Making Mind Matter More or Less



            There comes a time in every young philosopher’s career when he finds himself compelled to join some popular cause. Better late than never, I have decided to join the mental-causation-literature deforestation program. The issue, of course, is the causal efficacy of mental properties, and whether or not the current outbreak of epiphobia can be adequately treated if not cured.[1] As I read through the increasingly vast literature on this topic, I found myself bounced around a fair bit, wavering between dismissing mental properties as irredeemably epiphenomenal and rejecting the arguments for this conclusion as specious. As a piece of intellectual biography this is hardly interesting -- who am I, after all? But my self-diagnosis is, perhaps, worth noting. Not only are there at least three distinct argumentative strategies for establishing the inefficacy of the mental, there are at least four broad categories of views regarding the relation of mental properties to physical properties. And how effective the various argumentative strategies are depends on which account(s) of the relation of the mental to the physical is (are) on the table. I do not mean to suggest that anyone is confused on this point. I only mean to suggest that because different authors, and individual authors in distinct articles, deploy varying strategies against varying theories of mentality, the literature as a whole is hard to assess.

Part of my purpose in this paper is taxonomic: I want to distinguish among the various accounts of mental properties and determine which are subjected to the risk of epiphenomenalism by which arguments. And I want to organize the various responses to these arguments along similar lines. But I have a less modest goal as well. I want to defend a solution to the Exclusion Argument – in my view, the most fundamental argument for mental epiphenomenalism – that is available even to those who hold that the relation between mental and physical properties is extremely weak.


I: Mental and Physical Properties


            Before discussing the various accounts of mentality, a little stage setting is in order. I am going to be assuming that the relata of causal relations are events, where events are to be construed along Davidsonian lines as basic particulars. Moreover, I am going to assume that some but not all properties of events are relevant to the causal relations in which events stand to one another.[2] Finally, any talk of properties should be understood as referring to types. When I wish to refer to tokens, I will talk of “property instances” or “tropes” to make this explicit.

            Most theories of mentality take mental properties to stand in one of the following sorts of relations to physical properties: identity, functional, supervenience, and independence. Identity theories take mental properties to be identical to physical properties. Paradigmatically, they take mental concepts to be functional concepts, and the corresponding mental properties to be whatever physical properties occupy the requisite functional roles.[3] Suppose a physical property P is such that events which instantiate P stand in the sorts of causal relations specified by some mental concept MC. According to the identity theory, P would be identical to the property M determined by MC. As a result of worries about the multiple realizability of mental properties,[4] however, only species-relative versions of the identity theory remain popular.[5] On such views, because different physical properties – P1Pn – occupy the functional role specified by MC in distinct species – S1Sn – there is no unique property M determined by MC. Instead there are a number of distinct species-relative properties – MS1MSn – and each of these properties is identical to whatever physical property occupies the functional role in the species in question. 

            Functionalist theories, in contrast, identify functional roles not with mental concepts but with the mental properties themselves. This enables functionalism to reconcile the fact of multiple realizability with the existence of non-relative – i.e, species-independent – mental properties. But this prevents functionalists from identifying mental properties with the physical occupants of those roles. Instead they take the relation between these properties to be that of 2nd order to 1st order. A 2nd order property is the property of having a (1st order) property which satisfies some specification or other. And in the case of (2nd order) functional properties, the specification is that of occupying a certain causal role.[6]

            Supervenience theories claim that there is a logical (or, perhaps, metaphysical) relation between mental and physical properties but one which falls short of identity. Supervenience theories are characteristically motivated by the desire to reconcile materialism with the putative irreducibility of mental properties to physical properties. As such, they are regularly conjoined with material constitution theses.[7] Supervenience relations come in weaker and stronger varieties. Consider a pair of sets of properties M={M1, M2, …} and P={P1, P2, …}, where the Mi are the supervening properties and Pi are the supervenience base properties. The following are characteristic accounts of the nature of the relation in which these two sets of properties might stand:

Global Supervenience: For all pairs of worlds, w1 and w2, if w1 and w2 are P-identical then w1 and w2 are M-identical.  

If M stands in a relation of global supervenience to P, there are worlds in which two (or more) individuals are P-identical but M-distinct.

