The view that everything that actually exists is material, or physical. Many philosophers and scientists now use the terms `material' and `physical' interchangeably (for a version of physicalism distinct from materialism, see physicalism). Characterized in this way, as a doctrine about what exists, materialism is an ontological, or a metaphysical, view; it is not just an epistemological view about how we know or just a semantic view about the meaning of terms.


Materialism is a general view about what actually exists. Put bluntly, the view is just this: Everything that actually exists is material, or physical. Many philosophers and scientists now use the terms `material' and `physical' interchangeably (for a version of physicalism distinct from materialism, see physicalism). Characterized in this way, as a doctrine about what exists, materialism is an ontological, or a metaphysical, view; it is not just an epistemological view about how we know or just a semantic view about the meaning of terms.

Materialism versus Dualism

There are two prominent construals of `material'. First, according to many philosophers, something is material if and only if it is spatial, extended in space. One might thus propose that what it means to say that something is material is that it is extended in space. This construal of `material' is inspired by Descartes's influential characterization of material bodies, in Meditation II. Given this construal, materialism is just the view that everything that exists is extended in space, that nothing nonspatial exists. This portrayal of materialism is attractively simple, but may be unilluminating.

The problem is that the relevant notion of spatial extension may depend on the very notion of material in need of elucidation. If there is such dependence, conceptual circularity hampers the proposed characterization of materialism. The main worry here is that the notion of spatial extension is actually the notion of something's being extended in physical space, or the notion of something's being physically extended. It seems conceivable that something (perhaps a purely spiritual being) has temporal extension, in virtue of extending over time, even though that thing lacks extension in physical space. It does not seem self-contradictory, in other words, to hold that something is temporal (or, temporally extended) but is not a body. If this is so, the proposed characterization of materialism should be qualified to talk of physical space or physical extension. In that case, however, the threat of conceptual circularity is transparent. Even if there is no strict circularity here, the pertinent notion of spatial extension may be too closely related to the notion of material to offer genuine clarification. At a minimum, we need a precise explanation of spatial extension, if talk of such extension aims to elucidate talk of what is material. Perhaps a notion of spatial extension is crucial to an elucidation of materialism, but further explanation, without conceptual circularity, will then be needed. (Cf. Chomsky 1988.)

If there is indeed a coherently conceivable distinction between minds and material bodies, we must reject the view that materialism, understood as entailing mind-body identity, is conceptually, or analytically, true—that is, true just in virtue of the meanings of `mind' and `body'. Given such a coherently conceivable distinction, we can also challenge any version of materialism implying that psychological concepts (for example, the concepts of belief and sensory pain) are defined in terms of the ordinary physical causes of belief states and pain states. (Such materialism has been proposed by D. M. Armstrong 1977, and David Lewis 1966.) If `pain' is defined in terms of the ordinary bodily causes of pain, then it will not be coherently conceivable that there is pain without bodies. The concept of pain will then depend for its semantic significance on the concept of a bodily cause.

Materialists do not share a uniform view about the nature of psychological properties, such as the properties of being a belief, being a desire, and being a sensory experience. In particular, they do not all hold that every psychological property is equivalent or identical to a conjunction of physical properties. Only proponents of reductive materialism hold the latter view, and they are a small minority among contemporary materialists. Proponents of nonreductive materialism reject the latter view, and affirm that psychological properties can be exemplified even in an immaterial world. Such nonreductive materialists include functionalists about the mind, who hold that psychological properties differ from material properties in virtue of the special causal or functional roles of the former. Functionalists differ from behaviorists in acknowledging the psychological relevance of causal relations among not only stimuli and behavior but internal states as well. A third prominent version of materialism, eliminative materialism, recommends that we eliminate most, if not all, everyday psychological discourse, on the ground that it rests on seriously misguided assumptions about human psychology—assumptions that will disappear with the advance of science.

According to many functionalists, the causal roles that determine psychological properties are specified by the taxonomies, or systems of classification, found in the best contemporary psychology. These causal roles can depend on relational considerations that are independent of considerations about the composition of what exemplifies psychological properties. Functionalism allows that psychological properties can be "multiply realizable," realizable in compositionally different systems. Both carbon-based and silicon-based physical systems (and even nonphysical systems), for example, might support such a psychological process as thinking. Thinking, according to functionalism, does not require a specific physical composition for thinkers. Physical composition can vary as long as the appropriate causal-relational features obtain.

Nonreductive Materialism

Nonreductive materialists generally follow functionalists in emphasizing the importance of multiple realizability. Before we can appreciate the evidence for multiple realizability, we need an appropriate vocabulary. A type of state, property, process, object, or event (hereafter, simply "object"), is a class or kind of object that admits of instances. An egg, for example, is a type of single-celled organism. There are many concrete instances of eggs: for instance, in humans, in hatcheries, and in many refrigerators. These individual eggs are tokens. We understand an object as a type or a token relative to a taxonomy, a means of classification—although it does not follow that the existence of all tokens depends on language. A particular mastiff may be a token of the type mastiff, but it is also a token of the type dog, mammal, animal, domesticated animal, and slobbering thing. Types may be scientifically taxonomic, but are not so automatically; whether they are such depends on their role in a scientific theory. With these concepts in hand, let us turn to the issue of type-type, or "smooth," reduction.

