DERK PEREBOOM
WHY A SCIENTIFIC
REALIST
CANNOT
BE A FUNCTIONALIST
• . . there is no denomination so extrinsic that it does not have an intrinsic denomination as its basis. Leibniz to De Volder, April 1702
ABSTRACT. According to functionalism, mental state types consist solely in relations to inputs, outputs, and other mental states. I argue that two central claims of a prominent and plausible type of scientific realism conflict with the functionalist position. These claims are that natural kinds in a mature science are not reducible to natural kinds in any other, and that all dispositional features of natural kinds can be explained at the type-level. These claims, when applied to psychology, have the consequence that at least some mental state types consist not merely in relations to inputs, outputs, and other mental states, but also in nonrelational properties that play a role in explaining functional relations. Consequently, a scientific realist of the sort I describe must reject functionalism.
1.
INTRODUCTION
During the sixties, functionalism about the mental arose as a response to problems with behaviorism and other contending philosophies of mind. Although functionalism retains the behaviorists' relational or dispositional analysis of mental states, it allegedly avoids the problems of behaviorism by specifying that mental state types consist not only in relations to sensory inputs and behavioral outputs but also in relations to other mental states. Yet this revision does not address one of the deepest criticisms of behaviorism, advanced by Hilary Putnam in 'Brains and Behavior' (1975b). In this classic scientific realist essay, Putnam argues that we should identify mental states in the same way as we identify natural kinds in other sciences. In the case of polio, for example, there are explanations for the dispositional features (symptoms) of this disease, and we identify the disease with the underlying properties that provide this explanation. 1 Behaviorists, by contrast, identify natural kinds in psychology, like mental states, with dispositions to behave in certain ways, and not with underlying properties that explain these dispositions. I shall argue that since functionalism also Synthese 88: 341-358, 1991. © 1991 Kluwer Academic Publishers. Printed in the Netherlands.
342
DERK
PEREBOOM
identifies mental states with dispositions, it should be rejected in favor of a theory that comes closer to meeting Putnam's original specifications. My argument proceeds from the type of scientific realism (SR for short) developed over the past three decades by Putnam, Richard Boyd, and Jerry Fodor, among others. Functionalism, I believe, conflicts with the following two theses of this sort of scientific realism: (1)
No mature science is reducible to any other; in particular, natural kinds in one such science are not reducible to natural kinds in another.
(2)
For all dispositional features of natural kinds there are properties that explain these features at the type-level (Boyd 1980a, 1980b, 1984; Fodor 1975, 1980; Putnam 1975a1975e).
These two claims, when applied to psychology, have the consequence that at least some mental state types consist not merely in relations to inputs, outputs, and other mental states, but also in nonrelational properties that play a role in explaining functional relations. It follows that scientific realism, as formulated by Putnam, Boyd, and Fodor, is inconsistent with functionalism, which is also held by each of these philosophers. 2 In view of this inconsistency, I shall argue that functionalism, rather than SR, should be rejected. 2.
FUNCTIONALISM
AND
EXTRINSIC
PROPERTIES
Functionalism is a metaphysical theory about the nature of mental state types. It is the view that mental state types consist solely in causal relations to perceptual inputs, behavioral outputs, and other mental states (Block 1980). Alternatively, functionalism identifies mental state types with type-level dispositions to cause behavioral outputs and mental states, dispositions which have mental states and perceptual inputs as causes. The thesis that mental states are relational is roughly equivalent to the thesis that they are dispositional. To use an oversimplified example, on the relational characterization, pain is that state which is caused by pinpricks and scrapes, and causes the thought "That hurt" and the utterance "Ouch." Stated in terms of dispositions, pain is a
A SCIENTIFIC
REALIST
CANNOT
BE A FUNCTIONALIST
343
disposition to think "that hurt" and to say " o u c h , " which is caused by pinpricks and scrapes. According to functionalism, then, all of the psychological properties that constitute mental state types are relational. I shall claim that according to SR, some properties of mental state types must be nonrelational or intrinsic. Properties are relational (i.e. extrinsic) and intrinsic relative to things that possess them: F is an extrinsic property of x just in case F is a relation x has to a thing wholly distinct from x. F is an intrinsic property of x just in case F is not an extrinsic property of x. 3 The earth's being 93 million miles from the sun is an extrinsic property of the earth; a pain's causing one to say " o u c h " is an extrinsic property of the pain. A mousetrap's having a certain mechanism is an intrinsic property of the mousetrap; a diamond's having a certain chemical structure is an intrinsic property of the diamond. Furthermore, properties (extrinsic or intrinsic) may be had by kinds or types of things: A kind K has (or is partly constituted by) property F just in case all possible token instantiations of K have F. The kind human being has the intrinsic property of having a certain genetic make-up just in case all possible token human beings have this property. The kind pain has the extrinsic property of causing thoughts about pain just in case all possible token pains have this property. Whether the properties of an individual or a kind are intrinsic or extrinsic is conceptually distinct from whether they are essential to the individual or the kind. Both intrinsic and extrinsic properties, for example, can be essential to a kind; having the chemical structure of H 2 0 is an intrinsic essential property of water, and if functionalists are right, then mental states have essential relational or extrinsic properties. 