www.richmond-philosophy.net - © COPYRIGHT 2002-17 RUTC - Designed and maintained by Mat Carmody.

Mental Causation and the Exclusion Problem

Paul Sperring

I have just typed a meaningful sentence. And now you are reading another. Could I have typed these sentences if I had no mind? Most people would say no, and with some justification.[1] To put the question in a different form, was my mind, at least in part, causally involved in that sentence having been typed? If you think that this is not only a plausible claim but overwhelmingly likely to be true then you think that minds can shape the world - or at least certain parts of it. What I will attempt to reveal in this paper is that a seemingly sensible view in the philosophy of mind - a view called non-reductive materialism - might actually be thought to entail the denial that minds make a causal difference to the world.[2]

Supposing that you agree with this proposition:

(a) Typing meaningful sentences is caused, at least in part, by having thoughts.

Now think about which of these other propositions you agree with:

(b) Physical properties are sufficient for the bringing about of all physical events - for example, the sulphurous properties of the match-head, in the right sort of combination with the properties of the match box surface, added to the properties of the surrounding oxygenated atmosphere, are sufficient for the match's lighting. If all of these properties are sufficient for the match's lighting, then nothing else is causally relevant to the match's lighting.

(c) It is by means of the operation of functioning brains that thinking is possible. Destroy my brain and you destroy my capacity to think.

(d) Thoughts, while having a physical basis, can't simply be reduced to physical states, or described purely in the language of physics, because they have some features, such as intentionality or qualia [3], which resist such reduction/description.

If you agreed with (b) then you know something about physics (you may even know something about the causal completeness of physics). If you disagreed with (c) then perhaps you are Descartes, or at least someone with similar views to Descartes - not a popular position to adopt these days, but one with a good deal of heavyweight philosophical support historically. If you are a person who agreed with (c) however, and also (d), then you are probably either a non-reductive physicalist, or a property dualist.[4] Perhaps you think it matters which of these you are called, and perhaps you are right. But you might find it hard to come up with substantive differences that mark out your (qua property dualist) position from your sworn philosophical enemy's (qua non-reductive physicalist) position, and so might they. This won't matter for our purposes.

It may, however, turn out to be something of a surprise, for both you and your sworn philosophical enemy, to discover that you are going to have to give up on your commitment to the eminently sensible (a) if you want to hold on to (b) and (c) and (d). It may surprise you, that is, to learn that you are epiphenomenalists [5] - or at least that you are committed to positions that entail epiphenomenalism. Of course, it is possible that you agreed with (b) and (c) and (d) but didn't feel very strongly attached to (a) - in which case you are very probably David Chalmers.[6] Since you are most likely not David Chalmers, then you will resist the idea that (a) is something your position entails the need to deny. Of course, you might be puzzled why your position entails the falsity of claim (a), so perhaps it would be a good idea to make plain the reasons why you are committed to epiphenomenalism if you are committed to (b), (c) and (d).

Non-Reductive Physicalism and the Exclusion Problem

Before attempting to show why it might be thought that non-reductive physicalism entails epiphenomenalism I need to say a bit more about claim (b) above. Claim (b) has two distinct, although related, claims within it. The first, recall, is about what is causally sufficient for something else. Physical events, the claim said, require only the presence of physical properties in order that they come about (those properties are sufficient). We can call this claim, the Physical Determinism claim, which can be stated more precisely in the following way:

Physical Determinism (PD): For every physical event y, some physical property X is causally sufficient for y's occurrence.[7]

Recall our match bursting into flames in (b). The only properties we need to specify as being the "bringers about" of the bursting into flame of the match are physical properties. These physical properties are simply enough for the event to come about. However, we also said in claim (b) that whatever is sufficient for the cause excludes elements other than those causally sufficient elements that are responsible for the effect. This is a sort of "is enough" principle. To illustrate, if I have shot you through the temple with a very powerful gun and then, shortly after, run you over, then since the shooting was sufficient for the killing of you, then the running-you-over is not causally relevant to the killing of you. We can exclude the running over in our causal story (at least with respect to the story we tell about the cause of your death). We can call this claim the Exclusion claim, and it can be stated as follows:

Exclusion (E): If a property X is causally sufficient for an event y's occurrence, then no property X* distinct from X is causally relevant to y's occurrence.[8]

Now it might suddenly occur to you that there is an obvious counterexample to (E). Suppose as I shoot you some highly poisonous and fatal serum that I injected you with prior to the shooting is just taking effect and stops your heart just at the point that the bullet enters your brain. (E) seems wrong in this case since it isn't quite right to say that the shooting is irrelevant to the killing even though the heart-stopping poison is sufficient for it. This case involves what is called overdetermination. Overdetermined effects are those that have multiple and separate causes, such that even if one of the causes were not present the effect would still come about given the presence of one or more of the other causally sufficient sets of properties. Overdetermined effects are probably not all that common (as my example perhaps suggests). What seems most likely is that they are exceptional rather than the rule - in short, it seems intuitively sound to think that there is no systematic overdetermination of effects.[9] This would mean that (E) is general enough to serve our purposes. So let's move on.

