I

Can the mind change the world? NED BLOCK

Hilary Putnam origi nated the idea that mental states are computational sta tes. At first (Putnam, 1960), his view was that although mental states are not identical with computational states (or " logical states." as he then called them), there are useful analogies between them. Later (Putnam, 1967). he argued in favor of the identity on the grounds that it was more plausible to suppose mental states are function al states (as he then called them) than that they are behavioral or physical states. This doctrine - funct iona lism - has dominated the philosophy of mind for over twenty years. Shortly after proposing functional ism. Putnam re­jected it again (J 973), and he has maintained this position ever since (Putnam. 1988).

Putnam was my teacher during both my undergraduate and gradu·ate days, and I fear I have absorbed his ambivalence toward functionalism. My teacher has had a habit of changing his mind, but never has he done so within a single essay, and so in this chapter I have surpassed him. My chapter starts out as an argument for functionalism, but it ends up suggesting an argument against it. The issue is whether we can avoid epiphenomenalism, which I here understand as the doctrine that what we think or want has no causal relevance to what we do. I propose functionalism as a way of warding off arguments for epiphcnomenalism, but then I argue that functionalism may bring epiphenomenalism in its wake.

The orientation of the chapter is toward the sciences of the mind, and their relation to intentional content, that is, what is shared hy the

I am graleful for support IO the America n Counci l or Learned Sociclies and lo the Nalional Science Foundation. granl numhcr DTR88 12559. I am also grateful to Jerry Fodor, Paul Horwich, Frank Jackson, Gabriel Segal. and Marianne Talbot for !heir commenls on an earlier draft; to Thomas Nagel, who was the commentaror on the paper on which lhis chapler is based at a conference al Columbia University in December 1988; to Simon Blacktiurn, who was !he commentalor on the paper at Oxford in February 1989; to Martin Davies. Dorothy Edgington. and Oarry Smith for discussion or 1he issues while I was at Birkbeck College. University of London; and to the audiences al a number of philosophy colloquia.

f i

I I

138 NFD ULOCK

hclief that grass grows and the desire that grass grows, the that grass grow.~ that buth sraces are directed toward. The question at hand is whether the scie nces of the mind prc..:cludc int en tional content from causal relevance to behavior. One argument that the intentional con­tents of our beliefs, thoughts, and the like have no e ffects on our behavior could he put this way: The processors in the head are not sensitive to content, so how could content have any effect on the outputs or changes of state of the system of processors? And if content can't affect the operation of this system of processors, how could it play any role in producing behavior? This argument seems formidable whether one thinks of the processors as neural devices reacting lO

neural inputs or, instead, from the cognitive science point of view, as computational devices processing representations. ' Jn this chapter, I confin e myself to the prohlcm as it arises in the cogni tive science approach that is dominated by the computer model of the mind. I assume a very specific picture of cognitive science and its relation to the commonsense conception of intentional content, namely, the view ac­cording to which there is an internal system of representation from whose meanings our intentional contents derive (Fodor, 1975; Pylyshyn, 1984 ). One of my reasons for couching the discussion in terms of this view is that although those who adopt this view are motil'ated by the aim of showing how our commonsense beliefs about cont ent (including our belief in content 's causal efficacy) are vindicated by the computer model of the mind, the problem of thl' epiphenomenalism of content arises within this view in an extremely simple and straightforward (and poignant) way. The viewpoint assumed throughout the chapter is that of a supporter of the computer model in cognit ive science who also would like to believe that the contents of our thoughts arc indeed ca usally relevant to what we do.

The problem I have in mind might be put in terms of The Paradox of the Causal Efficacy of Content, namely, th<Jt the fo llowing claims all seem to be true, yet incompatible:

I. The intentional content of a thought (or other intentional state) is causally re levant to its behavioral (and other) effects .

11. Intentional conteAt reduces to mean ings of internal representa­tions.

II J. Internal processors arc sensitive to the "syntactic forms" of · internal representa tions. not thl'ir meanings.

The first claim is meant to he part nf the communscnse view of the mind. The third is plausibly taken to be a basic claim of the computer model of the mind, and the second i~ a useful and plausible way of thinking how Commonsense rsycholO!-,')' meshes with lhe comru ter

.. · · :;

model. T qucstion i useful pt. J think ti premise

The r. Turing r such as , sensitive Uu1 if 1h a gate. h gates? S se ntat io1 influcnc• soning a computl mind (c

assumrn My r '.

I. t· I I. l

II

111. s (.'

r IV . I

g

A suht I relcvarn of a lav. a count

The fir: propL·r1 SOllll' a and ge name : Bureau United contc111

c: thar grass at hand is

intent from Jtion al con­:cts on our !ad arc not feet on the;: cl if content :ould it play

fo rmiJable reacting to of view. as

; chapte r, I live scie nce the minJ. I lation to the he view ac-

·tation from '5; Pylyshyn, erms of this •ared by the tt (including le compute r 1 of content orwan.l (and tcr is that of ) also would eed causally

e Paradox of g claims all

1tional state) cts. I represe nta-

c forms" of

view of the he computer .siblc way of le comrn1ter

Ct111 rh1· 111i11d change the ll'orld? IJl)

ml>Je l. This s<.:eonc.J daim is by for the most conc rove rsial , hul I won 't be questioning it here . My reasons are that I think it is true. that I see no usdu l purpose to dividing meaning and content in this cn11text. and that I think the best be ts fo r resolving the paradox are to que:-.tion the third premise amt whether the reasoning that leads to the paradox is right.

T he reasoning behind the paradox goes something like this: Any Turing machine can be cons tructed fro m simple primit ive prnces~l) rs

such ;.is 1111<1 ga tes. or gates, and the like . (See Minsky, llJC17.) CJates are sen!'> itive to the syntact ic forms of representations, not tl1eir meanings. But if the meaning 1>f a rcprc:!sent ation cannot in liuence tl1c behavior of a ga te. how coulJ iL influence th t.: hehuvior of a computer - a system of gates? Since intentional content reduces lO meanings of internal rep re­sentat ions, and since meanings of internal re pre!>t!nt ations cannot inllut.:nce behavior, conte nt cannot influence hehavior t.:it her. The rea­soning assumes that at leas t as far as our thinking is con('erneJ , we are computers. This idea - which is s imply the computer model of the mind (o ur cognit ive mind, that is) - may he wrong, but I will be assum ing it m explore where it leads.

My plan for the chapter involves

I. Expl<J ining each premise II. Examining and rejecting a putative solution based nn a nomolog­

ical conception of causal re levance 111. Suggesting a solution based on a functionalist conception of

cont i.;nt and rncaning and a countcrfactual thl my 1>f causal relevance

IV. Discussing a probkm with the proposed solution. one lhat sug-gests that functionalism actually hreec.Js epiphenomenalism

A suhtheme of rhe chapte r is that a nomological theory of causal relevance (a theory tha t explains causal relevance in terms of the not ion of a law of nature) has more of a problem with epiphenomenal ism than a counterfacwal approach.

l . The premises

The firs! prcmi"c uses Lhe notion of a causally relevanr f"Opl.'rty. Some propl'rties of a cause arl' releva nt to the production of an c lfec t, and Sllmt: a re not. I lurrica111.: EliLa hrokc my window. Elii:1's wind speed and geographical pa th are ca usa lly relevant to the hr~aking, bur its name amJ the location of its records in the United ~.ta l es Wea1her Bureau un.: not. 1\ ccord i11g to th~ fi rst premise, if my hel ie f that the Unitcu States is a tfangcrous place causes me to leave the country. the content of the hclid is causally relevant to the behavior; ;i property that

140 NED £31.0CK

is not causa lly relevant to the behavior is the last letter of the name of the city in which the belief was formed.

Note that the point is not that beliefs, thoughts, desires, and the like (mental sta tes or events that have content) are causes, for example, of behavior. (I assume that they arc.) Rather, the point is that when mental events have effects, they typically have those effects (rather than different effects) because the mental events have the contents that they have, rather than some other contents. Typically, if the beliefs, thoughts, and so forth, had had contents that were appropriately different, they would have had quite different effects. For example, had I believed that everywhere except the United States is a dangerous place, then I wouldn't have left the country.

My metaphysical stance in this chapter is one in which mental events are the causes of behavior, and their contents are properties of those events that may or may not be causally relevant to the events' effects. I shall say that property P of event c is causally relevant to effect e and that c causes e in virtue of P more or less interchangeably. Also, I shall put the claim that content is causally relevant to something by saying that content is causally efficacious, and not epiphenomena!. Of course, my use of 'epiphenomenalism ' is importantly different from the tradi­tional one. I'm not raising any possibility of content being a property of a distinct mental substance, events in which are caused by events in the brain even though events in the mental substance never cause anything. And epiphenomenalism in my sense does not entail that content is itself caused hy an underlying physical state that also causes behavior. The reason I use this old word ('epiphenomenalism') for this newer problem is that the problem I raise is the modern heir of the old problem.

The second premise - that content reduces to meanings of internal representations (symbols in the head) - is much more tendentious than the first, and is certainly not part of commonsense wisdom about the mind. But it is a straightforward way of making commonsense realism about content compatible with the view that the machinery of the mind is one of computations on internal representations. According to the computer model, the mind (or its cognitive aspect) can be thought of as a system of processors that take representations as inputs, transform them in various ways, and then send them to other proces­sors, as in computers. One can think of the representations in such a system as being in certain computa tional relations to the whole system. These computational relations are determined hy the ways the system would treat the representation given various different states of its component processors. A sample computational relation is that of storing a representation.

Of cou remembe1 computer: hering mi that it ca1 storing a doctrine . premise. thought t

internal r propositi r thinking t

the head. Talk of

talk. The situation, the syster differs frc tionalism ~

cont rover: tional.

The th syntactic their mca and gate i inputs yic

I. Is 1

11. Do 111. [)('

e name of

1d the like :ample, of hat when 1ther than . that they thoughts,

rent, they ieved that e, then I

tal events ; of those · effects. I 'ect e and so, I shall by saying )f course, the tradi­ropcrty of nts in the anything. nt is itse lf tvior. The : problem Iem. f internal ndentious om about monsense :hinery of \ ccording ) can be as inputs, ; r proces­in such a le system. i e system ies of its s that of

.. ~:~: . . '

Can rhc mind change rite ~vorld ? l ~ l

Figure I

Of course, computers tha t you can buy don't currently think or remember. But if computers can be programmed to think, or if we are computers, this story can be extended to thinking things. Thus, remem­bering might consist in the storing of a representation in such a way that it can later be accessed. Remembering that grass grows would be storing a representation that means that grass grows. This is the doct rine of reduction of content to meaning stated in the second premise. More generally, us is argued in Fodor (1975), to have the thought that p is to be in a certain computational relation to an internal representation tha t means that p, and likewise for other propositional attitudes. The slogan - and it is only a slogan - is that thinking tha t grass grows is having ' Grass grows· in the thought box in the head.

Talk of the thought box in the head is (thinly) disguised functional talk. The sentences that are in the thought box share a computational situation, a role within the system - in other words, a function within the system. T his function al theory of what a thought is (and how it differs from a desire) should be flrmly distinguished from other func­tionalisms to be mentioned in this paper, especially the much more controversial it.lea that the meaning of a represent ation is itself func­tional.

The third premise is that internal processors are sensitive to the synt ac.: tic form of the internal representations that they process, not their meanings. Consider the and gate of Figure 1. What makes it an and gate is that it emits a 'I ' if and only if both inputs are 'l's'; all other inputs yield a ' ()' as outpu t. The a11d gate

I. Is not sensitive to I I. Does not react to

llJ. Docs not det ect

· · ..,.Z,;~~···~;i'_.~~mBDDBJEmmm•m----------

. i

• • ""'I .,-- - ~ • ~r~~ •• .., --:: ,· , . ' • ,'{~~ ~-·.- '":""i•41 ·, .. ~~ •.• '- ;- .. r, '•~

' ' , , I • ;' • : I' I • ' ·~

142 N E I> BLO C K

whether the 'l's' represent truth or the number 1 or nothing at a ll. Rather, the and gate is sensitive to, reacts to, and detects only whether the inputs are both ' l's' or not. Thus it is sensitive to the syntax. not the meaning of its inputs, and likewise for the primit ive processors postu­lated by cognitive science accounts of the mind. (The primitive proces­sors of a system are the ones for which there is no explanation within cognitive science of how they work; their operation can he explained only in terms of a lower level branch of science, physiology, in the case of human pr!mitive processors, electronics in the case of standard computer primitives.)

Note that the sense of 'syntax' I am using here (somewhat mislead­ingly) means fom1 class. 'I' and 'O' arc different syntactic objects in this sense. It is important to he aware that syntax in this sense of the term is another functional notion. English orthography is also funct ional, al­though this may be obscured by the rigidification of function hy conven­tion . For %xampl %, you wi ll hav% littl % trouhl % figuring out what 1%tt %r of th % alphab%t is th % on % to which th% unusual symbol in this s%nt%nc% should b% tr%at %d as functionally %quiva1 %nt.

Consider an input -output system whose input and output registers are bi-stable, and take on values of either 7 volts or 4 volts. Suppose that if both input registers arc at 4 volts, then the output is 4 volts, and every other input yields the 7-volt output. Then (1) the system is an and gate, (2) the 4-volt va lue counts as a ' l ' for this and gate, and the 7-volt va lue counts as a 'O '. A different ly constructed and gate might he one for which 7 volts counts as a ' I ' . ' 1' is conven tionally assigned to states of computer registers using this type of consideration. The functional roles of the bi-stable slates of registers simultaneously determine our identifications of the devices in the system (e.g., as adders and gates) and our identifications of the states of the registers as symbols. So it is having a certain functional role that makes a stale satisfy a sym acric description, in the sense of syntax used here.

Note that it would be a mistake to say that 4 volts in the first gate mentioned has the same meaning as 7-vo lts in the second gate. We don 't know what the meanings of the ' l 's' in ei ther gate are until we see other aspects of the ir function . These ' l's ' could be used to mean one, or true, or green; the input-output function does not choose among these and other possibilities. The aspects of function relevant to syntax are different from, though overlapping with, the aspects of function relevant to meaning.

