Witt Gen Stein Philosophy and Empirical Psychology

8/3/2019 Witt Gen Stein Philosophy and Empirical Psychology

http://slidepdf.com/reader/full/witt-gen-stein-philosophy-and-empirical-psychology 1/23

Wittgensteinian Philosophy and Empirical PsychologyAuthor(s): Richard RortySource: Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition, Vol. 31, No. 3 (Mar., 1977), pp. 151-172Published by: SpringerStable URL: http://www.jstor.org/stable/4319124

Accessed: 06/08/2009 02:16

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=springer.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Springer is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Studies: An

International Journal for Philosophy in the Analytic Tradition.

http://www.jstor.org

http://www.jstor.org/stable/4319124?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/action/showPublisher?publisherCode=springer

http://www.jstor.org/action/showPublisher?publisherCode=springer

http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/stable/4319124?origin=JSTOR-pdf



RICHARD RORTY

WITTGENSTEINIAN PHILOSOPHY AND

EMPIRICAL PSYCHOLOGY*

(Received 29 March, 1976)

The confusion and barreness of psychology is not tobe explained by calling it a 'young science'; its state is

not comparable with that of physics, for example, inits beginnings. (Rather with that of mathematics. Settheory.) For in psychology there are experimental methods

and conceptual confusion. (As in the other case concep-tual confusion and methods of proof).

The existence of the experimental method makes usthink we have the means of solving the problems whichtrouble us; though problems and method pass each otherby.

(Wittgenstein, Philosophical Investigations, II, xiv)

I

One central theme in recent philosophy has been a rejection of what

Quine calls the 'idea idea' - the notion that regularitiesn linguistic(or other)

behaviorare to be explainedin terms of mentalentitieswhicharethe imme-

diate referentsof words, or the immediateobjects of knowledge,orboth. It

is widely thought that a certain picture of the mind, common to Descartes

and Locke, has been graduallyerasedby the work of suchwritersas Dewey,

Ryle, Austin, Wittgenstein,Sellars,and Quinehimself. However,this picture- the one which gave rise to the 17th-centurynotion of a 'veilof ideas',and

thus to epistemological skepticism - has not been replaced by a new and

clearerpicture. On the contrary,there is widespread ratricidaldisagreement

amonganti-Cartesians bout what, if anything,we should say about the mind.

Ryle's magic word 'disposition'is no longer in favor, but more up-to-datenotions such as 'functionalstate' are offered as substitutes. Anything that

smacksof 'behaviorism's looked at askance,but it is agreed hat theremust

be some way of avoidingbehaviorismwithout falling back into the sort of

dualism which engenderedthe traditional'problemsof modernphilosophy'.

One radicalway of doing so is suggested by such admirersof Wittgenstein

PhilosophicalStudies 31 (1977) 151-172. All RightsReservedCopyright 1977 by D. ReidelPublishingCompany,Dordrecht-Holland



152 RICHARD RORTY

as Malcolm,who in the course of a polemic againstmuch recent work by

psychologists,says:

Thus, it is the facts, the circumstances surrounding the behavior, that give it the propertyof expressing recognition. This property is not due to something that goes on inside.

It seems to me that if this point were understood by philosophers and pshychologists,

they would no longer have a motive for constructing theories and models for recogni-tion, memory, thinking, problem solving, understanding, and other 'cognitive processes'.'

If onefollows out this line of thought, one can decidethat the whole empirical

science of psychology is based on a mistake, and that there is no middle

ground for empirical research between common-sensicalexplanations of

behavioron the one hand andstraightforward europhysiologyon the other.

The notion that there is middle ground will, on this view, be a product of

Malcolm's 'myth of cognitive processesand structures' what Ryle called'Descartes'Myth'. Malcolmseems inclined to take the matter to this point,

as when he describesthe Chomskyannotion of an 'internalizedsystem of

rules'as typical of the root mistakeof the 'traditionalheory of Ideas',viz.,

The assumption ... that in speaking a person must be guided. There must be somethingat hand that shows him how to speak, how to put words together grammatically andwith coherent sense ... What is being explained is knowledge - both knowing that andknowing how. The presence in him of the structure of the language or of its system of

rules is supposed to account for this knowledge - to explain how he knows.2

If we once see that "our understandingof human cognitive powers is not

advanced by replacingthe stimulus-responsemythology with a mythology

of inner guidancesystems,"3 Malcolmthinks, we shallnot think that there

areany explanations o be looked for in this area.

In this paper, I want to considerthe suspicions about psychology which

seem to flow so naturallyfrom Wittgenstein'sattack on modem philosophy.It is certainly true that the samemodel of the mind which led Descartesand

Locke to construct the 'traditionalphilosophicalproblems' was held by

Herbart,Helmholtz, and Wundt,and was built into the terminologyof the

young science of psychology.It would be surprising f the overthrowof this

model did not have some effect upon work in that science. Yet it would be

equallysurprising f a disciplinewhich is now severalgenerationsaway from

its philosophicaloriginsshould not be able to stand on its own feet. There

must, one feels, be psychological researchprogramswhich cannot be endan-

gered by a philosophicalcriticismof the vocabulariesused by theirdesigners.I shalldiscuss hree ines of argumentwhichWittgensteiniansaveemployed



PHILOSOPHY AND EMPIRICAL PSYCHOLOGY 153

to show that there is something dubious about all or many 'psychological

explanationsof behaviour'.Theselines are,as farasIcansee, quiteindependent

of one another, and of different worth. So I thinkit may help to disentangle

them and evaluate them separately. But they do spring from a common

ground- a suspicion of the traditional pistemologyout of whichpsychology

arose - and I shall conclude by suggesting that the force of all three linesof argumentcan be broken by distinguishingmoresharplythanpsychologists

often have between questionsabout the predictionof behaviorandquestions

about the justification of belief. Underlining his distinction, I shallargue,

permits us to disjoin psychology from the Descartes-Locke radition of

foundationalistepistemology, andthus from the 'idea idea'.