We are nearly ready to reveal why the non-reductive physicalist (who accepts (E) and (PD)) is committed to the denial of claim (a) set out at the very beginning of this paper - to wit, that mental events, or properties, have some causal efficacy. But first we should state what non-reductivism is committed to rather more carefully.

Non-reductivism (NR): For every physical property X and mental property X*, X is distinct from X*.

A couple of questions need asking here. Firstly, what does it mean to "reduce" something to something else? Secondly, why can't one reduce mental properties to physical properties?

In answer to the first question, when we talk of reducing things of type 1 to things of type 2 we might mean that things of type 1 just are, suitably understood, things of type 2; or we might mean that things of type 1 are better explained by appealing to type 2 type things. The first sort of view of reduction is what is called an "ontological reduction" - that at bottom the things of seemingly different types are the same sort of being. A well-worn example of an ontological reduction would be the reduction of heat to mean molecular kinetic energy (heat just is mean molecular kinetic energy). The second is called "explanatory reduction" - that the two types of things could be brought under one single type of description or explanatory discourse. An example here would be the attempt to describe biological entities (such as cells, for instance) in chemical terms - biological talk is reduced to chemistry talk. Of course, these two types of reduction needn't be thought to be wholly separate. The reason why things of type 1 can be explained by appeal to things of type 2 could just be because there is ontological sameness here. Of course, even where there is ontological sameness of things of type 1 and 2, it needn't follow that they can be explained by appeal to one sort of discourse. And this brings us our second question about (NR).

Mental properties could be thought to be either ontologically or explanatorily irreducible. To deny ontological sameness is to commit to something like a dualism (or perhaps pluralism) regarding properties. However, some philosophers who do not deny the "at-bottom-ontological-sameness" of mental and physical properties do not think that an explanatory reduction is possible. This may be because of a semantic incommensurability of the two sorts of predicates used to describe mental and physical states - the idea here is that the sentence "Bob desires a cup of tea" and some other sentence, S (where S is a very long sentence about the atomic structure of both Bob and a big enough portion of Bob's physical environment to include cups, tea-leaves, and so on), just couldn't possibly mean the same thing, no matter how one tries to square it. It might be that talk at the physical (say, micro-structural) level just leaves out lots of important features seemingly present at the level of the mental, or the level of persons (even though both mental entities and persons might just be physical beings). So to say, as (NR) does, that mental and physical properties are distinct, is to say that no mental property can be reduced to a physical one.

Having sketched out, albeit in a rough and ready way, the reasons why someone might agree with claims (b) - (d), we are ready to explain why these claims, taken together, appear to rule out acceptance of claim (a).

Recall that in accepting claim (b), about the match's bursting into flames, we were actually committed to accepting two principles, (PD) and (E). (PD) said physical properties were sufficient for causing physical events, and (E) said that nothing else, apart from the sufficient cause, could be relevant to some event y's occurrence. Claim (c) was a commitment to the physical basis of thought, and claim (d) pulled back from commitment to a full blown physicalist thesis that identified mental and physical properties. Now, let us suppose that (a) is true, that my thought causes the typing of a sentence. The typing is undeniably a physical event. Given (PD) it must, therefore, have a sufficient physical cause, and given (E) nothing else is relevant to the causing of the event. Now, one might be tempted here to say that the thought must be just one of the physical bits that make up the cause, but (NR) forbids this move, since, it says, no mental property is a physical property. But (a) now looks in trouble, because the thought that we said was the cause of the typing, at least in part, is ruled out completely as irrelevant to the causing of the typing. In short, commitment to(E), (PD) and (NR) gives us:

Epiphenomenalism of the mental (EM): For every physical property X, no mental property X* is causally relevant to y.

So, says (EM), no mental property can make a difference to the physical realm. Thus, on this view, my thinking of typing this sentence is not, after all, what makes this sentence come to be.

