Journal of Philosophy, Inc.

The Problem of Empiricism

Author(s): Roderick M. Chisholm
Source: The Journal of Philosophy, Vol. 45, No. 19 (Sep. 9, 1948), pp. 512-517
Published by: Journal of Philosophy, Inc.
Stable URL: http://www.jstor.org/stable/2019108 .
Accessed: 13/01/2014 10:06

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
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.

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Philosophy.

http://www.jstor.org

This content downloaded from 129.62.12.156 on Mon, 13 Jan 2014 10:06:46 AM

All use subject to JSTOR Terms and Conditions
512 THE JOURNAL OF PHILOSOPHY

COMMENTS AND CRITICISM

THE PROBLEM OF EMPIRICISM

Professor C. I. Lewis, in Book II of his Carus Lectures, An

Analysisof Knowledgeand Valuation,defendsthe thesis that the
meaningof any statementwhichrefersto a materialthingmay be
fully conveyed in statementswhich refer solely to sense-data or
the sensible appearances of things. His account is perhaps the
clearest and most careful defence of this empirical thesis which
has yet appeared and, in consequence,it enables us to state, more
clearly than has been possible before,the characteristicdifficulties
of empiricism.
According to Professor Lewis, an analysis of the meaning of
any ordinary thing statement,such as "This thing is red" or
"That is a doorknob," will show that the statemententails an
unlimited number of statementsreferringsolely to sense-data.
The sense-datumstatementsare "analytic consequences" of the
thingstatement;that this is so in any particularcase "is certifiable
a priori,either by referenceto logical rules or by referenceto
meanings involved or by both together" (p. 249). The relation
which thing statementbears to sense-datumstatementis similar
to that which "T is red" bears to "T is colored" (ibid.). Since
the full meaning of any thing statementmay be conveyedin the
sense-datumstatementswhich it entails, thing statementsmay be
said to be "translatable" into sense-datumstatements(p. 181).'
The principle difficultywith this view concernsthe firststep: the
problem of showing that any ordinary thing statementhas, as
analytic consequences,statementswhich refersolely to sense-data.
are the familiar facts sometimesre-
The roots of the difficulty
ferred to as "the relativity of sense perception." Whether a
material thing will ever present,say, a red appearance or sense-
datum depends partly upon the thing and partly upon the con-
ditions under which it is observed. If one knew that the thing
were red and that the lightingconditionswere normal,one could
predict that, to a normal observer,the thing would present a red
1 Seldom,of course,does anyoneever formulateany sense-datum state-
mentsand, as ProfessorLewis admits (p. 173), it may be questionedwhether
the terms of ordinarylanguage are adequate to conveyjudgmentsabout
sense-data. Althoughhe findsit convenientto presenthis conclusionsby.
referenceto relationsbetweenstatemeents,his theorydoes not purportto be
a linguisticanalysis. The problemis an epistemologicalone, concerningour
beliefs or judgmentsabout materialthings,and could be discussedwithout
referenceto statementsat all. The presentissues,however,may be brought
into clearer focus if we follow ProfessorLewis and discuss the relations
betweenstatements.It shouldbe noted that ProfessorLewis does not use
the briefterms"thing statement"and "sense-datumstatement."