The threelines of argument havein mindcan be called(1) the 'no private

language' argument, (2) the 'infinite regress' argument,and (3) the 'nofoundations'argument.Drasticallysimplified, they look somethinglike this:

(1) 'No privatelanguage':traditionalpsychology hasassumed hat we can

identify mental entities apart from the behaviorandcircumstances ttendant

upon them - as if we could simply introspect and christenthe occupantsof

the mental arena. Once we realize that suchchristening s impossibleandthat

it is not introspectible qualiawhich make something count as a thoughtor a

belief or a recognition,we see that there is nothing 'inner'to investigate.(2) 'Infiniteregress':This line is nicely formulatedby Malcolm:

If we say that the way in which a person knows that something in front of him is a dogis by seeing that the creature 'fits' his Idea of a dog, then we need to ask, "How does heknow that this is an example of fitting?" What guides his judgment here? Does he notneed a second-order Idea which shows him what it is like for something to fit an Idea?That is, will he not need a model of fitting? ... An infinite regress has been generated andnothing has been explained.4

The point can also be put as follows: any postulatedmentalentity will standto some 'innerself' as somethingin the physicalworld (a paradigmobject, a

rule-book)standsto aperson.So no relationof the lattersortcan be explained

by a relation of the former sort without generating ust the same problem

all overagain.

(3) 'No foundations':this is theargument,developedmost fully by Seilars,

that the notion of 'immediatelyknownparticulars'e.g., sense-data, learand

distinct ideas) is inherentlyconfused. For nothingcan be knownsavea fact,

and to speak of knowinga particulars to speak obliquely of knowingsome

fact about a particular.5Since knowledgeof a fact requiresabilityto justify

one's belief in the fact, one will not be able to know anythingwithout the



154 RICHARD RORTY

'mediation' of alot of otherknowledge.6 So the notion of 'elementarydataof

consciousness' is based on a confusion between a particular the physical

stimulus, or some physiological state produced by that stimulus) and some

fact about that particular.So there can be no such activityas the mind'sap-

plication of rulesto non-propositionaldata,thus becomingawarefor the first

time of a proposition. But without this notion of 'immediatedata of con-sciousness'there will be nothing for psychological processes' o process, and

no content to fill psychological 'structures'.So the notion of suchprocesses

and structuresmust be rejected for the same reasonthat we rejectthe notion

of 'a foundation of knowledge'.All three of these lines can be found in Witt-

genstein'sInvestigations, although none are original with him.7 I shall use

'Wittgensteinian hilosophy' as a label for the joint use of these linesof argu-

ment. I shall not go into detailabout the variouscontroversies o which eachhas given rise amongphilosophers,but rather focus as closely as possible on

theirrelevance o researchprogramsn psychology.

II

The 'no private anguage' ine of argument,I have arguedelsewhere,8comes

down to the claim that nobody who did not know a good dealabout humanbehaviorand social conventions could understand he meaningof suchterms

as 'thought', 'sensation', 'mental state', and the like, nor of terms which

referred o particular houghts,sensations,etc. Even if this claim is granted,

however,it is hardto see the relevanceof the point to psychologicalresearch.

Still, it mustbe admitted that Ryle andMalcolm,among others,havethought

it relevant, and have concluded from this argumentalone that thereis some-

thing dubious about ookingfor 'psychologicalmechanisms'. t has often beensuggested hat they think this becausethey move fromsaying

(I) The meaningof terms referring o the mental is to be explained

in termsof behavior where 'behavior' s short-hand or 'functions

relatingcircumstancesand stimulito behavior')ratherby interior

ostention

to saying

(II) Psychology can only concern itself with empirical correlations

betweenbits of behaviorand externalcircumstances.




The inference, as successivecritics of Ryle have pointed out, has no greater

force that theparallelnferencemade by operationalistphilosophersof science

about physics.9 Fodor, for example, remarks that the psychologistis quite

prepared to admit that certainfeaturesin behaviorand in the social matrix

are necessaryconditions for the occurrenceof thoughts, recognitions,emo-

tions, etc., but he urges that there may be lots of equallynecessary'inner'conditions as well.10 So if the psychological nvestigatorhassense enoughto

avoid defining,e.g., 'act of recognition' in terms of purely inner events,

presumablyhe can take advantageof behavior,etc., to identifyhis data. What

morethanthis can be required o avoidthe chargeof mythologizing?It might,

of course,turn out thattherewereno "interveningariables"worthpostulating

- but this could presumablyonly be discovereda posteriori, y tryingand

failing.As. P. C. Dodwellsays:

On Malcolm's view, psychologists would have to restrict themselves to investigations of

simple empirical relationships such as that which might obtain between memory and

sleep deprivation. But it is extremely difficult to see how this restriction could be

justified. The factors which psychologists investigate in human memory are empirical

relationships, although usually of a more complicated sort than those just mentioned.

Who, then should make the decision on which empirical relationships are to be cleared

for investigation? Not philosophers, surely.11

Even, however,if one regards uchrepliesas Dodwell'sand Fodor's as conclu-

sive (as I do), one can profit from asking why operationalismseems more

palatable n regard o the psychicalthan in regard o the physical.Thereare

at least two reasons.The firstis simplysociological:psychologyhas not been

as successfulas physicsin makinga differenceto our world.In the face of the

hydrogen bomb, to insist that sub-microscopicparticlesare 'a mere way of

speaking'beginsitself to seem a mereway of speaking.But none of psycholo-gy's various nterveningvariableshave made suchan impactupon the popular

consciousness,and to refer to them as 'convenient fictions' or 'conceptual

shorthand'does not disturbour sense of fitness. It is only when theoretical

constructsturn out to be so useful as to affect practicethat we beginto feel

ontological about them. The second reason that operationalismseems

appropriatehere is thattheslogansof Wittgensteinian hilosophycallup mem-

ories of severalgenerationsof behavioristsneers at the notion of introspec-

tion, and associatedmemoriesof the need for 'unifiedscience'.If one thinks

of 'psychologicalmechanisms'as opposedto 'physicalmechanisms', hen one

will think that psychologistsshould content themselveswith either straight-



156 RICHARD RORTY

forward behavioristic input-output functions or with neurophysiological

inquiry.

