in the quotation—the latter, where attention turns from the object of
thought to the expression of thought, and where the limit to be drawn is the limit
that separates those signs that are used to express thoughts from those signs that
are not, is perfectly possible. For there is no suggestion in this latter case that the
limit to be drawn is a limit between what has one kind of subject matter and what
has another. It is a limit, rather, between what has any kind of subject matter and
what has none. And we can certainly say what constitutes this limit—without hav-
ing to say anything about a subject matter about which, in the nature of the case,
nothing can be said. Whatever lies on the “wrong” side of this limit “will simply be,”
as Wittgenstein says, “nonsense.”9
Here is another way of approaching these issues—in terms of different kinds
of possibility. Intuitively, some kinds of possibility are strictly subsumed by others.
Thus whatever is technologically possible is physically possible, but not vice versa;
whatever is physically possible is mathematically possible, but not vice versa. The
natural picture here is that of a series of concentric circles, in which larger circles

329
include possibilities that smaller circles exclude. To be sure, there are some kinds of
possibility, notably those that have a cognitive element, that may not fit this picture.
These kinds of possibility may cut across the others. For instance, a given physical
possibility may qualify as such for reasons that are too complex or subtle for us fully
to grasp and may thereby count as an impossibility of another, cognitive kind that
in other cases extends beyond the physical: say, the imaginable.10 Furthermore, it
may be that none of these kinds of possibility admits of a precise characterization
and that what separates any two is always contestable. But still, the picture of a
series of concentric circles, applicable in at least some central cases, is an intuitively
compelling one. And it is often the means we use to indicate what a given kind of
possibility excludes, or at least some of what it excludes: we say that certain things
are not possibilities of that kind, by first identifying them as possibilities of some
more inclusive kind. Thus a politician may say, adverting to what is technologically
possible, “There are some ways of improving the safety of our railways that are
unaffordable.” A botanist may say, adverting to what is physically possible, “There
are some temperatures below which plant life is unsustainable.” (The politician is
not vindicated by the technological impossibility of a completely failsafe automated
signaling system; nor the botanist by the physical impossibility of any temperature
below absolute zero.)
Now it is natural to suppose, further, that there is one kind of possibility that
subsumes all the rest. We might call this “ultimate” possibility, or “logical” possibil-
ity. True, there are various reasons why this idea of an all-inclusive kind of possibil-
ity is not as straightforward as it appears, even when complications concerning
possibilities of a more or less cognitive kind are set aside. Thus consider the fact
that the impossibility of something’s being both green all over and red all over,
though logical in a loose sense, does not turn on the meanings of what would stan-
dardly be recognized as logical constants.11 But, to the extent that we are entitled
to think in terms of one all-inclusive kind of possibility, we have a further com-
pelling illustration of the aporia that afflicts the drawing of the bounds of sense.
For we obviously cannot say, except as a kind of joke, that what this all-inclusive
kind of possibility excludes are possibilities of such and such another kind, as it may
be the “illogical” possibility that grass both is and is not green. This all-inclusive
kind of possibility is not just another circle in the space we have been considering.
It is the space we have been considering. To delimit it requires something of an alto-
gether different kind. It requires an ascent, however indirect, to the metalanguage,
whereby certain combinations of words can be said not to represent possibilities at
all. They can be said, in one good sense of the phrase, not to make sense. This accords
with Wittgenstein’s shift of attention in the preface to the Tractatus away from
thought to the expression of thought. It also accords with various remarks in his
later work. At one point, in Philosophical Grammar, commenting on the infinitude
of the sequence of cardinal numbers, Wittgenstein insists that we should not say,
“There is no last cardinal number”—as though we were excluding some possibility
—but should rather say, “The expression ‘last cardinal number’ makes no sense.”12

330
And in Philosophical Investigations, having remarked that “essence is expressed by
grammar,” he says of a puzzle that he is wrestling with there, “The great difficulty
here is not to represent the matter as if there were something one couldn’t do.”13
This shift of attention from the object of thought to the expression of thought,
or from possibilities to the representation of possibilities, does seem to provide a
non-self-stultifying way of drawing the bounds of sense. (Of course, there remains
the exegetical puzzle of why Wittgenstein should nevertheless have stepped beyond
those bounds when trying to execute this project in the Tractatus; but we have
already seen how new readings of the Tractatus address this puzzle.) The basic pro-
cedure of drawing the bounds of sense by distinguishing, not between that of which
sense can be made and that of which it cannot, but between that which has sense
and that which does not, appears unimpeachable. But is it?
I alluded earlier to the fact that the project is supposed to be a philosophical
project, not a merely lexicographical one. It will of course count as a philosophical
project in so far as it provides for a general account of what it is to have sense and
does not merely, and literally, issue in a combined lexicon and grammar for some
particular language. Nevertheless, if it is to have the kind of philosophical point that
drawing the bounds of sense had for Kant, it must do more than that. For one
thing, it must combat various illusions of sense.
According to new readings of the Tractatus, that is precisely what the project,
as executed there, does do; or at least, that is precisely what Wittgenstein intends it
to do. The way in which Wittgenstein tries to realize this intention, according to
new readings, is by so presenting various illusions of sense that they eventually dis-
appear. But is there not more to it than that? Should there not be more to it than
that? Surely our philosophical aspirations are not going to be satisfied except in so
far as we have some sort of diagnosis—some sort of explanation for why various
assignments of meanings to signs appear to confer sense where they do not—and,
more generally and more significantly, except in so far as we have some general
philosophical understanding of what assignments of meanings to signs can achieve
and what, despite appearances, they cannot.14 How clear is it that we can attain and
express such an understanding, itself a project in drawing the bounds of sense,
without self-stultification? How clear is it, for instance, that we can attain and
express such an understanding without identifying various things which cannot be
expressed, and which cannot therefore be suitably identified, no matter what
assignments of meanings to signs we make? How clear is it, for that matter, that this
is not Wittgenstein’s project in the Tractatus?15
At the very beginning of his book Wittgenstein writes, “The world is all that is
the case. The world is the totality of facts, not of things.”16 This is connected to his
subsequent remark that “propositions can only say how things are, not what they
are.”17 Thus although we can name things, and thereby speak about them, we can-
not put them into words, or express them by means of propositions, in the way we
can facts. These remarks help to combat the temptation, real enough in my view, to
express what one understands in knowing the meaning of a word by casting it in

