1. science
2. social science
3. linguistics

Chapter50
What Is InnateandWhy : Commentson the Debate
Hilary Putnam
398
HilaryPutnam
data as (1) it is reasonableto assumethat an " unstructured" child would assume
that HI is valid. In fad , as we know, it is not, and H2 is (more nearly) corred.
Thus considerthe data of (2):
(2) The man who is hereis tall.- Is the man who is heretall?
The manwho is tall willieave .- Will the man who is tall leave?
These data are predided by H2 and refute HI , which would predid rather the
interrogatives(3):
(3) Is the man who hereis tall7
Is the man who tall will leave?
Now the question that arisesis this: how does a child know that H2 is corred
(nearly), while HI is false? It is surely not the casethat he first hits on HI (as a
neutral scientistwould) and then is forced to reled it on the basisof data suchas
(2).
'
Chomskys conclusionHornall this is the following :
Suchobservationssuggestthat it is a property of So- that is, of LT (H,L)- that
rules (or rules of somespecificcategory, identifiable on quite generalgrounds by
somegenetically detenninedmechanism
) are structure- dependent.The child need
not considerHI ; it is ruled out by propertiesof his initial mentalstate, So.
I wish to discussthis example by considering two different questions: (1) can we
accountfor the child's selectionof " structure- dependent" hypothesesand conceptsin
the courseof languagelearning on the basisof generalintelligence, without postulating
that the preferencefor H2 over HI is built in, or that a template of a typical human
languageis built in, as Chomsky wishes us to do; and (2) can we account specifically
for the preferenceof H2 over HI without assumingthat such a specificpreferenceis
built in? Before discussingthese questions, however, I want to consider the vexed
'
'
question, What is a grammarf
TheNatureof Grammars
A grammaris somesort of systemwhich- ideally- generatesthe " grammaticalsentences
"
of a languageand none of the ungrammaticalones. And a grammaticalsentence
is one generatedby the grammar of the language (or by any adequateone, if one
believesasZellig Harris doesthat there is no suchthing asthegrammarof a language).!
This is obviously a circulardefinition. But how doesone breakthe circularity?
"
"
Chomsky suggestedlong ago (in Explanatory Models in Linguistics ) 2 that a child
"
"
"
"
hearspeople classingsentencesas grammatical or ungrammatical not, of course,
in those words, but by hearing them correct each other or the child- and that he
projects a grammaras a simplest extrapolation from such data satisfying some innate
constraints.
The trouble with this view is that the factual premise is clearly false. People don' t
. If they object at all, it is to deviant
object to all and only ungrammaticalsentences
- but they do not, when
sentences
they correct each other, clearly say (in a way
that a child can understand) whether the deviancewas syntactic, semantic, discoursetheoretic, or whatever.
assertsthat the child is, in effect, supplied with " a list of grammatical
Chomsky
"
"
sentences and " a list of ungrammaticalsentences
and has to extrapolate from these
What Is Innate and Why : Comments on the Debate
399
two lists. But this is surely false. If anything, he is suppliedrather with a list of acceptable
sentencesand a list of sentencesthat aredeviant-for -some-reason-or-other; a grammar
of his languagewill generate(idealizing somewhat) all of the acceptablesentences
in the first list, but unfortunately, it will not be the casethat it generatesnone of the
deviant sentencesin the other list. On the contrary, the grammaticalsentenceswill be a
, which is not disjointfrom the (finite
supersetof the (finite list of ) acceptablesentences
.
list of ) deviant sentences
Moreover, the secondlist does not have to exist at all. Chomsky hascited evidence
that childrencanlearntheir first languagewithout being corrected; and I am surehe also
believes that they don't need to hear anyone else corrected either. Chomsky might
"
"
this by scrappingthe hypothetical secondlist (the list of ungrammatical, or at
reply to
"
"
least, unacceptable sentences
). He might say that the grammar of an arbitrary language
is the simplestprojection of any suitablefinite set of acceptablesentencessatisfying
someset of innate constraints. This throws the whole burden of defining what a
grammaris on the innate constraints. I want to suggesta different approach: one that
says, in quite traditional fashion, that the grammar of a languageis a property of the
.
