An A Priori Argument For Realism PDF

Journal of Philosophy, Inc.
An A Priori Argument for Realism

Author(s): Colin McGinn
Source: The Journal of Philosophy, Vol. 76, No. 3 (Mar., 1979), pp. 113-133
Published by: Journal of Philosophy, Inc.
Stable URL: https://www.jstor.org/stable/2025524
Accessed: 28-09-2019 12:47 UTC
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/2025524?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend
access to The Journal of Philosophy
This content downloaded from 131.220.249.56 on Sat, 28 Sep 2019 12:47:55 UTC
All use subject to https://about.jstor.org/terms
THE JOURNAL OF PHILOSOPHY
VOLUME LXXVI, NO. 3, MARCH 1979
C L = - -4:,.F F i+ C -; - i: :_:o
AN A PRIORI ARGUMENT FOR REALISM *
E) XCEPT in the vulgar sense, one is not a realist tout court:

one is a realist with respect to some or other type of subject
matter-or better, with respect to particular classes of state-
ments. Nor, similarly, is it feasible to be an unqualified anti-realist.
Nevertheless, specific realist and anti-realist theses are apt to ex-
hibit certain interdependencies. For example: realism about scien-
tifically posited theoretical entities is likely to go with realism about
macroscopic material bodies; realism about values will naturally
encompass both ethical and aesthetic values; realism about num-
bers may encourage a general acceptance of abstract objects; anti-
realism about the semantical and the mental may go hand in hand;
and so on. That is to say, particular philosophical arguments for
or against realism with respect to specific areas may call for parallel
conclusions in neighboring areas. Such interdependencies as those
cited are, however, of a relatively local and unsurprising kind. The
really interesting question to raise is how extensive such interde-
pendencies might be; for it may turn out that philosophical inves-
tigation of apparently disparate areas will disclose interdependencies
on a more global scale, and an appreciation of the lines of connec-
tion that define the scope and limits (if there be such) of global
realism and anti-realism may help in the resolution of the dispute
in particular areas. Thus we might inquire whether some preferred
formulation of scientific realism requires mathematical realism or
precludes it, whether insistence on the irreducibility of ethical or
* This paper was prompted by an aside of Michael Dummett's in "The Re-

ality of the Past," Proceedings of the Aristotelian Society, LXIX (1968/9): 239-
258, p. 250, and bears the mark of his discussions of realism throughout. I have
also been influenced by unpublished work of Marie McGinn and Christopher
Peacocke, and by conversations with Anita Avramides, W. D. Hart, and Arnold
Zuboff.
0022-362X/79/7603/0113$02.10 (D 1979 The Journal of Philosophy, Inc.
I I3
II4 THE JOURNAL OF PHILOSOPHY
aesthetic thought presupposes the reality of values, whether our

ordinary understanding of the world of spatiotemporally distrib-
uted bodies requires realism about causal modalities,' and so forth.
These are not, however, the questions I am going to discuss in
this paper; I mention them and the global question they instan-
tiate in order to locate in a wider perspective the question I want
to address. That question is this: Can we discover interdependen-
cies between realist and anti-realist conceptions of, on the one
hand, the external world of material bodies and, on the other, the
internal world of mental states and events-to put the matter ten-
dentiously-such that we find ourselves compelled to adopt certain
realist or anti-realist positions with respect to these two areas? I am
going to argue that we can indeed expose such interconnections
and that they force us to recognize that realism about both kinds
of statement is the only viable position. The general structure and
strategy of the argument is as follows. In respect of our two classes
of statements-which I shall for the sake of brevity hereafter des-
ignate as M- and P-statements, respectively-there are four possible
combinations of view: (i) anti-realism about both M- and P-state-
ments, (ii) anti-realism about M-statements combined with realism
about P-statements, (iii) realism about M-statements combined with
anti-realism about P-statements, (iv) realism about both M- and
P-statements. I aim to show that (iv) is true by eliminating (i)-(iii)
by appeal to global considerations designed to uncover the incon-
sistency of those positions. Since (i)-(iv) exhaust the alternatives
and all save (iv) are inconsistent, we must acknowledge that (iv) is
the only available position. The argument is thus indirect, like a
proof by reductio, in that we set out to establish joint realism by
supposing its negation, this giving three possibilities, each of which
is claimed to embody a sort of inconsistency. No direct argument
for realism in either area is essayed (though I think we shall see
that the eventual upshot has a strong intuitive appeal): rather,
certain interconnections are claimed and systematically exploited.
The result is (intended as) an a priori argument for realism about
M- and P-statements.
To proceed we need a preliminary formulation-to be refined later

-of the general import of realist and anti-realist doctrines. Follow-
ing Dummett, I shall take it that realism and anti-realism are best
1 As argued by Peacocke in "Causal Modalities and Realism," forthcoming

in Mark Platts, ed., Reference, Truth and Reality (Boston: Routledge & Kegan
Paul, 1979).
AN A PRIORI ARGUMENT FOR REALISM I 15
understood as contrary theses about the relation, with respect to

various subject matters, between truth and the recognition of truth.2
That is, what distinguishes a realist from an anti-realist attitude
toward the interpretation of a given class of statements is a differ-
ence in one's conception of how that in virtue of which the state-
ments are true or false relates to the means or route by which we
come to know their truth value. In yet other words, what defines
realism in contradistinction to anti-realism is a disagreement as to
the relation between truth conditions and assertibility or verifica-
tion conditions. The heart of a realist view of a given class of
statements is that their truth conditions in a certain sense tran-
scend-and so cannot be reduced to-their assertibility conditions.
The anti-realist view, correspondingly, is that the subject matter
of the given class does not thus transcend the grounds upon which
the statements are asserted. Notice that, under that general formu-
lation, realism and anti-realism, which are, strictly speaking, meta-
physical or semantic theses, are already characterized in epistemo-
logical terms: for the content of realism is precisely that there
obtain, or could obtain, recognition-transcendent facts, whereas
that of anti-realism is that there could not. As will emerge, this
epistemological formulation will play a crucial role in the argu-
ment I shall present. What I want now to register is that the state-
ment of realism in terms of recognition-transcendence has the
important consequence that a realist interpretation of a class of
sentences inevitably introduces the possibility of a skeptical chal-
lenge concerning our knowledge of the propositions thereby ex-
pressed, whereas an anti-realist interpretation evades that challenge.
Indeed, I think that a prima facie vulnerability to such a challenge
should be regarded as a condition of adequacy which any formula-
tion of realism is required to meet; and anti-realisms should cor-
respondingly be seen manifestly to foreclose the threat of skepti-
cism. In respect of M- and P-statements, this general formulation
may be specialized as follows: realism about M-statements is the
thesis that the truth conditions of these statements transcend the
experiential grounds on which they are asserted, whereas anti-
realism about M-statements denies this; and realism about P-state-
ments is the thesis that the truth conditions of these statements
transcend the behavioral grounds on which they are asserted,
2 See, for example, Dummett, "What Is a Theory of Meaning? (II)," in