4 I mention this because there is a potential source of confusion here: 'intrinsic' and 'essential' are sometimes used as synonyms. Whenever I use the term 'intrinsic', however, I shall use it as I have defined it above. Functionalism's motivation for attributing only extrinsic properties to mental state types is evident from two of the most important arguments
344
DERK
PEREBOOM
advanced in its favor. The first is tied to the reason that philosophers were attracted to logical behaviorism. Shoemaker (1984) asks us to imagine that pain is definable in terms of its causal relations to inputs, outputs, and certain beliefs and desires, but that those beliefs and desires are not definable in this way. Pain, he says, would then be a functional property, but only in an attenuated sense, because what functionalists have meant in claiming that it is a functional property would not be true: What they have meant is that it is expressible by a functional predicate which contains no mental predicates whatever, and not merely that it is expressibleby one that contains no predicate which is necessarily coextensive with 'is in pain'; and if the other mental properties by relation to which it is defined include beliefs and desires, this requires that those properties likewise be expressibleby such functional predicates. (Shoemaker 1984, p. 263) According to Shoemaker, functionalism shares with behaviorism the reductionist goal of eliminating mental predicates from scientific characterizations of mental states. But if some mental state types have intrinsic psychological properties, then a genuine reduction could not be achieved in the way the functionalist proposes. If, in addition to the relations, there were intrinsic psychological properties underlying and explaining the relations, then the elimination of mental predicates from functional characterizations would not amount to a reduction with any deep significance; for then only the functional relations, and not all of the psychological properties, would have been made respectable from a naturalistic point of view. A second argument for accepting functionalism hinges on the multiple realizability of mental states, and this argument also indicates that the functionalist attributes only extrinsic properties to mental state types. The octopus and the human being, he argues, can be in the same mental state. Yet their brains and bodies are not similarly constituted. How can this be? The functionalist's solution is to raise psychological description to a level that abstracts completely from any facts about constitution whatsoever, and he does this by restricting psychological descriptions to functional relations alone, prescinding completely from anything, afortiori any intrinsic psychological property, that underlies the functional relations. His view may not be that nothing underlies the functional relations, but rather that if something does, it must be characterized at a nonpsychological level of description. Thus, the
A SCIENTIFIC
REALIST
CANNOT
BE A FUNCTIONALIST
345
functionalist's response to his own criticism of type-type identity theory entails that all properties constituting mental state types are extrinsic. One must therefore also understand the dispositional characterization of functionalism as attributing only extrinsic properties to mental state types. The kinds of dispositions with which the functionalist identifies mental states are extrinsic properties of mental states. In what follows, I shall use the term 'disposition' accordingly. 3.
SCIENTIFIC
REALISM
AND
INTRINSIC
PSYCHOLOGICAL
PROPERTIES
I shall now argue that SR requires that dispositions of psychological natural kinds be explained by intrinsic psychological properties and that, hence, SR is inconsistent with functionalism. In this argument, I assume that psychology is, or at least can be, a mature science and that mental states are among the psychological natural kinds. The argument begins with SR's anti-reducfionist claim: (1)
No mature science is reducible to any other; in particular, natural kinds in one such science are not reducible to natural kinds in another.
In 'Special Sciences', Fodor (1980) provides an argument for this view. Consider a psychological law: PlX
causes
P2x
where P1 and P2 are kind predicates in psychology. The standard model of reduction, applied to psychology and neurophysiology, requires that every kind term that appears in this psychological law be correlated with a kind term in neurophysiology. The kind terms are correlated by means of bridge principles. They either translate kind predicates in one science into those of a more basic one, or identify the kinds themselves of one science with those of a more basic one. But kinds in psychology are multiply realizable in an indefinite number of ways at the neurophysiological level. Consequently, bridge principles for relating psychological kinds or kind predicates to neurophysiotogical ones will involve indefinitely long disjunctions, like: P l x = N l x v N 2 x v . . . N,~x,
where n is indefinitely large.
