Anomalous Monism

1. Introduction

Donald Davidson's anomalous monism is a theory of mind that can be regarded as emerging from two pressures on its predecessor, the type-identity theory. The type-identity theory, usually attributed to J. J. C. Smart  (Smart, 1959) and U. T. Place (Place, 1956), claimed that kinds of mental states are identical to kinds of brain states. Sensations of pain, for instance, were said to be identical to the firing of C-fibres or some such type of neurological state. According to this view, then, pain, conceived as a kind of mental state, is said to be reduced to a certain kind of neurological state. The reduction envisaged here was modelled on the kind of reduction seen in other areas of the sciences. For instance, lightning can be said to be reduced to a rapid discharge of electrons in the atmosphere. When such a reduction is made scientists are not saying that there are two phenomena that are correlated, but rather that lightning is nothing more than a rapid discharge of electrons in the atmosphere. Similarly, the claim made by the type-identity theorists was that pain is nothing more than a certain sort of neurological state. This theory can be said to be reductive in two senses. First, and most obviously, it is ontologically reductive, meaning that it claims that all it is to be in pain is to be in a specific kind of brain state; that's what pain is. Second, it is conceptually reductive, meaning that the mental predicate "pain" can be regarded as definitionally equivalent to a physical predicate describing the relevant neurological state. This follows from the claim that the bridge laws connecting the mentioned predicates -- laws of the form y iff j -- prove that the predicates are coextensive and therefore equivalent.

The model of reduction involved in the type-identity theory was associated with a movement in the philosophy of science called the "unity of science." Advocates of the unity of science, spurred on by the success of the reduction of certain biological phenomena to chemical phenomena, anticipated the reduction of psychological states to physical states through a series of inter-theoretic reductions from psychology to neurology, neurology to chemistry, and so on to the most fundamental physical level of description. By establishing a series of "point reductions" (such as the reduction of pain to the firing of C-fibres), it was thought that higher level theories such as Psychology could be modelled in the more basic reducing theories. Unfortunately, it soon appeared as though the anticipated unification of science was overly optimistic, and here we come to the first pressure on the type-identity theory. As Ian Hacking (Hacking, 1986) has suggested, the practice of science has shown more of a tendency toward fragmentation and compartmentalisation, with different areas of science paying little attention to one another. The envisaged unity of science soon began to appear to be little more than a philosopher's "idle pipedream." Given this, there was little reason to expect the kind of reductions envisaged by the type-identity theory, and hence, little reason to accept the idea of the definitional equivalence of mental and physical predicates.

The second pressure on the reductive theory of mind was that it is chauvinistic or speciesist. By identifying types of mental states with types of brain states the theory seems to preclude the possibility that other forms of life with brains or central nervous systems that are very different from ours could have mental states like us. If pain just is a particular kind of state of the human brain, then a creature without a human brain cannot, by definition, experience pain. This seems to be an unreasonable implication. We should be willing to allow for the possibility of what has come to be called "multiple realisation": the idea that pain (and other mental states) can be physically realised in a variety of different ways. For instance, while it may be the case that in humans pain is identical to a certain kind of neural state N, in Martians it is identical to a completely different kind of physical state, perhaps some sort of silicon state N*. In light of this, there was little reason to expect even the ontological reduction of mental states proposed by the type-identity theory. Given the principle of multiple realisation it seemed as though it could not be said that pain can be reduced to a particular kind of physical state such as a kind of neurological state.

Davidson's theory of mind, known as "anomalous monism," emerged as a form of identity theory that is sensitive to both of these pressures. According to anomalous monism, mental events are identical to physical events but the mental is not reducible to the physical. Hence, Davidson proposes an identity theory without the reductive bridge laws associated with the type-identity theory. Davidson achieves this is by maintaining that the identity holds between mental and physical tokens (particular mental and physical occurrences) rather than types (general kinds of events). So, according to Davidson, when I experience a pain that pain is identical to some particular physical state in me, and when you are in pain that particular pain is identical with some physical state in you, but this does not mean that when we are both in pain there necessarily exists some physical state that we share. Pain is physically realised in a different way in me than it is in you, and furthermore, it is possible for that physical realisation in each of us to change over the course of time. In this case it is not even true that when I am in pain at two times I am necessarily in the same physical state on both occasions. Since Davidson denies that an identification can be made between mental and physical types, he denies that mental concepts, such as pain, can be reduced to physical concepts. Davidson's theory, then, represents a form of nonreductive materialism.

