ag Tard de Ghari's The Phenomenon of Man
position in dhese matters been decisively refuted?” But anyone who
Adopts this position muse ager that any discussion of Telhards|
work by sclentiats must wait wpon the question whether the
fnceptual system of Cartesianiem ix selfeonistent or see
contradictory And this a manifestly philosophical question,
9
‘The Homunculus Fallacy
In the Philsphical Imestiations, Witgenstein says: ‘Only of a
Thumm being and what resembles (behaves He) living human
being cam one say: it has sensations; it sees is blind hears dea
i consious or unconscious” (1, §281). This dictum soften rejected
in practice by psychologists, physiologists and computer expert,
chen they take predicates whose normal application eto complete
human beings or complete animals and apply them 0 parts of
animal, suchas brains, ort electrical systems. This is commonly
‘defended as harmless pedagogieal device 1 wish to argue that tis
‘dangerous practice which may lead to conceptual and methodol
‘Seal confusion Isl all the reckless application of umarebeing
Predicates to insuficienly human-like objects the homunculus
Elling’ since its most naive frm is tantamount so the posteltion
fof fle man within a man to explain human expetience and
behavione
‘One ofthe first philovophers to draw attention to the hormancae
Tus fallacy was Descartes. In his Dips, he dserbes how “the
‘objects we lok at produce very peeec images in the back of the
eyes" He encourages his readers to convince themselves ofthis by
taking the eye of a newly dead man, replacing with paper or
‘eagshell the enveloping membranes atthe back, and plating i
inside a shutter so ast let igh through ft into an otherwise dark
room. "You will sce (I daresay wih surprise and. pleasure) 2
Dietre representing in natural perspective al he objects outside.”
You eansot doubt, he continue, that quite similar pete x
provduced in a living man’s eye, on the Tning membrane
Frther, the images are not only produced in the back of the eye
"Daca Dip ell Wy rnd Amo |
Gun amas16 The Homes Fallacy
but also sent on to the bain... aud when tis thus transite to
the inside of our head, the pict sll retains some degree of it
fesemblance to the object om which it urginates’ But he
concludes with a warning ‘We must not think tht itis by means
ofthis resemblance thatthe picture make us aware ofthe abjects~
ts though we had another pair of eyes to see it inside our brain?
(op. 245-6)
To think ofthe bran as having eyes and seeing the retinal image
would be one way’ of committing the homunculus fallacy. Bt in
Spite of warning us against the fallacy at this point Descartes
himself commits it when he comes to discus the relationship
between the soul and the pineal gland:
{we se some animal approch the ight reflected fom it aly
depicts two images of tone i each of cur eye, and thee 0
images orm two tars, by means ofthe opt nee, in he erie
race ofthe brain which faces cavities then fom thee by
‘means of the animal spine with which caves are led thee
age 9 radiate towards the He gland which surounded by
‘hese sprite tha the movement eich nach psn of ane fhe
lager tens towards the mame ott ofthe land towards which
‘erste movement which se pit of he other age whieh
fepresente the same part f this anima By thin means the two
Inne which aren the ban fr butane pon the gland, which,
cng mediately pn Ue sol, causes it sce the frm of hit
To speak of the soul encountering images in the pineal land iso
commit the homunculus fallacy for pace Descartes, 4 sot is 90
‘more a complete human being than a brain i. In ill there
nothing philosophically incorrect in speaking of images in the
brain; Descartes himself is anxious to explain that they are very
schematic images and not petres except in a metaphorical sense
No images have resemble the objects they repent in all epee
‘semblance in ew fatre enough, and ery ofen the
Paecion oan image depends on tat reeling the abject at
2 Des, Pain he She Mephl Wi of De
ES Manet Rn Camo tp rh oe
‘The Homes Fllgy uy
much ait mig rinatance, engraving, which consist merely of
lik prea over pape, represent ses owns, meh and
‘en bas and semper (Dapp. 248)
There would be nothing philosophically objectionable in the
suggestion that these schematic images might be observed by a
Tain surgeon investigating the gland. What ix misleading it the
sggeston that these images are vibe to the sel, whose peep
tim of them constitutes secing, What is wrong i that exactly the
‘same sorts of problems arise about Descartes’ explanation as about
his sapcandim, To the Arstoelians who preceded Descartes,
seeing necessitated a non-mechaniste phenomenon taking place in
the eye, Desearesinrexuced new mechanisms, but inhi system
the ‘non-mechanisic event in the eye is replaced by 2 new
ronmechanistic reading of patterns in the pineal gland. The
interaction between mind and mate is philosophically as puzzling
2\fr inches behing the ey a tin the eye tse
‘One danger, then, of the homunculus fallacy is that in problems
concerning peteeption and Kindred matters it conceals what i Left
to be explained. In the cae of Descartes, we ate put on our guard
by the quaintess of some ofthe physnlogy, 30 that we have 20
Aiicaly in discovering the gaps in his account but the phil
topical hieus can coexist with much more sophisticated physiol
fea information
‘contemporary expert on perception, Profesor R.L. Gregory,
sc the beginning ot his bok The Bond he Bra, echoes Descartes
‘warning against the homunculus fallacy
We are w familar with sein, tha take eap of imagination to
reais that here ae problems to besled, But conser We are
‘Bren tny dored pide down images inthe cjon and we 0
‘Spare told bjs in suroundn space. Bom the patter of
‘mulation othe retinas me peeve the world ef objet, abd his
‘eohing short of «mice
‘The oe i fen dover tke a camera, but the quite