Weak Supervenience: Necessarily, for all pairs of individuals, x and y, if x and y are P-identical then x and y are M-identical.

If M stands in a relation of weak supervenience to P, individuals in distinct worlds can be P-identical but M-distinct.

Strong Supervenience: Necessarily, for all individuals, x, necessarily for all individuals, y, if x and y are P-identical then x and y are M-identical.

One might insist, of course, that none of these accounts of the relation between mental and physical properties suffices for materialism unless the supervenience relation is supplemented with appropriate metaphysical underpinnings.

            There are two fundamentally different theories of the relation between the mental and the physical which imply that the mental and the physical and logically independent. Property dualists claim that mental and physical properties are fundamentally distinct kinds of properties. Following Kim, we might recursively characterize physical properties as follows: (i) the properties and magnitudes that figure in basic physics are physical properties; and (ii) properties micro-based on physical properties are physical properties.[8] Given this characterization, property dualism can be cast as the conjunction of the following two theses: (a) there are mental properties; and (b) mental properties are not micro-based on physical properties.[9] The second sort of theory which implies that mental and physical properties are logically independent consists of those which embrace indeterminacy theses. According to such views, intensional facts – facts about beliefs, meaning, etc. – are unfixed by physical facts. And, since physical facts are the only facts that could fix them, indeterminacy of the intensional results. Indeterminacy views differ from property dualisms in two respects. First, they need not be committed to the full-blown reality of mental properties, although they are typically committed to the practical utility of mental predication. And second, property dualists can allow that modulo natural laws, intensional facts are fixed by physical facts despite being ontologically independent of them.[10]

II: Arguments for Epiphenomenalism


            There are central sorts of arguments that have been utilized to for the causal impotence of mental properties: the argument from mental-physical anomalousness; the argument from content extrinsic-ness; and the exclusion argument. My focus here is going to be on the exclusion argument; it is, in my view, the most fundamental and the most worrisome of the arguments. But it will prove to be instructive to contrast it with the alternatives. In order to make the contrast sharper, I will formulate each argument in terms of three (putatively) incompatible theses.

The argument from mental-physical anomalousness (AMPA) relies on the incompatibility of the following three assumptions:

Mental Property Relevance: mental properties are (sometimes) causally relevant to the physical properties of physical events.[11]

Mental-Physical Anomalism: there are no strict psycho-physical laws.

Strict-Law Causation: whenever two properties stand in a relation of causal relevance to one another, they are subsumed under a strict law.

The AMPA proceeds from this incompatibility and the (putative) truth of Mental-Physical Anomalism and Mental-Physical Anomalism to the falsity of Mental Property Relevance. It is worth noting that this argument poses no problem for the causation of physical events – events with physical properties – by mental events – events with mental properties. The problem is with whether the mental properties of mental events can be causally responsible for the physical properties of physical events. And that is what Mental-Physical Anomalism and Strict-Law Causation jointly prohibit.

In my view, this argument is the weakest of the three against Mental Property Relevance. The reason is that both Mental-Physical Anomalism and Strict-Law Causation are more contentious than Mental Property Relevance. The strict law account of causation puts not just the causal relevance of mental property at risk, but also renders impotent all physical macro-properties. And the motivation for Mental-Physical Anomalism – our practice of holistic mental property attribution subject to the constraints of rationality[12] – arguably has more to do with the epistemology of attribution than it does with the metaphysics of the attributed mental properties themselves.

            The argument from content extrinsic-ness (ACE) relies on the incompatibility of the following assumptions:

Mental Property Relevance: mental properties are (sometimes) causally relevant to the physical properties of physical events.

Methodological Solipsism: the only properties causally relevant to the behaviour of an intentional system are the intrinsic (i.e., non-relational) properties of the system.[13]

Wide Content: mental properties are relational properties of intentional systems.[14]

As above, the ACE proceeds from the truth of the latter two assumptions to the falsity of Mental Property Relevance. While certainly troubling, this argument is, in my view, less fundamental than the exclusion argument discussed below. Content properties are best thought of as 2nd order properties of 1st order mental properties. As a result, the (putative) fact that content properties are relational is compatible with the intrinsic-ness of their 1st order bearers and, hence, the compatibility of the causal relevance of these latter properties with Methodological Solipsism. Moreover, even if one insisted that content properties are themselves causally efficacious, any attempt to establish this fact would have to presuppose the causal relevance of 1st order mental properties. And this is exactly what is put at risk by the exclusion argument. 