G. Evans and J. H. McDowell, eds., Truth and Meaning (Oxford: Clarendon
Press, 1976).
i i6 THE JOURNAL OF PHILOSOPHY
whereas anti-realism about P-statements denies this.3 The realist

theses thus imply some sort of recognition-transcendence with re-
spect to the truth of the statements in question, and it is precisely
because of this epistemic gap between evidence and truth that the
traditional skeptical problems as to the existence and nature of the
external world and of other minds arise. Let us now, equipped
with these rough preliminaries, turn to a criticism of the first of
the combined positions I identified, namely joint anti-realism.
II
The anti-realist doctrines I wish to consider may be labeled-I take

it with some historical precedent-phenomenalism and behavior-
ism. I shall understand the form of these doctrines to be defined
by the thesis that M- and P-statements have the truth value they
do have in virtue of the truth or falsity of statements drawn from
certain other classes, not trivially different from the given ones:
M- and P-statements are thus said to be subject to a reductive thesis.
A reductive thesis is to this effect: a sentence s of a given class K
is reducible to (true in virtue of) some sentence s' of a class R if
and only if necessarily s is true (false) just in case s' is true (false):
it is a logically necessary and sufficient condition for a sentence of
K to be true (false) that some sentence (or set of sentences) of R be
true (false).4 Instantiating for M- and P-statements, the reductive
anti-realist theses under consideration claim that these statements
are true in virtue of statements about experiences (E-statements)
or statements about behavior (B-statements). We may gloss the re-
ductive thesis definitive of phenomenalism as the claim that any
language, such as our own, which contains both M- and P-state-
ments exhibits a hierarchical ordering with respect to the relation
of reducibility phenomenalism defines over its member statements.
The basal statements of the hierarchy are the E-statements, a sub-
class of P-statements, and the distribution of truth values over
these determines the truth value of any M-statement in the lan-
guage (if it has one): the truth values of the M-statements cannot
3 In saying that M-statements are asserted upon experiential grounds and

that P-statements are asserted upon behavioral grounds I do not deny that
material objects (or indeed others' mental states) may be directly perceived.
Direct perception is compatible with the beliefs thus formed being inferential
in some sense: the significant point is that we shall be equally prone to form-
ing M- and P-beliefs in nonveridical cases as we are in veridical cases.
4A reductive relation between sentences may, for certain reductionist theses,
hold a posteriori. Since anti-realism is a thesis about meaning, however, the
reductions it advocates are better viewed as semantical and therefore knowable
a priori; hence my use of the traditional phrase 'logically necessary and
sufficient'.
AN A PRIORI ARGUMENT FOR REALISM 117
vary if the assignment of truth values to the E-statements stays

fixed. Similarly, behaviorism has it that the basal statements are
uniformly M-statements, where any P-statement of the language is
rendered true or false by some B-statement, these comprising a
subclass of the M-statements; again, the truth values of the B-state-
ments once determined, the truth values of the P-statements are
thereby fixed. What distinguishes phenomenalism from behavior-
ism is thus a selection of distinct kinds of statement as reductively
basic, i.e., a characteristic ordering of statements according to a
particular reductive thesis. (The orderings preferred by the two
doctrines often correlate with some relation of epistemological
priority, but we need not go into this aspect of the doctrines now.)
For both anti-realisms, then, the important point is that the truth
of a statement of the given class consists in nothing other than the
truth of some statement of the reducing class.
Before indicating why it is that these positions are not jointly
occupiable, it will be useful to distinguish three ways in which the
reducibility relation defined over the relevant classes of statements
may be understood. First, we might say that for any statement s
of the given class (M- or P-statements) there is some statement s'
of the reducing class (E- or B-statements) such that s reduces to s'
and s is determinately either true or false; this reduction may or
may not be translational, but it does (or would) preserve the law
of bivalence for statements of the given class. Second, we may say
that whenever a statement of the given class has a determinate
truth value this is conferred upon it by some statement of the re-
ducing class, but there is no presumption or guarantee that the
former statements will be subject to bivalence: it may be that
neither s nor its negation is made true by some reducing statement
S'.5 Third, it may be alleged that a reduction with this latter prop-
erty obtains but that the proper response to it is that the sentences
that suffer thus from truth-value gaps should be regarded as ex-
pressing no proposition, as strictly meaningless, so that bivalence
is in effect respected. Which of these three positions one adopts
will obviously depend upon the precise character of the anti-realism
in question; but a main consideration here is what logical type of
statement is assigned to the putative reducing class. Classically,
phenomenalism and behaviorism have wished to include counter-
factual subjunctive conditionals among their reduction statements,
5 This is Dummett's preferred formulation of anti-realism: see the works of

his elsewhere cited in this paper.
i i8 THE JOURNAL OF PHILOSOPHY
principally because of translational ambitions; for statements re-