2. The Theory

Davidson proposes and defends this theory in his article "Mental Events" (Davidson, 1980) and does so in order to remove an "apparent contradiction" that seems to follow from three principles he holds to be true about mental events:

At least some mental events interact causally with physical events. Events related as cause and effect fall under strict deterministic laws. There are no strict deterministic laws on the basis of which mental events can be predicted and explained. Given certain natural assumptions, premises (1) and (2) would appear to deny the truth of premise (3). If mental events interact causally with physical events, and if there are strict deterministic laws wherever there are causal relations, then it seems as though there ought to be psychophysical laws.

Davidson's means of eliminating the apparent contradiction depends principally upon two ideas. First, he argues for a weak interpretation of the principle of the nomological character of causation. Second, he excludes mental descriptions from participation in strict causal laws. The anomalism of the mental that is entailed by the exclusion of mental predicates from strict laws, together with premises (1) and (2) entails the truth of Davidson's nonreductive brand of identity theory.

3. The Nomological Character of Causality

In "Mental Events" Davidson suggests that we adopt a weak reading of Hume's claim that a causal law "covers" every singular causal claim. According to Davidson, by this we should not take Hume to mean that the statement of the relevant covering law is necessarily formulated in the same terms as the singular causal claim; instead, we should take him to mean that the statement of the law incorporates some true description of the events related as cause and effect. That is, where we have a true singular causal claim such as "a caused b" it is not necessarily the case that the relevant causal law is formulated in terms of the mentioned descriptions "a" and "b". For instance, consider the claim that the event reported on page 8 of Tuesday's Globe and Mail caused the event reported on page 12 of Wednesday's Tribune. This could very well be a true singular causal claim but we would not expect the underlying causal law to be formulated in terms of the mentioned newspaper headlines; instead, we would expect the law to be formulated in terms involving a more precise physical description of the events in question. So while it is true that where there is causality there are causal laws, such laws might be formulated using very different predicates than the ones used in the singular causal claim. In light of this, it follows that where we have singular causal claims involving mental events we need not expect the relevant covering law to make use of mental predicates. The causal claim He ducked because he noticed the projectile may be true, but there may not be a strict causal law connecting noticings and duckings. It is possible that the law will be formulated in quite different terms.

4. Mental Predicates and Strict Causal Laws

According to Davidson, the only predicates that are suited to the formulation of strict laws are those that would be employed in what he calls the "closed system" of an ideal physics. An ideal physics constitutes a closed system because the descriptions of events possible in that language are fully extensional and express exceptionless laws free from intrusion by intensional concepts. At this level of description events can be characterised in a precise and determinate manner, yielding repeatable results. The mental predicates used to describe the behaviour of rational agents are unsuited to the formulation of strict laws because they do not constitute a closed system. The reason for this claim lies in Davidson's holism about the mental. According to Davidson, mental states and events are not ascribed to individuals one by one, but are ascribed against the background of a larger set of mental states. It is this character of the mental that created such a problem for behaviourism. The behaviourists attempted to define mental states in terms of dispositions to behave in certain ways. One could plausibly identify the belief that it is raining with a set of dispositions, such as the disposition to carry an umbrella or to respond by making the utterance "Yes" when confronted with the vocalisation "Do you believe it is raining?" only by making further assumptions about the agent's other mental states, in which case mental vocabulary slips back in, defeating the behaviourist's attempt to eliminate it. For instance, the person in question will only respond by saying "Yes" if he understands the question and wants to tell the truth, and will only carry his umbrella if he has a desire to keep dry, remembers that he has an umbrella, and so on.