, not a property of the brain of Homosapiens
language
Calculus
Propositional
Let us start with a simple and well-understood example: the artificial languagecalled
"
"
propositional calculus with its standardinterpretation. The grammarof propositional
calculuscanbe statedin many different but equivalentways. Here is a typical one:
A propositional variablestandingaloneis a well-formed formula.
(II) If A and B are well-formed formulas, so are ' " A. (A &: B), (A v B) and
3
(A ::) B).
(I )
(III) Nothingis a well-fannedfonnulaunlessits beingsofollowsfrom(I) and(II).
400
Hilary Putnarn
-sponding to them there exist parallel inductive definitions of truth in propositional
calculus. But if we limit ourselvesto those that are computationally feasible(that is, the
correspondingdecisionprogram is short, when written in any standardformat, and the
typical computation is also short), not a great many are known, and they are all extremely
similar. In this sense,propositional calculusas an interpreted systempossess
es
an intrinsicgrammarand semantics.
Let me elaborateon this a little . If Martians exist, very likely they have hit upon
calculus, and it may be that when they use propositional calculustheir
propositional
'
'
logicians brains employ different heuristics than our logicians brains employ. But
that doesnot meanthat propositional calculushasa different grammarwhen usedby a
Martian and when used by a Terrestrian. The grammar is (anyone of ) the simplest
inductive definition(s) of the set of strings in the alphabetof propositional calculusfor
which truth is defined- that is, the simplestinductive definition(s) with the property
that there exist parallel inductive definitions of truth. Given the semanticsof propositional
calculus(and no information about the brainsof speakers
), the classof reasonable
is
fixed
that
semantics
not
the
structure
of
the brains that do the
,
by
grammars
by
.
processing
It
seemthat I have begged too many questionsby introducing the predicate
" truemay
"
it is not essentialto my argument. Supposewe do not define " true," but
but
;
"
rather follows &om." Any reasonably simple definition of the relation " x follows
"
&omy in propositional calculuswill have the property that it presupposesa syntactic
analysisof the standardkind. In other words, checkingthat somethingis an axiom or a
proof, etc., will involve checkingthat strings and componentsof strings have the forms
(p &: q), ' " p, (p V q), (p ::>q). The grammar (I), (II ), (III ) not only generatesthe set of
"
"
it in a way
strings over which the relation follows &om is defined, but it generates
that correspondsto propertiesof strings referredto in the definition of " follows &om."
Coming to natural language: supposewe think of a natural language as a very
complicatedformalized languagewhose formalization is unknown. (This seemsto be
how Chomsky thinks of it.) Supposewe think of the speakeras a computer that,
whether certain strings are " true," given certain inputs,
among other 'things," computes
"
or if you don t like true, as a computer that computeswhether certain sequencesof
"
"
"
"
strings are proofs, or computesthe degreeof connnnation of certainstrings, and so
forth. The fad is that anyone of these semantic, or deductive logical, or inductive
logical notions will have an inductivedefinitionwhose clausesparallelor at leastpresuppose
a syntacticanalysisof the language.
To come right out with it: I am suggesting (1) that the declarativegrammar of a
language is the inductive definition of a set of strings which is the set over which
semantic, deductive-logical, inductive-logical (and so on) predicatesare deAned;4 (2)
that it must be in sucha form that the inductive definitions of thesepredicatescaneasily
"
parallel" it ; (3) that the correspondingdecision program must be as computationally
feasibleas is consistentwith (1) and (2). If a languageis thought of in this way- as a
, with a deductive logic, with an inductive logic, and
systemof strings with a semantics
so on- then it is easy to seehow the grammarcan be a property of the languageand
'
not of the speakers
brains.
TheNatureof Language
Learning
Let us considerthe linguistic abilities of Washoe(the chimpanzeebrought up to use a
certain amount of deaf-mute sign languageby Alan and BeatriceGardner). No doubt
Chomsky will point out that Washoelacksmany of the syntactic abilities that humans
What
Is
and
Innate
: Comments
Why
on
the
401
Debate
'
have
and
,
on
these
he
grounds
claim
would
it
that
is
to
wrong
the
tenn
what
is
apply
-
ian
"
to
guage
what
. What
1 .
has
she
is
important
There
a
is
Washoe
class
of
with
( in
words
( classes
- or
more
There
is
- less
)
I
which
,
of
.
things
) with
language
,
is
term
, and
( to
gives
will
For
- or
more
bananas
stereotypical
a frame
not
important
call
which
,
associates
the
"
stereotypical
forth
,
grapes
"
banana
.