‘uneamerafike features of preeption which are most intresting
How i infrmation fom the ee cde ito newal ens, it he
Uioguage of the rain, ad seconsiaed nto experience of ur
‘ouding object? The tak of eye and brain site dillrent omuo The Homa Fllcy
citer photographie or # telovton camera converting ajets
Ineely info images, Thee ia temptation, which must be voded,
tony that the eye rodic pictures nthe bai A pice fa the
Train suggest the ocd of sce Kind oir eye ose but cis
‘would need a farther eye tose pitare- and von nan enlee
‘ere of eyes and pictures. This abd What the ye do
{eed the bain with nkrmaton ded incur acy = chan
‘Seta impulses ~ which bythe ide and he ptr brain
cy, represen object. We may the an analy Irom writen
guage the eters nd words this page have tin meng
10 thove who know the language. They aft the reader's bain
ppropritly, bat they are not pictures. When we lok 8 sme
thing, the ptirs ef acral sey represent the object and tthe
brain theaject, No Internal pte inven
“The warning against the fllacy is excellent; but the fallacy is
itelfmpliedin the suggestion tha the bran know language and
thar it has an object Ike the objects of perception, A converse
fallacy is committed when itis said that we are given tiny,
distorted, upride-down images in the eyes and that we perceive
patterns of stimulation on the retina, Heres nota bogus subject
of perception which is being aupplid, but 2 bogus object of
perception.
"The reader may fel that tis is completely uni criticism, The
words [have erticzed ae taken fom the fist page of popular
‘book: Whats the harm in personfving parts ofthe body inorder to
dramatize aclentf information hich ean be tated in completely
neutral metaphoriee Language?
Whether dramatization is good pedagogy depends on whether
the important events happen on or offstage, The overall paychlo-
seal problem of perception could be stated a flows: how dors |
human beng cope with the available sensory information, sd how
des he act on {0 Or, in one of Gregory's own formulations, how
does information control behaviour? Now tis ia problem which
‘would stil remain tobe slved even if we knew every detail of the
process of election and storage of information; snd one cri
pect of tis the ame whether the information isn the world, in
‘he retinas, inthe CNS. The problem thi what tthe rlation
PRL Ge, he En ei Me 3
The Hommel Fallacy 19
between the presence of information in the technical sense of
communication theory and the possession of information in the
non-technical sense in which one Can acute information about the
word by looking?
For if having information i the same as knowing, thea contain
ing information isnot the same as having information. An airtine|
schedule contains the information about sine departures but the
rine schedule doesnot knw the time of departes ofthe fights
The literate slave on whose shaven scalp the tyrant has tattooed
his state secret doce not fnow the information which his head
‘Acategory dierence is involved here. To contain information is
to be ina certain stat, while to know something isto possess a
‘certain capacity. A rate (such at being a certain shape ozs, or
having a certain multplicty or mathematical structure) is some
thing deserbable by its internal properties a capacity (rach asthe
bility to an afour minut mile otto speak French) is dscrbable
‘nly by specification of what would count a the exercise of the
fapaciy. Sates and capacities are af course connected: in the
Simplest case there san obvious connection between being around
og (state) and being able to ft into a round hole (eapaciy). But
the connections are not always (asin chat ease) analyte and many |
Forms of expertise consist in knowing which rates go with which
capacities (eg. what types of mushroom are poisonous, which
alloys wil stand whieh stains).
Knowledge isnot a state but a capacity, and a capacity of a
unique ind. The state of containing certain information is no
‘doubs connected with the capacity which is knowledge ofa certain
fet but the two are not identieal a the earlier examples show. We
‘may wonder what extra involved inthe knowing that p over and
above containing the information that p.-What is Knowing’ a
fapaciy to do, and what counts as an exercee ofthat capacity?
Clearly, there a no simple answer- One cannot specly behaviour
typical of knowing as one can specify behavior typical of anger.
(One eannos even specify behaviour typical of knowing that por 3
ven pt owhat behaviour the knowledge that pill lead will
‘depend on vshat one wants. For instance, knowledge that the
tvindow is open will lad to different behaviour in the case of130 ‘he Hommenkr Fall
someone who want it open and i the cae of someane who wants
Ftshut Tobe sure, the verbal utterance of isan aesiity which i
"uniquely expressive of dhe knowledge or belie that ps but even x,
this doesnot at all mean that anyone who knows that wil ever sa
tha
‘There is, chen, no simple way of specifing how knowledge ets
ceapresed in behavioue and why some pices of knowledge do not
seem to affect one's behaviour a all. Sil, to hnow isto have the
Ability to medi one's behavior in indefinite waye relevant tothe
pursuit of one's goals. Its because the aiineachedle doesnot
hhave any behaviour to be mosified by what is writen on it that i
doesnot know what the ight tines are
Tet us return rom knowing to secing. Seeing, when not lary,
Involves knoving: vision might be defined, erly, icuary, but
‘ot uninformatvely, asthe acquisition of knowledge in the vial
‘mode. Tn the Aristotelian tradition, prior to Descartes, i used tobe
‘id that it wat not the eye that saw, nor the soul, but the whole
‘nganism. This was because the normal way to discover whether an
fmunism sees is not just to study ite eye, but to investigate
‘whether its behaviours affected by changes oflight and ealour, te.