The insight gained from the failure of behaviourism, as Davidson sees it, is that beliefs and desires are attributed to rational agents on the assumption that such agents have mostly true and consistent beliefs. The identification of another's mental states must therefore cohere with his or her other beliefs and should preserve truth and consistency. Thus, a rational principle guides what mental states we ascribe to others. This guiding principle is known as the "constitutive ideal of rationality." It is this feature of the mental that renders the vocabulary of psychology unsuitable for the formulation of strict causal laws. Since our ascription of mental states and events to agents is, in light of the rational ideal, always open to reinterpretation over the course of time in light of new evidence and behaviour, mental events can be said to cause behaviour only as they are mediated by other mental events (namely, those ascribed to an agent at a later time in light of new evidence) "without limit." Because strict laws, being exceptionless, require fixed and determinate descriptions, the ever-changing and indeterminate behaviour of mental descriptions renders them inappropriate for participation in strict laws. This means that the law covering causal claims involving mental events must be formulated in physical, not psychological terms.

Davidson is careful to point out that although there can be no strict causal laws at the level of psychology, this does not mean that there cannot be lawlike generalisations involving events characterised under mental descriptions. Davidson distinguishes between "heteronomic" and "homonomic" generalisations. Homonomic generalisations are those we have reason to believe could be sharpened into strict laws with the simple addition of further caveats and qualifications, and as such do not require a radical shift in the vocabulary used to describe the events related as cause and effect. Heteronomic generalisations, on the other hand, are lawlike in the sense that they are confirmed by their instances, but they cannot be sharpened into exceptionless strict laws by means of additional ceteris paribus clauses; instead, a radical change in vocabulary is required. Thus, although Davidson denies that there can be either psychophysical or psychological laws, he retains an explanatory role for psychology by recognising the existence of psychological generalisations. This is significant since such generalisations are central to the task of explaining intentional action in psychological terms. In order to explain someone's behaviour in accordance with principles of rationality it is inevitable that one will make use of various psychological generalisations. And while such generalisations suggest that a strict causal law is at work behind the scenes, the relevant law cannot be formulated simply by becoming more exact in one's psychological descriptions. The relevant law can only be captured in the more precise terms of a closed physical theory.

5. Token identity

The truth of Davidson's token identity theory can be seen to follow from the above principles and from the assumption that at least some mental events interact causally with physical events. If a mental event M is the cause of a physical event P, then these events must, under some description, instantiate a strict causal law. Given the anomalism of the mental, the relevant law cannot employ the mental description of event M, and so must characterise M under a physical description, in which case M is a physical event. Thus, all mental events are identical to physical events even though there can be no psychophysical laws.

6. Supervenience

In addition to developing an alternative mind-body theory, a further innovation in "Mental Events" is Davidson's introduction of the concept of supervenience to the philosophy of mind. Davidson says,

Although the position I describe denies there are psychophysical laws, it is consistent with the view that mental characteristics are in some sense dependent, or supervenient, on physical characteristics. Such supervenience might be taken to mean that there cannot be two events alike in all physical respects but differing in some mental respect, or that an object cannot alter in some mental respect without altering in some physical respect (Davidson, 1980).

There has been considerable discussion about exactly how to understand Davidson's version of supervenience. In particular, there has been debate about the proper modal force that should be attributed to the word "cannot" in the above passage. That is, should Davidson be understood as saying that in all possible worlds physically indiscernible events are also mentally indiscernible, or is he saying that this relation holds merely within a world?

These two ways of characterising the supervenience relation correspond to the distinction made by Jaegwon Kim between strong and weak supervenience:

1. A weakly supervenes on B if and only if necessarily for any property F in A, if an object x has F, then there exists a property G in B such that x has G, and if any y has G it has F (Kim, 1993).

2. A strongly supervenes on B just in case, necessarily, for each x and each property F in A, if x has F, then there is a property G in B such that x has G, and necessarily if any y has G, it has F (Ibid., p. 65).