, which
)
- Washoe
for
Washoe
,
example
- less
so
-
nouns
Washoe
has
( for
acquired
"
'1
Alan
,
example
sign
this
of
application
:
following
certain
grape
with
the
"
word
2 .
the
associates
"
. But
learned
is
gives
to
apple
Trixie
) .
II
3 .
She
she
can
will
new
project
out
figure
uses
of
herself
this
the
. If
frame
use
she
is
her
teach
you
to
expected
a
make
new
word
of
"
date
, say
11
-
gives
,
date
( to
out
the
)
"
I'
4 .
can
She
use
the
word
"
and
to
combine
.
sentences
She
can
figure
5
use
expected
of
and
p
from
q
the
uses
of
far
beyond
and
p
.
separately
q
'
s
Washoe
Actually
for
these
now
"
.
go
only
these
a
internalized
'
s
people
a
rule
the
rules
dependent
effect
( of
effect
X
p ,
has
is
- Washoe
for
so
without
is
A
benefit
consider
just
Washoe
, B , and
are
C
"
X
gives
and
p
us
that
is
, and
'1
A
then
person
sentences
let
occurred
-
a
learned
this
really
.
surprising
Let
us
introduce
an
B
to
. And
q
of
tiny
of
fragment
avoid
the
Washoe
accusation
s
-
of
that
(I )
these
innate
a
is
are
template
things
"
C
Here
is
),
begging
question
If
p ,
b , c
the
and
are
the
describe
the
"
-
Ilshudder
the
,
say
-
predicate
inserted
are
quotes
II
to
corresponds
that
Washoe
is
b
the
not
- names
people
gives
the
q
of
K
kind
to
the
in
to
corresponds
'
~
gives
X
that
condition
that
learning
)
the
obtain
both
q
( to
c .
to
to
corresponding
interested
really
X
and
,
corresponds
condition
:
question
B , C , then
to
something
and
, p
and
p
in
fragment
corresponding
that
to
the
are
C
people
- Washoe
for
for
B ,
condition
sentences
submit
rules
- Washoe
for
K , and
are
q
Ilsemantic
corresponding
I
,
-
noun
corresponds
( II )
the
to
condition
Now
a
are
kind
of
to
predicate
the
"
.
X
If
semantic
( where
Illanguage
"
condition
a
"
'
above
are
q
-
noun
) as
has
Washoe
a
is
course
if
that
which
if
that
but
;
capacities
what
of
.
language
Nor
the
Washoe
to
-
structure
to
counting
and
,
rule
-
names
sentence
to
four
account
plausible
"
has
for
abilities
The
.
uninterpreted
certain
"
II
of
strings
is
and
( II ) .
But
(i )
If
that
gestures
interested
She
have
for
a
uninterpreted
-
reasons
practical
(I )
learning
certain
and
( II )
called
property
reward
, and
, approval
involves
automatically
so
-
forth
the
learning
.
grammaticality
in
(I )
learning
facts
grammatical
:
B ,
- for
sentence
( ii )
are
C
If
p , q
- names
people
- Washoe
are
and
X
the
set
the
which
the
by
of
- for
sentences
sentences
- Washoe
, so
syntactic
, and
"
~
,
X
gives
by
the
by
syntax
and
she
be
to
able
will
intelligence
( II ) , and
will
ascribe
fail
;
to
or
a
is
C
)
"
( i ) , ( ii ) is
the
precisely
"
( II ) ;
and
a
only
means
the
the
clauses
(I ),
that
Washoe
is
to
this
that
condition
end
( II )
presuppose
to
trying
, there
_
are
be
her
Washoe
too
low
intelligence
"
implicit
to
internalize
will
be
set
is
the
the
semantics
two
only
of
-
:
possibilities
dependent
, and
enough
rules
as
a
like
we
corollary
"
knowledge
over
defined
"
structure
high
-
precisely
learn
"
her
either
( to
.
q
grammar
to
( i ) , ( ii ) . Given
is
p
"
(I ),
given
the
is
corresponds
definition
analysis
- ese
Washoe
'
- Washoe
for
II
generated
predicate
inductive
-
noun
"
-
semantic
a
.