porting the occurrence of actual sense experiences scarcely suffice
to confer upon every M-statement a determinate truth value, and
similarly for actual behavior and P-statements. There are a number
of (by now) well-worn objections to such reductions which seem to
me very powerful, but which need not be evaluated in detail at the
level of generality set by this paper. For the record: it is hard to
see how either the phenomenalist or the behaviorist reductions can
succeed without tacit circularity, since the consequent of the sub-
junctive will not hold unless its antecedent introduces conditions
of the very sort claimed to be reducible-material-object conditions
or mental conditions, as the case may be; 6 and, connectedly, one's
strong conviction is that the proffered subjunctives are not barely
true, in Dummett's sense, but rather are true precisely in virtue
of (in part) the categorical M- and P-statements they are designed
to reduce.8 Impressed by such objections, one might prefer to con-
fine the reducing class to suitable categoricals recording the occur-
rence (possibly for all times) of actual experiences and episodes of
behavior, and then face the consequence that vastly many M- and
P-statements will have no determinate truth value, thus taking up
one or other of the second and third positions I distinguished. But,
whichever option one takes, a claim of reductive ordering will be
made, and that is my chief concern here.
Once the general form of phenomenalistic and behavioristic anti-
realism is clearly set out, as above, it is, I think, pretty evident why
it is that the doctrines cannot be jointly affirmed. The reason is
simply that they offer competing proposals as to what statements
comprise the basal truths: phenomenalism takes E-statements, a
subclass of P-statements, as basic, while behaviorism takes B-state-
ments, a subclass of M-statements, as basic. The result is that, where
one reductive thesis represents a statement as not itself requiring
6 Gilbert H. Harman explains the point tersely in Thought (Princeton, N.J.:
University Press, 1973), p. 10 f, and Peacocke makes much of it in Holistic Ex-
planation, hitherto unpublished.
7 The notion is fully explained in "What Is a Theory of Meaning? (II),"
sec. iII; roughly, it is the property of not being made true by some other (more
basic) statement.
8An interesting asymmetry of attitude is worth remarking here. Told that
material objects are "permanent possibilities of experience," one readily suspects
the claim to have dispensed with the objects, since it is extremely plausible
that it is precisely their independent existence that sustains such possibilities;
but the parallel claim for mental entities-they are "permanent possibilities
of behavior," i.e., behavioral dispositions-has not been apt to provoke a paral-
lel response. One would like to see a good reason for taking the cases differ-
ently, however.
AN A PRIORI ARGUMENT FOR REALISM II9
the application of a reductive operation, the other insists that such

a reduction be performed. Since a statement that is basal for one
anti-realism is derivative for the other, it is plain enough that a
vicious regress is generated by the conjunction of the two doctrines;
and this, of course, effectively frustrates the reductive ambitions
definitive of each anti-realist thesis. More explicitly, suppose we
take a certain M-statement s. Then phenomenalism will deliver as
its reducing statement some E-truth s'. But, since s' is a P-statement,
behaviorism offers up some B-statement s" as its reduction. Now s"
is itself an M-statement; so it demands from phenomalism some
further E-statement s"' to reduce it. And the cycle begins again,
ad infinitum. So the two doctrines simply contradict each other on
the crucial question of what statements make what other statements
true." It follows that position (i) is unoccupiable. And this already
suggests a limitation on any would-be global anti-realism; it ap-
pears that there is going to be something irreducibly realist in our
language and system of the world.
III
I imagine that the elimination of (i) will be fairly readily conceded:

few philosophers would think idealism and (behavioristic) material-
ism compatible, and the reasons traditionally prompting the two
views have not been such as to establish one of them if and only
if they establish the other. What is not at all immediately apparent,
however, is that (ii) and (iii) exhibit any hint of self-destructive
internal tension. To appreciate why it is that, as I claim, realism
about one area requires realism about the other, let us articulate
further the content of the realist interpretations of M- and P-
statements.
The fundamental thesis of realism about the two classes of state-
ments is captured in the notion of independence, i.e., the denial
that the truth of suitable E- and B-statements constitutes logically
9 Rudolf Carnap in The Logical Structure of the World, R. A. George, trans.

(Berkeley: University of California Press, 1967), remarks en passant (in sec. 57)
that it is possible to reduce all physical objects to psychological ones and also
possible to reduce all psychological objects to physical ones. Since he does not
confront the question which of these reductions is correct, the significance of
his constructions is hard to assess. If my argument is right, neither can be
correct. Nelson Goodman, too, in The Structure of Appearance (Indianapolis:
Bobbs-Merrill, 1966), p. 136 f, raises the topic of the competing claims of
"phenomenalistic and physicalistic systems," but eschews the metaphysical ques-
tion of which of them correctly characterizes the general nature of the world,
and so again their philosophical significance remains uncertain. No doubt the
pragmatic tone evinced by both writers reflects a familiar antipathy toward the
metaphysical question of which, if any, is actually true.
I20 THE JOURNAL OF PHILOSOPHY
necessary and sufficient conditions for the truth of corresponding

M- and P-statements, and similarly for falsity.'0 In other words,
M-facts are not reducible to E-facts, and P-facts are not reducible
to B-facts. The general notion of independence is reflected in a
cluster of assumptions and practices which we customarily fall in
with as uncritical realists and which would have to be abandoned,
or at least radically re-interpreted, if the anti-realist were right.
Thus the realist has it that it is possible for M-facts to obtain and
no experiences be had as of their obtaining, so it is not a necessary
condition for an M-statement to be true that some corresponding
E-statement be true; and further, since no set of purely experiential
statements ever logically entails the truth of some M-statement, it
is not a sufficient condition for an M-statement to be true that some
corresponding E-statement be true. Similarly, in respect of P-state-
ments the realist view is that it is possible for P-facts to obtain in
a person and he not behave in some manifesting way, so it is not
necessary for the truth of a P-statement that some appropriate B-
statement be true; and further, since no set of purely behavioral
statements ever logically entails the truth of some P-statement, it
is not sufficient for a P-statement to be true that some correspond-
ing B-statement be true. And if the truth of the relevant statements
can come thus apart, there can be no possibility of reduction, of
maintaining that one kind of statement is true in virtue of the
other. (The being of an M-fact does not consist in its being per-
ceived, and the being of a P-fact does not consist in its being be-
haved.) And it should be clear enough that this property of two-
way logical independence from experiences and behavior on the
part of M- and P-statements is precisely the source of the skeptical
questions realism is prone to invite, on account of the alleged
epistemic transcendence of subject matter (hence no necessity) and
the correlative possibility of error (hence no sufficiency).
As a natural corollary of the independence thesis your typical
realist will hold to a certain conception of perception and action:
viz., that a genuinely perceptual experience is caused by some ex-
ternal object ontologically distinct from the experience, and sim-
ilarly that a piece of intentional behavior is caused by a mental
10 Carnap formulates realism about material objects in the phrase "inde-