346
DERK
PEREBOOM
Since an indefinitely long disjunction of kinds in neurophysiology is not a natural neurophysiological kind, psychological kinds cannot be reduced to neurophysiological kinds. According to Fodor, such indefinitely long disjunctions are not natural kinds because they cannot appear in laws. They cannot appear in laws because "laws" involving such disjunctions are not explanatory. Such "laws" are not explanatory because they do not meet our interests in explanation, and, more deeply, because they leave behind the kinds and properties required to make laws genuinely explanatory. For example, genes are information-bearing entities. If one were to try to reduce the laws of classical genetics to chemical laws, one would abstract from the information-theoretic features of genes and focus instead on their many possible constitutions. In the process, the force of the generalizations is lost and one is left without the explanation one originally possessed (Kitcher 1984). Given this kind of anti-reductionism about the relation between psychology and more basic sciences, one must deny that psychological dispositions can be explained at a level more basic than the psychological. Consider the attempt to explain types of psychological dispositions in terms of neurophysiological properties. Since mental states are multiply realizable at the neurophysiological level, their type-level dispositions can only be "explained" in virtue of indefinitely many types of neurophysiological properties. The "explanation" of a type of psychological disposition must appeal to an indefinitely long disjunction of neurophysiological state types. But according to SR, this does not amount to a genuine explanation. This argument applies to any attempt to explain types of psychological dispositions by properties in any science more basic than psychology. Hence, since thesis (2) of SR does demand genuine explanations for the dispositions of natural kinds at the type-level, there must be psychological properties of mental state types that explain types of psychological dispositions. One might object to this argument by denying that scientific realism requires such type-level explanations of the dispositions of natural kinds. Explanations of such dispositions on a token-by-token basis will do. But SR rejects this position for two related reasons. First, in the Humean and positivist anti-realist view, natural kinds are characterized in terms of observable features and dispositions, just in virtue of human conventions and interests, and perhaps more fundamentally, in virtue of human mental structure. Boyd argues that this view of natural kinds
A S C I E N T I F I C R E A L I S T C A N N O T BE A F U N C T I O N A L I S T
347
does not explain successful inductions; it does not account for the projectibility of natural kinds. Rather, there must be underlying causal properties that explain these observable features and dispositions of kinds and, hence, explain why inductions over these kinds are successful. These causal properties must explain the dispositions of kinds at the type-level, for it is types, not tokens, that are projectible. In other mature sciences, dispositions of natural kinds are explained by underlying properties at the type-level. In biology, the dispositions of a species are explained by way of its species-wide genetic structure. In chemistry, the dispositional features of compounds like H20 are explained at the type-level by their structural properties. As Boyd argues: Kinds characterized by "explanatory essences" are also kinds from the point of view of inductive generalization; indeed, in mature sciences, kinds which are explicitly characterized in terms of explanatory essences are the overwhelmingly typical cases of inductively natural kinds. Kinds natural from the point of view of successful induction need not always be explanatorily natural kinds, but they must correspond in relevant respects to the (perhaps unobservable) properties and mechanisms which causally determine the observable properties of the subjects of empirical generalizations. (Boyd 1980b, p. 30; cf. Putnam 1975e)
The dispositional features of all natural kinds are explained by underlying, causally determining properties, whether or not these kinds are explicitly characterized by such underlying properties. Further evidence that Boyd's view of natural kinds is correct is provided by actual scientific practice. When scientists discover that there is more than one type of property underlying the dispositional features of a purported natural kind, the boundaries of the kind are typically readjusted. For example, in 1869, the term 'neurasthenia' was introduced to designate a nervous disease which results in severe fatigue. The term was established world-wide, but "like most descriptive terms, where basic organic or psychological understanding was lacking, it tended to be overinclusive and a receptacle for many diverse conditions" (Diamond 1977). Several different underlying causes were discovered for the dispositional features of this purported natural kind and, hence, the term 'neurasthenia' became obsolete by about 1930 (Kornblith 1979, pp. 120-23; Bing 1915; Chatel and Peele 1970). This case also illuminates the second reason for explanations of dispositions of natural kinds at the type-level. When cures for neurasthenia were sought, researchers discovered that several different causes had
348
DERK PEREBOOM
to be treated. Although some fairly superficial causal, inductive generalizations could be made about neurasthenia, such as "neurasthenia, whatever it is, causes fatigue," deeper generalizations require a different way of cutting up the world. The moral of this story is that when we classify kinds in ways that correspond to underlying causes of dispositions, richer and more fundamental inductive generalizations are made possible. According to Boyd: knowledge of the determining microstructural properties of matter is required for successful inductive generalizations about the sensible properties of matter, if such generalizations are to have the scope and precision typical of mature science. (Boyd 1980b, p. 30)