I'
For
is
of
the
syntactic
rules
(i ),
( ii )
-
(I ),
will
not
402
Hilary Putnam
"
"
"
"
becauseshe knows (I), (II) and in addition knows (i), (ii ), but becausehaving the
"know -hGw" that constitutes
implicit knowledge of (I), (II) includesimplicit knowledge
of (i), (ii ).
But the samething is true of the child. The child is not trying to learn a bunch of
syntacticrules as a kind of crazy end-in-itself. He is learning, and he wants to learn,
semanticrules, and these cannotbe stated without the use of structure- dependentnotions
. There aren't even plausiblecandidatesfor structure-independentsemanticrules.
So of course(given that his intelligenceis high enough to learn language), of coursethe
child " internalizes" structure- dependentrules. And given that he must be building up an
"
" inner
, noun, verbphrase
,
representation of abstractstructural notions suchas sentence
and so on in learning to understand the language, the mere fact that H2 uses such
notions and Hi doesnot, doesnotmakeH2 so much lessplausiblethan Hi .
"
"
, pulled a fast one on us. He presentsus with a picture of
Chomsky has, so to speak
the child as being like an insanelyscientisticlinguist. Both are looking at languageas a
streamof uninterpretednoises; both are interestedin an occult property of " grammati"
. From this (crazy) point of view, it is not surprising that HI seemsinfinitely
cality
"
'
't
simpler" than H2. So Chomsky springs his carefully preparedtrap- Why doesn
'
the child try t:, e simpler but falsehypothesisHi beforethe correcthypothesisH2f
But this isn't what-children(or sanelinguists) arelike at all. The child is in the process
of trying to understandEnglish. He has already tumbled (if Washoecan, so can he!) to
the fact that he needsto internalizestructure- dependentnotions to do this. So the mere
fact that H2 usessuchnotions doesn't at all makeit implausibleor excessivelycomplex.
The point is that the learningof grammaris dependent
on the learningof semantics
. And
there aren't even any candidatesfor structure-independentsemanticrules (if there are,
'
they get knockedout pretty early, even by a chimpanzees brain).
HI ConsideredMore CloselV
So far I have argued that H2 is not nearly as weird &om the point of view of the
by an innat.etemplateof languageasChomsky wants to make
intelligent brain unaided
it seem. But I haven't arguedagainstHI . So still the questionremains, why doesn't the
child try H 11
Let us try applying to this problem the conception of grammar we just sketched
as, so to speak
, semanticsminus the semanticpredicates). HI will only be
(grammar
"
tried" by the child if the child " tries" some semantichypotheses that correspond
"
"
to HI . The child wants to understand
questions, not just to Rag them asquestions. But
it is plausible to assume(and Chomsky himself would assume
) that understanding
questionsinvolves recoveringthe underlying declarative. This meansthat the questiontransformationmust have an inversethe child can perform. HI is indeed simple, but its
inverseis horribly complicated
. Moreover, its inverseusesthefull resources
of thegrammar;
"
all the notions, suchas noun phrase," that HI doesnot employ have to be employedin
recoveringthe declarative&om the
output of our applicationof HI . So it is no mystery
that the child (or its brain) never " tries" suchan unworkablesemantictheory, and hence
"
"
never tries HI .
itself employs " abstract" notions, sinceit containsthe phrase-structure
,
Incidentally
" HI
"
concept declarative, and applying" it , if it were a" rule of English, would therefore
involve working with notions such as noun phrase, since these have to be used to
declaratives.And somelanguagesdo have question-transformationsthat are
recognize
"
as structure-independent" as HI is; for example, in Hebrew one can form a question
What Is Innateand Why : Commentson the Debate
403
from a declarative by just prefixing na im. But this preAxing operation doeshave a
simpleinverse, namely, deleting na im. '
I would likenow to discussChomskys more abstractremarksat the beginning of his
paper. Let me begin with what he saysabout intelligence.