Whether or not Davidson had in mind strong or weak supervenience when he wrote "Mental Events" is significant in light of the fact that he treats supervenience as a relation of dependence. (In fact, Davidson seems to treat supervenience and dependence as equivalent concepts in the passage quoted above.) Kim has argued that weak supervenience is too weak to express the dependence of the mental on the physical. In his view weak supervenience lacks the proper modal force required for genuine dependence between the related properties. In Kim's words:

Determination or dependence is naturally thought of as carrying a certain modal force: if being a good man is dependent on, or is determined by, certain traits of character, then having these traits must insure or guarantee being a good man (or lacking certain of these traits must insure that one not be a good man). The connection between these traits and being a good man must be more than a de facto coincidence that varies from world to world (Ibid., p. 60).

Without a necessary connection between the supervenient properties and the supervenience base, then, it seems there is little reason to think of the supervening properties as depending on the base properties. Davidson has said that he accepts something like weak supervenience (Davidson, 1985), in which case it appears that, if Kim's intuitions about the modal force necessary for dependence are correct, Davidson's characterisation of supervenience cannot be regarded as a kind of dependence.

A further problem with Davidson's conception of supervenience is that it is consistent with the possibility that two people who are physically indiscernible with the exception of one small detail, such as that one has one eyelash that is slightly longer than his counterpart's, could differ radically in their mental states. While we might expect significant mental differences in light of different neurological structures, it seems very unlikely that two individuals could differ in their mental states in virtue of physical differences such as eyelash length.

The implications of these problems are far-reaching. Kim has argued that since Davidson's conception of supervenience is modally weak, it allows for the possibility of worlds that are physically indistinguishable from our own but which are completely lacking in any mental life whatsoever. This does not appear to capture a robust form of physicalism since the connection between mental and physical states is not sufficiently strong.

Davidson has not offered a clear response to this problem. However, he could respond in the following way (Campbell, Forthcoming). One could claim that talk of other possible worlds is not necessary to ground the dependence of the mental on the physical in this world. In fact, there is a clear sense in which Davidson's account of the relation between mental and physical descriptions captures the idea of psychophysical dependence.

In a passage where Davidson elaborates on his conception of psychophysical supervenience, he claims that the relation is best understood as one between predicates rather than between properties:

The notion of supervenience, as I have used it, is best thought of as a relation between a predicate and a set of predicates in a language: a predicate p is supervenient on a set of predicates S if for every pair of objects such that p is true of one and not of the other there is a predicate of S that is true of one and not of the other (Davidson, 1985, p. 242).

The fact that Davidson formulates supervenience as a relation between predicates identifies a significant difference between Davidson's understanding of the concept and Kim's. Kim conceives of events as being ontologically composed of properties. Thus, from Kim's perspective, supervenience is a metaphysical relationship between the properties that constitute events. Davidson, on the other hand, has always been reluctant to analyse events into property exemplifications. In this case it seems as though supervenience is, for Davidson, a linguistic thesis connecting mental and physical descriptions. The mental characteristics that we ascribe to agents are to a large extent determined by the behaviour of those agents and their interactions with the environment. If two speakers make the utterance "gavagai" under the same physical conditions (for example, they both point at a rabbit) then if we are to ascribe the belief that's a rabbit to one speaker, we must, to avoid arbitrariness, ascribe the same belief to the other speaker. Similarly, if we are to ascribe a change in belief to such a speaker so that we now say he believes that's an aardvark and not that's a rabbit, we must have physical evidence to do so, such as a change in behaviour or a change in the physical circumstances under which the utterance was made. The ascription of mental predicates to a speaker are therefore dependent on and determined by the physical predicates that can be ascribed to a speaker. So despite the fact that Davidson's conception of supervenience is modally weak, there is nevertheless a clear sense in which it captures the idea that the mental is dependent on the physical.

These considerations go a long way toward defusing Kim's objections. First, since a clear sense of dependence is captured by Davidson's conception of supervenience in this world it appears that questions about the modal force of the relation are beside the point. Second, the worry involved in the eyelash example appears to be misguided. The problem with that objection is it assumes Davidson accepts a form of local supervenience, whereby the mental supervenes on a narrow set of physical characteristics (such as ones describing the brain and central nervous system). Given the way Davidson thinks mental predicates depend on physical ones we have seen that this is not the case. Mental predicates supervene non-locally on a variety of physical states of the speaker and on features of the environment. Hence, the objection misses its mark since it presupposes a version of supervenience Davidson does not accept.