In the next section , I will explain the picture common in cognitive science of sensible or" rat ional " relations among contents deriving from the correlation between rational relations among contents on the one

hanJ . an the other cogni1 ive putc111ial shall see ··going r worries ; section i ~

to syntax soluliPn .

2 . .

The tit le example. hanc.lk o only knm represent decimal following

(J I ()

I , 0

U -t I [ I

The fir Ii.isl is t rn

Herc i ~

·o· in thi~

gate. 1"11c

inpu h a1 illus t1 ;1tet

hing at all. 1ly whet her ca.x, not the ;sors postu­tive proces­H ion within e explained in the case

.>f standard

1at mislcad­•jects in chis · the term is lCtional , al -1 by conven-1g out what ii symbol in al% nc. ·ut registers ts. Suppose 4 volts, and m is an and 1d the 7-volt ight he one ed to states ! functional te rmine our ; and gates) Jots. So it is

a syntactic

he first gate .d gate. We are until we ;ed to mean not choose

1 relevan t to aspects of

in cognitive eriving from on the one

Cun the J11ind change tht! H'orld? 143

Figurt; 2

hand, and rroct:'>sing rt:latio11s among syntact ic objects in the brain on the uthi.:r . This idea will illu'o!ratc hoth the s trength and wc;1kness of the cognitive scie nce riccure I wil l he assuming. The strcng! h lies in its potential fur a mechanical account uf thought and reason, but as we shall see, it Joes so in a v•ay that rdega te;;s syntactic properties to "going proxy" f11r semantic propaties. and that opens i t up to the worries about cpiphcnomt:nalism discus:>ed here. My a in1 in the ne . ..;t section is to mut iva1c prcmis<.: 3 - that internal proces:>or'> are sensitive to syntax, 1101 rnc;rning - ~ince it will take the heJt in rny proposed soluti1 1n.

2. The brain as a synlactic engine driving a semanlir engine

The title you just read ca n be understood by atten tion to a simrle example, a cornnwn type ur compute r adder stripped d11wn so as to ham.lie on ly one-digit addenda. To understand !he example, you nec:!d only know the following simpk fact s ahout hi na ry notation: 0 and l are reprc"cntec.J alike in binary and decimal. but the binary 1 ranslation of decimal ·2· is · tll'. The adde r pictured in Figure 2 \\ill solve the following four prnhkms:

0 1 0 = 0 I t U = I 0 t I

~ I = Ill

The first thrcl' equations ;m:: true in both binary and dc1·imal, but the last is 11 uc only in binary.

I krc is huw 1lic ;1cJdcr wur ks. The two digits to be acJdnJ. a ' I' amJ a ·o· in this case, ;1rc co1111ectcd both to an und gate anti J.11 exclusive-or gate. The lattt:r gate i'o a .. dillcre ncc detector ... i.e., it ouq uts a · 1 · if its

inputs arc difkn:nt . anu 11 '()' if thl.!y arc che same. In the case illuslratcd in FiFt11 L· 2. tile 1·1dusi1 ·1·-or ga te secs a ditfcrc ice, and so ic

l l I

144 NE D BLO CK

: 8J : 0

Figure J

outputs a ' l ' to the rightmost box of the answer register. The and gate outputs a 'O', and so the device computes the answer. The e.xclusive-or gate does the "work" in the first three problems, and is needed only for carrying, which comes in only in the last problem as illustrated in Figure 3. Both inputs are ·J's', and so the and gate outputs a ' l ' to the leftmost hox of the answer register. The other gate makes the rightmost hox a ' O', and so we have the answer.

Seeing the adder as a syntactic engine driving a semantic engine requires noting two funct ions: one maps numbers onto ot her numbers, and the other maps symbols onto other symhols. The latter function is concerned with the numerals as symbols - without atten tion to their meanings. Here is the symbol function:

'(r, ·o· -+ ·o· 'O', , 1 ~ - ') '

' I ' . 'U' - ·r • 1 ·, 'I' - ' l(J'

This symbol function is mirrorec.I by a function that maps the numbers represented by the nume rals on the left on to the numbers represented by the numerals on the right. This function will thus map nurnhers onto numbers. We can speak of this function that maps numhers onto numbers as the semantic functi on. since it is concerned with the meanings of the symhols, not the symbols themselves. (It is important not to confuse the notion of a semantic function in this sense with a function that maps symbols onto what they refer to; J shall discuss this latte r function short ly.) Here is the semantic function (in decimal notation - you must choose some notation to express any fu nction):

o. 0 - 0 0, 1 -t l I, 0 -+ 1 I , I -t 2

The fi rst maps the onto the

The ke two fun ct

I. A II . Sy·

II J. u~

The symt meanings useful re functi on described these two usual sen how it i ~

numbers. semantic of symbo The idea nature ha tions amc of the sy "syntact ic and 1·s ti meaning­- proces· a correla1 their fon drive our

Now t1

driving a inte ntion our cogn• the synta causal we

In this s making ti

)

and gate xclusi1Je-or :d only for stratecJ in ' '1' to the rightmost

tic engine numbers,

:unction is •n to their

e numbers !presented nbers onto 1hers onto

with the important

nse with a Jiscuss this in decimal unction):

Can rlre 111i11d change rhe world?

The firs t function maps symbols onto symbols; the second function maps the numbers referred to by the arguments of the first function onto the numbers referred to by the va lues of the first function.

The key idea behind the adder is that of a correlation between these two functions. The. Jcsigncr has joined toget her

I. A meaningful notntion (binary notation) 11. Symbolic manipulations in that notation

Ill. Useful relalinns among the meanings of the symbols

The symboli c.: manirulations correspond to useful relations among the meanings of !he symbols - namely, the relations of addition. The use ful re lations n111ong the meanings arc captured by the semantic function above, and the corresponding symholic relations are the ones described in the symbolic function above. It is the con-elation between these two functions (which establishes a semantic funct ion in the more usual !'ense of a function from words to their referents) that explains how it is that a device that manipulates symbols manages to add numbers. Now the idea of the brain as a syntactic engine driving a semantic engine is just a genera lization of this picture to a wider class of symbolic activities. namely. the symbolic activi ties of human thought. The idea is that we have symbolic structures in our brains. and that nature has seen to it that there are correlations between causal interac­tions nmong these structures and sensible relations among the meanings of the symbolic structures. The primitive processors " know" only the " syntactic" form of the symhols they process (e.g., what strings of O's and 1 ·s they see), and not \\.·hat the symbols mean. Nonetheless, these meaning-hlind rrirnitive processors control processes that " make sense" - processes nf decision. problem solving, and the like. In short, there is a correlation between the meanings of our internal representations and their forms. And this explains how it is that our syntactic engine can drive our semantic engine . 2

Now the picture just ske tched of the brain as a syntactic engine driving a semantic engine n:vea ls how it is that a mechanistic theory of intentionality can invite the charge of epiphenomenalism. It seems that our cognitive processes exploit a correlatio11 between the semantic and the syntactic. The syntactic.: properties of the representations do the causal work, and the semantic properties come along for the ride.

3. The appeal to laws

In this section, I will exnmine a putative solut ion, that is, a way of making the cogniti\t: science ric.:ture (premises 2 and 3) compatihle with

' 1

I I

146 NED llL OC!-:

the causal relevance of content (to the behavioral and other effects of contentful mental states, of course - I'll be leaving the prepositional phrase out often, just speaking, e lliptically, of the causal relevance of content). Actually, I shall start by briefly mentioning a rcJuctionist proposal just to set it to one side. If content properties could be identified with, say, neurophysiological properties, then there would be no opening for epiphenomenalism. If content properties are simply identical to neurophysiological properties, then the causal efficacy of the neural would guarantee the causal efficacy of content. Whatever the merits of physiological reductionism, it is not available to the cognitive science point of view assumed here. According to cogni tive science, the essence of the mental is computational, and any computational state is " multiply realizable" by physiological or electron ic states that are not identical with one anothe r, and so content ca nnot be identified with any one of them:l

Note that in reject ing this putative solut ion, I am not rejecting a "physicalistic" point of view. If all the nomologically possible things that can have computational properties are physical things, then the computational point of view, embracing th is idea, is itself physicalistic. Even if the computational prope rties that characterize mentality are in this sense physical, that c..loes not make them physiological or electronic (or syntact ic, for that matter), and so physicalism in this sense does not lead to any suggestion that processors in the head can Jetect content or meaning.

There is anotha putative solution 10 which I will devote more allention, one that appeals to a nomnlogical view of causaiion. The idea is that there arc non-str ict psychological laws involving content , and that law - even non-strict law - makes for causa l relevance. Fodor ( l 987h, forthcoming) argues that intent ion al laws provide non-strict intent ional sutllcient conditions for behavior, and that is what makes content causally relevant to behavior. It will pay us to examine a simple version of this nomist perspective: Fis causally relevant to an effect e if the instantiation of F is nomologically sutticient fo r e - even if the nomological sut1iciency holds only ceteris parihus.

The trouble with this simple version of the nomist approach is familiar: there can be correlation without ca usation - even nomologi­cal correlat ion of F, with G without a causal relevance relation betwee n F and G. And nomological correlation can involve nomological suffi­ciency. Suppose C (for ca use) is nomolngica lly sufticient for and uncon­troversially causally relevant to E (for e fl cc t) . Suppost: X is nomologi­cally correlated with C bt:cause X and C share a causal source, and so X is not causally relevant to C. Then, X is nomologically sufficient for E

without I ·•fork ·· c;

Herc i fire to a enough explosirn enough I

will cxpl· Nov,· t.

and c lcc electron~

electrica. cient for rnnducti

I take 11utin11 u

the n(mJ

crcast: d explo:-,io1 rising d relevant that allo lo explo'

The o avoid tit cases h c.l cases. a1 t erm~ nf way !l> g about wl he one I

r effects of cpositional .: levance of ·eJuctionist s could be e would be are simply efficacy of

hacever the 1e cognitive ~cicnce, the inal state is hat are not ed with any

rejecting a sible things ,s, then the 1h ysica l ist ic. tality are in Jr electronic .se Joes not t content or

cvotc more in. The idea ont ent. a nd 1nce. Fodor e non-strict what makes ine a simple .n effect e if even if the

approach is n nomologi­ion between logical suffi­. and uncon­is nomologi­urce, and ·so ;hcient for E

(·an rht 1J1i11d clumi;e rh e wvr/J? 1-l 7

C;iusa.1

Figure -I

without being cau ~;tlly rclc.:' ant to E. See Figu re ..f, which pictures this "fork .. case.

Herc is an example. Consider a setup in which a metal roJ connects a fire to a bomb. So long as tit<.: thermal conductivity of the rod is low, not enough heat is tr;1nsferred from the fire to the bomb to cause an explosion. But if the thermal conductivity of the rod is increased enough (say, by altering it..; compDsi tion ). then the heat from the fire will ex plot.le l he humh. -i

Now· then:.: is a l:1w - the Wiedemann-Franz Law - linking thermal and electrical i.:ont.luctivity under normal conditions. (The same free electrons carry hu1h charge and heat.) Hence for this setup, ri sing elec1rical conJuc1i ' ity, logcther with other things being equal, is suffi­cient fo r an t.:xplosion . Hence by the nomist criterion, rising electrical conductivity is cau~a lly rekva nl ll > the explosion.

I take it that 10 1lic cxtcnl thac we have a prc theoretical grasp on !he notion uf causal n.:lcvancc , I his consequence l>f the simple ve rsion of 1he nom ist criterion is simply wrong. The eleclrical Cl>!lductivity in­crease does not (;1usc 1hc explos ion; certainly it does not cause !he explosion i11 1·irt11c· uf hein.t: a 1ise i11 c:lertrical c:o11cluc1ii·ity. Rathe r, the rising electrical condue1ivity is an inactive concomitant of the causally relevant rising the rmal conductivity. It is the ri sing thermal conductivity that allows more heat tu be conducted 10 the bomh, causing the bomb to explode:.

The obviuus numist response \vould he to reformul!ltc the thesis to avoid these fo rk c:1ses. The idea would he to formulate what the fork cases have in wmmon that uistinguishcs tht.:m from genuine causal cases. and use this result 1u produce a definition of causal relevance in te rms of .. non-fork'' numnlogical sufficiency. I think this is a promising W'JY 111 go ab(lut cliaracteriLing causal rdcvancc, hut I have two doubts about whether the resulting nomist conception of causal relevance will he one 1hat saves us frnm cpiphcnomenalism.

i

l I I i

l • '

i I

148 NED BLO CK

First, recall that the problem of epiphcnomenalism arose to begin with because the cognilive science point of view held that nature had produced a correlation between syntax and semantics. only the former being directly causally related to behavior. What reason is there to think that an independently motivated definition of " non-fork'' nomo­logical sufficiency would count semantics as any more causally relevanl to behavior than risi ng electrical conductivity is to the explosion in the example considered earlier? Indeed, the correlation between semantics and syntax via a common cause is a good example of a fork .

Second, - the issue of the epiphenomenalism of the semantic may simply resurface in a different form in an independently motivated account of causal relevance in terms of nomologicity. Horwich (I 987, Chapte r 8) frames a definition of causation in terms of the idea of a basic law. A cause is linked to its effects hy chains of clircct nomological determination, and <..lircct nomological determination is de termination via basic law. "A direct cause of some effect is an essential part of an antecedent condition whose intrinsic description entails, via basic laws of nature, that the effect will occur" (ibid.). Cause c causes effect e if there is a set of events, r 1 ••• e,, linked by basic laws, and if c = e 1, and e = e,,. The account concerns event causation, not the causal relevance of properties. But assuming that it could be supplemented to form an account of causal relevance using the notion of a basic law. we can see that although it shifts the terms of the discussion of epiphenomenalism, it does not solve the problem. Is content epiphenomena!? That depends on whether the laws !hat dete rmine behavior in terms of content are basic. But one's decisions on whether these laws are basic - and. indeed, what basicness is - will be (and shou ld he) conditioned hy one's decisions on what is causally relevant to what.