This second reason for being operationalist n psychology is, of course,

no more forceful than parallelconsiderationsaboutMendel'suse of 'gene'.12

As long as psychological mechanismsare thought of as place-holders or

neurophysiologicalarrangements,unified science need not be endangered.But further, suppose we had achievedcomprehensiveand powerfulways of

predicting and controllingof behavior,ways which took us as far beyond

presentnotions of how to cope with people as physicshas taken us beyond

pre-nineteenth-century otions of how to cope with things. No physicalist

would be greatly bothered if no neurophysiologicalcorrelatesfor any of the

entities mentioned in this theory had turned up. No geneticist became

disenchantedwith the notion of 'gene' simply because DNA was a longtime coming, and psychologists' faith in physicalismcould doubtlessendure

centuries of similar waiting. It is true that if we felt that we had carried

micro-structural xaminationof the brain about as far as it could go, and

could find no brain-descriptionswhich seemed to varyconcomitantly with

psychologicaldescriptionsoffered by the theory, then we mightfeel that the

ideal of unified science wasendangered.But so we should - because t would

indeed have turnedout to be anunattainabledeal.No amountof philosophi-cal reinterpretationof the meaningor referenceof the terms of the psycho-

logical theory wouldmake it look anybetter.13

I conclude, then, that the 'no private anguage'argumentand any similar

line of argumentagainst the possibility of interiorostention does nothing to

impede any imaginableprogram of research n psychology. It would do so

only if one could infer from a doctrineabout how psychologicalterms got

their meaning something about the non-physicalcharacter of the entities

mentioned in psychologicaltheories(in some strongersense of 'non-physical'

thansimply'psychologicalas opposed to physiological').

III

I turn now to the infinite regressargument.Unlike the 'no private anguage'

argument,this is not simply a specific applicationof a very generalandvery

dubious philosophicaldoctrine(suchas operationalism).Rather, t is directed

againsta specific feature of psychologicalexplanations- their tendency to

explaina person'sabilityto do X by postulatinga certain othersort of ability,




as much in need of explanationas the first.("Howcanhe recognizeX's? By

recognizing heirsimilarity o theabstract dea of Xhood.") One canplausibly

claim that this is not just an occasional,but necessary,featureof psychologi-

cal explanation. Let us restrict ourselves to those explanationswhich do

involve what Dodwell calls 'models' - those which postulateentities rather

than simply correlatingpublicly observablecircumstancesor behavior.Whatsort of entities will we want in psychology? Presumablyonly those which

Fodor describesas 'abstract'.Considerthe following passagesfrom Fodor's

discussionof how we can recognizethe same tune when it is played at half-

speed, on different instruments,in differentkeys, etc., etc. (with sufficient

variation so that there is no overlap n 'auditory stimulus', but only 'family

resemblance').Ryle had dismissedthis question as simply the acquisitionof

a set of expectations, no moreproblematic han the acquisitionof the abilityto ride a bicycle. Fodor replies:

But if what the variouswaysof performingLilibulero' havein common is somethingabstract, t wouldappear o follow that the systemof expectations hatconstitutesone'srecipefor hearing he song mustbe abstractn the samesense...

... The relevantexpectations must be complicatedand abstract, since perceptualidentities are surprisinglyndependentof physicaluniformities mongstimuli.Sinceit isprecisely hisperceptualconstancy' hat psychologists ndepistemologists ave radition-ally supposed unconscious inferences and other para-mechanicalransactionswill be

neededto explain,it seemsrelevant o remark hat Ryle's treatmenthas beggedall theissuesthat constancyraises.4

One may agreewith Fodor that if there are 'issuesthat constancy raises' hen

Ryle hasbeggedthem. But Ryle could easilyreplythat the notion of 'compli-

cated andabstractexpectations' e.g., a set of unconscious nferences nvolving

reference to some rules, or some abstractparadigms)s what makes it appear

that there are issues here. Only the picture of a little man in the mind,

applyingrules drawnup in non-verbalbut still 'abstract' erms, makesus ask'How is it done?' Had we not had this picture imposed on us, Ryle can say,

we would respond with something ike "It's only possible because of havinga

complicated nervous system - doubtless some physiologist will someday tell

us just how it works." The notion of non-physiological models' would not

occur, in other words, if one did not alreadyhave the whole Cartesianbag of

tricksin hand.

This reply can be restated a bit more precisely. Suppose we agree with

Fodor that recognition of similarity among potentially infinite differences

is recognition of something 'abstract'- Lillibulerohood, say. What does it

mean to say that "one's recipe for hearingthe song must be abstract n the



158 RICHARD RORTY

same sense"?Presumably hat it mustbe able to distinguish imilarityamong

potentially infinite difference. But this makes us see that there is no use for

the notion of 'concreterecipe', since any recipemustbe able to do this. The

possiblequalitativevariationsamongthe ingredientsfor a particularbatchof

chocolate chip cookies are also potentially infinite. So it we are talk about

'complicatedsets of expectations'(or 'programs'or 'systemsof rules') at all,we shall always be talking about something 'abstract' precisely as abstract,

in fact, as the characteristicwhose recognition or the task whose accomplish-

ment) we wishto explain.Butthenwe are in a dilemma:either the acquisition

of these sets of expectations or rules tself requires he postulation of new sets

of expectationsor rules,or they areunacquired. f the formerhom is grasped,

Malcolm's nfinite regressreally will be generatedby Fodor'sprinciplethat

recognition of the abstractrequiresuse of the abstract, for what holds ofrecognitionshould hold of acquisition.If we grasp the latter horn, then we

seem to be back with Ryle: to say that people have anunacquiredabilityto

recognize similarity among infinite differences is hardly to say anything

explanatoryabout 'the issues that constancyraises'.

So, Ryle can conclude, these issues are either 'conceptual'issues about

the sufficient conditionsforthe ordinaryapplicationof terms ike 'recognize',

or else issues about physiologicalmechanisms.Thelatter sort of issuesinvolveno problemsabout regresses,since nobody thinks that 'constancy' requires

postulating'abstract'machanismsn photoelectriccells or tuningforks.Yet is

there any difference between middle C and 'Lillibulero', ave that we have

dubbed the former a 'concrete acoustic quality' and the latter an 'abstract

similarity'? One could specify a thousand accidental freatures (timbre,

volume, presence of light, color of the object emittingthe sound)which the

tuning fork ignores,just as the Lillibulero-recognizeroes. Since the abstract-

concrete distinctionis as relativeto a given data-baseas the complex-simple

distinction, it looks as if, in saying that psychologicalexplanation requires

reference to abstract entities, we are simply claiming that explaining the

sort of thing humans (or, at least, rats) can do requiresreference o different- categorically different - sorts of things than explaining what amoebae,

tuning forks,cesiumatoms, and starscan do. But how do we know that? And

what does 'categorically'mean here? Once again, Ryle seems in a position to

say that if we did not already have the Cartesianpicture (of an inner eye

looking at the rulesposted on the walls of the mental arena),we should not

know what to make of the claim.