Consequently, an explanation of seeing mat be an explanation not
nly ofthe acquisition and storage of information, but alo of what
‘makes dhe containing of thi information into knowledge eit
Felation to behaviour
Tn his paper “On how so litle information controle so much
behaviour, Gregory well sys
Peshaps the most fundamental quotion in the whole Bld of
experimental peychaogy it how fare behavior contd. by
aren availabe sory nfrmation nd hw far by lormasion
ttrady aored in the cental nevus spon?”
Bat in that paper he petenta a theory of seting as selection of
internal models without ssyng how the fnernal models ar elated
‘Reeach Report 1, Ap 1968p (Te page wan so plied name «
Tw! Bog ed Mae es Dey ee)
The Home ale ast
to behaviour, He speaks of a model “ling up the appropriate
Inusle powerlifting a certain weight and fades mediating
Spproprate behaviour (p 8), but he sowhere shows how these
ieaphors might be toca fot Real language. What he realy
txplaneis hw infrmaton of etn ype might reach he ral.
Tow let ws suppose that hi expanatin fh proves completly
correct. Even he ercial problem remains: and hat stil 0
te dane masked forthe reader, fo fr Gregory Kimsel, by the
Ue of homenculs predicates of the brain ad the se of meno
tr vepresetaonal or symbolic predates of tems in the bai,
Constr th ellowng psrage om the same pape
In gener the e's images are inloealy impotan onlin fr
1 honcoptel ature ean be read om the internal model hey
‘Site tmage ae mriey patches fight wich cannon be etn or
te dangerous ~ ut they serve at symbole fr seketig eral
te which nea te oni fetus wal oso. es
(hi reading of bjt characters from mages that al
perception 5)
Bat even if this mechanism is estentil for visual perception its
‘ot vinual perception. Selection af internal models would. be
possible, as sexing would not, in an iolated optical system
Incspabie of behaviour, This ot jus the ordinary language pint
ree wouldn't call such & thing seeing’ ~ itis methodological
point concerning the nature ofthe problems wo be solved and the
Feasonableness of extsspolations fran acquired reslts, The sion
that what i described i visual perception i encouraged By the use
‘ol langsage sch ae eatores cam be ead" and symbol or selecting
Tater inthe same paper Gregory write:
‘On this general vw pescepton tnt dey of sensory informs
‘lo bar rater 0 he neral muse selected by senor informa
‘hn. Inden he cuvent prerpon i the reaing sto modes,
ea)
Clay timate ein what prepion sy saying
that ee perception noe of X but of Y: iC wonder what pectin is
how am I helped by being tld that it of ¥ rather than of XPra The Hamncaas Falla
Gregory sense thi: dat is why his ist statement of this thesis
flowed by indeed” followed by a statement of an incompatible
thesis: Pereepsioneannot both be ofthe models and be the models
‘So far my objection to the homuncls mee has ie hat es
pedagogically and methodologically dangerous, as helping to cloak
the nature of problems tobe solved. But there isa more dangerous
tflect ofthe model which alone telly deserves the mame alley
[Let us suppose that we waive our objections to the use of
sman-bring predicates for on-human-beings ike brains. Let us
allow i oe said thatthe brain iP where P i some predicate
‘whose natural application sto whole human beings (It may, afer
all, be used in quotes. It usually isthe Bet time) Ther i stil an
important temptation to be resisted: the temptation to argue from
‘This man ie P
to This man's brain is P
or se sea. Gregory doesnot always resist his temptation. At the
breinning of the quoted paper he argue that learning or stving
particular events always ontogenetic. Naturally stored informa:
tion, he says, has two origina ancestral disasters, and previous
experince of the individual stored as "memory (1, his quotes).
‘To prove that storage of particular events x always ontogeti, he
‘Whats ceria is hatintrmaton gine phylogenetically always
ofthe genera sil! Kind. We aren abe oral nda ett
‘experienced y or ancestors p)
‘And @ propor of learning sills such a tennis and piano playing, he
os
‘We maybe able to eal thee particular games or concerts, but
sly es ot nda pat vents which are sore bat ater
Sppropite behaviour and vente (p 1)
Here the homunculus fallacy i committed thus "X remembers
‘hat pis being treated as equivalent to "X has stored the event hat
The Home Fallacy 39
P The only reason given for saying that information about
parscular events isnot stored phylogenetically is that we cannot
Fecal individual events in our ancestors’ lives. But this to argue
from “this man is not Po thi man's brain is not P™ which
fallacious, even ifthe man’s brain's being Pisa necesary condition
for his awn beng P.