Chomskyon GeneralIntelligence
So far I have assumedthat there is such a thing as general intelligence; that is, that
whatever elseour innate cognitive repertoire may include, it mustinclude multipurpose
learning strategies, heuristics, and so forth. But Chomsky appearsto deny this assumption
explicitly . I quote:
More generally, for any species0 and cognitive domainD that have beententatively
identified and delimited, we may, correspondingly, investigate LT (O , D ),
'
"
the learning theory for the organism 0 in the domainD , a property of the
genetically determinedinitial state. Suppose, for example, that we are investigating
the ability of humans to recognize and identify human faces. Assuming
"
"
face-recognition to constitute a legitimate cognitive domain F, we may try to
specify LT (H ,F), the genetically determinedprinciples that give rise to a steady
state (apparently some time after languageis neurally fixed, and perhapsrepresented
in homologous regions of the right hemisphere
, as some recent work
.
other
domains
can
be
studied
in humansand other
,
)
suggests Similarly
cognitive
.
We
would
to
find
hardly expect
organisms
interesting properties common to
LT (O ,D ) for arbitrary O , D ; that is, we would hardly expect to discoverthat there
exists somethingthat might be called" generallearning theory." As far as I know,
the prospects for such a theory are no brighter than for a " growth theory,"
intermediatein level betweencellular biology and the study of particular organs,
and concernedwith the principlesthat govern the growth of arbitrary organsfor
arbitrary organisms.
The key notion in this argumentis the notion of a " domain." How wide is a domain?
Is all of mathematicsone domain? If so, what about empirical science
? Or are physics,
chemistry, and so on, all differentdomains?
If Chomsky admits that a domain can be as wide as empiricalscience(that there can
'
be a learning theory for empiricalscience"), then he hasgranted that somethingexists
that may fit tingly be called " general intelligence." (Chomsky might retort that only
,
exceptionally intelligent individuals can discover new truths in empirical science
whereaseveryone learnshis native language. But this is an extraordinarily elitist argument
: the abilities of exceptionally intelligent men must be continuouswith those of
ordinary men, after all, and the relevant mechanismsmust be presentat somelevel of
functioning in all humanbrains.) Evenif only physics, or just all of solid-statephysics, or
just all of the solid-state physics of crystals is one domain, the same point holds:
heuristicsand strategiescapableof enablingus to learnnew factsin theseareasmust be
extraordinarily multipurpose(and we have presentlyno idea what they are). Once it is
granted that such multipurpose learning strategies exist, the claim that they cannot
accountfor languagelearning becomeshighly dubious, as I argued long ago.6 (Consider
Washoef)
On the other hand, if domains become so small that each domain can use only
are highly specific in purpose (such as " recognizing faces,"
learning
strategies that
'
"
learning a grammar ), then it becomesreally a miracle that evolution endowed us
with all theseskills, most of which (for example, higher mathematics, nuclearphysics)
404
Hilary Putnam
were not used at all until after the evolution of the race was complete (some 100,000 odd years~ago ). And the analogy with organ growth does not then hold at all: the
reason there does not have to be a multipurpose learning mechanism is that there are
only limited numbers of organs, whereas there are virtually unlimited numbers of
"
"
domains.
TheProspects
of GeneralLearningTheory
"
"
"
"
Chomsky feels that the prospects of general learning theory are bad. I tend to
agree. I seeno reasonto think that the detailedfunctioning of the humanmind will ever
be transparentto the human mind.7 But the existenceof general intelligence is one
of it is another.
question; the prospectfor a revealingdescription
if
the
innateness
is
,
Incidentally
hypothesis right , I am also not optimistic about the
a
for
revealing descriptionof the innate templateof language. The examples
prospects
has
Chomsky
given us of how to go about inferring the structure of the template
(suchas the argumentabout HI and H2) are suchbad argumentsthat they cast serious
doubt on the feasibility of the whole program, at leastat this point in history (especially
if there exist bothgeneralintelligenceandan innate template).
On the other hand, we may well be able to discover interesting fads and laws about
generalintelligencewithout being able to describeit completely, or to model it by, say,
a computer program. There may be progressin studying general intelligencewithout
its being the casethat we ever succeedin writing down a " generallearning theory" in
the senseof a mathematicalmodel of multipurposelearning.