So much for the strengthof the infinite regressargument.Now consider

the sort of rejoinder to it which might be made by someone like Dodwell,

who arguesthat non-physiologicalmodel-buildinghas nothing right (Fodor)

or wrong(Ryle) about it a priori, but is to bejustifiedby its fruits. Dodwell

is impressedby the analogybetween brains and computers:"thesinglemost

powerful influence on psychologists' ideas about cognitive processesis thenexus of concepts which has been developedfor computerprogramming."5

Nevertheless,he amits that

It might be argued that the computer analogy is trivial, because a program merelycodifies a set of operations which are like cognitive operations, but no more explainsthinking than does writing down a set of rules for solving arithmetic problems ... To saythat a computer program can 'explain' thinking, then, would have about the same forceas to say that a set of logical formulas 'explain' the laws of correct deductive argument. 16

To this argumenthe replies that the computeranalogy only has force once

one distinguishesbetweenlevels:

Explanations of what goes on in problem solving by a computer can be given at differentlevels. The implementation of a program has to be explained in terms of computer hard-ware, just as, presumably, the implementation of thinking has in some sense to be ex-plained by processes which actually occur in the central nervous system. The subrou-tines by which particular computations are made can be explained by reference to the'machine language' and the step-by-step algorithms by which solutions are found ... The

principle of the subroutine operation is not itself to be understood and explained justby examining the hardware, in just the same way that the point of multiplication tablescould not be grasped by examining the brain. Similarly, an understanding of how thesubroutines themselves work does not explain the principle of solving problems in termsof a sequence of steps ... For that, one must look at the executive process, which in themachine embodies the overall organization and goal of the program, and in the human

being a less clearly understood 'goal directedness'.17

The importanceof levels is illustrated,for example,by the fact that experi-

mentation may give reasonfor sayingthat we recognizevisualpattems by atemplate-matchingprocessratherthan a feature-extraction rocess.18To say

this is neither a 'conceptual'remark(about 'the executive process') nor a

'physiological' remark (about 'hardware'),but a remark which may be

genuinelyexplanatory none the less. Thenotion of 'subroutine' eemsto give

us just what psychology needs - an explanationof what the middle groundbetweencommon-senseandphysiology mightbe good for.

But how does this notion help us avoid an infinite regress argument?

PresumablyMalcolmand Ryle would urge that either 'templates'or abstract

ideas of the features extracted (depending upon which model one picks)

themselvesproduce the same problems as the 'constancy'they are supposed



160 RICHARD RORTY

to explain. But Dodwell can reply that they would only do this if they were

supposed to be answers to such generalquestions as, 'How is abstraction

(recognition, constancy) possible?' To such questions, one can say, there

is no answer, except the pointless remarkthat nature has evolved suitable

hardwareto get the job done. For any of Dodwell's models will indeed be

anthropomorphic,n the senseof envisaging little manin the braincheckinghis templatesor tickingoff features.'9 Thislittle man'spowersof abstraction

or recognitionwill be as problematicas thoseof his host. Butanthropomorphic

models here are no moremisleading han the programmer'snthropomorphic

remark that "The machine won't understand he problemif you use Polish

notation, because it's programmed o recognize only ...." Such a remark s

thoroughly informative,even though no series of such remarkscan help us

understandhow machines can think, how thought is possible,how abstrac-tion works,or anythingof a similar evel of generality.

But these latter questions are like 'How is motion possible?'or 'Whydoes

nature follow laws?' or 'What s life?' To complain that 'templates'- like

Locke's 'ideas'- are a reduplicationof the explanandum s like claiming hat

the particleswhich make up the Bohratom are a reduplicationof the billiard

balls whose behavior hey help explain.It turns out to be fruitful to postulate

little billiard balls inside big ones, so why not postulate little people insidebig ones (or little rats inside big rats)? Every such "model" is, in Sellars'

phrase, accompanied by a 'commentary' which lists the features of the

modeled entity which are 'abstracted rom'in the model. It seemsreasonable

to suggest that the implicit commentaryon all anthropomorphicmodels in

psychology goes somethinglike this: as long as we stay on the level of sub-

routines we shall feel free to talk anthropomorphically ither of inferences

and other operations'unconsciously'performedby the person,or performed

(neither 'consciously'nor 'unconsciously') by braincenters or other organs

spoken of as if they themselveswere persons.The use of such phrasesdoes

not commitus to attributing ntellect andcharacter o braincenters any more

than speakingof a 'red sense-impression' s the common factor in various

illusions commits us to the existence of somethingboth 'inner'and red.20But once we get off the 'sub-routine' evel and onto the hardware evel,

anthropomorphism s no longer in place. For suppose that perceptionwere

simpler than it is, so that some specialsort of neutralcurrentcame down theoptic nervewhen andonly whenpsychological heorypredicted he occurrence

of a red sense-impression and so for all other perceptual situations).If we




knew this fact, we should simply skip the 'sub-routine'explanationand go

straight to hardware.21The notion of 'sense-impression'would no longerhavea role(unlesstherewereothertheoreticalentitiespostulatedby psycholo-

gical theory which requiredthis notion for their explication). If thingsdid

turn out to be as simpleas this, then the 'computer'analogywould no longer

seem particularlyrelevant - any more than it does for one-celledanimals,wherethe step from behaviorto physiology is too short to make the notion

of levels' seem in point. But this is to say that if physiologyweresimplerand

more obviousthanit is, nobody wouldhave felt the need for psychology.

This conclusion may seem odd, particularly n the light of Dodwell's

remark(quoted above) that "The principleof the subroutineoperation is

not itself to be understoodand explained ust by examining he hardware,n

just the sameway that the point of multiplication ablescould not be graspedby examining the brain." 2 But this remark is seriously misleading. It

embodies a confusion between

(a) If you did not know what multiplicationwas, looking into the

brain wouldnevertell you

whichis quite true,and the dubiousproposition

(b) If you did know what multiplicationwas, you could not tell that

someone was doing a certain bit of multiplication by looking at

his brain.

This latter is dubious becausewe just do not know whetherthere are quite

simple neurophysiologicalparametersassociated with certain mental opera-

tions. It is immenselyunlikely that there are, but there is no a priori reason

why some suitable brain-probe-cum-microscope ight not show somethingwhich the trained observer would report as 'Ah, you're multiplying forty-

sevenby twenty-five'(and be rightevery time). Moregenerally, he question

of what is best explained in hardwareterms and what in programterms

depends entirely on how ad hoc the hardwarehappens to be, and on how

perspicuously t is laid out. Ad hocness and perspicuousness re, obviou?ly,relative to choice of vocabularyand level of abstraction - but then so is

the hardware-software istinction itself.23 Given the rightsort of hardwareand the right parameters, t certainly is possible to 'understandand explain

the principleof the subroutine ust by examining he hardware'. ndeed,one

can easily imaginemachines in which it would be easierto find out whatthe



162 RICHARD RORTY

machine was up to by opening t up andlooking thanby reading he program.