331
propositional form, as though what one knew were that something is the case. Such
remarks also have the kind of diagnostic generality to which I was referring in the
previous paragraph. Yet the claim that the world consists of facts, not of things,
stands in direct violation of something that Wittgenstein says later in the book,
echoing his admonishment in the preface against drawing the limits of thought:
[We] cannot say in logic, “The world has this in it, and this, but not that.”
For that would appear to presuppose that we were excluding certain
possibilities, and this cannot be the case, since it would require that logic
should go beyond the limits of the world; for only in that way could it
view those limits from the other side as well.18

So Wittgenstein has found himself falling prey to the very threat of self-stultifica-
tion that he highlighted in the preface.19 Nor is it clear that he has any other way of
conveying his insights into the nature of language. True, he might urge upon us a
distinction between effable states of knowledge, such as someone’s knowledge that
grass is green, and ineffable states of knowledge, such as someone’s knowledge of
the meaning of the word “green”; and he might insist, without any obvious self-
stultification, that only knowledge of the former kind can be expressed by what has
sense, or in other words can be expressed at all. But if he did thereby manage to
avoid self-stultification, then he would do so only at the price of vacuity. He would
still not have done justice to his own insights into what an abortive attempt to
express knowledge of the latter kind would be an abortive attempt to do—insights,
that is, into what would motivate the attempt. It remains unclear whether he has
any way of conveying these insights without himself making that same abortive
attempt. The aporia remains stubbornly in his way, still to be negotiated.

III.

Let us return to the very idea of demarcating that of which one can make sense by
dividing some metaphorical territory into that of which one can make sense and
that of which one cannot. In taking for granted that there is something incoherent
about this, we are taking for granted a relatively undemanding conception of what
it is to make sense of something. For the thought, presumably, is that
(1) in order to effect such a divide, one must make sense of it,
and
(2) in order to make sense of such a divide, one must make sense of what lies on
both sides of it.
But there are more demanding conceptions of what it is to make sense of some-
thing which allow us to resist either (1) or (2). Thus on a conception whereby one
does not make sense of something unless one’s understanding of that thing has a

332
suitable grounding in experience, we can resist (1): one might be able to effect such
a divide by arriving at an understanding of it which is not in any relevant sense
experiential. Again, on a conception whereby one does not make sense of some-
thing unless one attains a significant amount of knowledge about that thing, we can
resist (2): one might be able to attain a significant amount of knowledge about such
a divide without knowing anything, or anything substantial, about what lies on its
“far” side. We need to consider ways of working around the aporia which are vari-
ations on one or other of these two themes.
It is instructive, first, to ask whether Kant himself might be exonerated in these
terms. Consider the opening section of John McDowell’s Mind and World.20 In that
section McDowell refers to the famous passage in the Critique where Kant argues
that we need both intuitions and concepts in order to know anything and insists
that the former in the absence of the latter are “blind” while the latter in the absence
of the former issue in thoughts that are “empty”.21 “For a thought to be empty,”
McDowell comments, “ . . . would be for it not really to be a thought at all, and that
is surely Kant’s point; he is not, absurdly, drawing our attention to a special kind of
thoughts, the empty ones.”22 But it seems to me that that is precisely what Kant is
doing, or at least what he takes himself to be doing. This is why, elsewhere in the
Critique, he insists on the distinction between what we can think and what we can
know. What we can think outstrips what we can know precisely because it includes
what we can think without intuitions: it includes our “empty” thoughts.23 Granted
this distinction, the distinction between what we can think and what we can know,
it is entirely reasonable to equate the bounds of sense that Kant wishes to draw with
the bounds of what we can know. It is entirely reasonable, in other words, to
accredit Kant with a relatively demanding conception of what it is to make sense of
things, whereby we can make sense only of what we can be given in intuition.24 This
in turn allows him to work around the aporia in the first of the two ways suggested
in the previous paragraph. Kant can freely admit that the distinction he has drawn
between that of which we can make sense and that of which we cannot—between
how things appear to us and how they are in themselves, basically—is not itself
something of which we can make sense but, along with other things of which we
cannot make sense, is something that we can quite legitimately think. Thus Kant
writes:
That we can . . . have no knowledge of any object as thing in itself, but
only in so far as it is an object of sensible intuition, that is, an appear-
ance . . . is proved in the analytical part of the Critique. Thus it does
indeed follow that all possible speculative knowledge of reason is lim-
ited to mere objects of experience. But our further contention must also
be duly borne in mind, namely, that though we cannot know these
objects as things in themselves, we must yet be in a position at least to
think them as things in themselves; otherwise we should be landed in
the absurd conclusion that there can be appearance without anything
that appears.25