pendence from the cognizing consciousness," op. cit. p. 281, and G. E. Moore
employs the same style of formulation in "A Defence of Common Sense," sec.
II, Philosophical Papers (Reading, Mass.: Allen & Unwin, 1959). Hilary Putnam's
realistic view of the mental is similarly stated in "Brains and Behaviour," re-
printed in Mind, Language and Reality (New York: Cambridge, 1975).
AN A PRIORI ARGUMENT FOR REALISM I2 I
state or event from which it is ontologically distinct. Those are

peculiarly realist theses because the causality claim implies the
distinctness of the objects of perception from the perception itself,
and of the mental antecedents of action from the caused action.
Indeed, appropriate causation is plausibly regarded as criterial for
whether an experience is a genuine perception or a piece of be-
havior a genuine action: these very distinctions are thus commonly
drawn in terms of realist materials."- Connected with this causality
claim, and reinforcing the irreducibility consequent upon logical
independence, is the activity of (causally) explaining the truth of
certain E-statements by citing appropriate M-statements, as when
we explain why a perceptual experience occurred and had the
intentional-phenomenal character it had by saying that it was
brought about by some external material object being thus and so;
and similarly we have the practice of explaining the occurrence
and properties of certain bodily movements by reference to pre-
sumed internal mental states and events. If the anti-realist views
were correct, and the truth of M-statements just consisted in the
truth of corresponding E-statements, while the truth of P-state-
ments similarly reduced to the truth of corresponding B-statements,
then it would be hard to see how statements of the former kinds
could possibly explain statements of the latter kinds. So it appears
that the ascription to M- and P-facts of such a causal-explanatory
role vis-A-vis E- and B-facts is bound up with a realist conception
of their status in the world.
If those are some characteristically realist contentions about M-
and P-statements, why should they imply the inconsistency of posi-
tions (ii) and (iii)? The first stage of my answer rests upon a simple
observation. Consider M-statements first. I said that under a realist
interpretation E-statements are neither necessary nor sufficient to
fix their truth value; but if so, we can equally well say that M-
statements are neither necessary nor sufficient for E-statements-the
independence is symmetrical. To say that an M-fact is not necessary
for an E-fact is to say that an E-fact is not sufficient for an M-fact
(delusive experiences), and to say that an M-fact is not sufficient
for an E-fact is to say that an E-fact is not necessary for an M-fact
(unexperienced M-facts). The independence cuts both ways, and
11 Cf. H. P. Grice, "The Causal Theory of Perception," Proceedings of the

Aristotelian Society, supp. vol. 35 (1961), and D. Davidson, "Agency," in R.
Binkley, ed., Agent, Action and Reason (Toronto: University Press, 1971). I do
not say that these authors explicitly view causal theories of perception and
action as peculiarly realist.
implies realism as much one way as the other. Similarly for P-state-
ments. To say that a P-fact is not necessary for a B-fact is to say
that a B-fact is not sufficient for a P-fact (deceptive behavior), and
to say that a P-fact is not sufficient for a B-fact is to say that a
B-fact is not necessary for a P-fact (unbehaved P-facts). In formu-
lating the thesis that the truth of an M-statement does not consist
in the truth of a corresponding E-statement we find ourselves say-
ing-what is anyway hard to deny-that the E-statement is not true
in virtue of the M-statement; and similarly for P- and B-statements.
(Notice that this observation does not crucially depend upon the
actual or possible occurrence of delusive experience or deceptive
behavior, though I do think these are important in understanding
the epistemological corollaries of realism; for no one would main-
tain that, when an experience is a genuine perception or an episode
of behavior has a mental description, the truth of the correspond-
ing E- and B-statements just reduces to the truth of the statements
that report, as the realist asserts, their M- and P-causes.) So it begins
to seem that realism about M- and P-statements implies realism
about E- and B-statements, under the independence formulation.
But now E- and B-facts are just subclasses of P- and M-facts, re-
spectively; and if we are prepared to admit these in unreduced
realist fashion, there can be no objection of general principle to
admitting the rest.12 The same result issues from the causal-explan-
atory formulations of the two realisms: if material bodies must be
distinct from the experiences they cause in episodes of perception,
then the experiences are symmetrically distinct from the bodies;
and if mental states and events must be distinct from the behavior
they cause in events of intentional action, then the behavior is
symmetrically distinct from the mental antecedents-the effects
must be as real, by this standard, as the causes. And parallel re-
marks apply to the explanatory relation, as the realist construes it,
between M- and E-statements and P- and B-statements: explanandum
cannot reduce to explanans. In short, the formulation of each re-
alism in terms of independence seems, on the face of it, to imply
an equally realist interpretation of the statements that comprise
the assertibility conditions of our given statements.
This argument will certainly seem too swift; for, as stated, it
12 Cf. Gottlob Frege's attempted refutation of idealism in "The Thought,"'

reprinted in P. F. Strawson, ed., Philosophical Logic (New York: Oxford, 1967),
where he argues that there must be at least one object that isn't an idea, viz.,
the bearer of ideas, and so opens the door to the reality of all objects of
thought.
AN A PRIORI ARGUMENT FOR REALISM I23
ignores a crucial maneuver open to the partial realist. Consider

again the realist who desires his basal statements to be uniformly
material; i.e., he wishes to be a behaviorist and a realist about the
external world. His response to the foregoing observations will be
as follows: begin by formulating realism about M-statements in
terms of independence, causality, and explanation, thus introduc-
ing the required E-statements; then claim, anti-realistically in re-
spect of such statements, that they are to be construed as true in
virtue of appropriate B-statements-so that the E-statements needed
to formulate realism about M-statements at large are in turn re-
duced to a subclass of M-statements. Analogously, the realist about
P-statements who wishes his basal statements to be uniformly psy-
chological will initially state his mental realism in terms of inde-
pendence from B-statements, but then go on to subject the intro-
duced B-statements to a phenomenalistic reduction, thus rendering
them true in virtue of appropriate E-statements. The first philos-
opher is a behavioristic realist about the external world, the second
a phenomenalistic realist about the internal world. Are these not
perfectly consistent positions? I think that ultimately they are not,
but the reason is somewhat subtle.
It is tempting to suppose that these reductive reformulations of
the realist's characteristic claims are vulnerable to a regress argu-
ment, along the following lines. Suppose the realist about M-state-
ments and anti-realist about P-statements were to present us with
his reductive B-statements. Then we would be entitled to put to
him the question, in what his realism about these B-statements
consists. To be constant he is obliged to offer an independence
formulation, thus introducing a new range of as yet unreduced
E-statements. He now proposes to reduce these to further B-state-
ments, and inevitably invites a repetition of the question. Evidently
this process of question and answer generates a chain of alternating
E- and B-statements with no determinate upper bound. Again, sup-
pose the realist about P-statements and anti-realist about M-state-
ments were to present us with his reductive E-statements. Then we
can ask him in what his realism about these E-statements consists.
Constancy requires an answer framed in terms of independence
from further B-statements. These would then in turn call for phe-
nomenalist reduction, and the cycle recurs, thus generating an
indefinitely extended series of B- and E-statements. However, the
resulting series, though in a certain sense regressive, do not seem
viciously so, since there is nothing in the general position of the
differential realist to prevent him, at any arbitrary stage in the