Of course the fact that the nomist conception of causal relevance will not solve this pruhlcm all by itself is not a reason to reject the nomist conception.

Fodor (forthcoming) says !hat if there is a causal law that F instantia­tions cause G instantiations ( cetcris pan"b11.r ), then F is causally relevant to the G instantiations that arc cause<..! by F instantia tions. The appeal to causation in this sufficient condit ion is harmless for Fodor's pur­poses, but not for mine. For how are we supposed to know whether the laws that relate conten t to behavior are camal laws? For example, consider the (very low 4uality) law to the effect that instantiations of wanting G and bel ieving that A is required for G cause instantiat ions of A, ceteris paribllS. To know whether the content prope rties arc causally relevant to the action. we must know if the law is causa l. But if Fodor's suffici ent condition is all we have to work with, the issue of whether the

law is causa no progress

We have jus of nomologi nalism . I k11< re le\·a nee. 11

section, I wi ' - the Coun vancc is con way of avoic.

Su pposc v a {'I'. ' ff} ru a 'O' ou tput , just !he sam and '()' hat.I these symho been using rrimit ivc f)f• object (say, ; another strir caused the s: philmophica the meanin~

conclusion i ~

nal, so arc t I I want to 1

tions arc thl for cpiphcrn·

The f uncti thinking, pl a bines with a hctween inp arise from ti fun ction a I ist ~

body. others these imping ists take fun partly int ern

arose to begin int nnture had nly the former on is there to n-fork" nomo­usally relevant <plosion in the :een semantics '.)rk . sema ntic may ltly motivatet.l lo rwich ( 1987, the idea of a

-r nomological determination tia l part of an via basic laws ses effe ct e if if c = e ,, and usal relevance ed to form an w, we can see enomenalism, That depends 1f content are basic - ant.l.

Jnditioned tiy

relevance wi ll ·ct the nomist

at F instantia-1sally relevant s. The appeal Fodor's pur­

v whether the For example, tantiations of ;tantiations of s are causally But if Fodor 's ,f whether the

< 1111 tlw 111111cl c:lrc111gt.' the 1rorlcl:1 l 4l)

law is causal just is t lic epiphcnorn<:nalism issue we are discussing, so no progrc: 1.,s has been 111adc.

~. A proposal

We have just hccn et111sidcring. an approach to causa l relevance in terms of nomologicity. and '"C have not succeeded in avoiding epiphc:nomc­nalism. I know of on l~ one othl'r remotely promising approach tl' causal relevance. namely. th ~· arpniach in te rms ()f cnunterfactuals. In this sec tion , I will first gi\l· a ve rsion of !ht: argument for cpiphcnomcnalism - the Cou nterfactu;d Argument I wil l ca ll it - in which causal rele­vance is construed countcrfactually. Then I will suggest a functionalist way of avoiding the L' <1nclusio11.

Surposc we have :1 compula! ion al dev ice (say. an and gatd to which a I' 1 ·. '()'} pair - rcprL·senting the numhers I and () - is input, yie lding a ·o· output. representing 0. Now the 1· 1 ·. '()'}input pair would have had just the same e ffect. \ iZ .• the production of 'O' as output, even if the ' I' and ' ()' had represented truth anu f:ilsi ty instead of I and II. or even if these symhols hat.l represented black and wh ite, or even if we hadn't been using them to represent anything at a ll. And what holds for a primitive processor a l'o holds for arty S}S tcrn of them. If a syntactic object (say. a string or 'O's· and ' l's"> is input to the whole system. and an tJ ther string of zerncs and ones is output. the input string would have caused the same output string 11 0 mat te r what it had meant. In cnmmon philosophical p:1rlancL·. the syntax Llf the rcpresc11tatit1n "screens off" the meaning from having any causal relevance to the output. The conclusiun is that since meanings of repres t:ntations arc cpiphcnomc­nnl. so arc the co11tc 11ts of the mental state~ that they ground .

I want to offer the claim that if the meanings of in terna l representa· tions arc their " funL·tional mks." then the Counterfactu;d Argument for epiph c: nnmcnali~111 can he ,·anquished.

The fun ctional role o f a representation is its causal role in reasoning, thinking, planning, and in general in the way the representation com­bines with and inter ;1cts with other representations so as to mediate between inputs and nutputs. The functional roles of representations arise from the ways pr ncessors manipul:ttc their syntactic forms. Some functionalists take inputs to he impingements on the surfaces of the body. others take i11 p11ts to be the things in the world th:ll produce these imringcments Lind likewise f(lr outputs). Thus some functional­ists take functional role to he internal. whereas others t:ike it to he partly int ernal. partl ~ external. My 11sc of the fun ctionalist perspective

'

l I

150 Ni.I> B l.< >C"-

requires no commitment on 1his matter. More speciflcu lly. nothing I say here !>hould be construed as an endorsement of" narrow conten t."

A functional role theory of meaning is a theory nf what meaning is tha t yields an account of what it is about representations 1hat gfres them their meanings, not a ·' semant ics" in the sense of a theory of particular constructions in particular languages. Thus, it would not be the job of a function al role theory of meaning to explain why it is that 'The temperatu re is 70°', and 'The tempera ture is rising' do not entail '70° is rising'. Theories of the sort tha t would Jeal wi th such questions often address th emse lves to what meaning is, but they do not yield an account o( what gives representations their mean ings. Usually, for the ir purposes C:t ny "surrogate " of meaning - anything that acts like mean­ing in the way tlrnt se t theoretic entities " act like'' numhcrs - will do. (See Hlm:k, 1986, 1987, for more on this distinction.) A funct ional role theory can he thought of as a " use" theory of meaning, where the uses are at least partly inside the head.

One motivation fur such an account can perhaps he seen more clearly if one reflects on a case in which one learns the meanings of words without anyth ing approximating an eliminative definit ion for them. Conside r, for example, the learning of a new scientific theOI)' wi th its new theore tical terms. e.g., 'force', 'mass', ' moment um ', 'e nergy'. These new terms are not definable in everyday language (or anything like an "ohsc rvation language"), though they arc definable in terms of one another. T he s1udc11t learns these terms by coming to understand how to use them in thought , in experiment and observation, and in solving problems on quizzes. If mea ning is functional ro le, then it is easy to see wh)' learn ing new te rms is acquiring thei r use . It is a plus for a theory of what meaning is if ii also tells us what it is to know and learn meanings.

That is all I will say he re to motivate the func.:til>n al role accoun t. 5

Now on to the Counterfactual Argument. You will recall that the Counterfactual Argumen t said that it ap­

peared that a given syntactic object would have caused the same output even if it had mean t something quite different from what it actually means or even if it had meant n"othing al a ll , so the syntact ic form of a representation scrl'l'llS uff its meaning from having any causal rele­vance."

Suppose, for example, that the scntcncl: " I here is danger coming this way' is in the "bclid hox" (with its normal meaning). causing one to flee . Accord ing to the Countcrfactual Argumc..:nt , the sentence would have caused the fl eeing even if it had meant that my long-lose fri end approaches, or !ha t Empedocles leaped. Bui if funct ional ru le seman­tics is correct. then ii is not at all guaranteed that 'There.: is danger

coming th1 Empedocl1

. meamng a functional functional

Perhaps includes a; here wher hence its · causal cha coming th i fri end arr fl eeing wo avoid a nee. sentence i been diffe1

Jn sum, tat ion wou been ditte1 different f1 Counterfac

In term~

can be se n Are the mi inputs '.'

Of Ct) Ur~

Thinking < processo r · represt: ntn involve di t represc.: nta would hav1 meaning. 1

for a gem proces~ur ~

compkx r wou ld han meaning. l possibili ty .

This pl 11

tonom1n1s 1980). Oh~

lly. nnthing I say iw conten t. " .vhac meaning is 1tions tha t gi1·es ~ of a theory of it would not be

1in why it is that ng' do not entail h such questions do not yield an

Jsually, for their acts like mean­

nhers - will do. \ functional role , where the uses

een more clearly anings of words 1ition for chem. .:; theory with its . 'energy'. These anything like an in terms of one understand how 1, and in solving 1 it is easy to see is for a theory of learn meanings.

ii role account. 5

said that it ap­che same output what it actually llacric form of a any cau~al rele-

nger coming this , causing one to sentence would long-lost friend

mal role seman-fhe rc is danger

Can the 111i11cl change the ll'or/il? 151

coming this way' would ha\L' l·ausctl fleeing even ii it had meanr that Empedode:-. leaped . For if it h<.1<l had a ditfcre111 meaning, or no meaning al all, its functional role would have been different, and since functional role is causal rnlc, abstractly construed, a difference in functional ro le typically will include a difference in behavioral effects.

Perhaps the functional role of 'There is danger coming this way' includes an inference to senrcnccs such as · tt would be best not co be here when the danger arri\(~s" (with its normal funcr ional role and hence its normal meaning), and perhaps this inkrL'nce is part of the causal chain char led to tke ing. Then if rhe roken uf 'There is danger coming chis way' that was in 1he belief box had mc;111t that a long-lost friend approaches. !he inferL'nce woulc..f not have rakcn pl:.tce, and th<: flee ing woulc..f not have uccurred. We would have approac h inst ead uf avoidance , and fricndlv words instead of fear. If Ilic mc;ining of the

; -sentence in 1he bdicf box had been different, its dfeccs woulc..f ha\'e been ditferent and required different sem;~n tic descriprions.

In ~urn , if meaning is f uncrional role, ch en ir is fol , e thac a represen· talion wouh.i have had just 1hi.: effects that it did have if its meaning haJ been different Ditkrenr meaning requires different funcrional role, anJ different functional role requires different causes and / or effects. So rhe Counte rfactual Argument is unsound if meaning is f11 nctional role.

In terms of the original paradox, the point is that internal processors can be sensitive to hmlz synt ax and semantics. But hl'W is this possible?. Are the meanings of a gatt.! "' outpurs depcndenr on 1 he meaninf!~ of irs inputs'!

Of course not - and 1hert.:in lies the fall acy of the original argument. Thinking of internal processurs on the mode l of gates mi:-,lcaJs. For a processor that is a gm11i11e imi:mional S)'Stt.'111 thi.: difkrence between a represe ntation 's meaning one as opposed to lruth or green, would involve differences in the internal part of the functional role of the represe ntation. Nor so for a ga te. lf we ask whcrhcr a representation woulu havt! had just 1he same behavioral e tfects had ic had a different meaning, then for many differences in meaning. the answer will be yes for a genuine in tent ional ~~seem. though not for . 1 simrle primitive processor such as a gate . Indeed, the criti.: rion of ickntjty natural for c.J

complex processor alluws 011c to consitkr whether a given processor would h;1vc processed a rcp t l·se ntation differently had it hud a uilfcrt:nc meaning. Bur the na1urnl c:ritl!rion of idc!nti ty for a gate rules <nil th e.; possibility of irs processing dilferencly {while remaining the same gate}.

This pl1inc is easier tu apprecia te if we tli~ c in!!uish l>e cween au­tonomous and nbserver-rcl;icive meaning (Searle. 1980; Hau);!eland, 1980). Ohserver-relati\'C! meanings are inherited rn ··anings, mean ings

NED 111 .0< ' K

that intentional systems assign (e .g .. to linguistil' i1 c 111s or slates of a machine). Au1on1>mous meanings arc meanings nf rcrrc"c ntat ion<; or representational stu tcs nf an intentional system - for that system. They get their meanings from their function in the system. The repre­sentations of gates have 011/y obse rver- rdali,·c meaninp.s. (l\1y represen­tations have autonomous meanings for me. but they can also have a variety of ohserver-rclative meanings for o thers.) We can decide, if we like . that a 'I' that is input to~ gate means one. whereas an output ' I ' mc;ins Richard Nixon. We have a free hand. Uut autonomous meanings arc not suhjcc·t to whim in this way. It is for autonomous meanings that the point I have been making applies. flacJ an inru1 symbol had a diflcrcnl ;-iutnnomous meaning (for a genuine intent ional system). then it C1.1ulcJ have liacJ a dillcrcnt functional role ant.I thtlS uifl'crcn! effects. The trick of the argument that originally got us into trouble is lo concentrate on !he example of a gate in which only observer-re lative meaning is rclcntnt, 111aki11g us forget that we do not have a free hand in this way with autonomous meaning.

()f C(lllr<;C the aUlOIHllllOUS mcaningo.; of my rcprCSClltatitlllS Cllllld have hccn diflere 11t in attain ways without any chan!?e in the move­ment s of 111) body. Thi~ ic;; what is imagined in the farnouc;; " twin earth" example'\ - rdcrenlial changes wilhout changes in internal funciional roles. Bui of cou1 se twin earth cao.;cs are elaborately artificial. and one should not conclude from them that hc.t d a representation meant sume­thinp. else. the same behavioral effects \\oulcJ have occu rred nonethe­less. The mi~lcading effect of the gale example is to trick us in to trea ting 11// hypolhctica l Jifferences in meanings of representations as if they arc twin canh cases. If we compare and p.atcs wl11>se ' 1 ·s· mean one versus truth , we arr ronsillering an analog of a !win earth case for ga tes. But for gares. !win earth cases come cheap. Tn re;1cl this chc;1p­ness of l\.vin canh c;.1 ... es hack on to genuine inlcntional system<; is in effe ct to suppose that any old c:ountcrfactual situation in \.vhich refer­ences would be cJiffcrcnt is a twin earth case. and this is ;1 bad mistake.

When I prco.;cn tcd the Paradox of the Causal Eflicacy of Content. I mentioned three premises: that inte ntional content is causally relevant to beha\'ior. that intentional content reduces to meanings of internal representations. anJ that internal processors arc sensitive to syntax. nut meaning. Thinking of these as the premises. 1hc argument is unsound because the third premise is false.