333
 Does this show that Kant does after all have a satisfactory way of drawing the
bounds of sense? Well, no; that conclusion would be precipitate, not least because
there is reason to doubt his distinction between what we can think and what we can
know.26 And even if we grant Kant that distinction, there is reason to doubt whether
his handling of it is as careful as it should be. For example, he claims to have “proved”
that we can have no knowledge of things in themselves. But would such a proof not
issue in knowledge of the very sort it is meant to preclude?27 (Or is there some con-
trast between the “negative” knowledge in which it would issue and the “positive”
knowledge with which it would implicitly be concerned?28) However that may be,
we are not yet in a position to say that Kant has a satisfactory way of drawing the
bounds of sense. But we can say that he avoids the immediate structural threat of
self-stultification that constitutes the aporia with which we have been concerned.

IV.

Let us now return to twentieth-century analytic philosophy, where there is a notable

attempt to draw the bounds of sense which looks as though it might retain this very
advantage. Indeed it looks as though it might retain the advantages of both the
principal attempts to draw the bounds of sense that we have considered so far while
avoiding the defects of either. It shares with what we found in Wittgenstein a met-
alinguistic focus on the distinction between that which has sense and that which
does not. But it also ventures a general philosophical account of why this distinc-
tion needs to be drawn where it does, and here it shares with what we have just
found in Kant a relatively demanding conception of what it is to make sense of
something. Each of these can, in ways that we have seen, serve to keep self-stultifi-
cation safely at arm’s length.
The position I am thinking of is the logical positivism that finds popular and
forthright expression in A. J. Ayer’s Language, Truth and Logic.29 The principle of
verification enunciated in that book, whereby a statement has “literal meaning” if
and only if it is either analytic or, in some suitably refined sense, empirically verifi-
able,30 is a way of distinguishing between that which has a kind of sense and that
which lacks it—which provides one sort of protection against the threat of self-stul-
tification. But the principle allows for all sorts of meaningful statements that lack
this kind of sense. Most notably, it allows for statements of value, whose meaning
is of an altogether different kind, namely to express feelings and/or to prescribe
courses of action.31 To this extent such positivism works with a relatively demand-
ing conception of what it is to make sense of something—which provides a differ-
ent sort of protection against the threat of self-stultification.
But how exactly is this latter sort of protection to be implemented? We saw two
models in the previous section: drawing the boundary around sense in a way that
does not involve making sense of that boundary, contra (1); and drawing the bound-

334
ary around sense in a way that does involve making sense of that boundary but does
not involve making sense of what is on the “wrong” side of it, contra (2). Which of
these do advocates of such positivism profess to be doing? This choice is, in effect,
a choice about how to answer the question, “What status does the principle of ver-
ification itself have?”, a question that is often posed as a challenge to such positivism.
Both alternatives are reckoned to be unattractive. But actually both alternatives are
(prima facie) attractive; and advocates of such positivism, if they experience any
embarrassment with the question at all, are liable to experience the embarrassment
of riches. The first alternative, whereby drawing the boundary around sense does
not involve making sense of that boundary, is to regard the principle as a literally
meaningless prescription about how to use the expression “literal meaning”. The sec-
ond alternative, whereby drawing the boundary around sense does involve making
sense of it, but in a way that is innocuous, is to regard the principle as an analytic
truth validated by how the expression “literal meaning” is already used, at least by
philosophers party to the relevant disputes. In Language, Truth and Logic it is
unclear which of these tactics Ayer takes himself to be adopting. He calls the prin-
ciple a “definition”, but insists that “it is not supposed to be arbitrary.”32 Later he
makes clear that he took himself to be adopting the first alternative. He says that he
was never tempted to regard the principle as either empirically verifiable or analytic,
then continues:
Happily not everything that the verification principle failed to license
was cast by me on the pyre of metaphysics. In my treatment of ethics, I
made provision for prescriptive statements . . . . Accordingly, in . . .
Language, Truth and Logic, I treated the verification principle as a pre-
scriptive definition.33
So far, one might think, so good. However, as Ayer also goes on to remark,
there remains the question of why the prescription should be obeyed. “I evaded this
awkward question,” he writes, “by defying my critics to come up with anything bet-
ter.”34 But why is the question awkward? More particularly, why is it awkward for
Ayer? Not because he has nothing to say in answer to it. There is plenty that he
might say, and that would be consonant with his overall view of these matters, most
pertinently that only when a statement has what he calls “literal meaning” is there
any such thing as determining its truth or falsity; indeed, only then that it is either
true or false. What makes the question awkward for him is something that Michael
Dummett forcefully argues in the essay to which Ayer is replying when he makes
these remarks: namely, that no answer he gives will be fully satisfactory unless and
until it is placed in the context of some general semantic theory that is of just the
kind that he wants to cast “on the pyre of metaphysics.”35 Any such theory must
include a philosophical account of what, if anything, enables the truth or falsity of
statements of various kinds to be determined, and of how this in turn relates to
whether statements of those kinds are true or false. If, for instance, the presence of
an evaluative element in a statement prevents its truth or falsity from being deter-
mined, and thereby ultimately prevents it from being true or false, then the theory