This has all sorts of worrying consequences. Norman Malcolm says in "The Conceivability of Mechanism" that if we accepted something like (EM) then we should have to say that, for instance, no one ever speaks or acts.[10] And Jerry Fodor at the end of his "Making Mind Matter More" says if (EM) were true, then "practically everything I believe about anything is false and it's the end of the world."[11]

So much for mental causation. There is, however, no end to the sorts of things that might be edged out of the causal efficacy sphere by applying a generalised version of the Exclusion claim (E).

Take a non-mental property such as the property of being crumbly. Let's suppose that we wanted to say that it was in virtue of being crumbly that the shortbread biscuit left a residue of crumbs in Bob's beard. Crumbliness is a property enjoyed not just by shortbread biscuits, but by other sorts of biscuits as well, and also by certain cheeses, bits of chalk, dried limestone, and so on. But can we impute to the crumbly things any causal powers qua crumbly things since crumbliness is surely realised in each of these cases by the underlying microphysical properties. Crumbliness is, akin to functional mental properties, not identical to some particular type of physical property.[12] Crumbliness is a multiply realisable property. But it seems then that we can run the exclusion argument on the biscuit rendering it inert. That's to say, neither the biscuit nor it's crumbliness get a look in since they are edged out of the causal scene by the basic microstructural features of the biscuit.

Now, you might not like the example - it might appear to be of no great consequence since we lose very little of import in sacrificing these sorts of macro properties, so perhaps we are prepared to be reductive here, or maybe even to embrace eliminativism about crumbliness.[13] But it looks just as easy to run the argument for colours, say, such that the redness of the ball isn't causally efficacious in you picking it up when your are asked to pass the red ball. Indeed, take any macroproperty you like that isn't strictly identical to its microphysical base and, given the application of (E), one will find causal exclusion pretty much everywhere we look.

So, how should one respond to the exclusion argument about mental properties (putting aside for now the broader version)? Maybe one ought to give up on the commitment to non-reductive physicalism - that is, if we reduce the mental properties to their physical bases then they retain their causal relevance, since they are just the same as the physical properties (which we know have some causal oomph, don't we?). Well, maybe. Let us, however, consider an attempt to grapple with the exclusion problem from the perspective of the non-reductive physicalist.

A Way Out?: Yablo and Determination

Stephen Yablo (1992) argues that anything that is to count as an adequate solution to the Exclusion problem will have to satisfy certain constraints. These are as follows:

Mental Realism constraint: Keep the higher level intact (i.e. remain committed to (NR))
Explanatory constraint: Explain how the higher level is related to the lower level.
No Causal Conflict constraint: Have the levels working harmoniously with respect to causal responsibility.

Stephen Yablo's account, on the face of it, scores highly on all three in his paper "Mental Causation."[14]

Yablo takes it that mental properties are irreducible to physical properties because multiply realisable (this meets contraint (1))[15]. He also thinks that the mental supervenes on the physical (which, given a full understanding of such a relation, would meet contraint number (2)). What is does it mean for something to supervene upon something else? Briefly, the supervenience relation can be characterised as a one-way necessitation relation - which means that if the subvenient set of properties P (physical properties) exist then the supervenient property M (mental property) cannot fail to exist, but it needn't follow that if M exists that there also exists the P properties. An illustrative example of supervenience in a different setting would be the aesthetic properties of a sculpture. The sculpture is made up of physical properties, P, and has the supervenient property, B (its beauty, say). Now anything that had identical physical properties to our original - that is, anything that is P - would also have B - would be beautiful too.[16] However, suppose something had just that particular sort of beauty; would it follow that it must have just the same set of physical properties P? No, since it could be composed out of a set of physical properties P1, not identical to P (although where P and P1 are indistinguishable to any observer), and thus no particular set of physical properties is necessitated by the existence of a supervenient property.

So, (NR) is maintained, according to Yablo's account of the mental/physical relation, and we at least have some sort of understanding of the manner in which they are related. But, this doesn't, on the face of it, help us, since it looks as if the Exclusion worry is going to threaten the so-called supervenient properties with causal redundancy. However, it is exactly the one-way relation delivered by the supervenience view that is going to be useful in explaining why the mental realm isn't going to get pushed out of the causal relevance sphere by the physical realm. Mental properties, Yablo argues, are related to physical properties as determinables to determinates.[17]

Determinates - such as being scarlet or azure necessitate their determinables - being red or blue. But being blue doesn't necessitate the property of being azure - something can be blue and cerulean. This clearly illustrates the necessitation relation's one-way ticket. To take a different sort of example, Socrates' guzzling of the hemlock (determinate) necessitates his drinking of the hemlock (determinable). But drinking the hemlock doesn't guarantee any guzzling - since Socrates might have sipped.