            The three assumptions on whose incompatibility the exclusion argument (EA) relies are as follows:

Mental Property Relevance: mental properties are (sometimes) causally relevant to the physical properties of physical events.

Non-Identity: mental properties are not physical properties.

Causal Closure: physical properties of causes are always causally responsible for the physical properties of their effects.

Strictly speaking, the truth of Non-Identity and Causal Closure does not imply the falsity of Mental Property Relevance. After all, in any given case both a mental property and a physical property could be causally responsible for a single physical property exemplification, each of which would have been causally sufficient in its own right. But if mental causes are uniformly and without exception overdetermined, then mental properties make no causal difference in the world. Any claim that they are causally relevant nonetheless is hardly plausible. As I said above, I take this to be the most fundamental argument against Mental Property Relevance, and as such, it will be the focus of the remainder of the paper.


III: A Motley Assortment of Responses


            There are two distinct strategies that might be taken in response to the exclusion argument:  rejecting one of the assumptions; or arguing that, despite appearances to the contrary, they are compatible. It is worth pausing to consider the various ways one might go about implementing each of these strategies. The response advocates of the exclusion argument recommend is to reject Mental Property Relevance and embrace property epiphenomenalism. One might try to soften the epiphenomenalist blow by conferring on the mental a role in rationalizing explanations of behaviour, perhaps rationalizing explanations, in lieu of genuine causal efficacy. Any such manoeuvre will, however, prove unsatisfactory both to those who reject non-causal explanation and those who advocate a causal criterion of the real existence of properties.

An alternative would be to reject Non-Identity. As above, worries about the multiple realizability of mental properties can be alleviated by retreating to species-relative identities. The central worry about this suggestion is that the identities in question run the risk of triviality. In order for a set of species-relative properties – human pain, aardvark pain, octopus pain, etc., – to fall under a single mental concept, they need to all share a 2nd order property of some sort, perhaps a functional property. But it is an empirical question whether a single 1st order property will be the bearer of the requisite 2nd order property for all (or even a sufficiently large majority of) members of a given species. It could even turn out that the property which bears the requisite the 2nd order property varies not only with different members of the species, but with a single individual at different times. The shift from pain in general to human-pain and aardvark-pain, for example, may well leave mental properties untarnished, but a subsequent shift to Mary-at-t1-pain and Fred-at-t2-pain robs them of any interest (although not, of course, causal efficacy). As a result, anyone who thinks mental properties are not at such risk of trivialization will be loathe to reject Non-Identity.

            One might also reject Causal Closure on the grounds that it runs afoul of the causal efficacy of properties invoked in the special sciences.[15] After all, the argument goes, geological properties and chemical properties are causally efficacious despite being non-identical to strictly physical properties. And by parity of reasoning, the causal efficacy of the mental ought to be secure despite Non-Identity as well. Kim has argued, however, that the analogy drawn between mental properties and the properties invoked in the special sciences is misguided.[16] The exclusion argument puts the causal efficacy of mental properties at risk because mental properties are competing for efficacy with physical properties that are instantiated by the very same events – events at the same level on the macro-micro hierarchy. The only physical properties with which geological or chemical properties might be thought to compete for causal efficacy are properties of the micro-constituents of their bearers – entities at different levels on the macro-micro hierarchy. But since the bearers of these properties are distinct, macro-properties and micro-properties simply are not competing causes, in any direct sense, of the same events. Moreover, even though macro-properties are in some sense determined by micro-properties, they have their own causal powers over and above the causal powers of their micro-constituents.[17]

            The alternative to rejecting one of the assumptions of the exclusion argument is to establish that, despite appearance to the contrary, they are in fact compatible. There are two central strategies that have been deployed to this end: the supervenience strategy,[18] and the trope strategy.[19] The supervenience strategy relies on the thesis that supervening properties inherit the causal powers of their supervenience base properties. Typically, this causal piggybacking is thought to occur only in the case of strong (or stronger) supervenience relations. And paradigmatically defenses of it invoke counterfactual analyses of causation. Suppose a physical property P1 is causally responsible for a physical property P2, and a mental property M supervenes (in a sufficiently strong sense) on P1. Given that M supervenes on P1, the argument goes, if M had not occurred P1 would not have occurred. And given that P1 is causally responsible for P2, if P1 had not occurred, P2 would not have occurred. On this basis, it can be concluded that if M had not occurred P2 would not have occurred and, hence, that M is causally responsible for P2. I will return to this argument in my positive discussion below.