335
must indicate why—which is as much as to say that it must engage with the “meta-
physics” of value. This is enough to constitute a significant ad hominem point against
Ayer. More importantly, when combined with the worries expressed above in §2
about the project of attaining and expressing a general philosophical understand-
ing of what any assignment of meanings to signs can achieve, it indicates that there
is still a threat of self-stultification to be negotiated. Logical positivism may not pro-
vide us with the best of both the Wittgensteinian world and the Kantian world after
all.

V.

In §3 I voiced a worry about Kant’s own way of drawing the bounds of sense. This
is a worry about his distinction between what we can think and what we can know.
The “empty” thoughts that Kant sanctions seem to me (just as they seemed to
Strawson36) to be too “empty” to do the work that he requires of them. Still, at least
in drawing such a distinction Kant indicated one way to avoid the immediate struc-
tural threat of self-stultification that afflicted his project. It was something of this
same general sort, specifically a distinction of meaning between statements of dif-
ferent kinds, which, momentarily at least, seemed to provide logical positivism with
protection against its own equivalent threat.
In the work of Quine there is a descendant of logical positivism which is as
hostile as its forebear to the excesses of metaphysics and as deeply committed to the
links between sense, verification, and experience, but which is also utterly impatient
with any such distinctions of meaning. On Quine’s view, if a statement has mean-
ing at all, then it is either true or false, and there is such a thing as determining its
truth or falsity. “Suppose,” Quine writes,
we think of truth in terms of Tarski’s paradigm. The paradigm works for
evaluations . . . as well as for statements of fact. And it works equally well
for performatives. “Slander is evil” is true if and only if slander is evil,
and “I bid you good morning” is true of us on a given occasion if and
only if, on that occasion, I bid you good morning . . . . There are good
reasons for contrasting and comparing performatives and statements of
fact, but an animus against the true/false fetish is not one of them.37
Furthermore, whatever else might distinguish determining that one statement is
true from determining that another is, there is a holistic interdependence between
such things which means that there is never any answer to the question, “What
empirical evidence is required to verify just this statement?”38 In particular, there is
no statement for which the answer is “None”—no statement which can be verified
irrespective of what empirical evidence there is. That is, there is no such thing as an
analytic statement. Hence even the distinction of meaning that Quine’s positivist
predecessors wanted to draw within the range of statements that have sense is

336
uncongenial to him. Indeed, it is his hostility to this distinction which is as emblem-
atic as anything of his own brand of positivism.39
Prima facie, then, Quine is in trouble. He is espousing a kind of positivism which
involves him in drawing the bounds of sense, but which lacks the very resource
which looked as though it might enable someone in his position to draw those
bounds non-self-stultifyingly. In fact, however, as I indicated in the previous sec-
tion, what really carries the threat of self-stultification is the attempt to attain and
express some general philosophical understanding of why the bounds of sense
should be drawn where they should. But Quine’s refusal to recognize various dis-
tinctions of meaning between statements goes hand in glove with a refusal to rec-
ognize various distinctions of aim and methodology between explanatory projects:
in particular, the distinction between trying to attain and express such a general
philosophical understanding and trying to account, in broadly scientific terms, for
how, as a result of interactions between us and our environment, some things come
to have sense while others do not. As long as understanding why the bounds of
sense should be drawn where they should is seen as part of this scientific enterprise,
and not as some philosophical propaedeutic to it, then it is not at all clear that it
carries any threat of self-stultification. And as long as it is not seen in this way, then
it is not at all clear either that Quine will want anything to do with it or that he
should.
To be sure, Quine may find it harder than he supposes to keep some of his
predecessors’ distinctions at bay. There is a revealing section in Pursuit of Truth40
where Quine addresses the question whether the empiricism that underpins his
semantic views is itself empirical. Unsurprisingly, he insists that it is. He writes that
“it is a finding of natural science itself, however fallible, that our information about
the world comes only through impacts on our sensory receptors.”41 And he later
adds, “It would take some extraordinary evidence to [testify to either telepathy or
clairvoyance] . . ., but, if that were to happen, then empiricism itself . . . would go
by the board.”42 Yet he also seems to acknowledge, as an issue quite different in kind
from the empirical issue of what elicits or is used to justify any given scientific state-
ment, the issue of why the statement counts as a scientific statement (and thereby
counts as having sense). He writes:
When I cite predictions [that is, predictions of sensory input] as the
checkpoints of science, . . . I see [that] as defining a particular language
game, in Wittgenstein’s phrase: the game of science . . . A [statement’s]
claim to scientific status rests on what it contributes to a theory whose
checkpoints are in prediction.43
Given the context, this last statement can readily be heard as a prescription, such as
Ayer took the verification principle to be; or, worse still for Quine, as an analytic
truth.44
The important point, however, is that Quine’s drawing of the bounds of sense
is to be seen as part of a scientific enterprise and, seen as such, it does not appear to
be under any special threat of self-stultification. His semantics is informed by his