Chomskyon Evolution
'
Chomskydismisses Piagets questionregarding how sucha thing asan innate template
for languagemight have evolved. But he should not dismissit. One answerhe might
have given is this: primitive languagefirst appearedasan invention,introducedby some
extraordinary memberof the speciesand learnedby the others as Washoe learnsher
fragment of language. Given such a beginning of the instrument, genetic changesto
enable us to use the instrument better (including the enlargement of the sc-called
speechcenter in the left lobe of nonnal humans) could have occurred, and would be
explained, if they did occur, by natural selection. PresumablyChomskydid not give this
answerbecause(1) he wants to deny that there exists such a thing as general intelligence
, and to deny that even the simplest grammar could be internalizedby general
'
intelligencealone; and (2) he wants to deny that Washoes performanceis continuous
with languagelearning, and to deny that it has any interest for the study of language
learning. But this is surely perverse. If the first languageuser alreadyhad a complete
innate template, then this could only have beena miraculousbreak in the evolutionary
, asPiagetin effect points out.
sequence
'
Chomsky remarksthat we don t know the details of the developmentof the motor
organseither, and this is surely true. We do postulate that they developedbit by bit.
This posesdifficulties, however, sincethere are no creatureswith two thirds of a wingl
But therehavebeenimpressivesuccess
esin this direction (for example, working out the
evolution of the eye). We have found creatureswith gliding membraneswhich are, in
a sense," two thirds of a wing ." And we have found eyeswith only rods (no cones) and
eyeswith only cones(no rods). Sincethe first draft of this paperwas written , therehave
beenexciting new suggestionsin evolutionary theory.8
What Is Innate and Why : Commentson the Debate
4OS
It is one thing to say that we cannotscientificallyexplain how certainstructureswere
produced(and the theory of natural selectiondoes not even claim that those structures
were probable
), and quite another to say that we now have scientificreasonto postulate
"
a largenumberof " mentalorgans asspecificasthe various domainsand subdomainsof
humanknowledge. Sucha mental organizationwould not be scientificallyexplicableat
all; it would mean that God simply decided to produce these structuresat a certain
the oneswe would needa half a million (or whatever)
point in time becausethey were
'
yearslater. (Although I don t doubt that God is ultimately responsiblefor what we are,
. And , in any case,
it is bad scientific methodology to invoke Him as adeusex machina
this is sucha messymiracle to attribute to Him! Why should He pack our headswith a
"
"
billion different mental organs, rather than just making us smart?) On the other hand,
"
"
if our languagecapacitydid develop bit by bit , even with jumps, a descriptionof the
first bit will almost certainly sound like a descriptionof Washoe. But then we will have
concededthat someinternalizationof linguistic rules (at leastin prototype form) can be
accountedfor without innateness
.
A BetterArgument
"
"
But this suggeststhat there is an argument for some innateness that Chomsky might
"
"
have used. Considerthe phenomenoncalled echo-location in the bat. The bat emits
"
"
reflectedfrom the prey (or whatever- for example, a
supersonic noises, which are
"
"
singleinsect), and the bat can steer by thesesound-reflectionsaswell asif it had sight
(that is, it can avoid fine wires, catch the mosquito that is trying to avoid it , and so
forth). Now , examination of the bat's brain shows that there has been a tremendous
of the centersconnectedwith hearing (they fill about seven- eighthsof the
enlargement
'
bat s brain), as comparedto other mammals(including, presumably, those in its evolutionary
"
'
past). dearly , a lot of the bat s echo- locating ability is now innate."
SupposeChomsky were to grant that Washoe has protospeech, and thereby grant
that general intelligence can account for somelanguage learning. He could then use
"
"
evolutionas an argumentfor (some
. In other words, we could argue that,
) innateness
given the enormous value of the languageability (as central to human life as echolocation is to bat life), it is likely that genetic changeshave occurred to make the
"
instrument better- for example, the development of the speechcenter" in the left
lobe. (But caution is needed: if the left lobe is damagedearly, speechcandevelop in the
right lobe.) This argumentis the only one I know of that makesit plausiblethat there is
someinnate structuring of human languagethat is not simply a corollary to the innate
(that is, genetically predetennined) structuring of human cognition in general. But the
argument is not very strong: it could be generalintelligencethat has been genetically
refined bit by bit and not a hypothetical language template. Indeed, even species
and
all
useless
of
human
could
be
the
of
functionally
specific
languages
aspects
product
unknown but geneticallypredeterminedaspectsof the overall functioning of the human
brain and not clues to the characterof a languagetemplate; so the mere existenceof
suchaspectsis no evidenceat all for the templatehypothesis.
I think there is an answerthat Chomsky canmaketo this objection; but I will defer it
until I have discussedPiaget.