7. The Standard Objection

Aside from the above difficulties with supervenience, the principal worry philosophers have with Davidson's theory is that it appears to entail a form of epiphenomenalism. Epiphenomenalism is the view that mental states are the effects of physical states but are themselves without any causal powers. Since Davidson has denied that the mental properties of events can figure in causal laws, it seems he is committed to the view that all events stand in causal relation in virtue of their physical properties. This means that mental properties are causally inert, for they make no contribution to the causal relationships between events.

The form of epiphenomenalism described here is called type-epiphenomenalism. This is contrasted with token-epiphenomenalism. According to token-epiphenomenalism, it is token mental events that are without any causal powers. This version of epiphenomenalism is ruled out by Davidson's token identity. Since token mental events are identical to token physical events and token physical events are, it is assumed, causally efficacious, it follows that token mental events are also causally efficacious. Type-epiphenomenalism, however, is another matter. This version of epiphenomenalism claims that it is mental properties, not mental events, that are without causal powers. It is difficult to see how Davidson can create a place for the causal role of mental properties given his account of strict causal laws. This is a serious difficulty because it appears as though anomalous monism does not offer an adequate account of mental causation. For if mental properties play no causal role in the production of behaviour, then it seems as though the fact that my mental state was the particular mental state it was (for example, a desire for a beer) has nothing to do with my getting up and going to the refrigerator. Since we ordinarily think that our mental properties have a crucial role to play in causing our behaviour, Davidson's theory becomes very unattractive in light of this problem.

Davidson remained silent about this particular criticism for years despite the fact that there arose a small industry of criticism based on this objection (Stoutland, 1980; Hess, 1981; Honderich, 1982; Honderich, 1983; Honderich, 1984; Stoutland, 1985; Klagge, 1990). In a recent article entitled "Thinking Causes," (Davidson, 1993) Davidson finally offered a response to this line of criticism. The main thrust of Davidson's response is that the objection is misguided because causation is a relation that holds between events no matter how they are described. In this case, it "makes no literal sense" to say that one event causes another in virtue of certain properties as opposed to others. Hence, the objection that anomalous monism entails epiphenomenalism is, in Davidson's view, completely unsubstantiated.

Davidson's critics remain unpersuaded by this line of argument and continued to insist that a proper account of causation must make reference to the properties of events, otherwise the account of causation implicit in Davidson's theory is not only mysterious but is also at odds with the nomological account of causality Davidson himself endorses (Kim, 1993; McLaughlin, 1995; Sosa, 1995). If we draw on some themes from the previous section, however, it seems as though there's good reason to suppose that Davidson's critics are in fact misguided (Campbell, 1997).

As we saw in the brief discussion of supervenience, there is a substantial difference in the way Davidson and Kim conceive of events. Kim thinks that properties are among the ontological building-blocks of events. In light of this it is only natural to suppose that events have the causal powers they do in virtue of the properties they possess. In this case, it is quite likely that some properties but not others are responsible for the causal efficacy of any given event. Those properties that do not make a causal contribution will therefore be epiphenomenal. Davidson, by comparison, is reluctant to treat properties as real items at all, never mind as ontological constituents of events. Talk of properties, from Davidson's point of view, is better understood as talk of predicates. That is, properties are linguistic items, are ways of describing events. According to Davidson there is nothing "in" events that makes it true that they can be described using certain predicates as opposed to ot hers. Thus, there are no recognition-transcendent facts about events that determine how they can be described. Since properties are not ontological parts of events, it makes no sense to say that events cause other events in virtue of certain of their properties. It seems, then, that the epiphenomenalist objection to anomalous monism is based upon a conception of properties and events Davidson himself does not endorse. In this case the objection begs the question against Davidson. For the objection to work one would have to show that Davidson must, in light of his other commitments, accept something like Kim's conception of events, but there is little reason to think this is the case.