337
general worldview and is proffered from a point of immersion within that world-
view. Thus just as he couches his empiricism in terms of impacts on sensory recep-
tors, ocular irradiation, and the like, so too he couches his positivist conception of
meaning in those same terms.45
There is much here to give pause however. I shall mention two worries in par-
ticular that Quine’s critics have had. First, there is a worry to which McDowell has
given celebrated expression, a worry in which we can hear muffled echoes of
Wittgenstein’s insistence that the world is the totality of facts, not of things.46 This
is the worry that, by construing our evidence for our worldview as a matter of
impacts on our sensory receptors, ocular irradiation, and suchlike, Quine is casting
entities that are external to our worldview in a role that ought to be filled by enti-
ties that are already part of our worldview, namely experiences we have of things
being thus and so; and, by the same token, he is representing what ought to be a
logical or rational relation, namely the relation between our evidence and the rest
of our worldview, as a merely causal relation.47
The second worry suggests, conversely as it may appear (but see further below),
that Quine has represented the relation between our evidence and our worldview
as something too intimate. This is a worry that John Campbell has expressed very
forcibly.48 It is the worry that, by construing our evidence for our worldview in
terms that depend so heavily on that very worldview, Quine has negated an impor-
tant principle whose importance, indeed, he himself would be the first to empha-
size, namely that our worldview is underdetermined by our evidence. “[Given that]
the patterns of ocular irradiation have to be described in terms of the physics of the
day,” writes Campbell, “how . . . could they be consistent with some rival to the
physics of the day?”49
Now one might think that Quine has a perfectly satisfactory riposte to
Campbell’s rhetorical question. What matters, one might think, is not how the pat-
terns of ocular irradiation are to be described, but what their content is. Here is an
analogy. Imagine a brain in a vat, in a classical skeptical scenario,50 whose subject
thinks that he is living the life of a medieval monk. And suppose we have to draw
on various principles of computerized neurotechnology to describe what is hap-
pening to the brain. It simply does not follow that what is happening to the brain
is enough to refute the subject’s impression of what kind of life he is living. Again,
suppose we have to use some realist theory about middle-sized dry goods to
describe the impact of Samuel Johnson’s foot on a stone. It simply does not follow
that this impact is enough to refute Berkeleian idealism.51 Similarly, if we have to
use the physics of the day to describe certain patterns of ocular irradiation, it sim-
ply does not follow that these patterns are enough to refute each and every rival to
that physics. If there were people who, on broadly the same evidence as ours,
accepted such a rival, then they could not acknowledge any such phenomenon as
ocular irradiation (which of course would be a deficiency by our lights). They
would have to tell their own rival story about what empirical evidence they had for
their theory. Yet, for all that, their evidence would in fact (by our lights) involve

338
ocular irradiation. There is nothing incoherent in this. Such is how the underdeter-
mination of theory by evidence is bound to be described from a point of immer-
sion in one of the underdetermined theories—the only kind of point, in Quine’s
view, from which it can be described.52 Nor does this mean that various pragmatic
forces cannot eventually bring us to a point of immersion in one of the rivals to the
physics of the day. Campbell suggests that, without an Archimedean point, Quine’s
view leads to an unacceptable conservatism.53 But the image of Neurath to which
Quine famously appeals is precisely meant to show that this is not so: we can
entirely rebuild our boat even while staying afloat in it, provided that we rebuild it
plank by plank.54
Is this a legitimate reply on Quine’s behalf to Campbell’s rhetorical question?
Only on one absolutely fundamental assumption: that it makes sense in Quine’s
terms to talk about the “content” of our evidence. If it does not, the question whether
our evidence refutes this or that theory cannot so much as arise for Quine. But this
now brings us back to McDowell’s worry. For McDowell’s worry is precisely that it
does not make sense, in Quine’s terms, to talk about the content of our evidence;
that Quine has construed evidence as a matter of events and episodes which can
enter into causal relations but not into logical relations. This is why McDowell
thinks that Quine needs a fundamentally new and more commonsensical concep-
tion of our evidence whereby our evidence is a matter of how we experience things
as being. But really that is Campbell’s point too. “Scientific theorizing,” Campbell
writes, “can never let go of the idea that it is ultimately our experiences [as of
macroscopic physical objects] that have to be explained.”55 The two worries, despite
an initial impression of disparity, are of a piece.
And they cut deep. For they suggest that there is after all room for some kind
of philosophical propaedeutic to science. They suggest that we can legitimately seek
a general philosophical understanding of what science is answerable to. Moreover,
when this idea is combined with the broadly positivist conception of meaning that
Quine favors, with its perceived link between what science is answerable to and
what makes sense, then it leads down the very path that I have already identified as
the main route to self-stultification: the path of trying to attain and express a gen-
eral philosophical understanding of why the bounds of sense should be drawn
where they should. So it seems that Quine has become yet another example of how
not to evade the threat of self-stultification.

VI.