series, from producing the reductive statement demanded by his
anti-realism as that in virtue of which the E- or B-statement yielded
by his realist half is true. The case is unlike that of the joint anti-
realist, because there we were presented with incompatible claims
as to what constitutes reductive bedrock; but in the case of posi-
tions (ii) and (iii) the realist component need not, for all that has
so far been said, insist upon the irreducibility of the statements
from which independence is alleged. I think, however, that the
intuition that encourages one erroneously to suspect straightfor-
ward vicious regress here does have considerable force. Part of its
force can be brought out by means of the following argument
against the proposed reformulations.
It was remarked earlier that realism implies a gulf between truth
conditions and verification conditions-between truth and the rec-
ognition of truth-and that this gulf is what allows skepticism to
get purchase on the area in question. In respect of the external
world, this involves the idea that how experience represents the
world as being may not coincide with how it really is-appearance
and reality may fail to match. In respect of the mental, realism
implies that what a true P-statement corresponds to transcends, and
so may diverge from, the behavioral evidence on the strength of
which we make psychological judgments. Thus it is that skepticism
about the external world and other minds arises. Now our question
must be: do the suggested reformulations of realism adequately
preserve those epistemological corollaries? For if they do not, they
fail to capture the content of the intended realist theses. The quick-
est way to see that they fail in this is to negate the realist views as
thus reformulated and then test whether the resulting propositions
fulfill our conditions on an adequate formulation of anti-realism
for the two areas. In the case of M-statements, then, the reformu-
lated anti-realism would be to the effect (strange as it is) that state-
ments about external material bodies reduce to statements about
behavior-a sort of behavioristic phenomenalism. For P-statements
the reformulated anti-realism about them would be that they are
true in virtue of statements reporting experiences as of behaving
in a certain way-a sort of phenomenalistic behaviorism. Just as
we are advised under the reformulations to state realism about each
area in a uniform vocabulary, so we state the corresponding anti-
realisms in that very vocabulary. But clearly these formulations of
the two anti-realisms do not have the epistemological consequences
AN A PRIORI ARGUMENT FOR REALISM I.25
we expect of anti-realism about material objects and about mental

states. For, first, anti-realism about M-statements should have the
consequence, enjoyed by standard phenomenalism, that our knowl-
edge of the external world has a firm and skepticism-free char-
acterization; but behavioral facts are in no way epistemologically
privileged among M-facts at large, and so cannot afford the episte-
mologically unproblematic foundation that anti-realism character-
istically promises us. And, second, anti-realism about P-statements
should have the consequence that our knowledge of other minds
is innocent of illicit inference precisely because P-statements are
seen, upon reductive analysis, not to go beyond the publicly acces-
sible facts of behavior; but if behavioral facts are reduced to ex-
periential facts no epistemological progress of the sort desired by
the anti-realist about the mental is made, since these are just as
private and inaccessible as the P-statements they are intended to
reduce. So negating the proposed reformulations of realism about
M- and P-statements does not lead to an adequate and consistent
statement of anti-realism for the two domains. Unsurprisingly, the
suggested reformulations also misrepresent the epistemological pre-
dicament of the realist about each area: for if we permit ourselves
unproblematic epistemological access to behavioral facts as the sur-
rogates for statements about how things appear in immediate ex-
perience, then we must already have resolved the question of our
knowledge of the external world-or else we simply deprive our-
selves of the materials with which to raise that question. On the
other hand, if we try to construe all B-facts as ultimately experien-
tial, then we fail to capture the idea, essential to a realist view of
the mental, that P-facts transcend and are distinct from the publicly
accessible facts of overt behavior: for, under such a reduction, the
assertibility conditions for P-statements turn out to be themselves
as private and subjective as the statements whose truth they evi-
dence. We cannot really claim to retain the distinctive features of
the realist conception of the relation between mental states and
behavior if we insist upon reducing the latter to suitable experi-
ences; room must be made for the public. The underlying point in
both cases can be put as follows. According to realism about the
external world, M-facts transcend their assertibility conditions in
such a way that information about the totality of E-truths does not
decisively settle what M-statements are true, or indeed whether any
are; someone possessed of this information can still ask skeptical
questions concerning M-statements. Similarly, knowledge of the
126 THE JOURNAL OF PHILOSOPHY
totality of B-truths should not preempt the skeptical question

whether there really are any mental states responsible for the ob-
served behavior. Even an ideal observer, in something like the
position of God, could raise these skeptical questions given only
information about assertibility conditions. But, if we reduce E-facts
to B-facts, this no longer holds, since knowledge of E-facts will just
consist in knowledge of appropriate B-facts, and so will presuppose
unproblematic access to a range of M-facts. Similarly, if our ideal
observer were in receipt of all the B-facts and these reduced to
corresponding E-facts, then he would already have settled the ques-
tion of whether the world contains mental states over and above
material objects. So neither reduction would have the required
consequence that the truth conditions of M- and P-statements gen-
uinely transcend their usual assertibility conditions; neither, there-
fore, gives an adequate characterization of that in which the reality
of M- and P-facts consists. The simple truth is-and I shall have
more to say on this later-that a proper statement of realism about
material bodies requires unreduced acceptance of experiences as
that from which the external world is allegedly independent, and
that a proper understanding of realism about the mental requires
unreduced acceptance of episodes of behavior as that from which
the internal world is independent. This point is just made more
vivid by tracing out the epistemological consequences required of
an adequate formulation of realism for the two areas. So I think
that in the end our initial crude argument from independence
must be accepted and positions (ii) and (iii) declared unoccupiable.
But first a certain line of objection to the whole set-up must be
disposed of.'3
IV
Anxious to avoid compulsory occupation of position (iv), someone