Classifying kinds merely according to dispositions, rather than causes that explain these dispositions, keeps a science from formulating these deeper generalizations and, hence, blocks its progress to maturity. Thus, if natural kinds are to function in a mature science, there must be properties that explain their dispositions at the type-level and these kinds must be classified correspondingly. Consequently, functionalism indeed seems to be inconsistent with SR. According to functionalism, mental state types are pure dispositions or relations, and have no intrinsic psychological properties. Thesis (2) of SR requires that there be properties that explain the dispositional features of these mental state types. Functionalists would argue that these properties are neurophysiological, but this view is ruled out by the reasons the realist has for accepting thesis (1) of SR, anti-reductionism. A wild disjunction of neurophysiological properties cannot explain the projectibility of psychological kinds and taking this route would preclude the deep generalizations characteristic of a mature science. Hence, the properties that underlie and explain psychological dispositions must themselves be psychological, and functionalism, it would seem, rules out such properties. 5 4. CAN F U N C T I O N A L R E L A T I O N S E X P L A I N F U N C T I O N A L R E L A T I O N S .9
In order to firmly establish the inconsistency of functionalism and SR, an important objection remains to be answered. The functionalist who also embraces SR might argue that his theory does allow for psychological properties that provide the explanations SR requires because functional properties can do the job. Fodor (1975) and Putnam (1975b,
A S C I E N T I F I C R E A L I S T CANNOT BE A F U N C T I O N A L I S T
349
1975d) hold that beliefs and desires are crucial to explanations of behavior and relations to other mental states, but, at the same time, they maintain that the nature of these beliefs and desires is functionally characterizable. Putnam writes, for instance: My own view is that psychological predicates correspond to functional properties of human beings and other sentient beings. The presence of these properties explains the clustering of what some have called the 'symptoms' and 'criteria' of the various psychological states and conditions. (1975d, p. 278)
This amounts to the thesis that dispositions of types of mental states can be explained by dispositions of the same or other mental states. One problem with this view is that given a finite number of mental states, there will be at least one dispositional property left without an explanation, and this violates thesis (2) of SR. The functionalist might suggest that there could be a (virtuous) circle of explanations among dispositional properties of mental states, or else that explanations of psychological dispositional properties by psychological dispositional properties could be holistic. However, no one has ever provided a model as to how such suggestions would work. More fundamentally, what would it be for a dispositional property at a given level to explain another dispositional property at that level? Take, for instance, the dispositions of pain, screaming "Ouch" and thinking "That hurt." Is there some psychological disposition that could play a role in explaining these properties? Certainly other dispositions of pain cannot provide the kind of explanation the realist seeks. If the functionalist wants to take this route, it would be incumbent on him to show how such an explanation would proceed. I doubt that it could be done. Our model from other sciences has non-dispositional properties of kinds at a given level explaining the dispositional properties of kinds at that level. These other sciences have shown us how such explanations work. Psychology is better off adhering to the model of other successful sciences. The functionalist, however, might extend the argument by questioning whether the underlying genetic and molecular properties that explain dispositions of natural kinds in biology and chemistry are really non-dispositional properties of those kinds. These apparently intrinsic properties, he might suggest, are dispositional because there are thoroughly functional descriptions that provide necessary and sufficient conditions for natural kinds in these sciences. The apparently intrinsic
350
DERK PEREBOOM
properties of mentai state types are actually dispositional properties for the same reason. But even if a description of the complete causal role of water, a description that lists only dispositional properties, supplies necessary and sufficient conditions for this kind, it does not follow that the nature of water is exhausted by dispositional properties. Having the underlying structure of H 2 0 molecules might nevertheless be an irreducibly intrinsic property of water. This is because the kind that fits the description might still have properties besides those explicitly listed in the description. Although the purely functional description may pick out a natural kind, it might well leave out the intrinsic properties of the natural kind that explain its dispositions. 6 In summary, SR is committed to providing explanations for dispositions of types of mental states, but because of its adherence to antireductionism, it can appeal only to psychological properties as resources for this explanation. Yet, the only psychological properties functionalism admits are dispositional, and dispositional properties cannot do the job. The only possible candidates are intrinsic psychological properties of mental states. Since functionalism eschews such properties, it is inconsistent with SR. Hence, either SR or functionalism, or both, must be rejected. 5.