' "
"
Piagets Constructivism
The view I havebeenputting forward- that everything Chomskyascribesto an innate
"
"
template of language, a mental organ specificallydesignedto enableus to talk, can,
406
Hilary Putnam
for all we know, be explainedby generalintelligence- agreesin broad outline with the
view of P..iaget. However, there seemto me to be seriousconceptualdifficulties with
'
this view when it is combinedwith Piagets specificaccountof what generalintelligence
is like.
Piagetsupposesthat humanintelligencedevelopsin stages, eachstagedependingon
biological maturation (that is, the age of the child) and on the successfulattainment of
the previous stages. At a certain stage, certain conceptscharacteristicallyappear, for
"
"
example, the concept of conservation. But what is it to have such a concept as
conservation?
I submit that the only coherentaccountpresentlyavailablefor having the conceptof
conservationis this: to have the conceptis to have mastereda bit of theory, that is, to
"
have acquiredthe characteristicusesof such expressionsas sameamount," and some
sentencesinvolving such expressions
, or equivalent symbolism
,
key beliefs
' expressedby
. I don t claim that all conceptsare abilities to use symbolism; an animal that expects
the water to reach the sameheight when it is poured Horn a pot back into the glass
might be said to have a minimal conceptof conservation, but I claim that anything like
the full concept of conservationinvolves the ability to use symbolism with the complexity
'
"
"
of languagein certainways. (I don t claim that this is a tautology ; rather that it
is the only coherentaccountpresently availablefor what full-blown conceptsare. And
I don' t claim to have argued this here, but I have discussedthis elsewhere;9 and, of
'
course, this insight is not mine but Wittgenstein s- indeed, it is the main burden of
.)
Investigations
Philosophical
But if a maturational scheduleinvolving the development
is innate, and
of
' concepts
are essentiallyconnededwith language
, ! 0 then Piagets hypothesis would seem
concepts
' "
"
"
"
to imply Chomskys; constructivism would entail nativism.
Of course, Piaget does not commit so crude an error. He does not supposethat the
"
maturational scheduleis given (that is, innate); what he takes as given is reflective
""
"
abstraction it is this that precedeslanguage and that is supposedto take us Horn
one " step" to the next.
! If
But " reflection" and " abstraction" have no literal meaningapart Horn language
"
"
reflective abstraction is not literally meant, it is either a metaphor for empiricist
"
"
generalization, which is insufficient to account for languagelearning and use, or a
metaphorfor we-know-not-what.
"
"
It seemsto me that Piaget should take the view that reflective abstraction is
,
somethinglike the useof languagein the making of hypothetico- deductiveinferences
as Chomsky and Fodor urge, and henceconclude that something like the use of language
"
"
is innate. This position would have brought him into convergence with
Chomsky, insteadof into an unnecessarysectariansquabble. Moreover, his own suggestion
in 1958 that formal logic is the best model for human reasoning11is very
consonantwith sucha position.
some
that
were
a
little
.
I
care
wan
to
rec
things
"
"
some
of
these
errors
not
for
the
sake
of
but
beca
the
dis
,
,
being
p
icky
becomes
confused
at
critical
if
we
let
them
stan
.
hopelessly
point
a
First
:
Fodor
and
are
when
tha
it
a
is
Chomsk
quibble
they
say
simply
wrong
"man
"
'
"
"
"
that
we
can
t
learn
unless
some
innat
are
b
u
,
t.It
autology
anything
p
reju
in
is
not
that
our
heads
should
be
as
as
the
Tin
Wo
logically
impossib
emp
'sand
we
should
still
so
on
be
an
extre
talk
,love
of
a
,and
;itwould
just
exa
'
"
Fodors " Tautology
In the discussionFodor said
What Is Innate and Why : Commentson the Debate
407
causalanomalyif it ever happenedthat a creaturewith no internal structure did these
'
things. I don t d~oubt for one moment that our dispositionsdo have a causalexplanation,
and of coursethe functional organization of our brains is where one might look for a
'
causalexplanation (although I myself think that we won t be able to describethis in
12
future). But this still is not a tautology .
very muchdetail in the foreseeable
'
Second,it is true that we cant learnhow to learnunlesswe have someprior leamingdispositions: we have to have somedispositionsto learn that are not themselveslearned
, on pain of infinite regress(however, the impossibility of an infinite regressin the
real world is hardly a tautology!); but that does not meanthat it is logicallynecessary
(a
"
" that the unlearned
be
innate.