Now, does the exclusion argument gain any purchase here? Perhaps if the guzzling did the work in killing Socrates (and it looks to be causally sufficient) then is the drinking excluded? Clearly not. If a monkey is trained to push a button when it sees a blue object and it pushes the button when it sees an azure object, does the fact that the object was blue drop out as unnecessary? No, says Yablo. By way of illustration, if I happen to tread on your toes, causing you pain, then the causal relevance of my treading is not ruled out because of the work done by my foot. In the same way that when my foot is occupying a particular space and filling it completely this doesn't mean that my toes are crowded out.

In this competition wholes and parts are not on opposing teams; hence any principle that puts them there needs rethinking. Likewise any credible reconstruction of the exclusion principle must respect the truism that determinates do not contend with their determinables for causal influence.[18]

This seems to satisfactorily deal with the third constraint, and we are well on the way to dealing with (E) and thus avoiding the dreaded (EM).[19] Further, the determinate/determinable relation looks to be useful in helping to satisfy Yablo's first constraint (preserving the distinctness of mental and physical properties). This is because if mental properties are determinables for their physical determinates then if we don't have a problem saying that an object is both really coloured and really red then it's not at all obvious why we shouldn't accept the mental properties along with the physical ones.

What about the second constraint? Does the determinable/determinate relation do the explanatory work? Well, to say that X and Y co-vary, one way and of necessity is not to explain why X and Y are related thusly. Perhaps there is no explanation required here - it's just a brute fact that whenever there's an X there's a Y. But just as it seems reasonable to ask for an explanation for the supervenience relation it seems perfectly reasonable to ask how determinates and determinables are related - and there is some controversy here.

Further, even supposing that there is an adequate explanatory account of such a relation there is a deeper worry. Are mental properties related to physical properties as determinables to determinates?

On the first worry. It isn't clear that supervenience and multiple realisability are enough on their own to establish a determination relation. Yablo himself admits that asymmetric necessitation does not entail determination, although it is "extremely suggestive."[20]

Also, if the mental and physical are related in this way we should expect them to behave in ways consistent with the ways that other determinables and determinates behave. Is this the case? Douglas Ehring has put forward a number of reasons for casting doubt here. One of those reasons is as follows.[21] Determinates are subject to an "orderability " which says that different determinates of the same determinable may be ordered according to their relative differences and similarities. So, a red object is more similar to an orange object than it is to a yellow object with respect to colour. Now, take the determinable, "being in pain". We could order the same level determinates in the following way, an aching pain is more similar to a strain pain than it is to a tearing pain. Now, each of these is realized by some physical property and it isn't at all clear how the comparable physical differences of the realisers correspond to the comparable differences in the pains with respect to the one determinable "being in pain". Which is to say, it will be by no means necessary that the ordering of the physical properties will correspond to the ordering of the mental properties (physical property types p1, p2, and p3, may be ordered such that p1 is more similar to p2 than p3 when considered qua physical, but when considered qua pain p1 might be more similar to p3 than to p2).[22]

All this needs to be dealt with if we are to accept Yablo's account of the way to avoid the Exclusion problem. It does seem, however, that if the mental is related to the physical in the way that Yablo suggests (and given a suitably illuminating explanation of the relation) then it promises to be a fruitful response to the exclusion problem (and perhaps much else besides).

So, perhaps you are not, as a non-reductive physicalist, committed to epiphenomenalism after all.

Paul Sperring
Richmond-upon-Thames College

References and Further Reading

Chalmers, D.J., (ed) Philosophy of Mind: Classical and Contemporary Readings (London: Oxford University Press, 2002)
Ehring, D., "Mental Causation, Determinables and Property Instances", Nous, Vol. 30, No. 4 (Dec., 1996), pp. 461-480 (reprinted in Chalmers, 2002).
Fodor, J. (1989), "Making Mind Matter More," Philosophical Topics, 17: 59-80, (reprinted in Fodor, A Theory of Content and Other Essays (Cambridge: Mit Press, 1992)
Malcolm, N., "The Conceivability of Mechanism", The Philosophical Review, Vol. 77, No. 1 (Jan., 1968), pp. 45-72 (reprinted in Watson, Watson, G., Free Will: Oxford Readings in Philosophy (London: Oxford University Press, 1982)
Yablo, S., "Mental Causation" The Philosophical Review, Vol. 101, No. 2 (Apr., 1992), pp. 245-280


[1]  Where typing is thought to be more than simply accidentally applying pressure to the keys which happen to generate just those letter strings.  [back] 