But when I spelled the argument out. I s:1id that primi1ivc processors are sensit ive Ill syntax, not meaning. I then argued that the meaning of a representation cannot influence the neha\'ior of a system of proces­sors without influencing any of the particular processors themselves. So

if meaning Ci

hchavior of . cognitive scit can he an in t the purposes argument is ~

s Dretskc (I l18~ is e<tusall\' re point has Iii I I been discuso.;i favor of I he i• what they "i· production of terms of the Ulock . t tJ8h,;

relevant to th point - in foi senrations· fu any actual he

Consider '~ call it !he fro role of thi!'i re its prnductirn1 cont rollin11 th functional rnl flies <1 nd thei1 causully re leq ancestors of 1

the world -product iu11 o! ancestor of th content "1th r

influence of t fitness on the sort of influe1 tion recruited information ;1 1

improving the were replaced

'\ lales of a nta tions or 1at system. The reprc­y rcprcsen-1lso have a ~cic..le, if we 1 output ' I ' s meamngs anings that 1hol had a .;tcm), then c 11 t effects. 1ublc is to .;e r-relative . free hanJ

ions cou ld the rnove ­win earth" functional

11. and one ~ant some­! noncthe­~ k us into ations as if • 1 ·s· mean th case for his chcap­tems is in 1ich rcfer­t.I mistake. Content. I 'y re levant if internal ;yntax. not ~ unsound

::> rocessors 1eaning of of proces­ise lves. Su

if mean ing ca n't inll1H: ncL' the behavior (if a gate. it c;i n't influence the behavior of ;1 systt:m nf gatL''\. ·1 his rcasn11ing is mi\la ken (given the cognitive science assumplion that a cornpuler - a system of ga tes -can be an intcntilm;tl system - an assumption that I am accepting for the pu rposes of this discussion) and so understood this way, the original argument is sound hu t invalid.

S. Internal functional role and external content

Dretskc ( 1988) has ;1rg11cd that 1hc semantic cnntent of a representation is ca usally relevant to heh;1vit1r . This sect ion will hril'fly note tha t his point has lirtlc to tin with the pwhlcm nf cpiphctHlllll'llalism as I have been discussing it. I think of' Dn.:tskc 's cunsrdcrations as counting in favor of the idea that th...: infPr111;1tional values nf our rL· prescntations -what they "i11dic;1 te .. abe.H1t the world - ;ire ca usa ll~ releva nt to the production of the purely "interna l" aspect of their fu11ctional roles. On terms of !h t..: "two factor" vers ion o f functional rnk semantics [sec Uluck. 1<m<1. aru.J furt he r rcfercm:c~ th t..:reJ. the exlt.:rnal fac tor is causally relevan t to the production of the internal foctor.) Hut - and this is my point - informational value can he causally responsihlc fo r our repre­sentat ions' functional roles without heing i11vo lve c..I in the " triggering" of any actual behavior (in the usual sense of 'he havinr·).

Consider whatever it is that the frog uses to rcpre ..,e nt flies - let's ca ll it the frog's fly represen ter.7 There is an aspect tif the functional role of this representer thar is rnmpktcly int e rnal - 111 ;1i11ly a matter of its production hy tlashcs of movement on the frog's retina and its role in controll ing the aim uf tongue launchings. In add it ion tu its in ternal functional role. this reprcse11tt.:r has informational c11 nten1 rega rding flies and the ir locations. The pt>int is that the latter pl;1usibly has been causa lly relevant to the form er. More exactly. the infm111ational value of ancestors of this represente r - what they have been indicating about the world - ha\'c hcen invoh ed. or so 1rne might '\ t1ppose, in the production uf lhc n.:prcscnlcr·s rurrc111 functiona l n1lc. Perhaps an ancestor of the frog had an int 1..: rnal sta te.: tha t had so 111c informat ional con tent '"'ith respect to ancestor~ of flies or othe r foml 1111 the wing, and influence uf this slate on tongue launching conferred extra inclusive fitness on the frog ancestors whose fly-i nformation sta le had the right sort uf in llue11ce on tongue launchings. We may spcc.:u latc that evolu­tion recruited a primi tive nwtion detector th at providcJ a modicum of in format ion about winged hug" to gu ide the prefrog·s tongue; thereby improving the prefrog ·s chances of a meal. As the hugs evolved into (or were replaced hyl nic". the dclcctur was turned by eq1fution to flies.

I I

N I : ! > 111 tH h.

Then the fact that the frog's fly reprcsentt.:r has been ca rrying informa­tion about flies has causally concributed to giving this represi:ntation the functional role that it has - being produced by retinal movement flashes and guiding zapping. More generally, the line of 1hought is that what our representations have been indicating about the world has had an influence - via evolution - on their having the functional roles that they have in our hcaJs.

I do not wish to gu into th is reasoning in any detail , consider objections to it, or talk ahout the extrapolation from frogs to people. I want to point out only thac even if the reasoning is entire ly correct in its own terms, jt does not show that the informational content of a representation is part of what is causally re levant to (in the sense of a ' 'triggering" cause) the behaviora l output that th<: representation causes, and so Dretskc's proposal is nnt a solution to the problem heing considered in this paper.

Suppose that the frog has a fly-word (' FLY' ), the informational content of which has been causally relevant tu the estahlishment of lhc intcrnul functional role of ' FLY', incluJing the guiding of the frog's tongut: zapping behavior. ls lhis informational Cl>rllenl thert.:by causally re levant to the! prouuclion of, i.e .. involveu in the causa tion of. any particular zapping? This is the cpiphcnomenalism i ~sut: of th is paper (applied tu this case). The answer is obviously not. X can have cau!)ally promoted the patte rn of Y ~ Z without in any way trigge ring the currenc token of z. Fm X can have promoted Y --+ Z without now causing Y or enabling Y to cause Z. The informal ion al content of ' FLY ' does not causally contribute lo the appearance of this loKcn of ·FL y · in the frog 's head. Thal is done hy the fly that caused it. And once 'FLY ' lws apreared in lhc frog, the informational conccnt does not enable or aid ·FLY' in proJucing a Lapping. To dramatize the point , suppose that thi.: current ·FLY' token is a mis reprcsi.:ntat ion caused, say. by a 8-B. This fly token nonethdess indicates Hies (a misrer>resentation of a fly is st ill a rcprcscnta1ion of a fly). The history of correlation of ·FLY' tokens with flies has contributed tO the f unc­tional rok of· FLY' tokens in the frog, but once tha t role i~ se t, the past correlation is irre levant to the process hy which the B-B now produces the current ·fly · wken, which in turn prnduces the zapping that pops the 13-8 into the frog's gut. We can tell the whole mechanistic story abouc this causal prncc~s without saying anything ahouc how the mecha­nisms that sub~erve it a rose. And it is this former question chat this paper is about : is contenl part of the causal proces~ by which behavior is produced?

The plot ~u ism and sla)

the victory <

that funct io can win a h the end be crucial ca~e section is tll nrn havi: lh1 tenta tive.

In hrid. 1

consist in ti properties) input ~ and non-fu11cti111 nut the tunt

To get al that of a fu 1 which I me: or other (sa· one anoche . prope rt ics 11

cnies.) Con~

ignoring the provokes th anger. The ' live, of havi1 some propt· 1

docs the pre li ve n e~!-. cau bull is coo provoke the

Another l '

property. t Ii· first-oruer <.:I

x is dormiti' (when x is it your nuticin1 falling aske1

'ing informa ­sentation the d movement ought is that ··orld has had ctional roles

ai l, consider . to peoplt!. I correct in its :ontent of a 1e se nse uf a ation causes, l)hlem being

nformational 1ment of the .>f the frog's reby causally .tion of, any if this paper 1ave ca usally iggeri ng the .vithout now I content of .his token of Jscd it. And ·ontcnt does amatizc the presentation ates flies (a The history

to the func­set, the past Jw produces ng th~1t pops anistic swry 1 the mecha­on that this ich behavior

(an r/((' 111111d «liu11gc rite 11 c>rltl :' 155

6. Functional properties

The plot so tar is: fun ctionalism nh.:ds arguments for cpiphenomenal­ism and slays them. II would be nice to stop here. but alas, the story of the victory of fun<.:tionalism m·cr epiphcnomenalism is fiction. I argued that functionalism docs defeat the Countcrfoctual Argument, hut you can win a battle and still lose the war. Fum:tionalism loses the war in the end because functional prnpcrt ics are causally inert in certain crucial e<.1-.e :-.. Or rather. I fear that i.1 11 this is !rue. The point of chis section is w raise a sh:cptical doubt. The issues iJre complex, and l do not have the "P<1ce It\ C'iplnrc thc111 :1dcquatcly. Su my claims must bt: tentative.

In brief. mv puinl is this. h1ncti1H1al propaties are properties that consist in the having tif sumc propert ies nr other (say non-functional properties) that ha\'c ccrta in causal relations to one another ami to inputs and outputs. In the production of those outputs, it is the no11-fu11ctio11a l pmpcrtiL:S that :ire s1;111dardly the causally relevant ones, not the funct ional 1mipcn ics.

To get at the poinl. let's consider a slightly more general noiion than that of a functional property. the mition tJf a second-orde r property, by which I mc:rn :.i property tha t consists in the having of some properties or other (say lirst-orJn properties) that have certain causal relations to one another. (The gri..:atcr gener;tlity here is jus1 that second-order properties needn ' t i11nilve inputs :.inJ outputs, as with function;d prop­erties.) ConsiJcr the bullfighter's cape. The myth (which wc will iJccepr, ignoring the inconvenient color-blindness of bulls) is that i1s red color provokes !he hull ; that is. redness is causa lly relevant to the bull's anger. The c .1pe also has the "l:CtH1d-urdcr propc:rty of hcing provoca­tive, of having some pniperty 1>r other that provt)kcs the bull, of having some property or other that is L·ausally relevant to the bull's anger. But does the prolDC(ltirencss llf till.' care provoke the hull? Is the provoca­tiveness causally rele\·:1 111 to the bull's anger? Ir would seem not. The bull is tno stupid for that. The provocativeness of the cape mighr provoke the ASPCA. hut not the bull.

Another example: c1111sider dormitivity construed as a second -order property, the po~st:ssit>ll uf some property or ot her (for example. a fi rst-order chemical prnperty) that is causally relevant to slcc.p. That is, x is dormitive = x ha:- some plllperty that is causally relevant w sleep (when x is ingested). II a dnrmitive pill is slipped inlO your food without your noticing. the property of the pill that is causally relevant to your falling aslct:p is a (p1 L·s 11111ahl~ lir!->t-order) chemical property. not, it

156 NED Ill <>CK

would seem. the <lurmitivity itself. Different dormitivc potions will ;-ict via differenl chemical properties. one in the cnse of Valium. another in the case of Seconal. But unless you know about the dormitivity of the pill, how could lhc dorrnitivity ilself he caus<tlly relevant to your falling t.1s leep?

Of course if you do know ahoul the Jormitivily. then it can be causully relevant to your sleep. just as the provocative ness of the cape can affect the ASPCA. In fact . there is a well-knmvn phenomenon in which dormitivity doC's cause sleep. namely, the placeho effect. If a dormitive pill is so labelc<..I , thereby causing knowledge of its dormitivity, this knowkuge can cause sleep (though the truth and justification of the knowledge are of course causally irrelevant). So dorrnitivity can be causally relevant to sleep. OnJcc<l . there can be a dormitivc pill that works without any first-order effect. a pill whose dormitivity requires its own recogni tion. Suppose I market a sugar pill as a dormitive pill, and it becomes popular and works well . I make a fortune and close my plant. Yc;us later. when all my pills have heen used up. one of my customers who had ha<l years or sound ~lcep as a result of taking my pills finds nut that they worked ,·ia the placebo effect and sues me. Surely I can point out that he "as nol cheated - the pills were genuinely uormi livc. there was no false advenising.)

My claim is that second-order propert ies are not always causa lly relcv;111t to the effects in terms of which they arc defined.>< The only cases that I can think of in which second-order properties -;eern to be causally c flicac iou~ arc those where an intelligent being recognizes them. That is why I keep mentioning " standard" cases, cases where a secm1d-ordc.:r property is defined in terms of an eflcct and that effect is pr<H..luced without any recognition of the second-order property hy an intelligent being. J ndc..l to the claim of the first sentence of this paragraph the mnre tentative claim that in these standard cases, the seco1H.l-ort.kr property is causally inefficacious.

But hov. ca n it be that second-order properties are inert in some cases. efficacious in others'! Think (temporarily) in terms of a norn ist theory of causal relevance. If dorm itivity of a pill were nornologically su fficient tor the ingester getting cancer (Jerry Fcldor keeps tl)1ing to convince 111c that such a thing could happen without ren>!!nition of dormitivit) by an in tell igent being). !hen (let us suppose) dormitivity would he causally relevant to cancer. Out such cauc;aJ relevance to cancer wnuld not show or even suuest that dtirmitivity is 1101110/ogically sulflcicnt Im slt•cp . Hence we might have cetusal re levance to cancer hut not lo sleep: non-standard causal re levance without standard causal rek,·;1ncc. ( MPrc 011 nomological suflicicncy in a moment .)

Second-< though it i first-order. procedure a theory ir. some sort (In the ca~ as theoreti oretical te leaving m define 'T

1 •

So far, t

order prn1 itivc consc arc, a nat1 taking sue the causal special sci be that we sleep, and principles two.

first. le that cone.:< the effect nomologi< sleep. Tht being a w and expa dormitive this defini - and I t

sufficient relation. · nornologit

Now. I propcrlie~

for prope my aunt's tracks c.:h; his is alw nomologi·

otions will act 1m, ;mother in mitiviry of the to your falling

hen it ca n be ·ss of the cape hennmcnon in ho effect. If a its dormitivity, ification of the

1itivity can be uitivc pill that 1ity requires its rnitivc pill, and

and close my up, nnc of my It of taking my

and sues me. the pills were

ilways causally 1ed.H The only ies seem to be . . rng recognizes cases where a

1d that effect is property by an ntence of this Jard cases, the

inert in some ms of a nomist ! nomologically keeps trying to rcnignition of

Jsc) dormitivity al relevance to is nomologically :c to cancer but 1tandard causal nt.)