We
)
tautology )
dispositions
might (logicallypossibly
for no causeat all, for example,
acquirea new unlearneddisposition every five minutes
"
or for some causethat does not count as a fonn of leaming." There just aren't any
significanttautologiesin this area.
The reason this is not just a quibble is this: once we pare down Fodor' s and
'
"
"
Chomskys big tautology to something like this: as a matterof fact (not logic!), no
learningwithout somelaws of' learning, we see that no one, least of all the empiricists,
hasever denied it. Chomsky s and Fodor' s claim that there is a big, mysterioustautology that no one appreciateduntil Nelson Goodmanand that everyonethey dislike fails
to appreciateis mererhetoric.
408
Hilary Putnam
"
First a point of terminology: Every computer does have a built -in computer language
"
"
" "
"
, bQtnot a languagethat containsquantifiers(that is, the words all and some,
or synonymsthereof). Let me explain.
A digital computer is a device that storesits own program and that consultsits own
program in the courseof a computation. It is not at all necessarythat the brain be a
as
digital computerin this sense.The brain doesnot, after all, have to be reprogrammed
"
an all-purpose digital computer does. (One might reply that learning is reprogramming
" but Fodor is
;
talking about the program' for learning, not about what is learned,
and this program might be stored as the brain s structure
, not as a code.) Waiving this
"
"
storesconsistsof instructions suchas
objection: the program that a digital"computer
"" add the two numbersin address12 and " back to
none of which usethe
6
go
step
"
"
" "
word all. So generalization(A ) cannoteverbestatedin machinelanguage, even if the
computer's program is a program for making inductive inferencesin some formalized
language(for example, if the program is that of the hypothetico-deductive machine
mentioned earlier). Moreover, machinelanguagedoes not contain (nor can one introduce
" "
"
"
" "
" "
definition) such notions as tree, cow, jumps, spontaneous
into it
,
"
"
"
"
"
"
"
"
"
"
" and by - it
contains such notions as add, subtrad, 0, 1, put
so on
pert,
"
"
" "only
result in address17, go back to instruction so-and-so, and print out contents of
"
addressblah-blah.
Let us suppose, however (what needsto be proved) that our brain is a hypotheticodeductive machine, and that it carriesout inferencein a formalized languageILL (for
Inductive Logic Language) accordingto someprogram for eliminative induction. And
let us supposethat Fodor is not really talking about the brain' s machinelanguagewhen
he postulateshis "language of thought," but about ILL. Evenif so strong an assumption
is conceded,his argumentstill doesnot work.
To seewhy it doesnot work, let us recall that when the speakerhasfinally mastered
the predicatep, on Fodor's model, he is supposedto have acquireda new " subroutine."
Even if this subroutine is describedinitially in ILL or in some special" programming
"
'
, or both, it has to have a translation into machinelanguagethat the brain s
language
"
'
"
"
compiler" can work out, or the brain won t execute this subroutine. Let S be the
; then even if we grant that
descriptionof the subroutinein questionin machinelanguage
the brain learnsP by making an induction, it neednot be an induction with the conclusion
(A ). It would sufficethat the brain insteadconclude:
I will be doing OK with P if subroutineSis employed.
And this can be stated in ILL provided ILL has the concept " doing OK with an item,"
and ILL containsmachinelanguage. But this doesnot require ILL to contain (synonyms
" "
"
for) " face," " cow," "jumps," " spontaneous
, pert, and so on. Fodor' s argument has
failed.
Fodor suggeststhat he would claim that the machinelanguagedescriptionof how to
use, say, " tree" is (a form of ) the predicatetree.,But this is simply an extension of use
"
"
designedto makehis thesis"an uninteresting tautology .
Of course, the predicate doing OK with P" may arousesuspicion. But it shouldnot.
The " machine" (the brain) doesn't have to understandthis predicateas linguists and
philosopherswould! The generalization(B) is simply a signal to the machineto add
subroutineSto its repertoireof subroutines. ( Weshouldkeepin mind Dennett' s caution
that talk of " machinelanguage" is dangerousbecausewe are tempted to confuseour
abilities with the formalismin questionwith the machine's abilities.)
(B)
What Is Innate and Why : Commentson the Debate
409