WHY
REJECT
FUNCTIONALISM. 9
Given the inconsistency between SR and functionalism, I believe that we should reject functionalism rather than SR. We have already canvassed the reasons for accepting SR, and they are impressive. To show that rejecting functionalism is the preferable option, I shall argue that the reasons for accepting it can be accommodated by the view that mental states have intrinsic psychological properties that explain their dispositions. Let us consider the following three reasons for accepting functionalism: first, it accommodates the view that mental states are at least partially relational; second, it allows for the multiple realizability of mental states; and third, it keeps psychology respectable from the naturalistic point of view. First, is a view which posits intrinsic properties of mental state types consistent with what seems true, that mental states are indeed importantly relational? There are plausibly two relational aspects to mental states and their contents. First, functionalists have thought that part of
A SCIENTIFIC
REALIST
CANNOT
BE A FUNCTIONALIST
351
what it is to be a mental state of a particular type is to stand in certain causal relations to inputs, outputs, and other mental states. It may well be, for example, that a state could not count as a belief having the content that it is raining unless it is coinstantiated with some other mental states. Second, according to the view developed by Putnam (1975e) and Burge (1978, 1982), beliefs are dependent for their content on facts about the social and physical environment. Burge argues that the contents of many types of belief are individuated partly with reference to facts about the individual's social and physical environment. The belief contents of an individual, holding her complete individualistic nonintentional description fixed, can vary if her social or physical environment varies. Given Burge's point, having a belief with a certain type of content is a relational or extrinsic property of an individual. It follows that the belief-type is partly constituted by extrinsic properties. If having a belief is an extrinsic property of an individual because the belief consists partly in relations to features of the social and physical environment, then the belief-type also has extrinsic properties, since relations to these features of the environment can be attributed to all tokens of the belief. The reply to these objections is that mental states can have both extrinsic and intrinsic properties. The belief that it is raining can have all of the extrinsic properties the functionalist attributes to it and have intrinsic properties as well. On the issue that Burge raises, suppose Madeline is counterfactually shifted from one linguistic and physical environment to another of a radically different type, while all of her non-intentionally and individualistically described physical, behavioral, dispositional, and phenomenal histories remain the same. In such circumstances it is reasonable to suppose that Madeline's belief contents change. But this thought experiment fails to show that nothing in her belief contents remains the same, so it does not show that every aspect of having the belief with that particular content is an extrinsic property of her. Thus, Burge's point cannot be directed against the view that mental state types have intrinsic properties (nor does Burge suggest that it can). 7 Consequently, the existence of intrinsic psychological properties is compatible with the two ways in which mental states are plausibly relational. These two types of relational features fail to provide a reason to prefer functionalism over SR's position on mental states.
352
DERK
PEREBOOM
The second reason to accept functionalism is that it allows for the multiple realizability of mental states. But multiple realizability, by itself, does not entail that mental states are functional kinds. In general, natural kinds with intrinsic properties can be multiply realizable. Consider the following possible, if not actual, state of affairs (Kitcher 1984). Two animals are typically classified as members of the same species if they are capable of interbreeding. What underlies these functional relations is a single type of genetic structure, an intrinsic biological property of a species. But because different molecular compositions can have the same properties relative to biological explanation, the genetic structure of the animals may have different molecular constitutions, s So although the species is multiply realizable at the molecular level, an intrinsic feature of its nature, its genetic structure, plays a part in the explanation of its functional properties. Mental states may be natural kinds of a similar sort. Mental states that have intrinsic psychological properties may vary in neurophysiological constitution from token to token, while at the same time these intrinsic properties play a role in explaining the functional characteristics of the mental state type. This is possible because token mental states with different neurophysiological constitutions may have the same intrinsic properties relative to psychological explanation. SR requires that something like the following is the case: psychological state P1 is actually realized by token neurophysiological states of various types, N1. •. Nn. At no level of neurophysiological description are these tokens of the same neurophysiological type. But certain aspects of each of these token neurophysiological states are relevant to psychological explanation and these aspects are the same for each of the tokens of types N1. • • N,. Not all of these aspects are extrinsic properties, as the functionalist would have it, but among them there is at least one intrinsic psychological property of P1. This scenario is compatible with the multiple realizability of mental states, since there is no reason to believe that mental state types that have intrinsic psychological properties are not multiply realizable and, moreover, other mature sciences provide models for the multiple realizability of natural kinds that have intrinsic properties. Thus, the multiple realizability of mental states also provides no reason to prefer functionalism to the position on mental states that results from SR. T h e third reason for accepting functionalism is that it is easily integrated into a naturalistic picture of the world. Is SR's view of mental
