Beginning the 1960s Hilary Putnam, Jerry Fodor, and Richard Boyd, among others, developed a type of materialism that denies reductionist claims. In this view, explanations, natural kinds, and properties in psychology do not reduce to counterparts in more basic sciences, such as neurophysiology or physics (Putnam 1967, 1974; Fodor 1974; Boyd 1980a). Nevertheless, all token psychological entities—states, processes, and faculties—are either identical with (Fodor 1974) or just wholly constituted of (Boyd 1980a) physical entities, ultimately out of token entities over which microphysics quantifies. This view was soon widely endorsed and since then has persisted as an attractive alternative to reductionist and eliminativist forms of materialism. Reductionists, notably Jaegwon Kim, have raised a series of serious objections to this position, to which nonreductivists have responded, thereby developing the view more thoroughly.
Irreducibility, Multiple Realizability, and Explanation
In his early argument for nonreductive materialism, Putnam adduces the phenomenon of multiple realizability as its main justification (Putnam 1967). Kinds or types of mental states can be realized by many kinds of neurophysiological states, and perhaps by many kinds of non-neurophysiological states, and for this reason they do not reduce to kinds of neurophysiological states. Multiple realizability also has a key role in Fodor's more general argument against reductionism in the special sciences (Fodor 1974). Consider a law in some special science:
S1x causes S2x
where S1 and S2 are natural kind-predicates in that science. A standard model for reduction requires that every kind featured in this law be identified with a kind in the reducing science, by way of bridge principles. Bridge principles might translate kind-predicates in one science into those of a more basic one, or they might specify a metaphysical relation, such as being identical with or being a necessary and sufficient condition for, between the kinds of one science and those of the reducing science. But in some cases, Fodor contends, the sort of bridge principle required for reducibility will not be available. If kinds in psychology, for instance, are multiply realizable in an indefinite variety of ways at the neurophysiological level, purported bridge principles for relating psychological to neurophysiological kinds will involve open-ended disjunctions. These purported bridge principles will be of the form:
P1 = N1 v N2 v N3 …
which states that a certain psychological state, P1, is identical with an open-ended disjunction of neurophysiological states, N1 v N2 v N3 … , or
P1 ↔ N1 v N2 v N3.…
which states that a certain psychological state is necessary and sufficient for an open-ended disjunction of neurophysiological states. Fodor argues that because open-ended disjunctions of kinds in neurophysiology are not natural neurophysiological kinds, psychological kinds cannot be reduced to neurophysiological kinds. Fodor's reason for denying that such disjunctions are not natural kinds is that they cannot appear in laws, and they cannot appear in laws because "laws" involving such disjunctions are not explanatory. Such "laws" are not explanatory because they do not satisfy our interests in explanation. Fodor's argument for irreducibility, then, appeals to the fact that purported explanations for psychological phenomena are unsatisfying when couched in terms of open-ended disjunctions.
One reductionist reply is that these open-ended disjunctions nevertheless constitute genuine laws and explanations, even if they fail to meet certain subjective requirements. If only we were capable of taking in more information at once, we wouldn't have any trouble regarding open-ended disjunctive "laws" as genuine laws (Jaworski 2002). That people fail to find laws satisfying when they contain open-ended disjunctions may simply show a failing on our part, rather than a failing of the putative laws. This standard argument for nonreductive materialism appears to rely on a certain formal prescription for laws and explanations—that they cannot contain disjunctive properties, or at least not wildly disjunctive properties.
But even if the formal argument fails, multiple realizability can still sustain an important component of nonreductive materialism. In general, whether or not a property is multiply realizable can indicate the level at which it should be classified. Is the kind corkscrew a kind of steel thing? No, for it also has a possible aluminum realization. Is the kind believing that cats are nearby a neural kind of thing? If mental states are also realizable in silicon, then no. Multiple realizability might then provide the key to precluding classification of mental states as essentially neural, or as essentially classified at some lower level yet.