This content downloaded from 129.62.12.156 on Mon, 13 Jan 2014 10:06:46 AM

All use subject to JSTOR Terms and Conditions
 COMMENTS AND CRITICISM 513

appearance. If one knew that the lightswere out, or that the ob-
serverhad a certaintype of color blindness,one could predictthat
the thing would present some other appearance. And so on, for
any other thing and its possible appearances. To calculate the
appearances with complete success, it is necessary to know both
the thing-perceivedand the (subjective and objective) observation-
conditions,for it is the thing-perceivedand the observation-con-
ditionsworkingjointly which determinewhat is to appear. Pro-
fessor Lewis believes that "This thing is red" entails as analytic
consequencesan unlimitednumber of statementsreferringsolely
to what might appear. But the facts of "perceptual relativity"
suggestthat it doesn't entail any statementabout sense-data; they
suggest that a sense-datumstatementis entailed only when "This
thingis red" is taken in conjunctionwith anotherthingstatement
referringto observation-conditions.The translatabilitythesis re-
quires that both observation-conditions and things-perceivedbe
definablein terms of what might appear. But the facts of per-
ceptual relativityindicate that it is the joint operationof things-
perceived and observation-conditions which determineswhat is to
appear; hence the task of the empiricistwould seem to be similar
to that of an economistwho hoped to definebothsupply and de-
mand in termsof possible prices.
The complexityof the problemwill become evident if, before
turning to the details of Professor Lewis's view, we consider a
simple example. Considerthe thingstatement
This is red (P)
and the sense-datumstatement
Rednesswill appear (B).
May we say that P entails R? Possibly it will be immediatelyevi-
dent that no contradictionis involved in affirmingP and denying
R. The followingconsiderations,however,may make the matter
clearer.
Taken in conjunctionwith some other thing statement,refer-
ring to observation-conditions,P does entail R. The other state-
ment could be
This is observedundernormalconditions;and if this is red and is
observedundernormalconditions,rednesswill appear. (Q)
But taken in conjunction,not with Q, but with still anotherthing
P entails not-B.
statement,also referringto observation-conditions,
This otherthingstatementcould be
This is observedunderconditionswhichare normalexceptfor the
presenceof blue lights; and if this is red and is observedunder
conditionswhichare normalexceptfor the presenceof blue lights,
rednesswill not appear. (S)

This content downloaded from 129.62.12.156 on Mon, 13 Jan 2014 10:06:46 AM

All use subject to JSTOR Terms and Conditions
514 THE JOURNAL OF PHILOSOPHY

So far as ordinaryusage is concerned,it is quite evident that the

statementS is logically consistentwithP; thereis no contradiction
involvedin affirming one and denyingthe other. But the conjunc-
tion of P and S, if they are logically consistent,must entail every-
thing that P entails and can not entail anythinglogically incon-
sistentwith what P entails. If P and S entail not-B, it is impos-
sible that P entail R. Hence "This is red" (P) does not entail
"Redness will appear" (RB).
We may draw a similar conclusionwith respect to any other
categorical sense-datumstatementR'. Although there may be a
statementabout observation-conditions, Q', such that "This is red"
(P) and Q' entail R', there is also a statementabout observation-
conditions,S', such that P and S' entail not-R'. Hence P does not
entail R'.
ProfessorLewis admits that no thing statement,such as "This
is red," entails any categorical sense-datumstatement,such as
"Redness will appear"; he admitsfurtherthat "This is red" does
not entail any sense-datumconditional,such as "If such-and-such
should appear, then such-and-suchwould appear" (p. 237). The
sort of sense-datumstatementwhich is entailed is considerably
more complicated. It is a conditionalof the form: "If such-and-
such a presentation(or sense-datum)should appear and such-and-
such feelingsof action (also sense-data) should appear, then in all
probability such-and-suchanother sense-datum would appear"
(cf. pp. 248 ff). For reasons essential to his account of knowl-
edge and action,the antecedentof the sense-datumstatementmust
refer both to some "sensory clue" and to a feeling of activity,
and the consequentmust be prefaced by some such phrase as "in
all probability." These complications,however,do not affectthe
principle now under discussion.
We may utilize the above letters again in discussingProfessor
Lewis's view. One of his examples is the thing statement
There really is a doorknob in front of me and to the left. (P)
One of the complicatedsense-datumstatementswhich are analytic
consequencesof this mightbe
If I should seem to see such a doorknoband if I should seem to myself
to be initiating a certain grasping motion, then in all prbbability
the feeling of contacting a doorknobwould follow.2 (B)

Accordingto ProfessorLewis, R is an analytic consequenceof P;

or,in otherwords,P entailsB. But, again, if P entailsR, thenit is
2 Cf. op. cit., pp. 240, 248-249. We are to suppose,of course,that the
wordswhichappear in this statementare in what ProfessorLewis calls the
"expressivelanguage," referringsolelyto sense-data.