            The trope strategy relies on two presuppositions: property instances (tokens) and not properties (types) are the primary bearers of causal powers; and mental and physical property instances can be identical even though the properties themselves are distinct. As a result, mental property-instances which are identical to causally efficacious physical property-instances are themselves causally efficacious. And mental properties, as with physical properties, acquire their causal relevance from the causal relevance of their instances. There are, of course, a number of worries that arise for the trope strategy.[20] Below I will take up the question of what relation properties must stand in to one another if their instances are to be identical.


IV: Positive Reflections


            In reflecting upon these matters, we will be well-served to have on hand a more or less uncontroversial example in which a mental event causes a physical event and, more to point, one which is paradigmatic of the sorts cases in which mental properties are putatively causally responsible for physical properties. In order to be able identify a specific mental event as a cause, the case will be one in which there is an occurrent desire in the context of a standing belief. The example also has the virtue of being non-hypothetical.


A Trip to the Seahorse Tavern:

In the mid-1980’s, a group of friends and I were regular patrons of the Seahorse Tavern in Halifax, Nova Scotia. And the only thing any of us would ever order was Oland Export Ale in the bottle. Moreover, this fact about us was well-known to Frank Domini, the regular bartender at the Seahorse. As a result, whenever Frank was working, all I or any of my friends needed to do to order an Oland was to raise a single finger (and if we wanted to order two Olandswe needed only to raise two fingers, and so on). And so whenever the desire for an Oland arose, this was exactly what we would do. As you might imagine, this came in quite handy on a Friday night when the Seahorse staff had the stereo blaring at full blast.


Let me re-describe one such occasion. I had, throughout the evening spent at the Seahorse, a standing belief: if I wanted a bottle of Oland, I could acquire (or, at least, order) one by means of raising a single finger. A mental event occurred – I formed the desire for a bottle of Oland. And this event caused a distinct physical event – one in which the index finger of my right hand became extended and pointed upwards.

            For present purposes, I am simply going to take it for granted that the first event really did have the property of being a desire for a bottle of Oland and that it – the event – really was the efficient cause of the event of my finger raising. The question is whether the property of being a desire for a bottle of Oland is causally responsible for the property of being a raising of my finger. The problem, of course, is that the first event instantiated some neuro-physiological property or other – call it the “NP-property” – which was by itself causally sufficient for the instantiation of this latter property.

            What I want to argue here is that neither the supervenience strategy nor the trope strategy will secure the causal relevance of the property of being a desire for a bottle of Oland (or the DO-property) to the property of being a raising of my finger (the FR-property). Moreover, I want to argue that this is unsurprising and, more contentiously, that it is compatible with the real existence of the DO-property, even if the causal criterion for the existence of properties is correct. This is because even if it the DO-property is not causally relevant to the FR-property, it does play a causal role in others’ thinking about me.

            Before proceeding on this course of argument, it is worth pausing to say something about the relation between the DO-property and the NP-property or, more generally, the relation between mental properties and the physical properties with which they putatively compete for causal efficacy. And in my view, this relation mirrors that which holds between actions and their constitutive behaviours. Consider: the effect of my desire for an Oland – my finger-raising behaviour – was at the same time an Oland-ordering action. It was an event which had both the FR-property and the property of being an action of ordering Oland (the AOO-property). There are two important things to note about this fact. First, this event counts an action at all only in virtue of being caused by my desire for an Oland, and it counts as an act of ordering Oland only in virtue of features of the social context in which it occurs. Second, the finger-raising event counts as an act of ordering in virtue of features of the FR-property of the event. It is because the FR-property was exemplified by a member of my group of friends (namely me) in the Seahorse Tavern on a night when Frank Domini was working that the event of my finger raising counts as an Oland-ordering action, regardless of whatever other properties it might have. If this is right, the upshot is that the relation of the AOO-property to the FR-property, and more generally of actions to behaviours, is that of 2nd order to 1st order property.[21] As noted above, it is reasonable to suppose that the very same relation holds between the DO-property and the NP-property. Moreover, as with the FR-property, we can distinguish between the features that make it true that the NP-property is (in the sense of predication) a desire at all – presumably its causal relations to other bearers of mental properties – and those features which make it a desire for a bottle of Oland Export Ale.