Since the brain is almostcertainlynot such a machine, the point here is

one of principle - but the principle has philosophicalimportance.For it

shows that the distinction between psychology and physiology may not

be a distinction between two distinct subject-matters n any stronger a

sense than is, say, the distinction between chemistry andphysics. It mighthave turned out that chemical phenomenasuch as the formation of com-

pounds never had anything to do with the sub-microscopicmakeupof the

elements in question. But in fact it does, and so now whether one uses

physicists'or chemists'terms to explaina reaction s a matterof convenience

or pedagogy. If it turns out that physiology has asmuch to do with multiply-

ing as electronshave to do with explosions, then the psychology-physiology

distinction will be equally pragmatic.So the paradoxicalconclusion offeredearlier - that had physiology been more obvious psychology would never

have arisen- can be reaffirmed.Indeed,we can strengthen t andsay that if

the body had been easierto understand,nobody would havethoughtthat we

had a mind.24

It is time now to sum up this way of dealingwith the infinite regress

argument.The central point is just that explanatoryentities postulatedby

psychologists would reduplicateproblems in the explanandaonly if theseproblemswere bad problemsanyway- suchproblemsas 'Howis recognition

possible?' Philosophers like Malcolm and Ryle are accustomed to bad

philosophical answers to bad philosophical questions. ('How is motion

possible?' - As the actualizationof the potential qua potential.' 'Whydoes

nature follow laws? - Because of God's benevolence and omnipotence.')

Consequently they tend to see such questions lurking behind even quite

specific and limited researchprograms.They are not always wrong, since

psychologists (or theirpopularizers) ften tout theirlatest 'model'as offering

'a new understandingof the natureof mind'or some suchdubiousbenefit.25

But suppose such models as those Dodwell has in mind - proposals about

sub-routineswhich the mind follows in ways which areneither introspectible

(like the 'executive process') nor physiologically decipherable(like the

'hardware')- are viewed neither as contributions to resolvingCartesian

pseudo-problemsnor as discoveriesabout some nonphysical sort of entity.

Then the infiniteregressargumenthas no force. For whether or not hardware-correlates for these sub-routinesever turn up, the success in predictionand

control of behaviorwhichwould ensue on verifyingthe existence of such sub-




routines experimentallywould be more than enough to show the reality of

the objects of psychologicalinquiry.26Dodwell'ssuggestionis that nothing

succeeds like suchsuccesswhenit is a matterof establishing he non-mythical

can 'scientific' characterof one's subject. This may be pretty well the last

wordon the matter.

Applyingthis point to the dilemma I sketchedabove about acquiredandunacquiredabilities, we can cheerfully admit that any such model-building

must grant that nature has wired in some unacquiredabilities to perform

higher-ordermental operations.At least some of those little men performing

sub-routines n various brain-centerswill have to have been theresincebirth.

But why not? If one givesup the notion that empiricalpsychologyis goingto

do what the British Empiricistsfailed to do - show how a tabularasagets

changed into a complicated information-processing eviceby impactsuponperipheralense-organs then one will not be surprised hathalf of the adult's

sub-routineswere wired into the infant'sbrain.Further, t will not strikeone

as important to our understandingof the nature of man or his mind to

discover ust whichwerewiredin then andwhichcamealonglater.27Finally,

it will not seem odd that something'abstract' like a capacityfor recognizing

similarity in difference) is as unacquiredas such 'concrete' capacities as

making a differentialresponseto C-sharp.For one can simplyremindoneselfthe latteris itself as 'abstract'as an abilitycould well be, andno more abstract

than any ability must be. The whole notion of concrete-vs.-abstractbilities,

which is accepted as uncriticallyby Fodor as by Kant, is of a piece with the

notion of the 'irreduciblyphysical' vs. the 'irreduciblypsychical'. Nobody

can say how to draw these linesexcept relative o the temporarypurposesof

inquiry. But the Cartesianattempt to draw them once and for all, and the

'empiricist'and 'behaviorist'attemptsto 'reduce' the one to the other, have

created the view that certain deep mysteries which had confounded the

philosophersmay yet be penetrated by psychological research.Malcolm's

and Ryle's incautious use of the infinite regressargumentshould be seen, I

believe, as an understandablereaction againstthis notion that psychology

can succeed in solvingproblems which philosophershad posed. Theirargu-

ment does show that psychologistswill not do this, but it cannot show that

psychologistsmay not do a great many other things, nor can it show that

'cognitiveprocessesandstructures'are'mythical'.



164 RICHARD RORTY

IV

We come now to the third sort of Wittgensteinian riticism of psychology-

what I have called the 'no foundations' line of argument.The centralpoint

here is that much psychological researchuncriticallyemploysthe concept of

'datum', and that this notion is itself a left-over from the British Empiricists'confusionbetweenthe stimulationof asense-organ nd belief in someproposi-

tion about that stimulation or its source.This criticism echoes a remarkof

T. H. Green a hundredyearsago.Greenspeaksof

the fundamental confusion on which all empirical psychology rests, between two essen-tially distinct questions - one metaphysical, What is the simplest element of knowledge?the other physiological, What are the conditions in the individual human organism invirtue of which it becomes a vehicle of knowledge?28

The former question - about the simplest element of knowledge - is no

longermuchaskedby philosophers.Theyhave become sensitiveto the notion

that simplicityis as relativeasabstractness,andthat 'element' in this context

must be given a sense by some theory which divides mental faculties into

'thought' (which interprets) and 'sense' (which provides material to be

interpreted).This division,which Wundt made the cornerstoneof introspec-

tionist psychology, is now so dubious in philosophicalcirclesthat the notionof 'an element of knowledge'has anold-fashioned ound.But the behaviorist

suggestion hat we concentrateonvariants f Green's atterquestion replacing

'element of knowledge' with 'stimulus' - is itself viewed with suspicion

becauseof such questionsas those raised or Skinnerby Chomsky about how

one delimitsa 'stimulus'.29

The obscurity caused by these variousdoubtsmakesthe notion of 'datum'

suspiciouson itsface,

and thesuspicion

sincreasedby raising he point citedearlier: knowledge can only be of the truth of propositions, so whatever