Kim argues that multiple realizability might fail to undermine reductionism for a different reason. He contends that a higher-level property is precisely as projectible as the disjunction that expresses its multiply realizable character at a more basic level, and thus a generalization involving such disjunctive properties is just as lawlike as the higher-level generalization that it was meant to reduce (Kim 1992). The reason is that a higher-level property is nomically equivalent to such a disjunctive property. Nomic equivalence might be defined in this way: properties F and G are nomically equivalent if they are coextensive in all possible worlds compatible with the laws of nature. If Kim is right, then Fodor's formal argument does not appear to be sound, for it relies on the possibility that generalizations involving a higher-level property be lawlike whereas those involving the corresponding disjunctive property are not. But furthermore, Kim contends that wildly disjunctive properties are not projectible, and hence higher-level properties that are nomically equivalent to such properties are not projectible either. As a result, such higher-level properties cannot figure into laws, and they are not genuinely scientific kinds.
The example of a disjunctive property Kim adduces to make his point is being jade. "Jade" is a category that comprises two mineralogical kinds, jadeite and nephrite, and hence being jade is the same property as being either jadeite or nephrite. As a result, being jade will not be projectible. But in reply, being jade might turn out to be projectible despite its underlying complexity. Ned Block points out that all samples of jade share certain appearance properties, similarities that give rise to a certain degree of projectibility (Block 1997). More generally, properties that are multiply realizable can yet be projectible with respect to properties of selection, learning, and design. Because there are typically only a few ways in which entities of a particular higher-level type can be designed and produced, one can expect relatively broad similarities among these things that would render corresponding higher-level properties significantly projectible (Antony and Levine 1997).
Thus the heterogeneity of the possible realizations of a property is compatible with their having significant features in common, features that will sustain the projectibility of the property to some degree or other. This point is consistent with Kim's claim that a higher-level property is precisely as projectible as the disjunctive property that comprises all of its possible realizations. One should not conclude from the heterogeneity of the possible realizations of a higher-level property that there is no feature that can undergird its projectibility—in fact, of both the higher-level property and of the disjunctive property that comprises all of its possible realizations. Indeed, the projectibility-sustaining feature of a kind could be a characteristic that is significantly homogeneous across its heterogeneous realizations, one that might instantiate a unitary causal power at the level of description of the kind (Pereboom 2002).
Functionalism and Mental Causation
By way of objecting to Kim's reductionism, Block asks: "What is common to the pains of dogs and people (and all other species) in virtue of which they are pains?" (Block 1980, pp. 178–179). In reply to this concern, Kim points out that nonreductive materialists typically argue from a functionalist perspective, and that functionalists characterize mental states solely in terms of purely relational features of those states. Functionalism identifies mental state types with type-level dispositions to cause mental states and behavioral outputs given perceptual inputs and mental states—with the understanding that these dispositions are purely relational: that they are to be analyzed in terms of causal relations to perceptual inputs, behavioral outputs, and other mental states, and no intrinsic mental components. Functionalists claim that what all pains would have in common, by virtue of which they are all pains, is a pattern of such relations described by some functional specification. Kim then argues that in providing an answer to Block's question, the local reductionist—the one who opts for species- or structure-specific reductionism—is no worse off than the functionalist. Both are committed to the claim that there is no nonrelational or intrinsic property of pain that all pains have in common, and both can specify only shared relational properties (Kim 1992).
Kim implies that a functional specification does not provide a genuinely satisfactory answer to Block's question (Kim 1999). On the nonreductive view, if M is a mental property and B is its neural or microphysical base, then realizers for M can be found in B (at the level of B). This position allows that nondisjunctive realizing properties might be found in B for individual species- or structure-types—as long as there is no well-behaved (not wildly disjunctive) property in B that realizes every possible instance of M. The nonreductive materialist claims that none of this entails a genuine reduction of M to properties in B. As Kim assumes, the standard strategy for preserving M as meeting these specifications is to envision M as a functional mental property. But in Kim's view, the problem with the functionalist picture is that the causal powers of any instance of M will be causal powers in the physical base—they will not, at the token level, be irreducibly mental causal powers (Kim 1992, Block 1990). Hence functionalism cannot preserve the view that there exist causal powers that are in the last analysis irreducibly mental, and it is thus incompatible with a genuinely robust nonreductive materialism about the mental. Furthermore, Kim points out that given the genuine multiple realizability of the property M, the causal powers of the realizers of M in B will exhibit significant causal and nomological diversity, and for this reason the causal powers of M will exhibit such diversity. Thus, in his estimation, M will be unfit to figure in laws, and is thereby disqualified as a useful scientific property. He concludes that the functionalist model cannot protect m as a property with a role in scientific laws and explanations.