V: The Supervenience Strategy


            My argument that the supervenience strategy is inadequate will rely on the following two claims: the supervenience strategy is tenable only if mental properties supervene specifically on the causally efficacious properties of mental events (call this specific supervenience); and mental properties do not (even weakly) supervene specifically on those properties. For the purposes of this discussion I am going to assume a counterfactual analysis of the causal relevance of properties:

Property P of event e is causally relevant to property Q of event f iff if e had not been P, f would not have been Q

In addition, I am going to utilize a possible-worlds analysis of the truth conditions of counterfactuals:

“If it had not been the case that P, it would not have been the case that Q” is true iff there is at least one [~P&~Q]-world which is more similar to @ (the actual world) than any [~P&Q]-world.[22]

Now let us suppose that (i) mental properties strongly supervene on physical properties but (ii) they do not supervene specifically on the causally efficacious physical properties of mental events. At the time of my desire for an Oland, I instantiated not only the NP-property, but also a set of physical properties, P1, P2, …, Pn. My physical state at this time can be represented as {NP + P1 + … + Pn}. Strong supervenience implies that no individual in any possible world can be in this physical state but fail to instantiate the DO-property, that is, there is no possible individual in the following state: {NP + P1 + … + Pn + ~DO}. But if specific supervenience is false, an individual can both instantiate the NP-property and fail to instantiate the DO-property, as long as they fail to instantiate at least one of my other physical properties. That is, there are possible individuals in states of the following sort: {NP + P1 + … + ~Pi + … + Pn + ~DO}.

            Since the NP-property is causally responsible for the FR-property, we know that if the NP-property had not been instantiated, the FR-property would not have been instantiated. And given the analysis of counterfactual on the table this means that there exists a [~NP + ~FR]-world that is more similar to @ than any [~NP + FR]-world.[23] In order for the DO-property to be causally relevant to the FR-property, it must be the case as well that if the DO-property had not been instantiated, the FR-property would not have been instantiated. And this requires that there be a [~DO + ~FR]-world that is more similar to @ than any [~DO + FR]-world. Now given the causal relevance of the NP-property, this condition will be met only if there is a [~DO + ~NP] world that is more similar to @ than any [~DO + NP]-world.[24] But given the sheer number of physical properties I have, any plausible similarity metric will imply that there is some physical property, Pi, of mine such that a world in which my state is represented by {NP + P1 + … + ~Pi + … + Pn + ~DO}is more similar to @ than any world in which my state is given by {~NP + P1 + … + Pn + ~DO}.[25]

            What remains to be established is the failure of specific supervenience. And this is not hard to do. All that is required is to show it is possible for someone (me or someone else) to instantiate the NP-property without instantiating the DO-property. (Moreover, one could establish the falsity of even weak specific supervenience by showing that someone in fact has instantiated the NP-property without instantiating the DO-property.) Exactly how one goes about showing this depends, of course, on the account of mental properties on the table. Suppose, for example, that mental properties are functional properties. In this case, one could establish the falsity of specific supervenience by noting the possibility of an individual for whom the NP-property failed to occupy the Oland-desiring functional role. An analogue of David Lewis’ madman would do the trick.[26] This possibility would not suffice, however, if an individual exemplified a mental property just in case she was in a state that occupied the requisite functional role for a sufficient large percentage of the population of which she is a member, whether or not it occupies the causal role in her own case. To falsify specific supervenience in this case, one would need to invoke the possibility of an individual who is (i) capable of instantiating the NP-property and (ii) is a member of a population in which the NP-property does not occupy the Oland-desiring functional role.[27]  There seems to be little that can be said in defense of specific supervenience.[28]


VI: The Trope Strategy


            The failure of specific supervenience undercuts not only the supervenience strategy but the trope strategy as well. The problem is that this failure renders untenable the hypothesis that mental and physical property instances are identical despite the distinctness of the properties themselves. My argument here will rely on two assumptions:

(1) For all individuals x and y, if x is identical to y then it is not possible for x to exist and y to fail to exist.