A SCIENTIFIC
REALIST
CANNOT
BE A FUNCTIONALIST
353
states as amenable to naturalism as is functionalism? As we noted, Shoemaker (1984) points out that (one type of) functionalism shares with behaviorism the reductionist strategy of eliminating mental predicates from characterizations of mental states. If mental state types have intrinsic psychological properties, then this goal cannot be achieved in the way that these functionalists have hoped. Suppose the functionalist manages to characterize all of the extrinsic properties of mental states in non-mental terms. He has then not yet given an account of the mental state's intrinsic psychological properties. Since these properties cannot be recast as mere relations, this functionalist's reductionist strategy will not succeed. Hence, if this reductionist strategy is required by naturalism, we have a reason to prefer functionalism to the realist alternative. But Shoemaker's strategy is an unattractive way to secure a naturalistic psychology. There are two motivations for eliminating mentalistic predicates in the way he proposes. The first derives from a worry about the reference of mentalistic terms. Like mathematical terms, which seem to refer to strange Platonic entities, mentalistic terms arouse suspicion because they appear to refer to the strange immaterial entities of the dualist or the idealist. But this worry about mentalistic terms can be dispelled without eliminating mentalistic terms. As Boyd (1980a) and Fodor (1980) argue, a materialist need not be a reductionist, either linguistic or metaphysical. One may deny that any mature science linguistically reduces to a more basic science and still affirm that the entities referred to in each of these sciences are wholly constituted by entities over which physics quantifies. The second motivation for eliminating mentalistic terminology is epistemological and has its roots in logical positivism. Sentences containing only "physical" predicates are empirically verifiable, whereas according to the positivists' own view it is not immediately obvious that those with mental predicates possess this distinction. "Physicalist" sentences are verifiable from the empirical standpoint of the third-person observer; they can be empirically verified by anyone. Mentalistic sentences initially appear to be verifiable only from the first-person point of view. First-person verification is less than ideal; it is not the successful type of verification of the natural sciences. If one could show that mentalistic sentences are translatable into "physicalist" sentences, or if one could just show that mentalistic predicates have a "physicalist" account, then mentalistic sentences would be ideally verifiable after all.
354
DERK
PEREBOOM
Even if aspiration to this positivist goal is laudable, the prospects of functionalism reaching it are slim, since it seems impossible to specify inputs and outputs independently of mentalistic terminology. No one has been able to suggest a way of describing either sensation or behavior non-intentionally. Thus the possibility of linguistic reduction lends no advantage to functionalism over the view that mental states have intrinsic psychological properties. As I have already indicated, Boyd and Fodor argue that functionalism accommodates naturalism in a different way. They claim that although psychology is neither linguistically nor ontologically reducible to a more basic science, the functional system of psychological states is actually completely realized in a physical medium. Equivalently, all token functional states are wholly constituted of entities over which physics quantifies. But the complete physical realizability of functional states also provides no advantage for functionalism over the realist alternative. The view that some mental states have intrinsic psychological properties is no less amenable to this naturalist strategy, since the actual constitutions of these states are just as plausibly physical. We have already encountered a model for the physical realizability of mental states that have intrinsic psychological properties. There may be properties that all of the various neurophysiological realizations of pain have in common, and these properties are not just relations to inputs, outputs, and other mental states but rather properties that underlie and explain these relations. Such intrinsic properties are as readily physically realized as functional properties of mental states. Thus, all of the reasons for accepting functionalism can be accommodated by the view that some mental states have intrinsic psychological properties. Consequently, in view of the inconsistency of functionalism with SR, the attractiveness of SR, and no loss being incurred by trading functionalism for the realist alternative, functionalism should be rejected. 6.