However, there is available a nonfunctionalist account of these higher-level powers that nevertheless remains nonreductive (Pereboom 1991, 2002). Functionalists typically maintain that the causal powers that have a role in explaining the dispositional features of mental states are nondispositional properties of their realization bases. For example, many suppose that nondispositional neural properties, which instantiate neural causal powers, would serve to explain why being pinched causes wincing behavior. But if these causal powers are all nonmental, a robust sort of nonreductive materialist account of the mental is precluded, for then none of the causal powers would be essentially mental themselves. By contrast, the nonreductivist might endorse intrinsic mental properties that instantiate specifically mental causal powers (Pereboom 1991, 2002; Van Gulick 1993). Such a view would be incompatible with functionalism. It need not deny that there exist functional mental properties, or, more generally, relational properties of mental states, but it would endorse nonfunctional mental properties that, by virtue of the causal powers they instantiate, play an important part in explaining dispositional features of mental state types.
Consider the example of a ball piston engine, the most recent version of the rotary internal combustion engine, which has a specific internal structural configuration. Characteristic of this engine is its having parts with particular shapes and rigidities, and these parts must be arranged in a particular way. These features are manifestly not functional relations that such an engine stands in; rather, they constitute intrinsic characteristics of this type of engine. At the same time, these characteristics are multiply realizable. The parts of the engine can be made of material of different sorts—as long as the material can yield, for example, the required shapes and rigidities. The ball piston engine, then, has nonfunctionalist intrinsic structural properties that instantiate its causal powers, but nevertheless admit distinct realizations.
Similarly, it might be that the heterogeneous physical realizations of the dog's and the human's belief that cats are nearby exhibit a structure of a single type that is intrinsic to this kind of mental state, a structure that instantiates the causal powers of this belief. This structure may be more abstract than any specific sort of neural structure, given that it can be realized in distinct sorts of neural systems (Boyd 1999). Perhaps this same structure can be realized in a silicon-based electronic system, and such a system could then also have the belief. Imagine a silicon system that replicates the capacities of and interconnections among neurons in a human brain as closely as possible, and suppose this system is excited to mimic as nearly as possible what happens when a human being has this belief about cats. It is possible that this silicon state would realize the same belief, and have a structure that, conceived at a certain level of abstraction, is similar enough to the structure of the ordinary neural system for both to count as examples of the same type of structure. In this case and more generally, one does not seem forced to retreat to mere functional resemblance prior to investigating whether the relevant similarities extend to intrinsic properties.
According to nonreductive materialism, an event such as Jerry's feeding the cat (M2) will have a psychological explanation in terms of a complex of mental states—beliefs and desires he has (M1). Each of M1 and M2 will be wholly constituted of microphysical events (P1 and P2 respectively), and there will be a microphysical explanation of P2 in terms of P1. The explanation of M2 by M1 will not reduce to the explanation of P2 by P1. Underlying the irreducibility of this explanation is that M1 is not type-identical with P1, and that M2 is not type-identical with P2.
This picture gives rise to a pressing question: What is the relationship between the microphysical and psychological explanations for M2? In particular, given that both sorts of explanation refer to causal powers, what is the relationship between the causal powers to which the microphysical explanation appeals and those to which the psychological explanation appeals? Here is where Kim's challenge from causal or explanatory exclusion enters in (Kim 1987, 1998). If a microphysical account yields a causal explanation of the microphysical constitution of M2, then it will also provide a causal explanation of M2 itself. How might there also be a distinct psychological causal explanation of this action? Kim argues that it is implausible that the psychological explanation appeals to causal powers sufficient for the event to occur, and at the same time the microphysical explanation appeals to distinct causal powers also sufficient for the event to occur, as a result of which the event is overdetermined. It is also implausible that each of these distinct sets of causal powers yields a partial cause of the event, and that each by itself would be insufficient for the event to occur.