(2) An instance of a property P cannot both exist and fail to be an instance of P.

Assumption (1) follows from the Indiscernibility of Identicals and as such is uncontroversial. Assumption (2), however, requires some comment. The central idea is that what it is to be a property instance just to be an exemplification of the property in question. If one wanted to reify property instances, as tropes theorists would have it, the point could be made as follows: blueness is an essential property of a blue trope.

            At a certain time during my visit to the Seahorse Tavern the NP-property and the DO-property were co-instantiated. Let’s label the instance of the NP-property “NPSH” and the instance of the DO-property “DOSH”. Given the failure of specific supervenience, I could have instantiated the NP-property without instantiating the DO-property. More strongly, however, I want to suggest that NPSH could have existed even if the DO-property had not been exemplified. Exactly how one goes about establishing this point depends, of course, upon how property instances are individuated. If property instances are thought of as triples consisting of properties, property bearers, and times, one need only establish the possibility of my instantiating the NP-property at the requisite time without instantiating the DO-property. The failure of specific supervenience by itself establishes this. And if property instances are thought to be individuated, in addition, in terms of their causal origins one need only establish the possibility of my instantiating the NP-property at the requisite time without instantiating the DO-property, where the exemplification has the requisite causal history. This is not hard to do even if one takes mental properties to be functional properties. After all, a property can fail to be the realization of a certain functional role in a given object even though a given exemplification of it by the object has a causal history that that conforms to (or is compatible with occupying) the role.

            So it has been established that there are possible circumstances in which NPSH exists but the DO-property is not exemplified. Assumption (2) implies that these are circumstances in which DOSH does not exist. And since there are circumstances in which NPSH exists and DOSH does not, assumption (1) implies that NPSH is non-identical to DOSH. Since the trope strategy relies on relies on such identities, the trope strategy fails.


VII: Benign Repercussions


            So where does this leave us with regard to the exclusion argument? If no reconciliation of the three assumptions is forthcoming, one of the assumptions will have to be rejected. And in my view, it is Mental Property Relevance that is false. What I want to suggest in this section is that not only is this entirely unsurprising but that it is also compatible with the real existence of mental properties.

            Mental properties are attributed to objects on the basis of (and in order to explain) patterns in their behaviour, without regard to their internal physical structure. And such attributions are accurate (in my view) if, and only if, the behavioural patterns are produced by other “…real pattern[s] roughly isomorphic to [them] within the brains of intelligent creatures.”[29] But this is compatible with the production of identical patterns of behaviour being by distinct sets of physical properties in objects with fundamentally different internal structures. To attribute a mental property to two such distinct physical events is not to suggest that they a single causally efficacious property, but to taxonomize the events in terms of similarities between their distinct causally efficacious properties. And, of course, none of this should be taken to suggest that mental events are not causally efficacious – they are just not causally efficacious qua mental.

            Consider, by way of analogy, genetic properties.[30] These properties were originally attributed with an eye towards explaining trans-generational patterns in the distribution of heritable traits in living organisms. Now as it turns out, genetic roles are occupied by the same sorts of properties – DNA-properties – in almost all (heretofore encountered) living organisms.[31] As a result, there is a temptation to simply identify genetic properties with DNA-properties. It could (epistemically) have turned out, however, that genetic roles were occupied by very different properties in different organisms, or we could in the future happen across organisms with the same patterns of heritable traits but different physical constitutions. Exactly what (would or) would have become of “gene” talk in such circumstances is an issue in the sociology of science. Scientists might have taken all the structurally distinct organisms to have genes, they might have retreated to talk only of species specific genes, or they might have decided that there are no such things as genes after all. But if the first option had come to pass, there would have been little grounds to say genetic properties are causally efficacious. As long as genes were causally efficacious, it would hardly matter whether they were efficacious qua genetic.