CONCLUSION
The view that some mental state types have intrinsic psychological properties is well suited for dealing with possibilities that have presented a problem for functionalism. Consider David Lewis' (1980) case of mad pain. Lewis thinks that it is possible that a madman feels pain, but that his pain exhibits virtually none of the functional relations that pains
A SCIENTIFIC
REALIST CANNOT BE A FUNCTIONALIST
355
typically have: "He feels pain but his pain does not at all occupy the typical causal role of pain" (Lewis 1980, p. 216). According to SR, besides functional relations, there is another element essential to some mental state types, intrinsic psychological properties that play a role in explaining functional relations. Perhaps to be in pain in the full sense one must instantiate a significant number of the typical functional relations of pain. Here the madman falls short. But he may yet possess a state with the same intrinsic psychological properties that our pain has. Perhaps they are intrinsic properties that appear to us in introspection as the feeling of pain. The existence of such properties would explain how the madman can be in pain in some sense. Thus, by contrast with functionalism, the view that the mental state type pain has intrinsic psychological properties can explain what Lewis acknowledges to be possible: "I needn't mind conceding that perhaps the madman is not in pain in quite the same sense that the rest of us are, but there had better be some straightforward sense in which he and we are both in pain" (Lewis 1980, p. 216). Consequently, this realist position on mental states has an additional, and I believe important, virtue. It can explain the pervasive intuition that there is more to some mental states than their functional relations, since it allows for an intrinsic psychological property that we and the madman have in common. But the deepest advantage of this view is that it accommodates two theses of a powerful scientific realism: that no mature science reduces to any other; and that there are properties that explain the dispositions of natural kinds. Functionalism is incompatible with this type of realism. Although many would like to embrace both positions, they cannot do so consistently. They face a choice between two options: one functionalist, the other realist. The functionalist alternative forfeits many of the advantages of the realism, whereas the realist option can retain everything of value in functionalism. 9 NOTES 1 According to one realist notion, dispositions are identified with the underlying properties that explain tendencies to produce certain outputs. In this paper, I do not employ the term 'disposition' in this sense but in the anti-realist sense in which the behaviorists use it. a P u t n a m was the main motivating force behind both SR and functionalism, although he no longer holds both positions.
356
DERK PEREBOOM
3 One might try modal definitions of these notions. The following are revisions of David Lewis' (1983) definitions: (i)
(ii)
F is an intrinsic property of x just in case x's having F does not logically entail or rule out there being a contingent thing wholly distinct from x (meaning that it and x have no part in common), F is an extrinsic property of x just in case x's having F logically entails or rules out there being a contingent thing wholly distinct from x.
(i) Without the restriction "wholly distinct from x," being partly constituted of hydrogen would turn out to be extrinsic to a water molecule. (ii) Without the limitation to contingent things, if there are necessary beings there would be no intrinsic properties, since the existence of a necessary being is entailed by everything. Robert Adams has suggested the following problem for these modal characterizations. If Kripke is right, having the parents a person actually has is essential to that person's existence. So a person's existence entails the existence of her parents. Hence, a person's existence, on these definitions, will turn out to be an extrinsic property of that person, which seems absurd. I believe that this will be a problem for any modal definitions of these notions. Perhaps the only alternative is to take them as primitive, as I do in this paper. 4 Although, as I have defined the notion of a property of a kind, all properties of a kind are essential to it. Individuals clearly can have both intrinsic and extrinsic properties. s I do not claim that SR assumes that there must be intrinsic psychological properties of mental state types that explain their dispositions, or that it assumes the more general thesis that for any kind in science S there are intrinsic properties in S that explain the dispositional features of that kind. Rather, I argue that these theses can be derived from both the theses (1) and (2) of SR and the reasons scientific realists have for endorsing theses (1) and (2). 6 The relational description the first son of Queen Elizabeth II and Prince Philip in the actual world is a necessary and sufficient condition for being Prince Charles, but it does not follow that all of his (essential) properties, like his genetic structure, are also relational. 7 Dennett (1987) believes that Bnrge's thought experiments show that no belief contents are intrinsic psychological properties of individuals and, furthermore, that similar thought experiments can show that there are no intrinsic psychological properties of individuals at all. It would seem to follow from this that no mental state types have intrinsic psychological properties. But even if Dennett is right, the view that there are intrinsic psychological features of mental state types isn't undermined, for extrinsic properties may have intrinsic aspects. The extrinsic property, Adam's being tall, contains the intrinsic aspect of Adam's having some height or other. Thus, even if all of the properties of a mental state type are extrinsic in the way Dennett suggests, it is yet possible that its nature consists partly in intrinsic features. 8 The thesis of Philip Kitcher's article (1984) is that classical genetics does not reduce to molecular biology. The following passage illustrates the general point: with regard to the limb-bud allele, the abnormal morphology in the limb is a result of abnormal tissue geometry. About this Kitcher writes:
A S C I E N T I F I C R E A L I S T C A N N O T BE A F U N C T I O N A L I S T
357
Reductionists may point out, quite correctly, that there is some very complex molecular description of the entire situation. The tissue geometry is, after all, a configuration of molecules. But this point is no more relevant than the comparable claim about the process of meiotic division in which the alleles are distributed to gametes. Certain genes are not expressed because of the geometrical properties of the cells in the tissue: the pertinent cells are too far apart. However this is realized at the molecular level, our explanation must bring out the salient fact that it is a presence of a gap between cells that are normally adjacent that explains the nonexpression of the genes. As in the example of allele transmission at meiosis, we lose sight of the important connections by attempting to treat the situation from the molecular point of view. As before, the point can be sharpened by considering situations in which radically different molecular configurations realize the crucial feature of the tissue geometry: situations in which heterogeneous molecular structures realize the breakdown of communication between the cells. (p. 372) 9 I wish to thank Robert Adams, Lynn Rudder Baker, David Braun, David Christensen, Hilary Kornblith, Arthur Kuflik, William E. Mann, and George Sher for helpful comments on this paper.