The smooth reduction of one theoretical description to another preserves causal/explanatory role. This preservation of causal/explanatory role is reflected in at least one of two ways: (a) the laws in the reduced and reducing theories are similar (this concerns whether they isolate the same covariations in the world) and (b) theoretical-predicate pairs across the reduced and reducing theories isolate, or pick out, the same objects. (Cf. Churchland 1989, chaps. 1, 3, and Hooker 1981.) Traditional accounts of reduction imply that theories, laws, and terms can be objects of reduction. One law, for instance, is reducible to another if the law targeted for reduction is logically derivable from the corresponding law in the reducing domain. Reduction, construed ontologically, is a relation between two theoretically characterized domains of entities, whether postulated objects, properties, processes, states, events, or laws. (A postulated entity need not, of course, be an actual entity.)

The type-type identity theory has the disadvantage that, as a formulation of materialism, there is inadequate evidence for it. For example, there is no evidence that it is generally the case that, for every type of psychological process (relative to the best psychological taxonomy), there is a corresponding type of neural process (relative to the best neuroscientific theory). Many materialists hold that there is evidence instead for a weaker, token-token identity theory, according to which any individual or token—a particular dog, a particular NaCl molecule, a particular cultural ritual—is entirely composed of physical phenomena. We might assure ourselves of this fact by a strategy of decomposition: Analyze all the constituents of the token, and determine whether any nonphysical phenomena are present. There is, of course, an epistemological question about how we might detect (and thus interact with) nonphysical phenomena. Awaiting evidence for nonphysical phenomena, materialists can perhaps be excused for withholding assent to such phenomena.

Endorsing the aforementioned multiple realizability functionalists may acknowledge token-token identity, but they challenge type-type reduction. Relative to dualism, acknowledgment of mind-body token identity may itself seem reductive because it rejects dependence of minds on nonphysical substances. Nonetheless, acknowledgment of just token identity is, as materialist doctrines go, a nonreductive formulation of materialism. Another nonreductive version of materialism, compositional materialism, casts even token-identity theories as too demanding.

Compositional materialism implies that physical (and thus, for the physicalist, psychological) events are not typically identical to their smaller constituent features. There is, according to compositional materialism, plasticity (or, multiple realizability) even within a single physical token, just as there is within a type susceptible to instantiation by different physical tokens. An example from some influential work on compositional materialism states that an individual car remains the same car even if its generator is replaced, at least on our ordinary criteria of car identity (Boyd 1980, p. 100). A difference in molecular constituents of the car in two possible worlds does not preclude, on this view, sameness of car. (For an analysis of this and other accounts, see Moser and Trout 1995a).

Materialism, Naturalism, and Explanation

The fates of naturalism and materialism are intimately related. Naturalism is the doctrine that the methods of philosophy are continuous with those of the natural sciences. The following two general views of the relation between naturalism and materialism are noteworthy: (a) Naturalism is ontologically neutral regarding materialism, and thus is logically compatible with ontological dualism; (b) Naturalism presupposes materialism, and thus is incompatible with dualism. If (a) is correct, the defense of materialism on the basis of naturalism must appeal to additional supporting evidence, presumably empirical evidence. If option (b) is correct, an appeal to naturalism in defense of materialism would be question begging, because materialism would then be part of the doctrine of naturalism. We suspect that most contemporary naturalists would prefer (a) to (b).

Materialism and Mind

Materialists have always had the difficult task of explaining how their materialism can account for such psychological phenomena as thoughts, beliefs, desires, intentions, and sensory experiences—or at least for familiar talk of such phenomena. A materialist's options, put roughly, are these: (a) explain how ordinary talk of psychological phenomena ("folk psychology," for short) can, at least for the most part, be reduced to language that does not commit one to any kind of ontological dualism; (b) explain how folk psychology is misguided to such an extent that it will disappear altogether with the advance of science; (c) explain how folk psychology is perfectly compatible with materialism even if the kind of reduction sought by (a) is unavailable—in particular, explain either (i) how psychological phenomena actually depend on physical phenomena, owing to nonreductive "supervenience" relations of some sort, or at least (ii) how psychological phenomena are just special relational (for example, causal/functional) features of wholly physically composed systems (see Kim 1992).

Apart from the relational, propositional states of belief, there are monadic states as well. The typical monadic mental states are conscious states, and their materialist character is also a topic of controversy. It is not clear how a purely material phenomenon can give rise to subjective qualities (see Chalmers 1996 and Jackson 1982), and this mystery is widely regarded as an obstacle to any fully materialist account of psychological experience. The reason for this "explanatory gap" (Levine 1983) between material composition and conscious experience may itself be illuminated by recent developments in the philosophical analysis of complexity (Bechtel and Richardson 1993; Godfrey-Smith 1996). The complex relation between the material substrate and conscious experience is likely to leave the explanation of this relation nontransparent, as it does in other cases of complex phenomena.


J.D. Trout and Paul Moser