Can thl' mind 1'!1ange the 1~·orlt/.? 157

Second-order properties involve having some properties or other, and though it is often hclpf ul to think of the propenies quantified over as first-order. actually they can he any properties at all. There is a general procedure (see Lewis. 1970) for defining a second-order property, given a theory in which the property plays a role so long ;1s the theory allows some sort of a distinction between theore tical and observational terms. (In the case of a psychological theory. the distinctilln would he cashed as theoretical = mental. and observational = input ,'output.) If the the­oretical te rms arc 'T1' .•. 'T,, '. we can write the theory as T(T1 ••• T,,). leaving out all mention ol' the Dbservational entities. Then we can define 'T , ' as follows: .r has T 1 == EF1 • •• EFn [T(F1 • •• f) & x has F1 ].

So far, the case I've mac.Jc for the limited causal inertness of second­order properties is based entirely on examples. But if the counterintu­itive consequences of this causal inertness are as had as I will claim they arc, a natural response will be simply lo live with rejecting my way of taking such ex<1rnplcs. That is, if I am convincing Liter when I say why the causal inertness of the second-order commits us to a view of the special sciences I hat is hard to sv..;allow, the rcason;1ble response would he that we should just suppose that dormitivity is causally relevant to sleep, and provocativeness does affect the bull. So I wi ll try to get at the principles that underlie our reaction to the examples. I can think of two.

First. let us return to thl.'. nomist conception of causal relevance. On that conception, second-order properties are not ~ausal ly relevant to the effects in terms of which they arc defined because they are not nomologically sufficient fur those effects. Consider <.lormitivity and sleep. The relation between the two is more like the relation between being a widow and having had a hushand than th;1t between, say, heat and expansion. If a pill is dormitivc in the following sense: x is dormitive iff x has some prl)perty that causally guorantees (this is where this definition differs from the one offered earlier) sleep if x is ingested - and I take the pill , it follows that I sleep. The L1ct that dormitivity is sufficient for sleep is perfectly intelligible in terms of this logical relation. What reason is there to suppose that there -must also be a nomological relation between dormitivity and sleep?

Now. I am very much not saying that a logival relation between properties pn·cludes a nomological relation. This is as much a fallacy for properties as fur Davidsonian token events. Suppose dormitivity is my aunt 's favorite property, and sleep is my uncle ·s, and that my uncle tracks changes in my aunt's favorite property, changing his own so that his is always entai led by hers. Then dormitivity and sleep will be hath nomologically and logically related. Logically related under one set of

158 Nl·.D 111 .0CK

descriptions, nomologically under another. My point is not that a logical relation precludes a nomological relation, hut rather that the logical relation be tween uormitivity and sleep tells us perfrctly well why dormitivity involves sleep. The re would have to be some 5pecial reason to postulate a nomological relation as we ll, and since the stOI)' about my aunt and uncle wasn't true, I don '1 sec any such special reason. The point that this example is me;rnt to make is that it would he amazing if there was always some special reason why a second-order properly was nomologically related to the effects in terms of which it is defined.

This conside ration is based on the fact that sect)nd-order prope rties are de fined in te rms of effects. A second argument is based on a different featu re of sc.;concJ-ordc r prnpenics, the quantification involved in them.

Suprmsing that provm:ativcncss provokes the hull would be supposing a strange ·sort of ovc rd c.; tc.;rmina tion of the bull 's anger. Of course overdetermination docs some times happen . (The placebo effect is an example.) But to suppose that it always happens would be to suppose a bizarre systematic uvcn.k tcrmination. Wheneve r wt: have a first -orde r causal relation we can a lways de fi ne a second-order property on the model of provocativeness. <rncJ so every firs t-orde r causally relevant property would jointly determine its effect together with a second-order property. Indeed, the procedure iterates (for the re is a third-order property that consists in the possession of a second-order property that is causally relevant to the effect l, anJ so whenever there is one causally relevant property the re woulJ be an infinity of them. Even if the first-orucr property is causally sufficient for the dkct, there would still he an unending series of o the r causally relevant properties. And we can define causally suftlcient highe r-order properties that wo1,ild also he causally sufficient hy the same reasoning.

The relation hc tween second -or<kr prope rt ies ant.I the effects men­tioned in thei r definitions is a "fork" relation, a hit like the one discussed earlier in the selection on corre lation and causation. Both heat anJ electricity are conc..luctcd hy free electrons in metals. Rising velocity of free electrons

I. Is responsible for rising the rmal conductivity, and thus causally re levant to I he explosion

11. Js responsible for the cpiphcnomcnal rising electrical conductiv-ity.

Similarly, there is a tirst-nrdcr chemical property of Seconal that

I. Is causally rdevant to sleep II. "Generates" an epiphe nomena! second-order property of possess­

ing some propen y that is causally relevant to slee p.

In othe r w. explosion \· conductivil 1

ing an epip The pict

analogy ju! rela tions n different, a (limited) er used to pr< chemical pr tivily illtistr nalism' anJ gu men l a ls1 overde tcrm coincitkncl' is a lways a coinciJencl' causally efl i causally effi the prublen

In the c1 relevance i: favor count as earlier in nalism , ~o j cal theory. though it s approach .

Let me ' propeny is occurs; my relevant to second-orJt defined (a11<

of them tak seco1H..l -urdL would not c epipher1< Hllt lion argumL order propc effects. The of content

hat a logical t the logical ly well why

iecial reason >ry about my reason . The e amazing if 1roperty was Jc fined. :r properties based on a ion invulved

>C surposing . Of cours<.: e ffect is an

tO suppose a a firs t-o rdt.:r >e rty on the .lily relevanc econd-order

third-order 1ropcrty that one causally Even if the e would still And we can ~lid also he

dfccts men­ke the one .ation . Both .: tals. Rising

nus causally

I co1H.Juctiv-

·nal that

y of possess-

C1111 the 111111</ clta11gc: tltc h'c>rlcl? I 5lJ

In o ther words. just as the ri~ ing velocity of free ckctrons causes the explosion while engendering an criphenomenal increase in electrical conductivity, so the chemical prope rly causes the sleep while engender­ing an epiphenomena! second -orJer prope rty of Jormitivity.

The picture just sketched is au ractive. but hardly compell ing. The analogy just mentioned is far from perfect. The two engendering relations ment ioned in the prccL'l.ling paragraph arc certainly very different. and besides the analogy only holds if I am right about the (lim iced) epiphenomcnality of secu11d-orde r properties, so it cannot he used to prnvc that cpipht.:numcnality. Oncide nl:.llly. the fact that the chc:mical properly " engender:-. " dormitivicy ralher than causing donni­tivity illuslratcs the c.Jilforenet.: betwec.:n what 1 am call ing 'epiphenomc­nalism' and tradic ion al c..:p iphcnomr nalism.) The ovc..:n.ktamina tion a r­gument also is far frum cunvi11cing. We arc norm;illy 1 eluctant to accept ovcrdeccrmi11atillll hc..:cau~e ii is wrung, o ther things equal, to postulale rninciuenccs. If a man di1,;s hy drowning, we cannoc suppose that there is always another cause nf death :is wel l, say. shouting. But no such coincidence would ht: imolvcd in Ille series of highc..:r -anJ-higher-order causally e fficacious propercies I mc11 cioncd. If accepci11g such a series of causally etfit.:acious properties is a price that must be paid for avoiding the probkms to he men tioned , it ca n be paid.

In the cnu. the argument hased on nomological theories uf causal relevance is the best one. l loweva. we can hardly expect those who favor couni crfactual theories uf causal relevance to he convinced. Here. us earlier in the papa, 1h1' less1 >11 is tllllJ if you a1a111 10 111-oid epiphenome­na/ism , go j(Jr u co1111rafac tLw/ !11eory of ctwsul n:lernnCt' , 1101 11 nomologi­cal 1heo1y. This is, l suppose. the main positive roint of the paper, though its significance derwnds llll the !'ale of I lie countc rfo ctu~tl

approach. Let me sum up 1hc skep! ical thesis. Suppose that a second-orde r

property is ins1antia1ed , and the effect in terms of which it is defined occurs; my cla im is that the ~ccond -ordcr prope rly needn't be causally relevant to 1he e ffec t. I ha\'t~ mentiuned one type of case in which a second-order property can alfcct the effects in te rms of which it is defined (a nd ot her things as \\cl )), namely. when intcllige ~ c recognition of tht::m takes pluce. as in !hr placc ho efkcl. Accepting other sorts of second-order effects (e .g .. c;i usal relevance of dorn1it ivi ty to cancer) would not change my claim. since the arguments l gave fo r the limited epiphenomenality - the nomological argument and the overdetermina­tion argumenl - were res tri r1ed lo the causal re kvance of a sccond ­order property on the effects in terms of which it is defined, not other effects. T he c piphenomenalism I am worried about is 1101 total inertness of content properties, but '' he tht.: r. for cxampk. tile c:un tent uf my

J l

f i t

f

•

1611 Nl . D Ill (lC"K

Figure 5

desire for an ice cream cunc is causally releva nt to my !!ning to the ice cream shop. 11

At this poi111 , I shall change gears ahruptly. /\lthn11gh . as l'\'c said, the case I've matle for the (limited) incflicacy of the second-order property is far from conclusive. I think I have said- enough for it 10 be taken seriously. What I now propose to uo is simply ns.mme that second-order properties are not always causnlly relevant lo the effects in terms of which thl'y arc defined. and go on to discuss 1hc conse­quences of this idea.

The skeptical point of this section depends on !he fact !hat the mechanisms that manipulate genes and the mechanisms that manipu­late the internal represe ntation ~ in the hrain are not inte ll igent, and so cannot recognize second-order properties. Hence these are just the kinds tlf cases in which it would seem that the scconu-nrder propert ies arc inefficacious with respect to the effects in terms nf which they arc defined. Likewise for other special sciences whose properties nrc rlausi ­bly second-order.

Uy way of an example. consider a simple finit e automawn, !he one specified hy the machine tahk of Figure 5. What the table requires of any machine that it ckscribcs is that it have two sta tes. S 1 anc.J 5 2 , such that when the machine is in S 1 a11d sees a 'I' it cries 'oc.lc.I' and goes into 52 : and when the machine is in S~ and secs a ' I ', it cries 'even· and goes hack in to S,. And when the machine is in S 1 /S ~ and sees no input ('-' symholizcs the null input). it cries 'cven·; ·odd' (respectively). staying in the same state. Thu<; the machine tells us whether it has seen an odc.J or even numhcr of· rs· (though 1101 entirely sunT..,sfully. since it is under the impression thar 0 is an even numher).

Now suppose we have a machine on the desk in front of us that sa tisfie s Figure 5 <i.e., is described hy the tahle) and is in S,: it sees a 'l', and so it cries 'odd·. Let us ask: is the property of hcing in state S1 (of a machine of the type of Figure 5) causally relevant to the production of the cry of 'odd'? If the points made about dormitivity a nc.J provocative­ness are right, it would seem not. If we think of S 1 ns a property, it is

the prnpnty rcla!cd to 1>11

5. These ot h< a gcars-and­compurcr irn tronic implc causally rclc' such imrlcm

The up<;hc what cu1111w implement al sl ril ngc con\ comrarahly tional scicnt

( 'onsider plausibly f 111

tancc and actual char;1 dominant 0 1

of the di~l 1

there arc n prorerty of chemical pr via the prm tional prop proccs~ors

tics such ; processor~

rrorcrt ies which thcv gene . 111

In sum. inconsistcn

I. Speri. scicrn

2. Speci ~- runcl

e ff cct

l huvc jt us is a d epiphcno11

1g to the ice

as I've said, ccoml-ordc r for it to he

1.rn11ne that ) the effects ; the consc-

lct that the ha t manipu­gcnt, and so are just the ·r prnpcrties ich thev arc s arc plausi -

i C.Hl, I he one ; requires of :.1nd S z. such nd goes into en· and goes ~es no input ·espcct ive ly ), :r it has seen fully. si nce it

1t of us th at ; it sees a ' 1', state S 1 (of a iroduction of provocative­

>rope rty, it is

( '1111 ""' 111111cl c/11111.~c the 11 ·orld:' lhl

the prnpcrty 1ha1 n111-;i~f'.. 111 h:l\·i11g twn other properties tha t are reli1tcd tu 1rnc a1111thu a11d Ill inputs and outputs as spccifk<..1 in Figure 5. These other propcrtici; would he mechanical properties in the case o f a gears-and-pulleys impkmL'ntaliPn. electronic properties in a usual compute r i111plcmcntatio11, ;111<..I so t>n. If we arc Jealing with an elec­tron ic implementation. then it is the electronic properties that are causally re levant to the cry of 'oud· . not the property of possessing some such implementation proper!~.

The up<\ hot is lhat comp11tt11io11al properties lu11 ·c 110 causal relel'lmce fO

wlim cumpurcrJ clo. these c.:fkcts hcin!! en tirely the product of the implementations of the cn111putatio11al prorertics. This is indeed a strange cnncl usio n. and if I i llll right about second-order properties. <.:t1111p<1r:1hly st1 ang.e C(l nl'lu~i11 11s l'lltild he 1t:acl1cd about olhn func ­tional sc ic11n:s.

Conside r the rrorc r1 y of !ici ng a brown-hair gene. This property is plausihly functionally characteriz:.ibk in terms of its patterns of inheri­tance and interaction \\ith u ther genetic properties in determining actual characteristics. for e:w 111ple. in the fact thal brown-hair ge nes are dominant Dvc r hlonc.J-hair genes. Such second-order properties are part of the distinctive conccptu;tl apparatus of Mende lian genetics. But the re arc 1H> ge netic 111ecl1 ;111isms that detect c.)r arc sensitive lo the property of being a hruwn -hair gene. Genetic mechanisms detect bio­chemical prnpcrt ies. anti the prope rty of being a brown-hai r gene arises via the processing hascd c>n hioc.:hernical properties. Similarly, computa­tional properties such as S 1 men tinned above arise via the activity of processors thal detect syntact ic properties. not computational proper­ties such as S1. And se nwntic properties arise via the activity of processors lhat c.ktcct syn tact ic properties. All these second-order properties arc epiphenomena! with respect Lo the effects in terms of which t ltcv arc dcfinccJ. such as l:irnwn hair in the case of the hrown-hair gene. 111

Jn sum . we want to 111;1i11tai11 al l of the fu llowing. but they arc inconsiste nt :

I. Special science rrnper1ie'> are c;1u<\a lly rek,·;111t to the effects those sc.:ienc.:es predict ;111J explain .

2. Special science prcirc r!iL'S arc often functional. J. Fune.: I iunal proper! ics arc sla nt.lard !~· causally irrelevant !o the

e ffe cts they are defined in tcrms (lf.