I have suggested more than once that, when drawing the bounds of sense is con-
strued as a philosophical enterprise, or in other words when the aim of the enterprise
is to attain and express some general philosophical understanding of why the bounds
are to be drawn where they are, then self-stultification looms. There is evidence for

339
this in the Tractatus. And we have seen evidence in the work of various positivists
for how hard it is to keep a suitable distance from just such a philosophical enter-
prise, even given the firmest of resolves.
That there are links here with Kant should be evident to anyone familiar with
the accusation of self-stultification levelled against him by Strawson (see above, §I).
But the links are more profound than that. It is not just that attempts by twentieth-
century analytic philosophers to draw the bounds of sense share certain structural
defects with Kant’s attempts to do something analogous. They actually lead in the
direction of transcendental idealism. When Wittgenstein declares in the Tractatus
that the world is the totality of facts, not of things, he is setting the limits of the
world as the limits of what can be thought or said.56 That grass is green is part of the
world, because it is possible to think and to say that grass is green; but neither grass
nor greenness is part of the world, because there is no such thing as either thinking
or saying either grass or greenness.57 This is a kind of transcendental idealism.
The worry, of course, is that, in as much as a construction such as “thinks
greenness” is nonsense, then so too is a sentence such as, “There is no such thing as
thinking greenness.” Or, to put the worry somewhat less accurately but with greater
rhetorical force, if there is no such thing as either thinking or saying something,
then neither is there any such thing as either thinking or saying that there is no such
thing as either thinking or saying that thing. Transcendental idealism itself is non-
sense.58
If these suggestions are even broadly correct, then it is impossible to attain and
express any general philosophical understanding of why the bounds of sense should
be drawn where they should. It is impossible, on that ambitious construal of draw-
ing the bounds of sense, to draw the bounds of sense; and any attempt to do so will
issue in nonsense. But what follows? Each time that I have referred to what it is
impossible to do, I have quite deliberately used the phrase “attain and express”: it is
impossible to attain and express such a general philosophical understanding. What
follows, then—or at least, one thing that follows—is that either it is impossible to
attain such an understanding or it is possible to do that, but it is not possible at the
same time to express the understanding, in other words the understanding is inef-
fable. My own view is that twentieth-century analytic philosophy provides the
resources to accommodate the latter alternative, and indeed to accede to it. But that
is a story for another occasion.59

NOTES

1. Strawson (1966), a study of Kant (1933). (I have specified Kemp Smith’s translation of the
Critique, even though, on the whole, I prefer the translation by Paul Guyer and Allen W. Wood
(Cambridge: Cambridge University Press, 1998). I have done so because Kemp Smith’s transla-
tion of one crucial passage that I shall be citing later is clearer for my current purposes: see below,
note 25 and accompanying text.)
2. Strawson (1966), 11.

340
 3. Kant (1999), 10:123, p. 127. Cf. another letter to Herz, written the following year, ibid., 10:129, p.
132, where he said that he had been “making plans for a work that might perhaps have the title,
The Limits of Sensibility and Reason.” Although Zweig translates the title differently in the two
cases, the original German is the same: the word that is rendered first Bounds and then Limits is
Grenzen.
4. Strawson (1966), 11.
5. Ibid., 11–12.
6. Cerf (1972), 601.
7. Wittgenstein (1961), 3.
8. I use these labels because they usefully signal two exegetical tendencies. I do not mean to suggest
that there is a simple polarization in the secondary literature. Cf. in this connection Sullivan
(2003), footnote 2 and 214–15. For a traditional reading, see Hacker (1986), esp. ch. 1; and for new
readings, see Diamond (1991) and Conant (1989). For an intermediate reading, see McGinn
(1999). And for a quite different reading, see Moyal-Sharrock (2007). I myself try to provide
something that is not readily classifiable as either a traditional reading or a new reading (or per-
haps rather that is classifiable, with suitable qualifications, as both): see Moore (2003).
9. Wittgenstein (1961), 3. Cf. Sullivan (2003), esp. 209–11, to which I am much indebted.
10. Kripke (1980) is a classic discussion of a variation on this theme. See also Edgington (2004): but
note her footnote 12, where she says that, although the metaphysically possible and the epistem-
ically possible cut across each other, “one has to search hard for examples, which are rather con-
trived and on the whole not very important or interesting, of the metaphysically possible which
is not epistemically possible.”
11. Contra Wittgenstein: see Wittgenstein (1961), 6.3751.
12. Wittgenstein (1974a), 465.
13. Wittgenstein (1974b), Pt. I, §§371 and 374, first and third emphasis his, second emphasis mine.
Cf. in this connection Wittgenstein (1961), 4.113–4.116, and Williams (1981), 159–60—where in
each case there is an image of working outward from within the space of sense toward its “edge”,
which, if it means anything at all, surely means (something like) producing combinations of
words that make sense with a view to registering the point at which similar combinations of words
fail to make sense. Note: the failures to make sense that are of concern here may quite properly be
distinguished from other failures to make sense. Thus, even in Wittgenstein (1961), where
Wittgenstein is impatient with various attempts to discriminate between ways of failing to make
sense (see 5.473 and 5.4733), he himself distinguishes between lacking sense and being nonsensi-
cal (see 4.461–4.4611). This observation may in turn help Wittgenstein to treat as an ally some-
one who is prima facie a foe: I am thinking of Deleuze and his comments in Deleuze (1990), 35.
14. Cf. Sullivan (2003), 211–12.
15. This paragraph is a summary allusion to ideas that I have expressed elsewhere. See Moore (2003),
esp. 189–90, where similar questions are posed.
16. Wittgenstein (1961), 1 and 1.1.
17. Ibid., 3.221, his emphasis.
18. Ibid., 5.61. An obvious reply on Wittgenstein’s behalf is that he does not in fact say anything in 1.1
that he proscribes saying in 5.61, since he uses the word “world” differently in the two cases: in the
former, to refer to the realm of the actual; and in the latter, to refer to the realm of the possible. I
incline to the view that he uses it to refer to the realm of the actual throughout the Tractatus; and
that what enables him to refer to the realm of the possible in 5.61 is his use of other words and
phrases, notably “limits” and “in logic”. But even if I am wrong about that—even if Wittgenstein’s
use of “world” is ambiguous in the way proposed—what he says in 1.1, with its clearly implied
application to any other possible world, is still surely offensive to the spirit of what he says in 5.61.
19. And, for reasons that I shall sketch in §VI, he has found himself endorsing a species of transcen-
dental idealism to boot. The links with Kant are profound.
20. McDowell (1994).
21. Kant (1933), A50–51/B74–75.
22. McDowell (1994), 3–4.