By the solution to this problem that Kim develops, real causal powers exist at the microphysical level, and so the microphysical explanations refer to real microphysical causal powers. Only if psychological explanations in some sense reduce to microphysical explanations does it turn out that the psychological explanations also appeal to real causal powers—these causal powers will then ultimately be microphysical. Psychological explanations that do not reduce to microphysical explanations will fail to refer to causal powers, and thus will have some diminished status—such explanations might express regularities without at the same time referring to causal powers. This strategy solves the exclusion problem because if the causal powers to which the psychological explanation appeals are identical with those to which the microphysical explanation appeals, then there will be no genuine competition between explanations, and if the psychological explanations do not refer to causal powers at all, there will be no competition either. However, this solution, which Kim believes is the only possible solution to the problem he raises, would rule out any nonreductive view about mental causal powers.
Various proposals have been advanced in the name of nonreductive materialism according to which mental properties are causally relevant or causally explanatory, without being causally efficacious as mental properties. Such views, like Kim's, claim that all causal efficacy is nonmental (for example, Jackson and Pettit 1990). As Kim points out, these proposals do not amount to a robust sort of nonreductive materialism, which would preserve the claim that mental properties, as mental properties, are causally efficacious (Kim 1998).
What sort of response might the advocate of the robust view provide? First, in Kim's conception, any token causal powers of a higher-level property at a time will be identical with some token (micro)physical causal powers. There would be no token causal powers distinct from token microphysical causal powers, and this would preclude any robust nonreductive materialism. Higher-level kinds and explanations would at best group token microphysical causal powers in a way that does not correspond to the classifications of microphysics itself (Kim 1998, Horgan 1997). Such a classification might be of value for prediction, but there would remain no sense in which there exist causal powers that are not microphysical.
However, is token mental state M identical with P, its actual token microphysical realization base? Suppose that M is realized by a complex neural state N. It is possible for M to be realized differently only in that a few neural pathways are used that are token-distinct from those actually engaged. One need not rule at this point on whether the actual neural realization N is token-identical with this alternative—it might well be. But it is evident that this alternative neural realization is itself realized by a microphysical state P* that is token-distinct from P. It is therefore possible for M to be realized by a microphysical state not identical with P, and thus M is not identical with P. But furthermore, this reflection would also undermine a token-identity claim for mental causal powers—should they exist—and their underlying microphysical causal powers. For supposing that the token microphysical realization of M had been different, its token microphysical causal powers would also have been different. Consequently, there is good reason to suppose that any token mental causal powers of M would not be identical with the token microphysical causal powers of its realization (Boyd 1980a, Pereboom and Kornblith 1991, Pereboom 2002).
On this conception, a token mental state would have the mental causal powers it does ultimately by virtue of the token microphysical states of which it is constituted (setting aside any fundamentally relational causal powers). For this reason it makes sense to say that token mental causal powers are wholly constituted by token microphysical causal powers. More generally, the causal powers of a token of kind F are constituted of the causal powers of a token of kind G just in case the token of kind F has the causal powers it does by virtue of its being constituted of a token of kind G.
And now, just as no competition between explanations arises in the case of reduction and identity, competition also does not arise in the case of mere constitution. For if the token of a higher-level causal power is currently wholly constituted by a complex of microphysical causal powers, there are two sets of causal powers at play that are constituted from precisely the same material (supposing that the most basic microphysical entities are constituted of themselves), and in this sense we might say that these powers coincide constitutionally. That they now coincide in this way might give rise to the thought that these causal powers are token-identical, but, as has been shown, there is a substantial argument that they are not. And because it is possible for there to be wholly constitutionally coinciding causal powers that are not even token-identical, it is possible that there be two causal explanations for one event that do not exclude each other and at the same time do not reduce to a single explanation (Pereboom 2002).
If identity and not just constitutional coincidence were necessary for explanatory noncompetition, then there would be features required for noncompetition that identity has and current constitutional coincidence does not. The candidate features would be constitutional coincidence at all other times, and constitutional coincidence at all other possible worlds, even now. But it is difficult to see how the token causal powers' constitutional noncoincidence at some past time, or at some future time, or their merely possible constitutional noncoincidence even now would result in explanatory competition, whereas actual current constitutional coincidence in absence of any features of this sort (i.e., identity) would guarantee noncompetition.