J have just argued for J . JI I am right. then we have what for many of us is a uilemma: we rnus1 ei ther abandon functionalism or accept epiphcnomc nalism. But am I right about Y? Recall that the case for J

162 Nl · I> 1.11.<> CK

depended mainly on the nomological approach to causation, and per­haps the way out is to reject that approach. Before accepting that conclusion, however. it would he best to examine another way out.

7. Because

The claim that second-order properties are standardly causally irre le­vant to the effects in te rms of which they are defined seems to fty in the face of a va riety of commonsense and scientific facts, for' example, true uses of the r ollnwi ng:

I slept because I took a slee ping pill. My muscle tension evapora ted bccau::.c I took u muscle relaxant. The vase broke because il was fragile. J have blue eyes hccause I have two bluc-cyt: ge nes

l shall argue that such facts only seem incompatible wit h the (stan­dard) causal inefficacy of the second order because of two ambiguities tha t easi ly escape notice.

Lrpla11atio11 and ctw.rn1io11

One ambiguity is that ht:twccn explanatory and causal usc5. of ·because'. It is easy to be unclear ahout \.\hcther a ·oecame' sta tement is true in an explanatory or causal sense of ' hccau!->c '. If the dispositional / func­tional terms in the sentences lisled above are understood to refer to second·order properties, then the sentences are only true on an ex­planatory, not ;1 causal, reacJi11g of· because· - i f the line of tlwught of the last seet ion is correct.

Audiences who have heard this chapter read as a paper have ofcen objected tha t al though thc.!re is a clear distinction between explanation and causation. the dist i net ion het ween the n wsal-expla111.1to1y rele1'w1ce and causal relc•1'a11ce of properties that I rely on here is a distinction without a difference. Herc is the issue: The kind of explanat ion in­volved here is ca usal explanation. which is one of a number of species of explanation . So the kind of explanatory relevance of properties at issue is causal-explanatory relevance. And what is the distinct ion be­tween causal-L·xplanatory relevance of properties and causal rdevance of properties?

Here is a way of seeing that there is a difference and what ii is: notice that we can ca usa lly explain a case of sleep via appeal 10 the sccond­order property of dormi tivity. I fell asleep because I took a pill that had the property of having ~omc prnperty or nt ht:r that c;wse!'> skcp. Th i~ is

causal-explar of my fallin! second-order appeal to du Dormitivity causally re lc' appeal to a 1 tion of sleep chemical pm second-order in the posse~ cause it invol property ca11

causally dlic fi rst-order or for the stand

Func tional le.:

One con'itru . dormitiv1 1y ""'" one equally j This is th<: n< dormitivc:: vir , const rual is l

lion that pid former const1 to cont ext. · 1 same SeL·o1H.I ditfcrc111 lir~l

potions. For diffaenl fror Similar!), ·I he lotteries (Lev

To sum u1 rekvant Lu !>I

the 'hecaust'. ascribed to tl

David Le~ Smart 's itka Functional l e

i, and per­~pting thal 1y oul.

;ally irrele­o tty in the 1mpk. true

1 l.

the: (stan-1mbiguities

· · bccau!'>c 1•

t is true in )nal / func­to rdcr to on an cx­thought of

have often xplanation v relei•w1ce dist i net ion tnation in-of spc:cies

.)perties at

. nct ion be­• relevance

it is: notice ne second-111 thal had ·er. This is

Can rlie 111111cl clwng!' rlie 11·or/11:' lhJ

causal-explanatory because it ruks out alternative causal explanations of my falling asleep - for l!xample, that I was gra~ing pa~ers. But second-order properties being (standardly) causally 1nefficac1ous, the appeal to dorrnitivity is not an appeal to a causally relev~nt prope~ty. Dnrmitivity is causal-explanatorily relevant tn sleep without being causally relevant to sleep. Why? T he appeal to dnrmiiivity im·oh·es an appeal lo a property that is genuinely causally relevant to th: produc­tion of sleep, namdy, the unnamct.I property (pn.:sumahly a lirst-order chemica l property) whose existl!nce is mentioned in the analysis of the sccoml-ort.ll.!r propl!rt y, thl.! onl.! that the tlmmitivity ol the pill consists in tht: possl.!ssinn of. Dormi1ivi1y has rnusal -explanawry relevance be­cause it involves a causully dlicaciuus properly. A causally inefficacious property c<.111 nonetheless be causal-explanawry if it "brings in"' a causally ellicacious property. Uut the causally rdevanl property is the first-order one , not the second-order one (assuming that my argument for the standard causal inertness of the second-order was right).

.·lmhig11i1y of f111 1c1io11al 1em1s

Functional terms arc ambigullUS. Ll.!t 's lakl! 'dorrn1tivi1y' as an example. One construal is thl! seconu-orucr o ne much di sc11"isl'd here already -dormitivity -"" pu.uess111g a prup('n)• ilwt causes sleep. The other construal, one equal ly justil ic..:<.J in ordinary usage, is rhc propeny rhat rnuses sl!'ep. This is the natural reading uf ·uormitive virtue' in the ;1sscrt i1>n that the durmitive virtue of Secon;.il is causally responsible for sleep. This latter construal is one in which ·Jormitivity ' is analyzed as a t.letinite descrip­tion that picks out the firs1 -orda property that is quantified over in the former cunstrua l. The two cunstruals arc vay dilh:rcnt in their relation to context. · Dormitivity', on the formt:r construal. always picks out the same seconu-nrc..l e r property. But on tht: latte r conscrual, il picks out dilferent first-order properties in the case of diffcrt:nl types of slcepi11g potions. For the prnpercy thac ca uses sleep in 1he c11se of Seconal is difforent from the property that ca uses sleep in the case uf Valium. Similarly, '1he winning number" picks out Jilferc..:nc numhers in dilferenl lotteries {Lewis's example) .

To sum up: In the second-order sense. uorrnilivity is not causally relevant tu skt:p; in the first-order ~cnse. it is. And the plausibility of the 'because' claims mentioned al the outse t of thi ' section can be ascribed to the first-orJc:r interprera tion. 11

David Lewis has long advocated a form of functionalism based on Smart's it.lea of topic neulral analyses of mental 1er111s (Lewis, 1972). Functional lerms arc defined by Lt.:wis in th l.! faller ol the two manntrs

•

IM NFI> 111.0C"I\.

mentioned above - in terms of definite descriptions. Many functional­ists adopt a f unctiunal state identity thesis, one that identifies mental states with second-order properties. Lewis, hy contrast, says that mental states are fimc:tional/y specifi<'d brain st at es. Many fun ctionalists take pain to be a second-order state, but Lewis takes pain to be the first-order hrain state that plays the role characterized hy the second­order state. Where some functionalists identify mental states with states that consist in causal roles, Lewis l<1kcs mental sta tes to be the entities that . have the causal roles. The issue is aboul how to regiment the language of the mental (and other 1hcorctical langu<igcs).

Now some may embrace Lewis's first-ord er reading of functional terms of the special sciences as a way of having their cake and eating it too. On Lewis's construal. it is correct to sciy "Dormitivity is causally relevant to sleep," "Content is causally relevant to behavior," and the likc.12 If we accept Lewis's reading, we can preserve the talk that would he appropriate if special science properties were causally efficacious. So it seems th;-it we can avoid epiphenomenalism. and not have to give up functionalism. And there is no need to reject a nomological theory of

· causal relevance. Unfortunately. the Lewis solution (as I shall call it) is only cosmetic.

To construe the functional terms of a special science in Lewis's manner is to construe them as picking out " lower level' ' properties, properties of an implemefllation science. 'Believing that grass is green' on the Lewis construal picks out a physiological property. (In us, that is; in the case of a hypothetical intelligent computer, it would pick out an electronic property.) And in a certain sense, 'be lieving that grass is green' does not pick out a psychological property: that is, it does not pick out a property that is part of the distinctive conceptual apparatus of psychology. On Lewis's construal, 'brown-hair gene' (a term used by Mendelian geneticists long before the advent of molecular biology) picks out a piece of D!'IA, the concept of which would have been utterly alien to Mendel. Consider the computational state S 1 of Figure 5. I said it was a computational state. and indeed the computat ional definition was given in the text. But on Lewis's construal. 'S 1' picks out an e lectronic state or a mechanical sta te. or some other implementation state, depending on the way the automaton is constructed. 'S 1' would be defined hy Lewis's method as the ( co11tcr;tuall.r relr!'am) state that has the

role specified in rlie table of Figure 5. So · S, · dues nor pick out a computational state at all. .

Of course, if a "psychological property" is just a pniperty picked out hy a psychological term, then 'believing that grass is green' does pick

out a psycl out a Mc; property. Pickwickia the concCJ part of thL

tinnal pro nothing ar tus of clc<.:

My poi1 properties properties strual doc state 5 1 h of" Even! tional stat · the causa \arly, sim psycho log. believing of the me

Constn make the distinct iv( way of r• Lewis's n But 'havi ·s,· pi ck~

tion stat< Computa sciences. to them enccs. T sciences conceptt: spcci<ll ~ propcrti~

cff ccts it In su1

epiphcn• allows u·

Many functional­itfentifies mental

. says that mental unctiona lists take

pain to be the ti hy !he sccond­s tatcs with states lo he the entities to regiment the

es).

ng of functional ·ake and eati ng it itivity is causally

· havior." and the e talk that would !ly efficacious. So t have to give up logical lhcory of

is only cosmetic. 1 Lewis 's manner ~rtics. properties s green' on the us, that is; in the dd pick out an ng that grass is it is. it does not ~ ptual appara tus (a lerm used by

>lecular biology) 1ave been utterly f Figure 5. I said tional definition , · picks out an implementation

~ d . 'S ,' would be start• that has the nor pick out a

:>erty picked out .rcen · does pick

-- - --

(·an the mind clwnge the world? 16.)

out a psychological property. and likewise 'brown-hair gene· does pick out a Mendelian property and ·5 1' does pick out a computat ional prope rty. But these: consl ru;.ils o f "psychological rroperty'', etc., are Pickwickian. Psychological properties, on this construal, are not part of the conceptual apparatus of psychology; Mendelian properties are not p:irt of the conccrtual apparatus of Mendelian genetics; and computa­tional propert ies are properties that a computer scientist may know nothing about . properties that arc really part of the conceptual appara­tus of electronics or mechanics or hydraulics.

My point is this: You tfo nol vindica te the causal e fficacy of the prope rties of a special science by construing its terms as referring to the properties of other sciences, impleme ntation sciences. The Lewis con­strual docs preserve a way of talking that we want. We can speak of the state S 1 being causally relevant to the production of the machine 's cry of "Even! " But the cost is that we no longer take S 1 to be a computa­tional state (except in a Pickwickian sense). So a Lewis vindication of the causal efficacy of the computational is certainly Pickwickian. Simi­larly, since believing that grass is green is a physiological, not a psychological , state, the preservation of talk of the causal efficacy of believing that grass is green is n(l real vindication of the causal efficacy of the mental.

Const ruing terms of the special sciences in Lewis's manner does not make the second-order properties I have been talking about any less distinctive of their special sciences. Rather, it merely forces a different way of referring to them. 'Believing that grass grows', construed in Lewis 's manne r, picks out a physiological state, not a functional state. But ' having tile hcl id that grass grows' picks out the functional state. 'S 1' picks out an electronic or mechunical or some other implementa­tion state. hut 'be ing in S 1' picks out a genuine computa tional sta te. Computational stat es are what are distinct ive about the computational sciences. and Lewis's way of talking does nol prevent us from referring to them or make them any less distinctive of the computational sci­ences. The same holds for the functional properties of all the special sciences for which functional properties are part of the distinctive conceptual apparatus. These properties remain distinctive of these special sciences even if we construe terms a la Lewis. and these prope rties remain causally im:rt (that is, causally inert with respect tu effects in terms of which they arc de fined).

In sum, moving to the Lewis way of talking does not solve the epiphcnomenalism problem, but only provides a way of talking that allows us to avoid facing it. The basic picture pro\·ided by the Lewis way

• I 00 NI.I> llLOC'K

of talking is the same as lhe one I have been trying to avoid: special sciences whose distinc:live properties arc causally inert but thac can be used to pick ou t causally cl!icacious properties of lowt.:r-lcvel sciences.

T his brings me to a second objection to che Lewis solu tion. Second­order properties in the sense discussed here art! properties that consis t in the having of properties wi th such and such relat ions to one another. As I mentioned earl ie r, the properties quantified over need not be first-order. In the case of our menrnl properties construed as second­order properties, the quantified propert ies are physiological. Phy~iologi­cal properties are biologicul properties, and biological properties gener­ally are prime· candidates for functional analysis. Thus ·believing that grass grows' construed in lhe Lewis manner, since it picks oul a physiological property, may pick ow another jimcrio11ally imlfritl11M~d

property. Perhaps biological terms themselves pick out first-order prop­erties, but perhaps nol. Perhaps only the most fundam ental of physical properties are ge nuinely firs t-order. In that case. even much of physics would be epiphcnomt.: nal! O r perhaps not evt:n the most fundamental physical properties arc first-order. (Maybe there isn ' t uny mllst fun<Ja­memal level.) Perhaps science is functional "a ll the way down .'' (Recall that there is a case made by Lewis himself for fun ctional analysi!:i of all theoretical terms.) This is nOl a matter that is easily resolved, hut without a resolution that ru les out the " fu nct ional all the way down" option, we cannot re ly on construing mental language in Lewis\ man­ner to rescue us from epiphenomcnalism.