'datum'means it cannot mean 'particularknown as a basis for inferenceto

the truth of propositions'. Since one can only infer from propositions to

propositions, t is tempting o say, with Sellars, hat allawareness s a linguistic

affair.30But if one says this, then the whole notion of leaming more aboutman'sknowledge by studying rats begins to seem silly. Even if one does not

go out on Sellars'particular imb, however, one canwonderwhether there is

anythingmuch in common between the sort of 'awareness'which rats and

babies have of smells and the sort of awarenessmanifested by, e.g., the

remark'It smells like a dead whale'. The latter, as Sellars says, can have a




place "in the logical spaceof reasons,of justifying and being able to justify

what one says"31 but the formerhas a place only in a causalstory (or in a

fantasy in which the rat is thought of as sayingto itself "It ratherreminds

me of the smell that morning at the seasidelast year, so..."). When one

remembers he epistemologicalconcernsaboutjustifying empiricalknowledge

which gaverise to the thought-vs.-sense istinction n the firstplace,one maywell feel, with Green, that the psychologists'notion of 'datum' is designed

preciselyto blur the differencebetweenaccountsof justificationandaccounts

of acquisition.A bit more preciselyput, the notion of 'beingan Xish datum'

bridges he gapbetween

(a) a particularwhich standsin a certainrelationto some organ,and

producesdiscriminativebehaviortypical of the behavior nduced

by the presenceof anX

and

(b) belief in, or a tendency to believe, some suchpropositionas thatthere is anX at hand.32

To get this difficulty into focus, consider once againthe organism-com-

puter analogy. When we want to describethe input to the computer, wehave a fairly simple choice between a 'hardware'description of particulars

(e.g., the presenceor absence of electricalcharges n variouswires)on the one

hand, andon the otherhand certainbeliefs ('the number n registerB is largerthan in registerC', 'the cash receipts yesterday were $ 238,645', etc - the

choice depending upon whether a sub-routineor the programas a whole is

being focused upon). Here there is no blurringof the line between a purely

causal set-up and an epistemicone. (Although, of course, there is plenty ofchoice within the first category about whereonce locates the 'interface',and

within the second among 'degrees of abstractness'.)But in the case of

organisms hings are more difficult. In language-usingrganisms, t is obvious

what to choose as input to the 'executiveprogram' where this is taken to

be theself-description y the subjectof his own activity,andhis ownjustifica-tion for it. The input will be, roughly, the beliefs accepted by the subject

without conscious inference. But when it comes to 'sub-routines'- thesubject-matterof psychology as opposed both to common-sense elf-descrip-

tion and to physiology - consciousness s obviouslynot a relevantcriterion.Nor is study of the brain,since even if one hadthe physiologydownpat, one



166 RICHARD RORTY

will still have to pick the place where the physicalstops and the psychical

begins. One will need to know where to drawthe lines aroundthe 'blackbox'

whose internal organizationone is trying to guess. Thus psychologistscan

view the input to the organism-as-visual-information-processing-devices (a)

visualqualitiesof perceivableobjects in the vicinity, (b) lightwavesreaching

the eye, (c) light wavesreaching he retina,(d) patternsof retinalstimulation,(e) neuralcurrents sent down the optic nerve from the retina, (f) patternsof

neural excitation in the visual cortex, or a lot of other things. Which one

chooses determinesa lot about what sort of experimentsone runs andwhat

sortof 'sub-routine' ne hypothesizesto explainthe experimentalresults.Yet

it is hard not to feel that there is a 'right'choice to be made, because it is

hard to avoid thinking of the placewherethe body ends andthe mindbegins

as something to be discovered, rather than arbitrarily elected for heuristicconvenience. But this is just the same difficulty as determiningwhere the

'true data' (the ones which are simply given, with no interpretation)are

found - at whatpoint sense('receptivity') eavesoff andthought('spontanei-

ty', 'processing')begins. It is obvious that if we take 'Xish datum' in the

causalsense - sense (a) above - one can cut up anymachineor organismn

an indefinitely large number of ways, calling events at the cut 'data' and

events further on 'processing'or 'interpretation'.But when 'Xish datum' istaken in sense (b) - as belief in a proposition,or a tendency to hold such

belief - the whole notion of 'the simplest element of knowledge' comes

back to haunt us. For it looks as if it were an importantquestionaboutjust

where - in the mind? in the brain? those elementsarefound. So it looks as

if there were such things as the 'right' sub-routinesto be found - those

chosen, so to speak, by Evolution as the Chief Programmerof the human

species. Yet what would count as 'right'here? Once both introspectionand

physiology are put out of court, how do we know that we are dealing with

tendencies to hold beliefs, rather than with causalantecedents or accidental

concomitantsof such tendencies?

So much for the puzzles which the notion of 'datum'cangenerate.I think

that these puzzles can be clearedup, insofaras they raisedoubtsaboutactual

psychological research programs, simply by noting that nothing in such

programsdependsupon whetherthe 'data'describedare 'trulydatal' or not.

Nothingdepends, ortheformulationof 'sub-routines'whichcan be confirmed

by experiment and used to predict, on whetheror not empiricalknowledge

has 'foundations'.No such researchprogramcould tell us what suchfounda-




tions mightbe. Once we get rid of the notion that some data are more datal -

more primitive, more basic, more pure - than others, the purportedly

'confused' notion of 'datum' can be seen as just a harmless and handy

metaphor. To say that the retina 'sees' or that the visual cortex 'knows'

or that a given braincenter 'infers' from what the visual cortex 'tells it' - or

any othersuchconvenientanthropomorphism is no moremisleading hanto

say that the record-changernfers(from the failureof its sensorto tell it that

there is still a record on top of the spindle) that the symphony is over and it

can turn itself off. It is also no better grounded.Wemight learn a greatdeal

about the brain by suchheuristicanthropomorphisms,ust as barbarianswho

-had never seen a record-changermight thus learn a great deal about record-

changers.But we shall not be helped to understand the nature of the relation

between the mind and the body', 'the mystery of consciousness', the natureof knowledge',or anythingelse which smacksof philosophy.