In another paper, ‘Percept illsons and brain model
Gregory considera whether the bran is est regarded a a digital or
san analogue device. He writes
1c ot imple pone he aa ci contain
tate seston When's as
‘anne supone thatthe strata problems are solved by cor
Pos mal ene, ining ep
better wo make the lane cai for children and anal: tha hey
She pmo ogo by ng anges of met
SSeccpeas wht iting sata chs
SORT aca h ie tinct Poel tarog
SSSy eso seq ering ot mama ree
Piet tp ie ella hie Sg
is rn pen obec lesa ey
rach twit ema ca”
Here the homunculus fallacy is committed in the sentence
“Perceptual lemming. surely cannot require the learning of
mathematics’, Irs the child that i ding the perceptual learning
what, if anything, is supposed to be learning mathematics isthe
Child's brain, Iti empansible that child building toy bricks
Should know advanced mathematics; but fom this noching at all
follows abou what information is contained in the chills brain
TVeondlude that there is good reason t heed the warning of
Wingensten with which this paper began. The moral is no that
the human-being predicates cannot have thee use extended at al,
‘bu that they must be extended cautiously and se-consciously, and
that Wf they are extended one may not argue fom the aplication of
> LG, oop an ei de, Pig of he 8,19
(ot yp tap134 The Homes Fallacy
such a predicate to whale hutnan bing othe application of the
transferred predicate to anything other than the whole fumaa
being
In brief postscript, wish to classify some of the points made
above, and to disown some ofthe theses atrbuted tome by some of
thore who read that part af this evay. Fist 1 do not accuse
Professor Gregory of mistaking mechanistic description or concep-
‘ual analysis; nor do Think that ether the philosopher's answer or
‘the neurophysiologis’s answer to the question “what i perception”
enjoys a privileged stat. Secondly, To not object to every
textension ofthe application af'a predicate ram a sentient whole to
1s part. Thirdly, {took up no position om the general question
whether conscious activites canbe said to br (nothing but) the
Imicro-trictral processes postulated to explain ther, Twill
fxpand each of these points, snd then briely restate why I all the
homunculie fallacy a allay.
dot think that Gregory i under any illusion tae he is ding
conceptual analysis. T think he is engaged in constructing, and
testing experimentally, hypotheses about the mechanisie neces
sary to explain the phenomena af visual perception. But concep
Analysis is relevant o what he s doing in evo ways, Fis, analysis
OF the concept of perapin ix necessary to delimit what are the
‘Phenomena to be explained; scondly, analysis of the concepts of
qt leguage show that such things a8 secing and decoding
annot be done by brains unless we can atte to brsins certain
types of behavior which we ean attribute to whole human beings,
To ataibute such activities to brane without suggesting how the
relevant behaviour might be attributable to brains iT aintaned,
{o mask empirial problems which remain tobe solved
‘The moral of my paper, Tsai, was not that human-being
predicates cannot have thei we extended a al, but that thir uae
Inust be extended cautiously. Consequently, Cam unmoved ite
Pointe ot hat hands can grasp and old: such extensions set to
‘me well within the hounds of caution, Moreover, my objection was
pot extenialy to predicates of wholes being attached wo predicates
Of part, but to predicates belonging to human beings being
‘Te Hommetet Fllgy 135
attached to non-human beings. The same illacy could be commit
ted (though my name for it would not be apt) by the ineations
pplication of human-being predicate wo wholes of which human
bring ae parts, such as communities and states. Populations, like
human beings, grow and shrink; but it would abviouly be
falleous to angue that a human being was shrinking because the
Population he belongs to ie shrinking, or that a population ix
rowing becaute every member of it i growing. And states may
Ihave intentions which none of thee ciaens have, (Cf. Wittgen
stein, Zt, pp. 148)
"The question whether perception can be said tobe ida! with
plysilogical procestes stems to me to lacks clear sense, and T do
rot wish to anwer itone way oF the other. My complaint against
Gregory’s identification of vstal perception with bis postalated|
selection of internal modes was not based on «general thesis chat
peteeption cannot be identical with a brain process. Though tates
find capacities sre conerptually diferent, it need nt be misleading
to aay, eg: that a pegs ability to Bt imo round holes tris
roundness. In the same way it may be that there ia physnlgical
procese~ the acquisition of physiological state ~ which can be
‘Said w be visual perception. But no one can claim to have identified
such a process ni he has Brought out is connection with the
types of behaviour which are the criteria for the occurrence of
‘visual perception. And this Gregory has not done
‘allay, riety speaking, sa orm af argument which can lead
from tue premises toa fale conclusion. The inappropriate use of
predicates not being form af argument, ot stely a fallacy, as
Totserved ny pape. Buti ead toa frm of argument, which T
‘aimed to detect in Gregory's articles, which i fallacious in the
frit sense of the word: the argument that because a certain
Thuman-being predicate attaches os human being it attaches to his
brain, or ee noua The mere inappropriate use of human-being
predicates may be calle fallacy im an extended sense, because
Inay siggrst conlasions which are unjustified notably the conc
‘lon that more ha been explained by a peyehological theory than
has infact been explained
‘Normally, in an adult human being, the ability to see carries
with ie the ability to say what ie seen, though of course not136 The Hommel Falay
everything which is actualy seen is actually talked about. The wae
fof language to report what is seen, like any use of language, is
emarkably fee fom stimulus contol ~ a point which har been
repeatedly’ made, in general tenms, by Chomsky. No account of
human pereepson can approach adequacy unless it inches an
explanation ofthis fact. Consequently, even ie knew every detail
‘of physiological proceses by which visual information reaches the
brain, and every detail ofthe physiological processes by which the
linguistic utterance of visual reports i prduced, the problem of
the telationship between the input and the output would be
‘completely untouched, This problem ia major part ofthe problem
‘ofthe physiological explanation of perception, adits existence
‘masked by tall ofthe Bain eading features af jects rom images
and calling up appropriste muscle power.