Construing special science terms in Lewis's manner makes epiphe­nomenalism easier to stomach, but epiphenomenulism with cont rolled heartburn is st ill cpiphenomenalism. Perhaps the solut ion is to reject functionalism, but if functionalism is rejected , we must he gin anew to

confront the arguments for epiphenomcnalism in the computer model all over again. functionalism at least has the virtue of disposing of epipht! nomenalism on a c.:ounterfactual unders tanding of causal rele­vance. However, if we prefer the nomological approach to causal re levance, I am doubtful that there is any way of avoiding epiphenome ­nalism.

Notes

There arc: . howc::ver, importa nt dilfc:rcncc:s that I Llon·1 have: the: ~pace: lo go i111 0. Brictly. it is more: plausihlc thal content supervc:nc) un the neural lc:vel than that content :.upc.:rvc:nn on the · ·~yntac:1i c'" leve l. lMu11: ahou1 the n11:a11111g or · ~yntac1i c' here shortly.) A 11 appeal lo supcrvenic:nce of con tent 1>11 lm,.c:r lcvcb can be u~c:d Ill a tt c:mpl to avoid 1he c.:piphcn11mc.:nalism aqwmenl.

2 The: idi.:a d Sec: abo [) hrst appc:a1 cau~al 1n1c: among 1h1.: if ii encou ~ymbob, at meaning~ ;1

itlcntifying nc:urolog1c;

3 This has lo oc:en di alk ~1.:c Kim ( I ' ( l980a) a1g and Blacl-. b

4 Ahernativc: tl.!mpc::ratu1 explode .... i parl will co IO bt U llCOI

the: explo~ i 1 ClllHJ Uc:l i\'11 ~ t.:1rndi1io 11 i ~

5 Sec: Ul11ck I Sc:c: Fod1>r 1

ti Li.:Pore am Coun1c:dd t.: syntacl 1c: ol have: cau~e 1

meanings a )(illlC:C), thl by a dilfc:r1 )Creen off 1

meaning~ ~1 the: syn1ac1i that i ~ ;,ib~u

( ·oun tcrf:tct S)'lllfllClric'u /,

ulht:r, ~II th duh agJ111 s1

Th is iss1H The: Le !'ore par1icula1 c: in my h c: ;.11

behavior . B agrc: i.: 1h;i1 !'­meaning ;.11

Bruc:c: c::vcn Cc:rtainl) 111

all. 1lien 11 · I hid 1111: 1111 WC may ~UJ l:hure · ltkc r a1cd ..:ausa ll

oid: special that <.:an be el scie nci.:s . rn. Seconc.J­tha1 c.:omisc ne another. ..:eu not be as scconc.J­Phy~iologi ­

nies gcner­iicving chac ic.:ks out a •ufi1 ·id1w111d m.Jcr prop­nf physical of physics

.ndamcncal 1osc fu nc.Ja­n." {l~ec;,i ll

1lysis uf all olvcd, but vay down" wis 's man-

es eriphc­con1 rolled s to reject n anew to 1ter rnouc:l ~ posing l) f 1usal rele-to causal phenomc-

to go 11110

::I than that >f . !>Yll I J CI u:' 1 he: U)lt:d Ill

I 6 7

2 The iJe:i d.:~..:11bt•J hc1c wa .;. as l.1r as I J..rww. tirst articu!Jtec..I in r:udur ( 1975. 1980). Sec ;1lso Dc 11nct1 ( 198 1 I in \\h1ch the: terms 'semantic engine ' anJ ·~y ntacti.: engine ' lirst appc: ;ir, .rnJ N..:well ( l'lKll) :111 d l'yl)shyn ( 1984). I splikc of :i w m :latiou between causa l 11Hcract1011s among ~ymh11lic s1ructlm.:s in o ur br;.iins and sc:nsiblc: rc:lations among the: mc:aning) nf the '} mhul structure,. This way l>f speaking c:in be: misk:iJing if 11 encourages the: picture 111 thc neurmc1c:ntist opening the: brain. ju!>I wcing the: symbols. and !hen figuring ou t ' ' h.11 1hq rnc:an. Sinu.: syntactic obic:cts (and pc:rhapi. meanings ilS \\.ell) .ire: func111111:.il enllties (though dilfc:ren1 1ypc:s of funcllonal en111ies). iJcnllfying citha of them 111 1hc hrJ111 will rc:4uirc rnnsic..lc:rablc: app1n:1:.11io11 ll f nc:urological functioning .

J This has long bc:c:n a cc:n tr:il tlogmJ of the \.·o mputalional point 1if vic:w. anJ ha) long hc:c:n challcngc:J hy ~rnnc philmuphc:r). h1r a rgumc:nls against the: claim )latc:d here, )CC: Kim ( 197:! ) RicharJ~un 1197111. [nc ( 1983). anJ Pa1nc1a Churchla11J ( 1'>86). lllod ( 19Hlla) J1gue) ;1ga111)1 K im. l'atrtua Kitcher ( 1980. 19X:!) ;ugues againsl R1chanhl1n, and Ulac.:kburn < f11rthc.:nn11ng l :.irg110 aga1n,1 Enc anJ C:hu1chlanJ.

.t Altcrna11vc:ly. v.c coulJ r:mc: the 1hc:r111al c1mduc1ivi1y of the: roJ by 1;.iising its lcmpc:ratu rc:. in v.hid1 -: a~c p.HI nl thc :1ddi1ional hc:;11 that cause) the bomb to explode will come fn>111 1hc hi:al ~11urcc Iha! rJi)c:s the 1cmpc:ra1urc: of the: roJ, aml pa rt will i:omc lr11m the: lirt.: \JJ lhc incrc:.l)t:d thi:rmal conduct ivity tlf !he r11J. J lake: ii to he um:ontrovcrsial that ri~ing thermal rnnd11c1ivity of the: rod is causally rc:lcvant lo the c:xplm,ion . Furthcr, 1h1~ c.1~c lit, the: n11111ist mudc:I. For in thi) sc tup. rising tht:rmal ..:onJ1u.:1i, ity. ci:1e1u punh1". 1~ -11 t1 i~· 1c111 fur an explo~illn, (:inJ !he: 1<'1<'111 punb11~ i:o11Ji t iu11 is ~ati:-.lieJl.

5 Sec: Uloi:k ( 1986. 1987) l'M 1llher r~·a::.on~ h>r ll.lking fu nct ion;i l rnlc lhc:mic:. se riously. Sec Fudor ( lll87al. ( ·1iap1 c:1 1. and Schi1ft:r ( 1987). Chapter 2. for oppo:.mg .irgumc:111s.

6 LcPore anJ Loc:wc:r ( 1987) g1'c: an 11bjc:u111n tin dkct} to Sosa\ (1984) \'t:r ~ion of the: Countcrfactual Argumi:nl. Pu 11 111g the: mailer in my terms. thc::y agree that if a syntactic objcct wi1h a cc:rta111 meanin!_? produces an o u1pu1. the syntact ic ohjc:ct woulJ have: cau!>c:d the same: output c:ven 1( it had had a ditferent meaning. But they JJd that meanings arc:: lawfully relatcJ 10 hc::havio r in !>Orne: circumstances, so in suc.:h ci rcum­stancc::s. the: mc:aning wou ld ha\ c:: cau~eJ the same: output even if 11 hiiJ hi:c:n carric:d by a c..lilfc:rc:nl syn tactic 1lhjec1 If 1he fi r, 1 countc:rfactual shows that ~yntactic forms screen olf meanings fro m cau)a l rc:leva ncc. then the:: sc::coml show!> the: converse:: , that mc:anings ~c1een olf s1n1ac11c form:-. from causal rc::lcvancc:. Bui 1f both .ire true, both the: syn tactic form and the mcan111g o f a rcp1esc: ntation arc: epiphe nomena!. anJ sinc.:c that is Jb)urd , the obv111u ~ conclu,iun i!> that some thing is wro ng with the f11rm of the Countc:rfac1u:tl Argu111i:111 :\ lco11.l111g 111 Lc Pore anc..I Loc'.1-cr, st:man1ics a11J ~yn tax arc S)'r11111 etm ully rc:bted to !he:: pmJuc.:1i11 n oi behavior; if one is c:piphenomc:n :il. so is the: 1llhcr. ~o the Cuuntcriai: tu<il :\rg111nen1 i:a11 111J1 use the cau~al rc:levancc of ~~ntax as a duh <igains1 thi.: i:ausal n.:kvant:c:: uf sc:m:111tics.

This issuc deserves much more:: di: taikJ lrcatmc:nt . but I c:in only !>umrnarilc briefly. The LePorc:: anJ Loc: wc:r .1rgumc:n1 fa ils to take in to account the Ji!>t1nclill11 hc:1wc:c:n a particu lar nc111 token . and s111111· 111kc::n uf that type:. Suppme I have: a !>y11t<Ktic objec t in my hcaJ. Sally. w11h c:c::rt ain ~yn1ac11c and semantic propc:rtic:s th:it producc::s behavior. Bruce. As I mc:n11oned 111 1he I.i sl paragraph. LePore anJ Loc:wc:r and I agree: I hat Sally w11ulJ have: i:au!>ed Urucc: even if ii had had a different mi:aning or 110 mc:aning at ;,ill. But can v.c make: !he >)'111111erncul clui111 that Sally wuuld have cau~cd Uruc.:c: c:vc::n if it hall had J diffc:rc:nl syn1.1c1ic identity or no syntactic illc:111i1y at all'' Certainly lltll~ If Sally haJ h.iJ J J1llerc:n1 -;yntactic identity o r no sy11tac1ic identity al all. thcn it ~imply \I.Ou lu mil he:: the )}'n l.1llic 11h1ect I o riginall}' pic.:kc:tl o ut .1!> ·Sall} · 11 .ul the 111 ii;111al 111c:a11111g hl·t:11 · , .1 rrn:J .. hy ,1 i.yntactic uhjcct not illc11 1icil to Sally, wc mdy :.uppose that the rnuh ' 'oulll hJvc: be: en a Brucc:-/ikt' be: ha\ ror Uut thh Bruce -like h1: havi11r wou ld 11111 he 1dc:r1 1icd to Bruce - given that cvt:n t ~ Jrt· 111divill11 · atcd ..::.111~.111). a~ a ll gu11J Da, 1d"1111.111s hu..:h a~ l.cPorc and Luc\\erJ hcllcH: Brucc \

l I $

l l f i

'

't ' \ t _.. •' .... • A• ..,; ' ' ' •

N I JI Ill fl! . h:

illc 111i1y r cquiro.:~ hc.-111!! i:;i11,cd by Sall) . l.d'o1r ;111d I <•ewer may led 1ha1 lhc " ·•r 0111 i' lt 1 lalk t1C1I ahnul lhc pr11d11r1i1111 of B1m·c . h11 1 ral ht•t o l Bruce lil..t· t·venh lypc'>. no r rok ens. <l ht· ic is ~C1mc hin l of !hi" 111 lhcir papc1. ) Bui rhc snt•t·11i11J: :11 g11mc111 only works wirh 111ke11s. Whal makes us \\ ;1111 lo say 1h;it lhc sy111 ai:1ir "neens nH the semantic io; thar the ver} ~amc hil o f hchavior would ha\'C hccn prmlurcd even 1f rhe o;vn1;i c.:1ic eve111 1ha1 p rodm:ed ir had h:uJ a dilfen:nr me;ining (See hie I lt>8<1) for a di-;,ecrion of a11111 hc r falla l'y 1ha1 f;1l b afoul of 1hc rypc / tokcn t.J1,1i11ctill11 lo~e ther wirh ca u"al individuation.) l.e l'rn c ant.I Loewe r ;11 t· righ l that the re is a m111111/u1w1il sr111n11•1ry· hc rwce n syntax and 'c rnan tic~. That i~. •here arc circ11m"ta11ces in whit:h ho lh rhe ~emanric and syn l;il' lic p1oper1ic5 .,f ;in c'c111 arl' m1m11l11~ic1 ll~ sulficienl for ;i cer1 ai n rype o f he havior. But what is al issm: hcrt· '' rnun1crfnc1ual \~ mmetr: 1 alht•r than a no n1t1logil'al symmet ~ . ant.I thi~ they have 1101 argue ti Im.

7 Nolt' rh:.it it is rrn part •>I tht· " Language of Thought " view 111 im i' I thal 1h1: fmg h;is a language of !hoiight. 1 he argumen t for a language of though! in human' tlerentls on lht' produclivity a nd srlema! icit v of human thoughl. anti lh i' <lrl!llfllClll u.jll IHll ;1ppfy to creatures who~e thought j~ Ill!( at lc;t ~I as r rml11e1ive ;111d ~.\ ~ t l' lll atic . Sn· the epi logue 111 F<1tlor ( 19K7aJ.