[2]  The arguments offered here are not my own, but have been suggested by a number of philosophers - I borrow most of this material from Stephen Yablo (1992).  [back] 

[3]  Intentional states, such as beliefs, can supposedly be related to non-existent things, such as goblins or phlogiston, in ways that purely physical things can't. I can believe that goblins burp pure phlogiston, whereas I cannot point my umbrella at a goblin. Qualia, or qualitative mental states, it is argued, resist explanation given even the most comprehensive physical account imaginable - the thought here is that there is something it is like to have a sensation of something red, or taste something chocolatey, which can't be captured by the language of best possible physics (that's to say, even if God is doing the physics).  [back] 

[4]  Property dualists reject the classic Cartesian, or substance, dualism, and claim that there is one basic sort of stuff in the world, but that sort of stuff can give rise to distinct kinds of property. Mental properties would simply not be the same sort of properties as properties of the brain on this view. Non-reductive physicalists argue that mental states are, at bottom, ontologically identical with brain states, but that the former cannot be explained by appeal to the latter (for various different reasons). Statement (d) could be construed as either property dualist or non-reductive materialist - depending on how the claim was cashed out.  [back] 

[5]  Epiphenomenalism is the view that mental states play no causal role at all - that all the causal work in us is done by our physical states. This sort of view denies that the sentences typed throughout this essay are not caused by anything that is mental. We shall explore this odd view in greater detail later.  [back] 

[6]  David Chalmers defends the view that we may just have to give up on the idea of mental causation if we accept a non-reductive account of the mind, allied to certain other commitments (to be spelled out later) in The Conscious Mind.  [back] 

[7]  Stephen Yablo (1992) sets out the problem in this way.  [back] 

[8]  Yablo sets out the problem in terms of mental and physical events, but says that properties X and X* are straightforwardly substitutable where events x and x* are used. He points out also that the exclusion argument raises separate difficulties (for particulars and for properties), but on the whole these have been treated in isolation. My focus will be particularly on properties, in setting out the problems.  [back] 

[9]  There are, however, defenders of the idea that overdetermined effects are widespread (for instance where minds and bodies are concerned), and although they are in a minority in the philosophical community this, of course, is no indication that they are wrong about this. Since my job is just to show why non-reductive physicalists are committed to something like epiphenomenalism, and most non-reductive physicalists reject systematic overdetermination of effects, then I am just going to assume that there is no such systematic overdetermination.  [back] 

[10]  N. Malcolm, in Watson (1982).  [back] 

[11]  From Fodor (1992), p156.  [back] 

[12]  Functional properties are those properties which, according to one version of the view called Functionalism, are defined in terms of their causal roles. Two things may have identical functional properties, but not be identical physically (although if two things were physically identical they would be functionally identical, supposing them to embedded in identical environments). So, for instance, two things physically distinct could be functionally identical insofar as they are both egg whisks, or clocks, or pains and so on. These properties are said to be "mulitply realisable" in that they can be realised in many different physical systems.  [back] 

[13]  An eliminativist about X denies that there are any Xs, whereas a reductionist says that there are Xs all right, it's just that they are at bottom Ys. Successful eliminations have been performed on phenomena such as evil spirits in modern medicine, and the ether in modern physics.  [back] 

[14]  Yablo (1992) Philosophical Review, reprinted in Chalmers (2002) (all references from Chalmers).  [back] 

[15]  This, recall, is how we characterised functional properties. Something may be in the same mental state as I am, but not be in the same physical state as I am, according to the multiple realisability of the mental thesis.  [back] 

[16]  I am ignoring, for the sake of simplicity, here the complications introduced by the original/copy distinction in works of art, whereby the latter might not be said to share in all of the aesthetic properties of the former, given the different possible relationships to the creator of the work, and so on.  [back] 

[17]  If something is shaped then it can be shaped in a particular (determinate) way. Thus a square object is related to a shaped object as determinate to determinable. However, the same thing can be both determinate and determinable, depending on the relation considered. So blue is a determinate relative to coloured, but determinable relative to azure.  [back] 

[18]  Yablo (1992) p.183.  [back] 

[19]  Interestingly, Yablo smells something like a reverse exclusion argument here which relegates physical phenomena, because they are almost always overladen with details "far beyond the effect's causal requirements."(Yablo (1992) p.181), to the position of the superfluous. Does this bring causal conflict in by the back door, but this time making the physical epiphenomenal?  [back] 

[20]  Yablo (1992) p.182.  [back] 

[21]  Ehring (1996)  [back] 

[22]  Ehring (1996) pp 472-3  [back]