            The question that remains, however, is whether mental properties ought to be taken to (genuinely) exist at all if they play no causal role is the behaviour of their bearers. After all, the causal criterion of the existence of properties is eminently plausible. It might be suggested that mental properties can satisfy this criterion as long as they satisfy some other causal role. They might, for example, play a causal role in observer thinking about the behaviour of their bearers. Even though the DO-property was not causally responsible for the raising of my finger, it certainly played a causal role in Frank Domini’s thinking about me; it was only because he recognized that I was instantiating this property that he proceeded to provide me with a bottle of Oland. The trouble with this suggestion is that, for all that has been shown, the deployment of mental concepts in our thinking about others may be inessential, or at best a convenient shorthand. If, as with genetic concepts (at least under the sorts of circumstances described above), the deployment of mental concepts is merely a matter of choice, the fact that we do deploy them can offers little solace to the realist about mental properties.

But maybe we can do a little better. Let me, for a moment, engage in an evolutionary fantasy. Suppose there was a world populated by a wide variety of different sorts of organisms with fundamentally different internal structures. And suppose that despite marked differences among the behaviour producing causal mechanisms in the various species, these mechanisms all follow a very similar pattern. If the members of these various species were to compete for scarce resources and engage in mutual predation, having the ability to predict the behaviour of one’s competitors would have survival value. Now if an individual organism had concepts corresponding to the causally efficacious properties in a given species, it could compete successfully with members of that species but not against members of the others. And if an individual possessed concepts corresponding to the causally efficacious properties of all the species, it could compete successfully with members of all of them but only at the cost of a large loss in cognitive efficiency. But an individual organism possessed (and could deploy) concepts corresponding to the pattern properties, it could efficiently and successfully compete with members of all the various species. As a result, the innate possession of such concepts might well be selected for in this world. And if it were, the pattern-properties would have been causally responsible for important features in the cognitive architecture of members of a certain evolved species. On this basis, pattern properties would have a strong claim to genuine existence, in my fantasy world. But if those evolved organisms are us and the pattern properties are mental properties, that is, if this is evolutionary fact not fantasy, then mental properties have an equally strong claim to real existence.


VIII: Concluding Remarks


Much on the ongoing discussion of mental causation can be traced to Davidson’s defense of anomalous monism, in which he showed us how mental events could cause physical events.[32] Central to his view was his claim that our attributive practices – holistic and subject to the constraints of rationality – resulted in the anomalism of the mental. Davidson was (and continues to be) subjected to much criticism because his view seemed to imply that the mental qua mental was causally inefficacious. We seem to have ended up, however, in a position much like Davidson’s, albeit via a rather different route. External behaviour is often the product of internal mental events. But the mental properties of these events are not responsible for this behaviour. The role of the mental qua mental is restricted to observer cognizing of this behaviour. My difference with Davidson is that I think all of this is compatible with the causal efficacy of mental properties. It is just that the causal role I think mental properties (may) play is orthogonal to that which they are ordinarily thought to play. But they matter, more or less.



Peter Alward

Department of Philosophy

University of Lethbridge

4401 University Dr.

Lethbridge, AB

Canada T1K 3 M4







[1] See Jerry Fodor, ‘Making Mind Matter More,’ Philosophical Topics 17, pp. 59-79, 1989, p. 59 for a definition.

[2] If event c causes event e, this is because c has property P: the fact that c causes e is explained by the fact that C is P. The explanation relation between these two facts should not be presumed to be a singular causal relation although it may be underwritten by causal generalizations involving event types.

[3] See, e.g.,  J.J.C. Smart, ‘Sensations and Brain processes,’ Philosophical Review 68, 1959.

[4] See Hilary Putnam, ‘Psychological Predicates,’ in Art, Mind, and Religion, W.H. Capitan and D. D. Merrill, eds., University of Pittsburgh Press, 1967, pp. 37-48.

[5] See, e.g., David Lewis, ‘Mad Pain and Martian Pain,” in Readings in the Philosophy of Psychology, Vol. 1, N. Block, ed., Harvard University Press, 1980, pp. 216-222.

[6] Jaegwon Kim, ‘Does the Problem of Mental Causation Generalize?’ Proceedings of the Aristotelian Society, 97:281-297, 1997, p. 290.