might try to question the very idea of formulating realism in terms
13 I should perhaps make it explicit that this paper does not, officially at any
rate, address the question of solipsism. Solipsism is not realist about P-state-
ments, in my sense, because it regards non-first-person ascriptions (what Carnap
calls the "heteropsychological") anti-realistically-other minds are logical con-
structions out of my experiences (the "autopsychological")-and thus it takes
all statements to be true in virtue of first-person P-statements. On the con-
trary, I assume the existence of a plurality of persons and ask after the relation
between their experiences and the material world, and their behavior and
mental states. You might helpfully conceive of the issue as a very general
question of radical interpretation: given that a speaker's language contains
M- and P-statements, what metaphysical schemes of interpretation are possible,
as represented by positions (i)-(iv)? My thesis is, then, that only scheme (iv) is
consistent. Refuting solipsism would demand further considerations.
of independence from E- and B-statements. The hope would be to

formulate realism for M- and P-statements in such a way that no
symmetrical condition on E- and B-statements results. The claim
I now wish to defend is that such alternative conceptions of realism
for the two areas as I can produce fail systematically either to pro-
vide necessary or sufficient conditions for an intuitive formulation
of realism.
A first proposal for doing away with explicit allusion to E- and
B-statements for a realist interpretation of a class of statements
might require that the class meet the following two conditions:
(a) the statements obey the law of bivalence, and (b) there is no
class of statements, genuinely distinct from the given one, such that
statements of the given class are true in virtue of statements of that
other class, i.e., statements of the given class are barely true. There
are two objections to this formulation of realism. First, it fails our
requirement that an adequate formulation display the distinctive
epistemological properties of the statements in question, in partic-
ular, the way in which the gap between truth and recognition of
truth is apt to invite skeptical challenge. This failure is especially
apparent if one tries to obtain a characterization of the intended
anti-realisms by negating the realist formulation: the operation
does not yield the specific doctrines of phenomenalism and behav-
iorism. Only if we instantiate the existentially general condition
(b) with experiential and behavioral statements do we get the right
results, but of course this just takes us back to my initial argument.
But second, failure of the given class to meet the condition of bare
truth is not sufficient to indicate an anti-realist interpretation. For
consider physicalism with respect to mental states, and elementary-
particle micro-reduction with respect to macroscopic material bodies.
Whether true or false, such theses are intuitively realist about P-
and M-statements; not all reductionism is anti-realist.14 These doc-
trines are intuitively realist, I would say, because they do not deny
a cleavage between truth and the recognition of truth. Unlike phe-
nomenalism and behaviorism, they do not embody a purported
reduction to assertibility conditions. So, because bare truth is not
a necessary condition for a realist interpretation, independence
from all other classes of statements is too strong a requirement for
such an interpretation.
In reply to this objection someone might propose weakening the
14 As Dummett himself insists in "The Reality of the Past," op. cit., and
in "Common Sense and Physics," forthcoming in a Festschrift for A. J. Ayer.
128 THE JOURNAL OF PHILOSOPHY
formulation by omitting condition (b) altogether. There would still

be the objection from epistemology, but there would now be the
problem that the resulting formulation is too weak. That mere
conformity to bivalence is insufficient for realism is shown by re-
calling the first and third of the anti-realist doctrines I distin-
guished in section II. Surely classical reductionist phenomenalism
and behaviorism, with their invocation of subjunctive conditionals,
count as anti-realist, yet both seek to retain bivalence; and the
positivistic third view a fortiori rates as anti-realist. (It seems to
me, incidentally, that these positions show Dummett's rather mono-
lithic formulation of realism-across-the-board in terms of conformity
to bivalence to be deficient; 15 one needs to say something specific
about the relation of the statements in question to their distinctive
assertibility conditions.) And if we ask in what the contrast between
such doctrines and the realist reductions I just mentioned consists,
I think the answer must be that the former deny, whereas the latter
do not, the features of logical independence, causal relations, and
explanatory role that I earlier set forth as definitive of a realist
view of M- and P-statements. This leads me to conclude that my
initial formulation in terms of these features uniquely captures the
content of the realist conception of the two types of statement. But
if the argument that starts from that style of formulation is sound,
we are left with global realism as the only really viable position.
v
Realism and anti-realism about M- and P-statements are theses

about the kind of meaning possessed by these statements, i.e., about
what their truth conditions consist in. According to realism, their
truth conditions are such as to transcend the conditions that we
recognize as verifying or falsifying the statements in question; this,
to repeat, is precisely why skepticism about the external world and
other minds seems in order. It is thus clear that the task of defend-
ing realism about a class of statements is an enterprise quite dis-
tinct from rebutting skepticism concerning that class: one is not
yet in the business of answering the skeptic unless one has already
assumed a realist interpretation of the statements. That these are
distinct questions is apt to be disguised by the fact that anti-realism
15 See the articles by him I have cited. I think, in fact, that the formulation
of realism in terms of irreducibility to assertibility conditions lies behind the
bivalence formulation: it is just that there seems no necessity for the asserti-
bility conditions of a given class of statements to be incomplete with respect
to the assertion of each statement of the given class-complete assertibility is
quite compatible with anti-realism.
is at one stroke a theory of meaning and an account of how state-

ments endowed with a meaning so characterized are known. Now
I have claimed to establish realism about M- and P-statements on
the basis of an a priori argument, by an appeal to definition in-
deed. This goes against the grain of the currently dominant view
sometimes labeled "empirical realism," this being some such thesis
as that metaphysical realism should be grounded upon much the
same considerations as realism about theoretical entities, or indeed
a realistic view of particular kinds of theoretical entities, viz. some
sort of simplicity or inference to the best explanation.16 That is to
say, one should adopt something like an a posteriori scientific ap-
proach to the realism/anti-realism dispute. I strongly suspect that
such an attitude arises at least in part from a conflation of the
questions of realism and skepticism. It is perhaps to be expected
that our knowledge of the truth of realistically interpreted M- and
P-statements should be a posteriori knowledge and that resistance
to skepticism should therefore assume an empirical form. But the
question of realism with respect to a class of statements is not
whether and how we might justify our claim to know the state-
ments to be true (though to establish it would presumably be to
show that we do know they have verification-transcendent mean-
ing); and, given this conception of the enterprise, it is also not very
surprising that it should be capable of a priori demonstration. (I
do not mean to deny that there might be an a posteriori defense
of metaphysical realism; I am only insisting upon a careful separa-
tion of questions.) Moreover, I think one's strong conviction is that
the falsity of phenomenalism and behaviorism, as theories about
the meaning of M- and P-statements, is of a much deeper and more
conceptual character than the standard talk of simplicity and the
like would seem to suggest. It does not appear to have the status
of a mere empirical fact, albeit a highly general one, that M-state-
ments are not true in virtue of E-statements and that P-statements
are not true in virtue of B-statements. If that is right, it seems pref-
erable to have a way of establishing it which adequately reflects its
intuitive status in our general conception of the world.
VI
Dummett has advanced a perfectly general argument, based upon