10
Language and the Mind
‘We may ssfly think ofthe lnguage ficult, the number ay
tnd eters, as'mentl organ’ napus othe heart othe val
‘atom othe jtem of motor corinatin and plang. There
peas ta be 9 clear dcmateaton line between phys organ,
perp nd motor ystems, and ent cules nthe respects
Ergun!
"To view the language faculty as an ongan like the heart involves a
cp pilosophial confusion, Chomhy’s description ofthe mental
structures that he investigates introduces am ielevant metaphysi-
fl element atthe interface betwen physiology and psychology. 1
Intend to justi this complaint by a detailed examination of some
‘cal partages in Chomshy’s latest book, Rules and Repetto
But before doing so let me, inorder to avert misunderstanding, list
‘number of points on which pilosopers have picked quarrels
‘with Chomsky and’on which T chink ie he, and not his
‘hilsophical rites, who i inthe right
have no quarrel with the iea that thee are faculis of the
‘mind, and thatthe mind in thst sense har = modular structure. 1
have no quarrel with the notion of deep structures, or mental
representations, in the only sense in which thee are rally relevant
to the exclting empirical inquiries that Choneky and his associates
fre engaged in T have no quarrel with the idea that in sing
Tanguage we display tacit knowledge, operating rules and prince
ples that cannot in the normal way be brought to conscious
formulation. Finally, Ihave no objection to innate mental suc
tures on the grounds oftheir innateness. Obviously, human Beings
are born with certain abies, including abilities to macure aswell,
"Nm. ud rts Ce Ua Pel1s Language nd he Mind
4 abilities to learn. Whether the ability to acquire grammars of
certain kind isan ability to lara or an ability to matute wade
‘certain conditions seems to me a philosophically open question
‘apable in principle of being sted by empirical ing.
Despite this range of agreement, I think that Chomsky employ:
in his writing a confued notion ofthe metal. T should perhape
bein by explaining what T think a non-confied notin of th
mental looks ike
"The mind isthe capacity to acquire intellectual sil. The chie
and most important intellectual kil is the mastery of language
‘Others, such x knowledge of mathematics, are acquired by humar
brings through the anguages tht they have mastered, So the study
ofthe aquisition and execs of language the way paral te
Study the nature of the human min,
‘Someone who has acquired language dros that language
Knowledge of language ivan ability! an ubiity that can be
exerised in many diferent ways, for fatance, by speaking the
language, by understanding what i std to one inthe language by
reading the language, by talking to oneself one’s head in th
language. To know a language ut to have the ait to do these
And sinilae things. Ti concep truth that to sty a ably
{you have to study the exercise of that ability: to investigate what
{he ability to is you have wo investigate what ging i, Soto study
Inoledge of fanguage you have a conser and examine what the
exercise of linguistic knowledge is, The exercise of linguistic
Knowledge can be called, you Hike linguistic behaviour. But
“behaviour must be vndersiond in broad sense, 30 that, for
instance, redting a poem to myself in my head imperceptibly to
hers will count as an instance of my lnuinic behaviour
‘We must distinguish between abilities and thelr postessorson
the one hand and their veces on the other, The ponsessor of
ability is wha he the ability, Lamm the poascsor of my linguistic
ability itis T (and not my mind or my brain) who know English
nd am exercising this capacity in giving this lecture. Similarys my
ar has the expaciy to decrerate can go slower in answe to my
touch onthe ot-brake. The sia the ears ability to decelerate
Js the brake mechani; similarly, my eye va prt ofthe vehile of
ry ability t see. The vehicle of an ability i that part of is
Language and te Sting 139
possessor in vet of which i able to exercise the ability. A
‘chil is someting concrete and more or less tangible: a ability,
fn the ether hand, has neither length nor breadth nor leston Kt
's,ifyou Tike, an abstraction fom behaviour.
An important instance of the distinction between possessor,
tility, and vehicle i the distinction between peopl, thir minds,
fan their brains, Human beings ar ving bodies of certain Kind,
fra have various abilities, The hurnan tind isthe capacity that
yhuman beings have to acquire imellectal abiltin: capacity i
itself an ably, but @seoned-order aby, the ability to acquire
ables. The vehicle of the man mind i, very likely, the human
brain. Human beings and heir brain are physical objects hee
‘minds are not, because they ae capacities. This does not mean that
they are spirits A round pes ability tf ints round hole sot
physical object ike the round piel but no one wl suggests
A spirit It isnot any adherence o spiritualism, but simply concern
for conceptual clarity, that makes uy insist dht a mind is not 3
physical abject and does no havea length and breadth,
a mind tr nota physical object, cami have a stuctare at all?