Imagine that a person's current token mental state M actually constitutionally coincides with token microphysical state P. Now assume with Kim that if M were identical with P, and if their causal powers were identical, there would be no explanatory competition. Then if mere constitutional coincidence without identity resulted in explanatory competition, that would have to be because at some time in the past or in the future, or at some other possible world even now, M and P and their causal powers are constitutionally noncoincident. Suppose that M would still exist even if a few neural pathways in its neural realization were token-distinct from what they actually are. These neural changes would render M's microphysical realization base distinct from P, and thus M and P would be constitutionally noncoincident in some other possible world, and, similarly, mutatis mutandis (that is, the necessary changes having been made) for their causal powers. How could a possibility of this sort introduce explanatory competition? It would appear that actual current constitutional coincidence alone is relevant to securing noncompetition, and thus for this purpose constitutional coincidence without identity would serve as well as identity. Consequently, it would appear that available to the nonreductivist is a solution to the exclusion problem no less adequate than Kim's own.
The Threat of Emergentism
Kim contends that nonreductive materialism is committed to emergentism (sometimes called strong emergentism, which he thinks is a radical and implausible view. In his analysis, emergentism claims a distinction between two sorts of higher-level properties, resultant and emergent, that arise from the basal conditions of physical systems (Kim 1999). The basal conditions of a physical system comprise (i) the basic particles that constitute the physical system, (ii) all the intrinsic properties of these particles, and (iii) the relations that configure these particles into a structure. The higher-level properties that are merely resultant are simply and straightforwardly calculated and theoretically predictable from the facts about its basal conditions—which presumably include the laws that govern the basal conditions—whereas those that are emergent cannot be calculated and predicted. Theoretical predictability contrasts with inductive predictability. Having regularly witnessed that an emergent property is realized by particular basal conditions, we would be able to predict this relationship, but this sort of inductive predictability is not at issue. Rather, according to emergentism, knowledge of the basal conditions alone, no matter how complete, does not suffice to yield a prediction of an emergent property.
Emergentism also endorses downward causation; it claims that higher-level states can have lower-level effects. Emergentism about the mental asserts that mental events can cause microphysical events. Plausibly, nonreductive materialism also countenances downward causation of this sort—M1 causes M2, but because M2 is wholly constituted of P2, M1 also causes P2. Kim thinks that by virtue of endorsing this sort of downward causation, nonreductive materialism is committed to emergentism.
However, the nonreductive view's allowing for downward causation is not by itself sufficient to render it emergentist. Endorsement of downward causation would indeed be radical if it also specified that mental properties could effect changes in the laws that govern the microphysical level independently of any emergent properties (call them the ordinary microphysical laws). Supposing that M1 were such an emergent mental property, M1 could cause P2 in such a way that P2 is no longer governed by the ordinary microphysical laws, but instead by laws that take into account the special characteristics of the emergent properties, or no laws at all. But nothing essential to nonreductive materialism entails this radical variety of downward causation (Pereboom 2002).
We might suppose that the capacity for altering the ordinary microphysical laws is what provides emergent properties with their distinctive nature. And this potentially explains why such properties would not be predictable from the microphysical base together with these ordinary laws. Information about the ordinary laws and the microphysical base might be insufficient to predict the law-altering behavior of the higher-level property. But there is no feature of the nonreductive model per se that renders higher-level properties any less theoretically predictable than they would be on a reductive model. In each model, holding relational conditions fixed, a particular set of basal conditions will necessitate the same unique higher-level properties. The nonreductivist is no more committed to some factor that threatens theoretical predictability, such as the capacity of higher-level properties to alter the ordinary microphysical laws, than is the reductionist.
Arguably, therefore, nonreductive materialism can respond effectively to the most serious arguments made against it over the last forty years, and as a result, it remains a viable position about the nature of the mental.
Antony, L., and J. Levine. "Reduction with Autonomy." In Philosophical Perspectives. Vol. 11, Mind, Causation, and World, 83–105. Oxford: Blackwell, 1997.
Block, N. "Anti-Reductionism Slaps Back." In Philosophical Perspectives. Vol. 11, Mind, Causation, and World, 107–132. Oxford: Blackwell, 1997.