[7] See, e.g., Lynne Rudder-Baker, Explaining Attitudes, [ref]

[8] Kim, (1997), pp. 292-4. A property, P, is micro-based on physical properties iff P is the property of having proper parts a1, a2, … an, such that P1(a1), P2(a2), … ,Pn(an) and R(a1, … , an), where P1, P2, …, Pn, and R are all physical properties.

[9]  Non-reductive materialists might object here that their views satisfy these two conditions but do not count as property dualism. To alleviate this worry, one might require in addition of dualism that that it involve the rejection of even global supervenience.

[10] Donald Davidson, in ‘Thinking Causes,’ in Mental Causation, John Heil, ed., New York: Clarendon Oxford Press, 1993, pp. 3-17, argues that anomalous monism is committed to the supervenience of the mental on the physical. This seems at odds with his earlier defense of indeterminacy. 

[11] See footnote #2 above.

[12] See Donald Davidson, ‘Mental Events,’ in Experience and Theory, L. Foster and J.W. Swanson, eds., University of Massachusetts Press and Duckworth, 1970.

[13] See, e.g., Jerry Fodor, ‘Methodological Solipsism Considered as a Research Strategy in Cognitive Psychology,’ Behavioral and Brain Sciences, 3, 1980.

[14] See, e.g.,  Hilary Putnam, ‘The Meaning of “Meaning”’, Mind, Language, and Reality: Philosophical Papers, vol. 2, Cambridge University Press, 1975 and Tyler Burge, ‘Other Bodies,’ in Thoughts and Objects: Essays on Intentionality, A. Woodfield, ed., Oxford University Press, 1982.

[15] See, e.g., Tyler Burge, ‘Mind-Body Causation and Explanatory Practice,’ in Mental Causation, J. Heil and A. Mele, eds., Oxford: Clarendon Press, 1993 and Lynne Rudder-Baker, ‘Metaphysics and Mental Causation,’ in Mental Causation, J. Heil and A. Mele, eds., Oxford: Clarendon Press, 1993.

[16] Kim, 1997.

[17] Kim (1997), p. 292.

[18] See, e.g., Nick Zangwell, ‘Good Old Supervenience: Mental Causation on the Cheap,’ Synthese 106:67-101, 1996.

[19] See, e.g., David Robb, ‘The Properties of Mental Causation,’ The Philosophical Quarterly 47: 178-94, 1997.

[20] See, e.g., Paul Noordhof, ‘Do Tropes Solve the Problem of Mental Causation?’  The Philosophical Quarterly, 48:221-226, 1998.

[21] This result prevents a defense of the efficacy of mental properties which claims that they are not causally responsible for physical properties but only for action properties. If action properties are 2nd order properties of 1st order physical properties, exemplifications of the former can only be caused by means of causing exemplifications of the latter.

[22] This is based on the analysis in David Lewis, Counterfactuals, Cambridge: Harvard University press, 1973.  Note: a P-world is a world in which P obtains.

[23] Note: a [P1 + P2]-world is a world in which I instantiate P1 and P2. 

[24] I am assuming that if the NP-property had not been instantiated, no other physical property causally sufficient for the FR-property would have been filled the causal breach.

[25] This remains true if we restrict our attention (as we should) to those physical properties not included among the causal conditions under which the NP-property works its causal magic.

[26] David Lewis, ‘Mad Pain and Martian Pain,’ Readings in the Philosophy of Psychology, Vol., Ned Block, ed., Harvard University Press, pp. 216-222, 1980. Lewis’ madman is someone for whom pain is caused by moderate exercise on an empty stomach and causes him to cross his legs and snap his fingers, aw well as turning his mind to mathematics.

[27] Of course, how one goes about specifying some such individual will depend on the account of population membership on the table.

[28] It might be thought that since mental properties are strictly speaking properties of neural events rather than persons, specific supervenience might be salvageable. The trouble is that since mental properties are relational (perhaps functional) properties, the supervenience of mental properties on the physical properties of neural events is a non-starter.

[29] Daniel Dennett, The Intentional Stance, Cambridge: MIT Press, 1987, p. 34. Dennett would disapprove.

[30] Similar observations could be made regarding valence properties.

[31] Viral genes, I am told, are an exception. They consist of strands of RNA rather than DNA.

[32] Davidson, 1970.