what is involved in understanding a language, designed to under-
16 See, for example, Putnam's discussion in "Other Minds," op. cit., and
J. L. Mackie in "What's Really Wrong with Phenomenalism?," British Academy
Lecture, 1969. The view is well expounded (though not endorsed) in M. Wil-
liams, Groundless Belief (Oxford: Basil Blackwell, 1977), chap. 4.
mine realism about the meaning possessed by certain sentences of

natural language.17 The purported upshot of the argument, which
will not be repeated here, is that truth cannot, as realism requires,
transcend the recognition of truth. Dummett suggests instead that
we might do better to adopt a verificationist or assertibility-condi-
tions theory of meaning, thus relinquishing (as he thinks) realism
about the subject matter of the relevant statements. We already
saw in section II that there must be some limit on such a general
argument, since the anti-realisms to which it leads are jointly in-
compatible. But if the thesis of the present paper is correct, then,
at least in respect of M- and P-statements, a realist interpretation
of their subject matter is obligatory, and so a verificationist theory
of meaning according to which the content of these statements
reduces to the content of some suitable set of E- and B-statements
cannot be acceptable. So I think that the considerations here ad-
duced constitute a direct demonstration, not predicated on any
particular theory of language mastery, that some sort of truth-con-
ditions theory, as opposed to a Dummettian verification-conditions
theory, has to be right.
Our considerations also bear upon another theme of Dummett's.
He insists that the semantic theorist produce some account of how
his favored meaning assignments to sentences are manifested in the
use a speaker makes of his sentences. This translates in the present
case into a demand to specify in what way the phenomenalist or
behaviorist differs from the realist in his employment of M- and
P-sentences. What we have already said about what makes a person
a realist suggests the outlines of an answer. Aside from commitment
to bivalence for M- and P-sentences, which is plausibly necessary
but not (I have suggested) sufficient for realism, the realist may be
said to differ from the anti-realist precisely in his acknowledging
the logical independence I have made so much of-he will assent
to E- and B-statements without automatically assenting to the cor-
responding M- and P-statements, and vice versa-and he will em-
ploy the latter types of statement in a practice of explaining the
former, as well as accept causal statements of the kinds I identified
earlier. Given knowledge of such linguistic dispositions, the radical
interpreter will be empirically warranted in construing his speaker
as a realist about M- and P-statements. Taken together, these two
points-the direct metaphysical argument for realism and the pro-
17 See especially "What Is a Theory of Meaning? (II)," op. cit. I have criticized
the argument in "Truth and Use," forthcoming in the collection cited in foot-
note 1; that paper and the present one may be read in tandem.
posed manifestation conditions for a realist semantics-seem to me

to add up to a way of confronting and deflecting Dummett's gen-
eral line of argument.
VII
A discursive argument unbacked by an intuitive picture seldom

convinces. I hasten in conclusion, then, to place the argument of
this paper in some sort of over-all perspective; what follow are
some more or less impressionistic gestures in that direction. Our
comprehensive realist conception of empirical reality is of an ob-
jective spatiotemporal world whose intrinsic nature is independent
of the local and relative peculiarities of the conscious beings that
form a part of it.18 We think of objective reality as causing changes
in the course of experience undergone by these beings, notably in
perception; and of changes in objective reality as being occasionally
wrought by the actions and movements of these beings, as a causal
result of mental states and events within them. Neither sector of
reality-external or internal-is a closed system; they interact in
various ways. Our picture of the world is in this way fundamentally
dualistic (which is not to say Cartesian). This dualism exhibits a
certain categorial difference which may be characterized as follows.
The external world of material bodies and events has the charac-
teristic of objectivity; i.e., it is to be conceived in an absolute way,
as not owing its intrinsic nature to the relative and subjective
sensory modalities and conscious experience of the sentient beings
that inhabit it. To conceive it aright, therefore, one needs somehow
to prescind from one's subjective and local standpoint and aspire
to what might be called an "absolute" conception of the objective.
By contrast, the internal mental world is distinctively subjective,
and so, to grasp how it is for a conscious being, one needs to project
oneself imaginatively into his subjective position, to ask what it is
like for that being.19 Rather than attempt to prescind from the
subjective, one needs precisely to recognize the relativity of mental
facts to a particular standpoint. The two kinds of fact-correspond-
ing to M- and P-statements-are thus categorially different: one
sort of fact is essentially observer-independent, the other essentially
observer-dependent. To put it yet another way, a proper compre-
18 Cf. Bernard Williams' discussion of the "absolute conception" of the world

in Descartes: The Project of Pure Enquiry (New York: Penguin Books, 1978),
and Dummett on "absolute and relative forms of description" in 'Common
Sense and Physics', op. cit.
19 I rely here upon Thomas Nagel's "What Is It Like to Be a Bat?," Phil-
osophical Review, LXXXIII, 4 (October 1974): 435-450.
hension of the objective external world requires a distinction be-