This content downloaded from 129.62.12.156 on Mon, 13 Jan 2014 10:06:46 AM

All use subject to JSTOR Terms and Conditions
 COMMENTS AND CRITICISM 515

logically impossible that there be a statementS, consistentwith

P and such that P and S entail not-R. Hence, if there is such a
statementS, P does not entail R. Is there,then,such a statement
S?
It should be remarkedthat we are not asking whetherthere is
such a statementS which is true. For ProfessorLewis's theory
concerns the meaning of thing statements,what is certifiablea
priori by referenceto logic and the meanings of terms. If there
is a statementS, which in conjunctionwith P entails not-R,then,
whetheror not S is true,the theory-at least in application to this
case-is mistaken.
Clearly thereare many such statementsS. One mightbe
I am unable to move my limbs and my hands but am subject to
delusions such that I think I'm moving them; I often seem to myself
to be initiating a certain grasping motion, but, when I do, I never
have the feeling of contacting anything. (S)

This statementin conjunctionwith P entails not-R. There is no

reason to suppose that S is inconsistentwith P. Hence it is false
that P entails R. Similarly,for any other complex sense-datum
statementR' whichmightbe thoughtto be an analyticconsequence
of P, it would seem to be possible to formulatea statementS', con-
sistentwithP, and such that P and S' entail not-R'.
Thus it remains to be seen in what sense any sense-datum
statementcan be regarded as an analytic consequence of "This
thingis red" or "That is a doorknob."3
It was suggestedabove that statementssuch as "This thing is
red" entail sense-datumstatementsonly whentaken in conjunction
with some other thing statementpertaining to observation-con-
ditions. Thus, in our earlier example, "This is red" (P) does
entail a sense-datumstatementwhen it is conjoined with our
statementQ: "This is observed under normal conditions; and if
3 A. J. Ayer, who also defends the translatability thesis, admits that no
thing statement, such as " This thing is red," entails any particular sense-
datum statement. But, he holds, the thing statement does entail numerous
disjunctive " sets" of sense-data statements, " where the defining character-
istic of the set is that all its members refer to sense-contentsthat fall within
a certain specifiable range" (Language, Truth and Logic, 2nd edition, p. 13;
cf. Foundations of Empirical Knowledge, pp. 240-241). E.g., each member
of a set might refer to some specific shade of red, not referredto by the other
members,but all of the sense-data referred to would fall within the "fairly
indefiniterange" of redness. The view that the thing statement entails such
a set, however, is subject to the difficultiesnoted above in the case of single
statements; for, again, it would seem to be possible to find another thing
statement which, taken in conjunction with "This thing is red," entails the
contradictoryof any set which Mr. Ayer might mention.