Yes, itean. Theset of abies through which the mental capacity
serie have relationships to eachother - there ar relationships,
for instance, between the ability to mulply and the ability wo take
square roots ~ and these relationships Between abilities form the
Hructre of the mind. Not only human beings have abilities that
fe structured in tis way: we ean discover the stuctre latent in
the operations ofa pocket ealeulator by ieatilying the algorithms
thar ie uss, To discover the algorithm that a calelator uses, ay,
for the extraction of square roots cals for mathematical, rather
than electron, ingury- When considering dhe human mind, the
Physiologist isin the postion analogous to the electronic engince,
bad the peychologiat sin the poston of mathematician tho i
trying to deduce from the form the calculators output takes (what
Kind of rounding errors it commits et) what ithe sgorthm that
itis using
mmsky makes a distinction tween capacities and their
vekicles, a2 U have done. He deseribes the objects of his atudy at
“human cognitive capacities and the mena structures that seve as
the vehicles forthe eneeie ofthese expacien’. But in erm of the40 Language and he Mind
Aistnetons that [have drawn the mental structures that Chomsky
is imerested in are capacities, not the vices of eapacte ti the
physiological hardware characterise of the exerci of the relevant
‘mental capacities that x the vehicle,
Tn fact Chomsky's mental suucture sem to belong at times to
the world of software, at times tothe world af hardware. The great
majority of what he sys, asa linguist, about the knowledge and use
‘of grammar by language uses, s pene itlligible in terms of|
mental structures, being particular abilities and ther exercises the
‘xerise ofthe ability to operate an algorithm, to diacover the value
fof-a particular grammatical function for a given grammatical
Argument. But from time to time Chomsky the philosopher inter
‘ens to tellus that what he alkng about is otto be understood
28 capacity or ability at all Iissomething quit diferent, whic |
Undeies the ability and its exercise, and ‘which yet docy not
underlie iin the way that the physiological structures and
proceses ofthe brain do.
“To show that iis ponsibe to know a language without having
the expacity o us it, Chomsky ofr the following argument:
marines pertan who knows Engish anders cerebral damage
doce no at he language eters tl a even we
in speech, comprehension, lett supe, even thou
Seppe thatthe fle the injury recede and with no forthe
‘experience or exponue the person reeves the iia apa
{se the Language In the intervening pred, he had apy
peak or understand English, even though, tough the eit
{limatly placa) structures that unde that eapacy were
‘damaged: De the penn know English doting the rng
prio? (P51)
‘The answer, Chomsky says “yer: that is shown by the fet of|
do not wish wo contest the answer that Chomsky gives to his
question: what does sem to me surprising is hi lear assumption
that there is a fact ofthe mater here, tobe settled by considering
empirical evidence. IF it eally were a factual matter whether
person inthe condition deseibed by Chomsky knew English or not,
Language end ie Mind ua
then a thought experiment would be a most inappropriate way to
settle the question. A thought experiment snot an experiment and
oes not provide empirical evidence: ts fneton in philoxophy i
rather to daw atention to the shape and structure of our concept.
This Chomsly'silustration eectvely docs: it shows the fuzzy
cges of the concept ming Beplish Inthe normal ease, a lange
‘numberof extra for the application ofthe concept ae preset: the
person in question can readily speak, understand, and think in
English. We imagine the erteria white away, 40 tha all hat
lefts thatthe person i going later to use English normally, Shall
swe say he knows English n the iterval? Well we ean tay what we
Tike a long as we know what we are doing it sup tous to decide
whether what islet is sufficient fr us 4 eal “knowledge of
Engi’. Perhaps Chomshy i ight thatthe mote natural decision
isto say that its suficent. Fine, then, let us say that the person
ows English But why should we not alo say that the person
retains the capacity to speak English? For exrancous reasons, he
‘cannot te oF exercise the capacity atthe moment, but since, ex
yet, he's going to use en ature without any ofthe normal
Acquistion processes, sit not natural to sy that hel hed on 0
icin the mean? The concept of espacio ae English has exactly
the same fuzzy edges as the concept of Anledge of Bgl and
CChomaty’s example does nothing to separate the two concepts
‘Of course, Chomsky denies the presence ofthe eapacity during
the intervening period, He says
Inthe intervening period he had no capacity tospek or understand
Engl, even in though
But why should he say that? I iy possibe that he thought that it
followed frm something else that on the hypothesis, genuinely
He nad no capacity to spakor understand Engl, ven in hough
‘nthe ivervening pei.
‘But, ofcourse, the first proposition doesnot allow rom the second,
And if Chomsky hinks th tds he is mistakenMe Language ond the Mind
‘Chomsky goes om to considera sein aphasic who is like the
frst but who never recovers speech To deny that his person lacks
knowlege of English would be perverse, he argues: we have agreed
tha the fret aphasic knew English, and this second one i in
exactly the same mental (ltinatly physical) state, as might be
shown’ on autopsy, Well, once again this would bea matter for
‘decision, ot fr discovery at surely this ime the natural decision
‘would go the other way. ‘The one remaining peop to support the
Application of the concept lanring English — that the unforanate
‘vas later gong to use the language again without relearing ~ has
Bren pulled away. So what grounds are left for saying that be
[Knows English? ‘But heii the same brain state as somebody who
does knove English” But that isto beg peesely the question at
Issue that ithe brai state and not the presence of absence
the capacity, chat sees whether someone knows English oF nt.