341
 23. E.g., Kant (1933), Bxxvi, footnote; B146; B166, footnote; and A771–72/B799–800.
24. Cf. ibid., B148–49; A239–40/B298–99; A247/B304; A696/B724.
25. Ibid., Bxxvi, his emphasis. (This is the passage to which I referred in note 1.)
26. Cf. Moore (1997), 139–40. See also ibid., 250–51.
27. If Williamson (1996) is right, then there is even a question about whether Kant is entitled to make
assertions about what we can know nothing about: Williamson argues that, in asserting some-
thing, one represents oneself as knowing it.
28. Cf. Kant (1933), B307 ff.
29. Ayer (1971).
30. See ibid., Introduction.
31. Ibid., ch. 6, passim.
32. Ibid., 20–21.
33. Ayer (1992), 149.
34. Ibid.
35. Dummett (1992), esp. 133–34. Ayer partially concedes this point in Ayer (1992), 150.
36. Strawson (1966), e.g., 264–65.
37. Quine (1981), 90.
38. Or hardly ever. Quine does acknowledge some rare and artificial exceptions: see Quine (1986),
620.
39. The locus classicus is Quine (1961).
40. Quine (1992), §8.
41. Ibid., 19.
42. Ibid., 21.
43. Ibid., 20. There are three things to note here. First, my gloss on “predictions” in the first pair of
square brackets is taken from ibid., 21. Second, my replacement of “sentence” by “statement” in
the third pair of square brackets is simply to bring the quotation into line with my usage elsewhere
in this essay: I hope it does not do violence to Quine’s intentions. Third, in the second ellipsis
Quine writes, “in contrast to other good language games such as fiction and poetry,” which indi-
cates that he is not impatient with all distinctions of usage between statements.
44. But perhaps Quine manages to stop it from sounding like the latter when he subsequently
observes that, given evidence for the falsity of empiricism, “it might indeed be well to modify the
game itself,” ibid., 21. For the significance of this observation, and its relevance to whether his own
earlier statement is analytic, see Moore (2002), §I.
45. It is largely on this basis that he famously draws the conclusion that, if a question about meaning
cannot be answered by adducing empirical evidence, then there is no fact of the matter concern-
ing what its answer is: e.g. Quine (1960), ch. 2, and Quine (1992), ch. 3.
46. For McDowell’s own explicit reference to Wittgenstein in this connection, see McDowell (1994), 27.
47. See esp. ibid., Afterword, Pt. I, §3. It is interesting to note in this connection how evasive much of
Quine’s language is. In Quine (1992) he speaks of “the flow of evidence from the triggering of the
senses to the pronouncements of science,” 41. The word “flow” here nicely straddles the very
divide that McDowell is trying to get Quine to recognize.
48. Campbell (2002), ch. 11, §5.
49. Ibid., 233.
50. See Putnam (1981), 5–6.
51. Johnson famously thought it was enough: see Boswell (1887), vol. 1, 471.
52. See e.g. Quine (1960), ch. 1, passim.
53. Campbell (2002), 234.
54. Quine (1960), 3.
55. Campbell (2002), 234. Cf. also Stroud (1984), ch. 6, esp. 250–54.

342
 56. I do not mean to deny that he is doing the converse as well. There may be reciprocal dependence
here.
57. This is connected with the fact, as I see it, that there is no such thing as saying what it is for some-
thing to be green, where this means saying what it is, in essence, for something to be green (cf.
Moore [1997], 134–35, 163–64, and 184). But surely there is such a thing as knowing what it is for
something to be green? Perhaps there is. But see in this connection Wittgenstein (1974b), Pt. I,
§78. (There is also a connection with the remarks made toward the end of §II above about the
ineffability of knowing the meaning of the word “green”.)
58. Cf. Wittgenstein (1961), 5.6 ff. The links between Wittgenstein’s early work and transcendental
idealism, and other related links, including the link between his later work and transcendental ide-
alism, have been a preoccupation of mine for some time. See e.g., Moore (1985); Moore (1997),
esp. chs. 6–9; Moore (2003), which deals in particular with the threat that accompanies any
attempt to categorize something as nonsense, namely the threat of uttering further nonsense (see
esp. §VIII); and Moore (2007). The inspiration for much of this is Williams (1981).
59. See again the material cited in the previous footnote.