REFERENCES Bing, Robert: 1915, Nervous Diseases, Rebman, New York. Block, Ned: 1980, 'Troubles with Functionalism', in Ned Block (ed.), Readings in the Philosophy of Psychology, Vol. 1, Harvard University Press, Cambridge, Massachusetts, pp. 268-305. Boyd, Richard: 1980a, 'Materialism Without Reductionism: What Physicalism Does Not Entail', in Ned Block (ed.), Readings in the Philosophy of Psychology, Vol. 1, Harvard University Press, Cambridge, Massachusetts, pp. 67-106. Boyd, Richard: 1980b, 'Scientific Realism and Naturalistic Epistemology', in Proceedings of the Philosophy of Science Association 1980, Vol. 2. Boyd, Richard: 1984, 'The Current Status of Scientific Realism', in Jarrett Leplin (ed.), Scientific Realism, University of California Press, Berkeley and Los Angeles. Burge, Tyler: 1978, 'Individualism and the Mental', in Midwest Studies 1978, University of Minnesota Press, Minneapolis, pp. 73-121. Burge, Tyler: 1982, 'Other Bodies', in Andrew Woodfield (ed.), Thought and Object, Oxford University Press, Oxford. Chatel, J. C. and R. Peele: 1970, 'The Concept of Neurasthenia', International Journal of Psychiatry 9, 3649. Dennett, Daniel: 1987, 'Evolution, Error, and Intentionality', in The Intentional Stance, MIT Press, Cambridge, Massachusetts. Diamond, Leon: 1977, 'Neurasthenia', in Benjamin Wolman (ed.), International Encyclopedia of Psychiatry, Psychology, Psychoanalysis and Neurology, Vol. 8, Aesculapius, New York, pp. 27-28. Fodor, Jerry: 1975, The Language of Thought, Harvard University Press, Cambridge, Massachusetts. Fodor, Jerry: 1980, 'Special Sciences,' in Ned Block (ed.), Readings in the Philosophy
358
DERK
PEREBOOM
of Psychology, Vol. 1, Harvard University Press, Cambridge, Massachusetts, pp. 12033. Kitcher, Philip: 1984, '1953 and All That. A Tale of Two Sciences', The Philosophical Review XCIII (July 1984), 335-73. Kornblith, Hilary: 1979, Knowledge Without Foundations: A Causal Theory, unpublished Ph.D, dissertation, Cornell University, Ithaca, New York. Lewis, David: 1980, 'Mad Pain and Martian Pain', in Ned Block (ed.), Readings in the Philosophy of Psychology, Vol. 1, Harvard University Press, Cambridge, Massachusetts, pp. 216-22. Lewis, David: 1983, 'Extrinsic Properties', Philosophical Studies 44, 197-200. Putnam, Hilary: 1975a, 'Dreaming and Depth Grammar', in Hilary Putnam (ed.), Philosophical Papers, Vol. 2, Cambridge University Press, Cambridge, pp. 304-24. Putnam, Hilary: 1975b, 'Brains and Behavior', in Philosophical Papers, Vol. 2, pp. 32541. Putnam, Hilary: 1975c, 'The Nature of Mental States', in Philosophical Papers, Vol. 2, pp. 429-40. Putnam, Hilary: 1975d, 'Language and Reality', in Philosophical Papers, Vol. 2, pp. 272-90. Putnam, Hilary: 1975e, 'The Meaning of 'Meaning", in Philosophical Papers, Vol. 2, pp. 215-71. Shoemaker, Sydney: 1984, 'Some Varieties of Functionalism', in Identity, Cause, and Mind, Cambridge University Press, Cambridge, Massachusetts, pp. 261-86. Dept. of Philosophy The University of Vermont 70 South Williams St. Burlington, VT 05401-3404 U.S.A.