Block, N. "Can the Mind Change the World?" In Meaning and Method: Essays in Honor of Hilary Putnam, edited by George Boolos, 137–170. Cambridge, U.K.: Cambridge University Press, 1990.
Block, N. "Introduction: What is Functionalism?" In Readings in the Philosophy of Psychology, vol. 1, edited by Block, 178–179. Cambridge, MA: Harvard University Press, 1980.
Boyd, R. "Kinds, Complexity, and Multiple Realization." Philosophical Studies 95 (1999): 67–98.
Boyd, R. "Materialism Without Reductionism." In Readings in the Philosophy of Psychology. Vol. 1, edited by N. Block, 67–106. Cambridge, MA: Harvard University Press, 1980a.
Boyd, R. "Scientific Realism and Naturalistic Epistemology." In Proceedings of the Philosophy of Science Association. Vol. 2. East Lansing, MI: Philosophy of Science Association, 1980b.
Clapp. L. "Disjunctive Properties: Multiple Realizations." Journal of Philosophy 98 (2001): 111–136.
Fodor, J. "Special Sciences: Still Autonomous After All These Years." In Philosophical Perspectives. Vol. 11, Mind, Causation, and World, 149–163. Oxford: Blackwell, 1997.
Fodor, J. "Special Sciences (or: The Disunity of Science as a Working Hypothesis)." Synthese 28 (1974): 97–115.
Heil, J. "Multiple Realizability." American Philosophical Quarterly 36 (1999): 189–208.
Horgan, T. "Kim on Mental Causation and Causal Exclusion." In Philosophical Perspectives. Vol. 11, Mind, Causation, and World, 165–184. Oxford: Blackwell, 1997.
Jackson, F., and P. Pettit. "Program Explanation: A General Perspective." Analysis 50 (1990): 107–117.
Jaworski, W. "Multiple Realizability, Explanation, and the Disjunctive Move." Philosophical Studies 108: (2002): 298–308.
Kim, J. "Making Sense of Emergence." Philosophical Studies 95 (1999): 3–36.
Kim, J. Mind in a Physical World: An Essay on the Mind-Body Problem and Mental Causation. Cambridge, MA: MIT Press, 1998.
Kim, J. "Multiple Realizability and the Metaphysics of Reduction." Philosophy and Phenomenological Research 52 (1992): 1–26.
Kim, J. "The Myth of Nonreductive Materialism." Proceedings and Addresses of the American Philosophical Association 63 (1989): 31–47.
Kim, J. "Phenomenal Properties, Psychophysical Laws, and the Identity Theory." Monist 56: (1972): 177–192.
Kitcher, P. S. "1953 and All That: A Tale of Two Sciences." Philosophical Review 93 (1984): 335–373.
Pereboom, D., and H. Kornblith. "The Metaphysics of Irreducibility." Philosophical Studies 63 (1991): 125–145.
Pereboom, D. "Robust Nonreductive Materialism." Journal of Philosophy 99 (2002): 499–531.
Pereboom, D. "Why a Scientific Realist Cannot Be a Functionalist." Synthese 88 (1991): 341–358.
Putnam, H. "Language and Reality." In his Philosophical Papers, vol. 2, 272–290. Cambridge, MA: Harvard University Press, 1975. This paper was delivered as a Machette Lecture at Princeton University, May 22, 1974.
Putnam, H. "The Nature of Mental States." In his Philosophical Papers. vol. 2, 429–440. Cambridge, MA: Harvard University Press, 1975. First published as "Psychological Predicates," in Art, Mind, and Religion, edited by W. H. Capitan and D. D. Merill, 37–48 (Pittsburgh: Pittsburgh University Press, 1967).
Shapiro, L. "Multiple Realizations." Journal of Philosophy 97 (2000): 635–654.
Sober, E. "The Multiple Realizability Argument Against Reductionism." Philosophy of Science 66 (1999): 542–564.
Van Gulick, R. "Who's in Charge Here? And Who's Doing All the Work?" In Mental Causation, edited by J. Heil and A. Mele, 233–256. Oxford: Oxford University Press, 1993.
Yablo, S. "Mental Causation." Philosophical Review 101 (1992): 245–280.
Derk Pereboom (2005)