REFERENCES

Ayer, A. J. 1971. Language, Truth and Logic, 2nd ed. Harmondsworth: Penguin.
Ayer, A. J. 1992. “Reply to Michael Dummett,” in L. E. Hahn (ed.), The Philosophy of A. J. Ayer. La
Salle: Open Court.
Boswell, James. 1887. The Life of Samuel Johnson, LL.D. Oxford: Oxford University Press.
Campbell, John. 2002. Reference and Consciousness. Oxford: Oxford University Press.
Cerf, Walter. 1972. “Review of The Bounds of Sense,” in Mind 81.
Conant, James. 1989. “Must We Show What We Cannot Say?”, in Richard Fleming and Michael Payne
(eds.), The Senses of Stanley Cavell. Lewisburg: Bucknell University Press.
Deleuze, Gilles. 1990. The Logic of Sense, trans. Mark Lester and Charles Stivale and ed. Constantin V.
Boundas. New York: Columbia University Press.
Diamond, Cora. 1991. “Throwing away the Ladder: How to Read the Tractatus,” reprinted in her The
Realistic Spirit: Wittgenstein, Philosophy, and the Mind. Cambridge: MIT Press.
Dummett, Michael. 1992. “The Metaphysics of Verificationism,” in L. E. Hahn (ed.), The Philosophy of
A. J. Ayer. La Salle: Open Court.
Edgington, Dorothy. 2004. “Two Kinds of Possibility,” in Proceedings of the Aristotelian Society Supp.
vol. 78.
Hacker, P. M. S. 1986. Insight and Illusion: Themes in the Philosophy of Wittgenstein, rev. ed. Oxford:
Oxford University Press.
Kant, Immanuel. 1933. Critique of Pure Reason, trans. Norman Kemp Smith. London: Macmillan.
Kant, Immanuel. 1999. Correspondence, trans. and ed. Arnulf Zweig. Cambridge: Cambridge
University Press.
Kripke, Saul. 1980. Naming and Necessity. Oxford: Basil Blackwell.
McDowell, John. 1994. Mind and World. Cambridge: Harvard University Press.
McGinn, Marie. 1999. “Between Metaphysics and Nonsense: Elucidation in Wittgenstein’s Tractatus,”
in Philosophical Quarterly 49.
Moore, A. W. 1985. “Transcendental Idealism in Wittgenstein, and Theories of Meaning,” in Philo-
sophical Quarterly 35.
Moore, A. W. 1997. Points of View. Oxford: Oxford University Press.
Moore, A. W. 2002. “What Are These Familiar Words Doing Here?”, in Anthony O’Hear (ed.), Logic,
Thought and Language. Cambridge: Cambridge University Press.
Moore, A. W. 2003. “Ineffability and Nonsense,” in Proceedings of the Aristotelian Society Supp. vol. 77.
Moore, A. W. 2007. “Wittgenstein and Transcendental Idealism,” in Guy Kahane, Edward Kanterian,
and Oskari Kuusela (eds.), Wittgenstein and His Interpreters: Essays in Memory of Gordon Baker.
Oxford: Basil Blackwell.

343
Moyal-Sharrock, Danièle. 2007. “The Good Sense of Nonsense: A Reading of Wittgenstein’s Tractatus
as Non-Self-Repudiating,” in Philosophy 82.
Putnam, Hilary. 1981. Reason, Truth, and History. Cambridge: Cambridge University Press.
Quine, W. V. 1961. “Two Dogmas of Empiricism,” reprinted in his From a Logical Point of View:
Logico-Philosophical Essays, rev. ed. Cambridge: Harvard University Press.
Quine, W. V. 1981. “On Austin’s Method,” reprinted in his Theories and Things Cambridge: Harvard
University Press.
Quine, W. V. 1986. “Reply to Jules Vuilleman,” in L. E. Hahn and P. A. Schilpp (eds.), The Philosophy of
W. V. Quine. La Salle: Open Court.
Quine, W. V. 1992. Pursuit of Truth, rev. ed. Cambridge: Harvard University Press.
Strawson, P. F. 1966. The Bounds of Sense: An Essay on Kant’s “Critique of Pure Reason.” London:
Methuen.
Stroud, Barry. 1984. The Significance of Philosophical Scepticism. Oxford: Oxford University Press.
Sullivan, Peter. 2003. “Ineffability and Nonsense,” in Proceedings of the Aristotelian Society Supp. vol. 77.
Williams, Bernard. 1981. “Wittgenstein and Idealism,” reprinted in his Moral Luck. Cambridge:
Cambridge University Press.
Williamson, Timothy. 1996. “Knowing and Asserting,” in Philosophical Review 105.
Wittgenstein, Ludwig. 1961. Tractatus Logico-Philosophicus, trans. D. F. Pears and B. F. McGuiness.
London: Routledge & Kegan Paul.
Wittgenstein, Ludwig. 1974a. Philosophical Grammar, ed. Rush Rhees and trans. Anthony Kenny.
Oxford: Basil Blackwell.
Wittgenstein, Ludwig. 1974b. Philosophical Investigations, trans. G. E. M. Anscombe, rev. ed. Oxford:
Basil Blackwell.

344