“The question berging nate of Chomeky’s procedure ie masked
by his tue of expressions auch as ‘mental (alimately physical)
structures’ and ‘mental (imately physical) stats. Chomaky ses
Such expressions to indicate that his mentalsm doesnot involve
ny sort of immateratisms mental structures are simply physical
Structures desrbed at a certain level of abstraction, But the
‘expressions are illchosen, whatever one may think of iamaterial-
fom, beeause they conceal the fact thatthe criteria a deity for
mental state are not the same as those fora physical sate. Two
People can be in the same mental state while Being ina dierent
‘Physical state and can be nthe same physical sate while Being ia
A diflrent mental state. To say tis doesnot beg any questions
About materialism, since is equally re of computers chat there
‘ho one ~ one correlation between software structures and hardware
‘Aw analogy may help here. A monarch is legal person in the
sense that whac i sto be a monarch i defined by'a set of legal
Felaionships. ‘Bat all actual monarchs, from Hammurabi to
Elizabeth MI, are physical persons, human beings af esh and
bod: We might therefore, in the style of Choma call momar
Tegal (ultimately physical) persone But f we do ao, we invite
‘confusion. Suppose that I met the King of England in 1937 andi
royo Tsay, "1 have just met again the same legal (alimatly
Language and be Mind ug
physical) personas E met in 2937 Whom did meetin 194% the
Duke of Windsor or King George VI? The confusion in aking of
the same mental (imately physical) sructute less obvi but
10 les serious,
Thdeed, the confusion it more serious. For monarchs are indeed
persons; whereas states ofthe mind and states of the brain are not
Sates of the same Kind of thing. I agree with Chomsky that to
describe a state of mind isto describe, at a certain degree of
abstraction, a physical objec; but the physial abject so described
isa human being and no a brain. The brain state characteristic of
speaker of English ifwe go alg with Chomsky in assuming that
there are such may, forall we know, be eproicibe in a brain
nit. However successfully dhey were reproduced they would not
constitte knowledge of English for itis poople, not brains, to
‘whom ic makes sense to attribute auch knowledge
etme emphasize ain that Tam not arguing lor immateiaiom
‘orspittualism, The conceptual points that Ihave been making can
bbe made about pocket calculators no les than about human beings.
My calculator works out che square root of 129436780. Tn a fash,
there comes the answer ¢4111.13106, Between my pressing the
square-root key and the display appearing, complicated events ook
place init electronic innards, Those event, whatever they were,
{ould have taken place in a dilrentealeulstor doing dillerent
job; and a diferent caleulator doing the same job might well have
taken an electronically totally diferent route, Moreover, the hard
‘ware might have been taken out of the case, separated (orn the
inpat keys and the output spay. Whatever electronic events then
took place inside ie would not have been the working out ofthe
square root of 123456789. Forin the sens in which eaeulators can
‘work out square root for us cs only whole calculators and not
portions of their electronic anatomy, however sophisticated, that
fan do the working out
CChomsky’s final argument against the identification of know:
ledge of English with the capacity to use it goes ike this:
Were we deny capac and knowledge, we would presumably
te Tod to say tht the aphase doc not hrow English when the|
|
Me Langage and the Mind
Tall Anowiedge of Engh can aie in mind etally lacking hie
Anowledge without any revit experience whatorer a he case
recovery sows, something that psy not ioe of he chs
‘mind and cems an exotic ei. fp. 52)
Such a claim would indeed be exotic; which makes it more
surprising that only forty pages later, on p. 3, Chomsky should
ommend Arthur Danto for pointing out the possibility that in
principle there might be a "Spanish il” with the property tht by
{aking it we should have been caused (adventtiosly) to be master
of Spanish without having learned the language’. Such a master of
Spanish would undoubsedly know Spanish, Chonky says, hence
We cannot rle out in prin
right bring aout the men
Spanish
pl he pty a ing «pi
In such a case, of course, knowledge of Spanish would arse in a
‘mind tally Tacking the knowledge withowt any relevant experi
"ence whatsoever: the claim that Chomsky dismissed as exotic nthe
fares passage. Indeed i sa more exotic aim, for in the ease of
the recovered aphasic it might well be claimed that there wat
felevant experience the experience at the time when the language
‘eas orginally acquired inthe normal way
T think that Chomsky i right not to reject a inconceivable the
roton that a pill might give us mastery of Spanish the inconsi
tency between this concrasion and his earlier poston merely rings
fut further the faiity of his atempe to separate knowledge
Enslish rom the ability to se the mastery ofthe language, The
idea is indeed exotic, but it iz not literally inconceivable, What
‘would be inconceivable would be the ea tha ill might give one
Knowledge of Spanish ata psig oe te coc tne the aguage.
‘The root of Chomsky’'s canfusion is his fur to distinguish
betwen two diferent Kinds of evidence that we may have forthe
‘obtaining of states of allaits: 0 diatinguish between crea and
‘smplms (lo use the terminology introduced by Wittgenstein.