This content downloaded from 129.62.12.156 on Mon, 13 Jan 2014 10:06:46 AM

All use subject to JSTOR Terms and Conditions
516 THE JOURNAL OF PHILOSOPHY

this is red and is observedunder normal conditions,redness will

appear." Without seeking to beg any questions, we might call
the latter part of Q a psychophysicalstatement,for it refers to
sense-data which will appear under certain physical conditions.
The sense-datumstatement("Redness will appear") is entailed,
not by the thingstatement("This is red") but by the conjunction
of: (i) the thingstatement; (ii) the statementabout observation-
conditions ("This is observed under normal conditions"); and
(iii) the psychophysicalstatement("If this is red and is observed
under normalconditions,rednesswill appear"). As we have seen,
the thing statement,"This is red," if conjoined with a different
statementabout observation-conditions and witha differentpsycho-
physical statement,may entail a differentsense-datumstatement.
It could be contendedthat the psychophysicalstatement(iii) is
analytic,that "If this is observedunder normalconditions,redness
will appear" is an analytic consequence of "This is red." This
fact would be of littleuse to the translatabilityprogram,however.
For this method of deriving sense-datumstatementsfrom "This
is red" proceedsby utilizingadditional thingstatements("This is
observedunder normalconditions, " referringpresumablyto light-
ing arrangements,ocular conditions,and so on). Instead of elimi-
nating thing statementsin this manner, therefore,we would be
multiplyingthem. And the new thing statementswould present
our old difficultiesagain.'
The translatabilitythesis, however,does claim to provide an
accountof the respectin whichour knowledgeof thingsis founded
in, and is verifiableand falsiflablein, sense experience. If we deny
this thesis we must provide an alternativeaccount of the manner
in which such experiencemay be said to justifyour knowledgeof
things.5 This is not the place to pursue this furtherquestion in
detail; but it is relevantto note that, in principle,the problembe-
comessimilarto that of the validity of memoryand that Professor
Lewis's own methodof treatingthe latter problemmay in fact be
applicable to both problems. The possibilityof our having any
knowledgeat all, he believes, requires that we make two assump-
tions about memory:"First; whateveris remembered,whetheras
explicitrecollectionor merelyin the formof our sense of the past,
is prima facie crediblebecause so remembered. And second; when
the whole range of empirical beliefs is taken into account, all of
4Cf. H. H. Price, Humne'sTheoryof the External World,pp. 183-188.
5 Similarly, we should have to reconsider the thesis, fundamental to Pro-
fessor Lewis's views (as well as to Mr. Ayer's), that a syntheticstatementis
meaningful only to the extent to which it refers to possible experience. But
it is reasonable to suppose that the acceptability of such a criterionof meaning
should be secondary to considerations such as those discussed above.

This content downloaded from 129.62.12.156 on Mon, 13 Jan 2014 10:06:46 AM

All use subject to JSTOR Terms and Conditions
 COMMENTS AND CRITICISM 517

them more or less dependent upon memorialknowledge,we find

that those whichare mostcredible can be assured by theirmutual
support,or . . . congruence" (p. 334).6 If thingstatements are
not translatable into sense-datumstatements,it may be that the
validity of our perceptual knowledge of things requires similar
assumptions. It may be that wheneverthe presence of a sense-
datum leads one to accept a belief about a material thing (e.g.,
whenever,as a matterof fact, the presence of a red sense-datum
leads one to accept the belief that one is observinga red thing)
the belief which is thus "perceptually accepted" 7 is prima facie
credible because it is so accepted. Indeed Mr. Price has said as
much. And it may be that whenthe wholerange of our perceptual
beliefs are taken into account,all of themmore or less dependent
upon our perceptualacceptances,we findthat thosewhichare most
credible can be assured by their mutual support. These assump-
tions do not claim any facultiesfor man whichare not involvedin
ProfessorLewis's defenceof memory.8 Whetherthey will suffice
for justifyingperceptual knowledge,however,is a question which
can be answeredonlyon the basis of a discussionas thoroughas the
one which ProfessorLewis devotes to memory.
RODERICK M. CHISHOLM
BROWN UNIVERSITY

PROFESSOR CHISHOLM AND EMPIRICISM

Professor Chisholmquestions the adequacy of the account of

perceptual knowledgewhich I have put forward,on the ground
that this account is incompatiblewith "the familiar facts some-
timesreferredto as 'the relativityof sense perception. "1
I have held that what the statementof a perceptuallylearned
objective fact, such as "This (seen object) is red" or "This is
square," means-in one specific,and specified,sense of the word
'meaning'-is explicable by some set of statementsrepresenting
predictionsof possible experienceand having the form,"If S be
given and act A initiated,then in all probabilityE will follow,"
where 'S', 'A', and 'E' each refersto some recognizableitem of
direct experience,and the colloquial phrase "in all probability"
is intendedto suggest a probabilityapproximatingto certainty.
8 Cf. Bertrand Russell, An Inquiry into Meaning and Truth, p. 202.
7 " Perceptual acceptanee" is Mr. Price 's term. Cf. Perception, pp. 139 fE.
8 Such assumptions are involved in the doctrine which C. S. Peirce called
" Critical Common-Sensism." Cf. Collected Papers, Vol. V, Book III.
Certain passages in Professor Lewis 's book suggest that he would not be dis-
inclined to make similar assumptions. Cf. p. 326.
1 Page 512, preceding.

This content downloaded from 129.62.12.156 on Mon, 13 Jan 2014 10:06:46 AM

All use subject to JSTOR Terms and Conditions