X My point diffe rs Imm lhe usual skept icism about the cau~a l ellicacy pf tl ispo,i1ior1al prnrcrlies in thi ~ rcspccl. and :tl~O in lfl;'I( it i ~ ahotr t :1 catq?Ot)' of proret1ics -srconu·mder properlies - pf which tli~positinnal pmpert ie" a re 11n ly a ' Pet:ial r:1~c

IJ Afrcr writ ing thi~ chapter, I rcau Jack"tin anti f>e !l ll ( l t>H8 l, \\ h il'h cla imed rhat sccontl·nr(kr r roptrl ies ;ire ine ll icaci1111 > hdurc I tl itl . 1 lowever. Jacbon ;iml Pe 11 i1 say lh;i t 'l't:nrH.l ·n1dcr propc rli t·s a rc irrcrt. 1101 just wit h rc, pccl 111 th t• d lecl' in terms of which they a re de fined. hul all e lfects. ·r hey don I give ;iny of rhc· ;i1g11mtnt' used here. thou~h Jaeko;on tells me (in corresponcknce > rhal the merde1crminat1on a rgu· ment is csse nr ia lly consonanr wi1h lht·ir wa~ of looking a1 rhing' They go further. comm ill ing them~elves. I th ink. lo the claim that all " mulriply rt-:dii'.;ihk" prope rt ies are cau~ally irrcrt 1 hus if an increase in rcmrcr aturc of a g:i.-.. in\ltk a scaled glass con!arner 1ec;ul!s 111 !hl' gla's 'ha tlering, the} rai..e rhc fact rhal there :ire many conligurn1ion5 of mnleeult'" thar rnuld 1cali7.c rh is pmcc.-.. .. lo 'hnv. thal 1he increa"e in tcmpernturc did no! t·au' c: thl' sha11t·1 ing. hut rather the c;i u-;e wa~ the specifi c molt:l'Ular i111e ral' li1111s. They tlo nlll note thill the same poinl apr lic:s to the molcrnlar intt: racliorrs. whid1 :.i re lht· m .. e lve~ mull iply n:a li7ahk m te rms of i111 e r acr iun~ a mong e lectrons. prnt1111, , ne 1111 1111 ~. c!c. And 11f u iu rsc. s11n1t· of these p:111 icle.; ;ire perhaps 1cali1.ahlc in difkre nl """Y' h\ st il l 'imalkr parlich:s The up<; hot 11f 1he11 pm i1in11 would he th at lht: nnly !!em1111t· causality inhtrl'' at lhl' lt:\'l'I of l'a ~i c ph} sics. 13111 what if !hc1c j, 110 suc h level" ·1 hi ~ i ~ :1 real phy~irnl po~'ihi l i t y . (St:e Dc hmt: ll. 191-llJ. 1990. ) 11rctr (10.1itiw1 tlu 11111 ·1 1'1111 11 r· tlr1 11 111 111111· /...111111 · /iu 111r1· 11"1c tho tltl'lt' 1111· <Ill}' c <111.111/fr

rffic11no11.1 pro1wrt1c'l 111 all. Ja<.:k,nn tells me llwt 1h1:y a rt' prep:irctl lo loca le all genuine causali1y at rhe hasir physics k "t:I. ~o l1J11).'. a~ one tlisti npuishe' hc1wet: 11 two rypes of causa l rekva nce : lhe nntion u~ed ht: rl' anti 1hc notion of ct u ~;i l rclevann· that corrc ~pomls 10 tht•ir idea ol a ca usal programming c .~pl:ina t ion ·1 ht· iJc;1 nf rau<;al programm ing c;in he: ill11slr ;1tcu hy lllt'ir t•xample tl l an expla na11011 of "hy lreec; grt1w foste r in Melbourne tha n in C:inhe rra : Why? Bernu\t' rhere arc more fro\!<; in Ca nht:rra . 1 he re ht•i ng m11rc fr11i.t' in ( anlwrra ra mall~ prn~1a111' the re he in~ fa~ ! e r ~rowrh in Mclbn11111t' . They hllltl 1ha1 ).t'Co11d ·111dcr prtipcrlics (and 111 hc 1 11111l11pl~ reali1ahle prope r! iec;) Can be C'1ll'<t lly relevant lo 'lllllc:l h ing in !hi' Catl\af r111gram · ming scn~e. but n11t in the sense appropri;11e only 1t 1 ha~rc phy!>in

Ill For tht· pu rpost'" 11f rhi" e:c1111plc . I am i!!norin!! lhc i'-'11t· of whether thl'rt io; anyth ing at the nwlccular lc\l' I 1h;11 uirre~pnnds ver) well 111 the g1:ne . St·e Philip Kitcher ( 19X2).

11 See Hlock (lt>XOa> tor ci mnrt: tletaikd d i~rnsc;inn 111 the dilfe re nc1: hel\\t'Cn rhc liro;t anu !'eC11nd·ortJe1 inlnprrlal HlllS

I ~ Block ( lllXhl. pp. h(iX-9. take' th i' way uul. ;" doe ' fad;,11n and l'c11i1 < l llRH>.

Rinck. Nell. I Ma.;c;,· I la

1980h. " Wh

198(). " /\d\'e eds .. Mui l\ t innCSOl i

1987. 'Tune Soc: i<'f,1'. su

Chu n:hland. P D chmclt. I Ian

sio11 '?" fin 19911. "Exrc·

Scir11cr. 2-Dennt: l I. O ;rn i

ell .. Rcd11< Drtlskc., rtcd .

bridge . 1\1. Enc, Berent. I '

80, 279- 9~

I 98(i. "Ess<:

vcnicnrc. Sr11dics i11 , 40.1- 2<1 .

f-ollor. Jerry, U niversity

I 980 . "I\ kt h nitivc l'syc

I 9H7a. / 's.1·(' /r I 987h. " l\1a ~

Loewn t 1• In press . " M

11 e1 u gc:I a ml. J t 1 I ioral a11d r

I <JK 1. Muul I Ho rwich. Pau I. Jackson. Frank

M111cl , 47 . . Kim. J ac!!"'·t111.

lurnt i1y Tl 1984. "Epipl'

ec.ls .• /lfid11 Minnc~ota

lhc \\:ty 1111t llt' . ly p l''·

'!! ;irE! llnll'll l

cc ns 11lf 1 he eve n 1( the

I l IJHfil for a 11 11 toge t ht: r flt JI/II 1/o~tni/

es i11 which ulfkient for nc;try ra ther

t: frog has a dc re11us on II 1101 a11ply ic Sn· 1 hc

ispo,i l io11;1l rope rl ies -: t:ial case. .1irned that

anti Pc ll it .,., in term'\

rrn:: n I ~ used ali11n argu· go further. propt:rt ics

caleu glass ;.ire many

i nne ase in he sp ecific molecu lar

o ns ;.imong re perhaps ir pn,i ti1111 .. Bur what 'llN. 19<>0.) II}" I Oll.\t11fy

locate all tween two va ncc thal

1 t>f c:.iusal 1 re es grow . rrusts in e ing faster ·r multiply I program-

1 ~ anything 1p Kitche r

1 the lirst -

!8).

Can till' 111i11d c'1"11gt' the 11·orltl? 169

Bibliogra phy

Ollll:k. Neu. I 1JXlla . Rcc1cfin1:s 111 /'l11lo.w11lt1· of P.frc hnlo.1..'1 ·. I ; ,f, I . Cambridge. Mass.: l larvard University Pn: .... -..

l!J80h. " What Is r u11r1irn1alis11i'.1 " In Block (198()).

l 98rl . .. Advertisement for a Semantics l<1 r Psychology." In I'. A French ct a l. , eds .. Mid11 t'.1·1 Stwlic.1 in l'htl11w11h.1'. Vol. X . t-.linm:apnlis: University of !\f inneso la Pres-., 6 15- 7H.

I 1187 ... Fu net ion al Role anu Truth Conditions ... /'111c t'<'d111g' of tlw / fm 1melinJ1 Sout·rr . .... up plcnientary \'Olu me. h i , I :'7-81.

Churchlc111d. P. S .. 198(1. Nl'11mp/11/11wph,1'. Cambr idge. Mass.: MIT Press. . Dchmelt, I Jans. 1989. "Triton . .. . elec tron . ... cosn11111 . ... : An infin ite rcgrcs­

siun'?" flm1 ·t•<'cl111g1 of'"" National / ln11lem,v of Sci<•11n·s. ~6. 8618. 1990. "Expnir11c11ts on the struc ture of an imJiviLlual clc1rn:ntary p;1rticle,"

Scfrna. 247, 5J9-·t'i. Denne tt . Danie l, 1981. " Three Kimi-; of l11ten1ional P-;yc hology." In R . Healy,

ed .. R£'c/11c rion. Time ancl Rcallfv. C rn1hridge: Camoridgc University Press. Dretskc. rrccl. 1988. F.xplt1111111~ /11-ltul'im · Uc·a.wnv"' a ll'orlt! of Ca11.'ic.:1. Cam­

bridge. Mass. : f\ t!T Press. Enc, Berent. 198.1. " In Defence of the Iden tity Theory." Jm1111al uf Philo.w phy.

811. 279- QH. 198(), "Essentia l ism with lnJi, id ual Essences: Causa tion, Kinds. Super­

vcnience, and Restricted ldc111ities." In P. A. French cl a l., eds .. .\fidwest Swdies 111 l'hilowphy_ Vol. X I. tin111.:apolis: University of Minnesota Press, 40J-2h.

Fodor. Jer ry, 1975. The !.a11.1(11a~" of 1'/w11xl11 . Camhri dg<.: . Mass. : I larvard Univl!rs ity Press.

1980. "Mct l10dologic;1' Solipsism Considered HS a Rt:sl!n rch Strategy in Cog­ni1ive Psyrhology.'· 8t·l1111 ·ioml nncl 11mi11 Sc:ienas, .1. 417 -24.

l987a. ' "'-'"·'11111.'11111111ics . CamhridJ!e. Mass.: MIT Prcso; . I 987b ... t-.faking Mind Mati e r More ." This is Fodm's reply to l.cp11rc and

Loewe r ( f ll87). The abstract appeared in Jvunwl of l'li il11sophy. 84. M2. In press. " Making Mind Matte r Mon:." Philo.wphical Top.in.

Haugcla nd . John . 1980. " Programs. Caus;il Powers. and l11! <.:r1tio nali1y." R<'ha/'· ioral anti Bmin Sciena.s, .1. 4J2-3.

1981 . ,\fine/ l ks1xn. Camhridge. Mass.: f\- llT Press. Horwich. Paul. 19~7 . . -tsy11111wmef 111 Timi' . Cambr idge. f\ 1 as-;.: M t:r Pre-;s. Jackson. Frnnk. and Ph ilip Pe ttit. 1988. ·-r-unctiurwli-;m anti Uroau Conten t. ''

,\f111tl. 97 . .187. 38 1- 400. Kim. J aegv.·1111. l <n2. " Phc no111e11.il Pm pi.: rt ics. Ps} rlmphys i('a l Laws. and the

ldrn1i1y Theory." Mo111s1. 56. 177-92. Partially rcprinh.:<l in Ulock I 1980). 1984 ... Epiphc nomcnal ;i nti Sup<.:rvcnirnt Causation." In P. A. French ct al.,

eds., ,\fidhest Sllldic•s in l'/11/moph.r . Vol. IX. Minncapnlis: University of Minnesota Press: '257-70.

t

i

I

t 170 Nl I> lll<>C !-..

Kilcher, Patricia. 1980. " l fow lo Rec.Jun: a Funclillnal Psychology." Pltifosophy

of Science, 47, 134-40. 1982. "Genes. HccJuc1ion ;1ncJ f u11ctio11;.il P~yd10logy. " l'hilow11l11· of Sc1t·11u»

49, 633-6. Kilcher, Philip. 1982. "Gene~." 8n11sh Jounwl for the P/11/u.wphy vf S(lmce. 33.

337-59. LePore, E., ancJ B. Loewer. 1987. "Mind Mailers," Jounwl oj Philv.1Uph.r. 84.

630-4 1. Lewis, David, 1970 . .. How lo Ddln\.'. Tht:on.:tical Terms." Jo11n111/ of I'ltilosoph)',

67, 427- 46. 1972. "Psychophysical and Thcoret ic;.il I dent i ficat ions ... . ·I 1111rn/cHi1111 Jo11mal

of l'hi/osoph.r. 50, 249-58. Reprinted in Block ( 1980). Minsky, Marvi n, 1967. Co111pwurio11 . Englewood Cliffs, N.J .: Prtnticc-l lall. Newe ll , Ahin, 1980. ' 'Physical Symhol Syslcms," Cog11iti1 ·e Science>, 4, 1.15-83. Putnam, llilal)1• 1960. "Minds and Machines." In S. llook. ed .. l>i111n1.1 io11s of

Mind. New York: New York University Pr<.:ss. 1967. "Psychological Prcdica1es." In \\'. H . Capiian and D . D . Merri ll, eds. ,

Art , Mind , and Rl'iigwn. Pittsburgh: University of Pillsburgh Press. Rc1i1led "The Na1urc of Mental S1a1cs" in Pu tnam\ collected pap<.:rs. Reprinted in Block (I 980).

1973. 1975. " Philosophy and Our Mental Life ." Moul, La111:1w1:c · and Reality: Philo.\Opl11rnl Pupers. Vol. 2. Camhridgc: Cambridge Univcrsi1y Press. Hcprirllcd in Block ( 1980), and in somewhat different form. in 1 laugcland (1981). Originally published in Cognition , 2 (1 973), wilh a scclion on IQ thal has hccn omitted from all of the reprin ted vcrsions.

1988. l kprese111a1io11 a11tl Rcaliry. Cambridge. Mass. : MIT Pre~s . Pylyshyn, Zenon, 1984. Co111p111a1io11 and Cogmtio11: !S.W<'S in the Fo111ula1w11s of

Cng11i1i1 ·r Science . Camhridge, Mass.: MIT Press. R id1ardson, Rohcrt , 1979. "J7unctionalism ancJ Rcductioni ~rn , " l'h ilowplty of

SL"i~nce. 46. 533 / 558. Schiffer, Stephen. 1987. Rc11111n11rs of ,\-feunin1: . Cambridge . Mass.: MIT Press. Searle, John , 1980. "Scarlc's Reply LO Critics of· Mind~. Brains and Programs',"

Belial'iorul and Brain Sciences, 3, 450-7. Sosa, Ernest, 1984. ·'Mind-Body lntt:ract ion and Supervenicnt Causa1ion.·· In

P. A. French c t al.. eds .. Midwcs1 Slluiies in I'hilo.w11hy. Vol. IX . Minneapo­lis: U11ivcr:..11y of Minnesula Press. 271-82.

F

Sc.:ic nt ilic.: re

istic rrmluc causal ~I rue ac.loplion o (fo_: 011 ii fl LJ I

causal ~11 uc the ca~c lo appare111 ly , involve 1 ek turcs; tilt.: c• tu he thCOl) J lJ8Ha, I lJ8'

Putnam l 9i that appeal

A variL'l )' 1

thcory-dt:p• i ~tic and p

. . variou~ sci edge liu t conccpb a linguist re c lhl!Ol) -illd l