tween appearance and reality; but to understand the region of
subjective reality-how it is consciously for a being-appearance
is all; we are here interested precisely in how things seem. If these
rough and intuitive remarks have verisimilitude, it is not hard to
explain why it is that phenomenalism and behaviorism should seem
to the realist so rebarbative to reason: for phenomenalism repre-
sents a refusal to register the objectivity of the material world,
since the materials it allows itself-viz. experiences-have the effect
of assimilating its essential character to the subjective; whereas
behaviorism for its part tries to assimilate subjectivity to objective
and publicly accessible behavioral facts, and so fails to do justice
to what the realist takes to be distinctive of the mental realm.20
Phenomenalism is objectionably anthropocentric; behaviorism is
objectionably nonanthropocentric.
These reflections help us appreciate better, I think, why realism
about one area ineluctably brings realism about the other: it stems
partly from the fact that our conception of the objective world is
founded upon a contrast with how things subjectively seem to us
or to other creatures-independence and this notion of objectivity
are thus two sides of the same coin, which is why the behaviorist
reduction of E-statements was inadequate to capture the realist
conception of M-statements; whereas, symmetrically, our notion of
the subjective world is defined by a contrast with the objective-
independence from behavior and this notion of subjectivity are
thus intimately related, which is (in part) why the phenomenalist
reduction of B-statements was inadequate to capture the content of
a realist view of P-statements. It is interesting also to observe that
the assertibility conditions for M- and P-statements invert the cat-
egories that define their truth conditions: I mean that subjective
experiences comprise the assertibility conditions for objective M-
statements, while (at least for other-ascriptions) objective items of
20 A large question raised by this sort of objection to behaviorism concerns

whether physicalism is equally cast into doubt by it. Though I cannot discuss
this issue here, I think that we should take a different view of the two cases.
My reason, roughly stated, is that if an organism is in the same internal phys-
ical state as another for which there is something it is subjectively like to be
that organism, then there is something it is like-the very same thing-to be
that first organism; but it is not true that the subjective is similarly super-
venient on the behavioral. To put it another way, corresponding physical
makeup gives organisms the same range of accessible subjective viewpoints,
but corresponding behavior does not. Not all versions of physicalism need to
deny the subjective; reductive behaviorism, it would seem, is another matter.
But clearly the question requires further investigation.
behavior comprise the assertibility conditions for subjective P-state-

ments. It is no surprise, therefore, to find that affirming realism
about one area in terms of independence from assertibility condi-
tions should place us squarely in the other area, and that the re-
spective assertibility conditions should resist reduction to facts of
the opposite category.21 What is more, it seems to me hardly to be
doubted that E- and B-statements play a unique role in our asser-
tion of M- and P-statements, a role which may be characterized by
saying that it is an a priori truth that E- and B-statements consti-
tute assertibility conditions for M- and P-statements: i.e., that these
classes of statements are related in this way is not something we
know merely empirically, but is such that it is part of the sense
of M- and P-statements that E- and B-statements comprise asserti-
bility classes for those statements. These assertibility conditions
may be contrasted in this respect with (say) statements about the
micro-structural properties of material bodies and statements about
conditions in a person's nervous system. We can readily conceive of
coming to learn empirically that such statements can function as
assertibility conditions for ordinary material-object statements and
mental attributions, but plainly knowing this to be so is not built
into understanding the sense of M- and P-statements; it could not
be, since we understood those sentences before we knew of mol-
ecules and neurons. Clearly such empirically discovered assertibility
conditions cannot play the same role in shaping our understanding
of M- and P-sentences as does a grasp of their semantic relation to
E- and B-statements.22 It is this fact, perhaps, that captures the ele-
ment of truth in phenomenalism and behaviorism; but this alone.
COLIN MCGINN
University College London
21 Recall the discussion of reductively reformulated independence in section iI.

22 This difference seems to correspond to one aspect of Wittgenstein's dis-
tinction between symptoms and criteria; see The Blue and Brown Books (New
York: Harper & Row, 1965), pp. 24/5. Notice that I claim (pace Wittgenstein)
an a priori evidential relation only between the classes of statements, not between
individual statements in those classes. A closely allied point about behavior and
the mental occurs in my "Mental States, Natural Kinds and Psychophysical
Laws," Proceedings of the Aristotelian Society, supp. vol. 1978: 195-220, ?6.

An A Priori Argument For Realism PDF

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

An A Priori Argument For Realism PDF

Caricato da

Copyright:

Formati disponibili

Journal of Philosophy, Inc.

An A Priori Argument for Realism

AN A PRIORI ARGUMENT FOR REALISM *

E) XCEPT in the vulgar sense, one is not a realist tout court:

* This paper was prompted by an aside of Michael Dummett's in "The Re-

0022-362X/79/7603/0113$02.10 (D 1979 The Journal of Philosophy, Inc.

aesthetic thought presupposes the reality of values, whether our

To proceed we need a preliminary formulation-to be refined later

1 As argued by Peacocke in "Causal Modalities and Realism," forthcoming

understood as contrary theses about the relation, with respect to

2 See, for example, Dummett, "What Is a Theory of Meaning? (II)," in

whereas anti-realism about P-statements denies this.3 The realist

The anti-realist doctrines I wish to consider may be labeled-I take

3 In saying that M-statements are asserted upon experiential grounds and

vary if the assignment of truth values to the E-statements stays

5 This is Dummett's preferred formulation of anti-realism: see the works of

principally because of translational ambitions; for statements re-

the application of a reductive operation, the other insists that such

I imagine that the elimination of (i) will be fairly readily conceded:

9 Rudolf Carnap in The Logical Structure of the World, R. A. George, trans.

necessary and sufficient conditions for the truth of corresponding

10 Carnap formulates realism about material objects in the phrase "inde-

state or event from which it is ontologically distinct. Those are

11 Cf. H. P. Grice, "The Causal Theory of Perception," Proceedings of the

12 Cf. Gottlob Frege's attempted refutation of idealism in "The Thought,"'

ignores a crucial maneuver open to the partial realist. Consider

differential realist to prevent him, at any arbitrary stage in the

we expect of anti-realism about material objects and about mental

totality of B-truths should not preempt the skeptical question

Anxious to avoid compulsory occupation of position (iv), someone

of independence from E- and B-statements. The hope would be to

formulation by omitting condition (b) altogether. There would still

Realism and anti-realism about M- and P-statements are theses

is at one stroke a theory of meaning and an account of how state-

Dummett has advanced a perfectly general argument, based upon

mine realism about the meaning possessed by certain sentences of

posed manifestation conditions for a realist semantics-seem to me

A discursive argument unbacked by an intuitive picture seldom

18 Cf. Bernard Williams' discussion of the "absolute conception" of the world

hension of the objective external world requires a distinction be-

20 A large question raised by this sort of objection to behaviorism concerns

behavior comprise the assertibility conditions for subjective P-state-

University College London

21 Recall the discussion of reductively reformulated independence in section iI.

Potrebbero piacerti anche