Where the connection between a certain kind of evidence an the
conclusion drawn fiom it is a matter of empirical discovery,
Language andthe Mind us
‘hough theory and induction, the evidence may be called amp
of the tate of alli; where the relation between evidence and
Conclusion isnot something discovered by empirical investigation,
bouts something that must be grasped by anyone who ponssrs the
concept ofthe state of ais in question, then the evidence nota
tere symptom, but sa cerin ofthe event in question. A red sky
Ac ight may bea symptom of good weather the along moran:
but the absence of los, the shining ofthe sum, ete, tomorrow are
‘ot jst symptoms but criteria forthe good weather. Simulaly, the
‘ccartence of certain electrical brain patterns may be, of may some
day come tobe, symptoms ofthe presence of knowledge of English
in the person whose brain is in question. But his ready use of
English snot just a symptom of tis a criterion of knowledge of
English
‘To grasp the importance ofthe distinction berween criteria in
symptoms in connection with knowledge af language, consider the
folowing case. Suppose that Profesor Chomsky were now to di
aad on opening bis skull we disrered that there was vothing
inside except sawdust, This i indeed an exotic suggentin: it
happened it would be an artonishing miracle. But sit happened it
Wwotld noc east the lightest doubt on what we all now know,
namely, that Chomsky knows English extremely well But on
‘Chomiy’s view we would have to say that infact i showed he
never knew English a al since on his view knowledge of English is
lulumately a certain phyical structure. But the supposition that
someone can use English as Chomsky does and yet not cw
English it not just miraculous supposition isa plee of Biter
‘Chommskys characterization ofthe mental structures tha interest
him makes them straddle uncasly the distinction between hard
‘ware and software, They seem to be to ghostly to be hardware
(irom time to ie herein us that ti part fhe theory hat
they shouldbe in the Brain rather than inthe iver); bu they also
‘scam tobe ton concrete tobe satware, otherwise they could ot be
‘haracterized as ultimately physical structures. Bue though is
‘characterization of mental structures i, T have argued, confured
land incoherent, what he is studying when he is studying mental
Structures is something genuine, important and fascinating. Tei6 Language ad the Mind
precisely the relationship between diferent capacities and sets of
apacties,
‘Chomaky would deny this If we eject his conception of mental
structures, he says,
Weare lft with a deep ty of behavior, potential behavior,
Alispasons behave and soon a ty han my epinion anna,
in cca be pursued tna eforent wy (09)
‘What in fet we are et with, and what Chomsky in fact studies, re
relations between dierent abilities: in particular the ability to
render the value of certain functions (eq. linguistic tansorma-
tions for given arguments. Its abilities ofthis kind that Chomsky
jn studying when he aims to uncover the structures underlying out
se of language
“There are many inellectwal tasks that we ean perem in more
than one way. When we multiply a umber by ten, for instance, we
an (a simply add a 0 (4) recite the appropiate part ofthe ten
times table () write the number down ten times and add up the
result. Now, wherever we havea cae of doing A by doing B thete
will be questions to be raised about the relations between the
Ability to do-A and the ability eo do Bs we ean only map by ten
inthe fret way, for instance, we know the decimal notation, and
in the second way fe know the ten mes table, And when we do
‘A by doing B it may well be that we kaow very well hat we are
doing A, but do not know without reflection that we are ding By
‘when we return a serve tennis there are many movements of|
hand and arm that we are not normaly aware af by which we make
the return. Similarly, in performing intellectual tasks ~ including
‘he comparatively modest ones of pronouncing a word or constrict
ing a sentence ~ there are many sub-tasks that we perform without
conscious advertence. When we atk what rules or principles we
tmploy in periorming thes task, we are asking what subabilies|
Wwe are exetisng when we exercise the ability fo we language.
Tn what sense, then, does the petformance of such tasks have
‘psychological realty’ when itis not conscious? The ability being
exercised isa peli veaity inthe sense that tian ability that
is being exercised in a tank that isan intellectual one and not a
Language en the Mind “7
merely physical one. Ie is a psychological rely in hat its open 10
‘esting which of the various posible algorithms that T might use in|
performing the task i the one that I'am actually using, Tes here
that iis evant to study the reaction time of subjects and simile
‘Phenomena to which Chomsky’ attitude has become increasingly
cavalier,
entirely agree with Chomsky in regarding the consciousness of
‘mental process as Being quite inesenil to the reality of the
Process. But this does not mean that payehological reality i
{tnimpertant, Once again, the pont can be made in terms of imple
‘non-human machines. It is 8 genie empirical question which
particular algorithm is being wsed by a computer or eaeulatr 0
Produce the solution to a problem, and the speed of computation
And the nature of rounding errors and the ike sight provide
‘mpircl evidence for oe against the use ofa particular algorithm,
Tewould inno way count against this that there was no presenta:
tion ofthe algorithm inthe ouput display of the calelator, of in
the monitoring file provided by the computer, Tt isthe monitoring
fle that isthe analogue, inthe computer context, ofthe conscious
‘count that we can give of the way in which we. perform
Jnellectal task.
"The philosophical confusions that I ave claimed to detect in
‘Chomaty’s presentation of his theories of knowledge of grams
‘rein no way onganie to the theories themselves, The theories i
fave understood them, can be stated in such a way 38 tobe quite
fice ofthe particular frm of mental in which they have Been
wrapped up and which has provided ineevant distraction to
philosophical and pychologia! eis,