This somewhatdisappointingandblandconclusion is, I think, all that there

is to be said about the psychologists'useof the notion of 'data'.If I amright

in saying that the soul or the mind would have been an unnecessaryand

unintelligible notion if we had had a more perspicuous physiology, it is

because the notion of 'elements of knowledge' would have been equally

unnecessaryand unintelligible.The notion of common andpublicdatawhichreceivepossibly idiosyncraticand wrongheaded nterpretations s a perfectly

reasonablebit of commonsenseepistemology. But, sinceAristotle,it has been

blown up by philosophers into a distinction between faculties, and since

Descartes nto a distinctionbetween'categoricallydistinctmodes of existence',

or 'spiritual'and 'material'substances.Yet from the point of view of actual

research- actualformulationof sub-routines,as it were - the common-sense

distinction s all we need. A distinctionbetweenthe propositionswhichalmosteveryone would accept ('There's somethingwhich looks red to Smith out

there') and those which aremore dubious('It's that rarespeciesof butterfly

we've all been looking for') is all that is needed as a basisfor constructing

'models'. When we anthropomorphize brain centers, or the retina, or

'consciousness'or 'the faculty of spontaneity', we think of this organ (or

faculty, or whatever)as standingto whatever we regardas its input as the

subjecthimself stands to non-controversial erceptualreports.The important

point is just that this model embodiesno confusionbetweencausalprocesses

in hardwareand nferentialprocesses nsub-routines.Forit iswholly concerned

with the latter.The'confusion'whichphilosophershaveindeedmade between



168 RICHARD RORTY

particulars Xish data in sense a) and beliefs or propositions (Xish data in

sense b) looks as if it were being made by psychologists.33 It looks this

way because the original mpulseto psychological research n the 19th cen-

tury was an outgrowthof this confusion, and becausecontemporarypsycholo-

gists, in revolt against behaviorism, ometimes like to think of themselves as

doing 'scientifically' what Locke andKant did in their armchairs.But there isall the difference n the world betweensaying

(1) We must isolate those non-propositional tems of awarenesswhich

are the foundation for belief in propositions

andsaying

(2) We can treat such items aspattems of retinalexcitation as if they

were beliefs in orderto use the metaphorof 'inference romdata'

in constructingmodelsof mentalprocesses.

Psychologistsneed say only the latter.If they confine themselves o this, they

can follow Putnam in treating the distinction between 'brainprocess' and

'mental process' as of no greater philosophicalinterest than that between

'hardwaredescription'and 'descriptionof the program'.34The temptationto

say the former- the epistemologically-motivatedemptationto 'discover helink between the mind and the body' - can be treated on a parwith the

temptationto raise the question"How can the computertell that that pattern

of chargescomingdownthe wireis the total dailycashreceipts?"

v

The replies which I have offered to Wittgensteiniancriticismsof empirical

psychology suggesta position which can be outlinedin the followingtheses:

(1) The source of Wittgensteiniananimus against psychology is the

subjects'sunfortunatephilosophical association: The real objec-

tion is to epistemology, and to psychology only insofar as it is

taintedby epistemology.

(2) Wittgensteinianshave confused criticismof foundationalepistem-

ologies - the sort of epistemology which engenders he notion of

private languages earnableby inner ostention, non-propositional

knowledgeof particulars, tc. - with criticismof the metaphysical

claim that mentalandphysicalentities are 'irreduciblydistinct'.




(3) Attackingthe notionof 'irreducibly istinctontologicalcategories'

has led Wittgensteinians o an uneasy alliance with traditional

behaviorism, and a hostility to all the 'interveningvariables'

postulated by psychologists.

(4) But the notion of 'irreducible'has been taken too seriously by

Cartesians,by behaviorists,andby Wittgensteinians.There arenocriteria for determining 'reducibility' except perhaps for the

discovery of necessary and sufficient conditions in 'reduced'

terms for statements about 'unreduced'entities. But no such

conditionshaveeverbeendiscoveredaprioriin anyphilosophically

interestingcase,andit is hardto see how they could be.

(5) If we cease to worry about reducibility and irreducibility, the

relation of psychology to other disciplines will cease to seemproblematic or interesting,as will the line between the body and

the mind. The relation between psychology and physiology will

be no more problematicthan the line (rough and pragmatically

drawn) between sociology and anthropology, or that between

economics and politics, and 'the mental' can be definedjust as

'whateverpsychologistsfind it useful to talkabout'.

In developing this position, I have avoided discussing such philosophical

slogans as 'mental events may tum out to be identical with brain events',

'psychologicalstates are functional states', and the like. I regard uchslogans

as, once again, the result of takingthe distinction between mind and body

more seriouslythanisreallynecessary.I havedealtonly with the quitelimited

question: are there relevantphilosophicalcriticismsof psychologicalresearch

programs,based on the 'Wittgensteinian'evolt againsttraditionalCartesian

notions? If I have succeeded in showing that there are not, then the sameconsiderationsshould show the converse: that no suchprogram ould, by its

success, advance or obstruct any doctrine in 'the philosophy of mind'. The

notions of 'philosophical problems raisedby psychologicaldiscoveries'and

of 'philosophicalcriticism of psychologists' methods and doctrines' must

standor fall together. It would be as well if they both fell.

PrincetonUniversity



170 RICHARD RORTY

NOTES

* This paper is based on research supported by NSF Grant #GS-33489, which I grate-fully acknowledge.' Norman Malcolm, 'The Myth of Cognitive Processes and Structures', in CognitiveDevelopment and Epistemology, (ed. by T. Mischel), New York and London, 1971,p. 387.2 Malcolm, p. 389.

Malcolm, p. 392.4 Malcolm, p. 391.

See Wilfrid Sellars, 'Empiricism and the Philosophy of Mind', in his Science, Percep-tion and Reality, London and New York, 1963, especially pp. 128-34.6 See Sellars, p. 168.

Ludwig Wittgenstein, Philosophical Investigations (3rd edition, New York, 1958). Forthe 'no private language argument' see Part I, Secs. 269-281. For the 'infinite regressargument', see, e.g., Part I, Secs. 86, 213-215. The 'no foundations argument' is suggest-ed rather than stated, but it can be viewed as a generalized version of the long and

oblique criticism of 'ostensive definition' which opens the Investigations. I have discus-sed the importance of Wittgenstein's use of these last two lines of argument in 'Pragma-tism, Categories, and Language', Philosophical Review, LXX (1961), 197-223, especial-ly pp. 214ff.8 Sce Rorty, 'Verificationism and Transcendental Arguments', Nous V (1971), 3-14.See also, J. T. Saunders and D. F. Henze, The Private Language Problem, New York,1967.9 The point was first made: I believe, by Hofstader in 'Professor Ryle's Category-Mis-takes', Journal of Philosophy XLVII (1951), 257-70.10 Jerry Fodor, 'Could There Be a Theory of Perception?', Journal of PhilosophyLXIII

(1966), 371.'" P. C. Dodwell, 'Is A Theory of Conceptual Development Necessary?' in CognitiveDevelopment and Epistemology (cited in n. 1 above), p. 382.12 See Fodor's remark (p. 373) that 'a single-minded adherence to Ryle's suggestionsuffices to 'eliminate' not only 'para-mechanics' but also automotive mechanics". Thepoint is that Ryle's operationalist reasoning rules out any 'construct' - 'gene', 'vaporlock', or what have you - which is postulated to explain 'observables'.13 I suspect that no amount of work in "meaning analysis" or "theory of reference"could help avoid Cartesian dualism, given this result of further empirical inquiry, andthat only this result would make Cartesian dualism something more than a philosophicalartifact.

4 Fodor, pp. 377-378.15 Dodwell, p. 370.16 Dodwell, p. 372.17 Dodwell,p. 372.18 Dodwell, p. 378.19 This would be denied by many psychologists. E.g., Gregory, citing with approvalHelmholtz's notion of 'unconscious inferences' involved in perception, says that

"we must be clear that there is no 'little man inside' doing the arguing, for this leads tointolerable philosophical difficulties. Helmholtz certainly did not think this, but hisphrase 'unconscious inferences' and his description of perceptions of 'unconscious con-

clusions' did perhaps suggest, at the time, to people unfamiliar with computers, somesuch unacceptable idea. But our familiarity with computers should remove temptationtowards confusion of this kind. For we no longer think of inference as a uniquely humanactivity involving consciousness." (The Intelligent Eye, New York, 1970, p. 30)




I disagree with Gregory that the little man leads to 'intolerable philosophical difficulties',mainly because I do not see that little computers raise fewer problems, or are less 'con-scious' than little men.20 See Sellars, p. 182 on 'commentaries' and pp. 192ff. on real sense impressions.21 Quine makes this point: "Fire one afferent nerve and chalk up one sensory atom,and awareness be hanged. Ironically, though, it is just here that we do best at last to dropthe talk of sense data and talk of nerve endings instead." (Grades of Theorecticity', in

Experienceand Theory, (ed. by L. Foster and J. W. Swanson), Amherst, Mass., 1970,p. 3.22 Fodor also has suggested that the distinction between 'functional' (or 'program')analysis and 'mechanical' (or 'handware') analysis in psychology is irreducible and notjust a matter of convenience. See his 'Explanations in Psychology', in PhilosophyinAmerica, (ed. by M. Black), Ithaca, 1965, p. 177. I have argued against this suggestionin 'Functionalism, Machines, and Incorrigibility', Journalof PhilosophyLXIX (1972),203-220.23 On this sort of relativity, see William Kalke, 'WhatIs Wrong with Fodor and Putnam'sFunctionalism',Nous III (1969), 83-94.

24 I do dot mean by this that we would not have thought of ourselves as having beliefsand desires, and as seeing, inferring and so on. But we would not have saddled ourselveswith the notion of a 'separable active intellect' nor a Cartesian 'immaterial substance'.Our concept of mind would have been much closer to Ryle's or Aristotle's than to theCartesian concept we presently have.25 See, for example, Seymour Papert's Introduction to WarrenS. McCulloch, Embodi-ments of Mind, Cambridge, Mass., 1965. Papert, in explaining the importance ofMcCulloch's work, tells us that "we need no longer be trapped by the dilemma" of

a split ... between psychology, which was based on mechanism but which was unable toreach the complex properties of thought, and philosophy, which took the properties of

thought seriously but could be satisfied with no conceivable mechanism (p. xiv)

The insight which is to resolve this dilemma has as its 'principal conceptual step'

the recognition that a host of physically different situations involving the teleonomicregulation of behavior in mechanical, electrical, and even social systems should be under-stood as manifestations of one basic phenomenon: the return of information to form aclosed central loop (p. xvi)26 It is tempting to think of 'intervening variables' postulated by psychologists (with'sub-routines') written in terms of them) as mereplace-holders for undiscovered neuro-logical processes. We usually do, indeed, assume that when neurophysiology reaches acertain point it will serve as a touchstone for choice

among competing psychological'models of the mind'. But it is important to see that even if we somehow discovered thatneurophysiology will never reach the stage we had hoped for, this disappointment wouldnot make the psychologists' work any more dubious, either 'methodologically' or 'meta-physically'.27 The notion that it is important to discover what is 'innate' comes out in such ques-tion as "Does all knowledge (information is the contemporary term) come through thesense organs or is some knowledge contributed by the mind itself?" (J. J. and E. J.Gibson, 'Perceptual Learning: Differentiation or Enrichment?', PsychologicalReviewLXII (1955), 32.) Gibson and Gibson take this question with entire seriousness, andurge that, pace Hume and Helmholtz, perceptual learning is not unconscious inference

from memory-traces, but simply "increased sensitivity to the variables of the stimulusarray" (Gibson and Gibson, p. 40).28 T. H. Green, CollectedWorks, ondon, 1885-1888, I, p. 19.29 See Noam Chomsky 'A Review of B.F. Skinner's VerbalBehavior'LanguageXXXV



172 RICHARD RORTY

(1959), 26-58, especiallySection 3. See also Paul Ziff, 'A Response o StimulusMean-ing',PhilosophicalReview LXXIX (1970).30 Sellars,p. 160. See also Toulmin'sdiscussionof Urmson'sversionof the same thesisin Toulmin'sThe Concept of 'Stages' n PsychologicalDevelopment', included n Cogni-tive Developmentand Epistemology,cited in note 1 above), pp. 42ff.31 Sellars,p.169.32 GeorgePitcher A Theoryof Perception,Princeton,1971) develops he view that one

could understand verythingabout perceptionby sticking o the latternotion. See espe-cially pp. 73-74.33 See J. 0. Urmson, 'Recognition',Proceedingsof theAristotelianSociety, N.S. LVI,1955-56, pp. 259-280.3 Hilary Putnam, ('Mindsand Machines',n Dimensionsof Mind, (ed. by S. Hook),New York, 1960, 138-164 (especiallythe concludingparagraphs)) as the firstphilos-opher to clearlypointout that the moralof the analogiesbetweencomputers ndpeoplewas not "Computershelp us understand he relationbetween mind and body" butrather"There annotbe any problemaboutthe relationbetween mind andbody."

Witt Gen Stein Philosophy and Empirical Psychology

Documents

Transcript of Witt Gen Stein Philosophy and Empirical Psychology