Chapter 21 Making Mind Matter More Jerry A. Fodor

An outbreak of epiphobia (epiphobia is the fear that one is turning into an epipheno- menalist) appears to have much of the philosophy of mind community in its grip. Though it is generally agreed to be compatible with physicalism that intentional states should be causally responsible for behavioral outcomes, epiphobics worry that it is not compatible with physicalism that intentional states should be causally responsible for behavioral outcomes qua intentional.

In this paper, I want to argue that epiphobia is a neurotic worry; if there is a problem, it is engendered not by the actual-or-possible successes of physicalistic psychology, but by two philosophical mistakes: (a) a wrong idea about what it is for a property to be causally responsible; and (b) a complex of wrong ideas about the relations between special-science laws and the events that they subsume.^[V] Here's how I propose to pro­ceed: First, we'll have a little psychodrama; I want to give you a feel for how an other­wise healthy mind might succumb to epiphobia. Second, ΓU provide a brief, sketchy, but I hope good-enough-for-present-purposes account of what it is for a property to be causally responsible. It will follow from this account that intentional properties are causally responsible if there are intentional causal laws. I'll then argue that (contrary to the doctrine called "anomalous monism") there is no good reason to doubt that there are intentional causal laws. I'll also argue that, so far as the matter affeds the duster of issues centering around epiphenomenalism, the sorts of relations that intentional causal laws can bear to the individuals they subsume are much the same as the sorts of relations that mm intentional causal laws can bear to the individuals that they subsume.

for now, what the vocabulary

tional terms.)

3. Stipulation: A property is “causally of things that have it. And (also by responsible are epiphenomenal.

But then, consider the mental event m (let's say, an event which consists of you desiring to lift your arm) which is the cause of the behavioral event b (let's say, an event which consists of you lifting your arm), m does, of course, have certain intentional properties. But, according to 2, none of its intentional properties is identical to any of its physical properties. And, according to 1, tn's physical properties fully determine its causal powers (including, of course, its power to cause b). So, it appears that m's being the cause of your lifting your arm doesn't depend on its being a desire to lift your arm; tn would have caused your lifting of your arm even if it hadn't had its intentional properties, so long as its physical properties were preserved.² So it appears that tn's intentional properties don't affect its causal powers. So it appears that tn's intentional properties are causally inert. Gearly, this argument iterates to any intentional property of the cause of any behavioral effect. So the intentional properties of mental events are epiphenomenal. Epiphobia!

Now, the first thing to notice about this line of argument is that it has nothing to do with intentionality as such. On the contrary, it applies equally happily to prove the epiphenomenality of any non-physical property, so long as property dualism is as­sumed. Consider, for example, the property of being a mountain; and suppose (what is surely plausible) that being a mountain isn't a physical property. (Remember, this just means that “mountain" and its synonyms aren't items in the lexicon of physics.) Now, untutored intuition might suggest that many of the effects of mountains are attributable to their being mountains. Thus, untutored intuition suggests, it is because Mt. Everest is a mountain that Mt.

No doubt there will be those who are prepared to bite this bullet. Such folk may either (i) deny that property dualism applies to mountainhood (because, on reflection, being a mountain is a physical property after all) or (ii) assert that it is intuitively plausible that being a mountain is causally inert (because, on reflection, it is intuitively plausible that it's not being a mountain but some other of Mt. Everest's properties—specifically, some of its physical properties—that are causally responsible for its effects). So be it; I do not want this to turn into a squabble about cases. Instead, let me emphasize that there are lots and lots and lots of examples where, on the one hand, considerations like multiple realizability make it implausible that a certain property is expressible in physi­cal vocabulary; and, on the other hand, claims for the causal inertness of the property appear to be wildly implausible, at least prima fade.

Consider the property of being a sail. I won't bore you with the fine points (terribly tempted, though I am, to exerdse my hobbyhorse³). Suffice it that sails are airfoils and there is quite a nice little theory about the causal properties of airfoils. Typically, airfoils generate lift in a direction, and in amounts, that are determined by their geometry, their

Making Mind Matter More 153 rigidity, and many, many details of their relations to the (liquid or gaseous) medium through which they move.

Now, the relative velocity of the airfoil may be increased by forcing the medium to flow through a "slot" (a constriction, one side of which is formed by the surface of the airfoil). The controlling law is that the narrower the slot, the faster the flow. (On sailboats of conventional Bermuda rig, the slot is the opening between the jib and the main. But perhaps you didn't want to know that.) Anyhow, airfoils and slots can be made out of all sorts of things; sails are airfoils, but so are keel-wings, and airplane wings and bird's wings. Slots are multiply realizable too: You can have a slot both sides of which are made of sailcloth, as in the jib/mainsail arrangement, but you can also have a siot one side of which is made of sailcloth and the other side of which is made of air. (That's part of the explanation of why you can sail towards the wind even if you haven't got a jib.) So then, if one of your reasons for doubting that believing that P is a physical property is that believing is multiply realizable, then you have the same reason for doubting that being an airfoil or being a slot counts as a physical property.

And yet, of course, it would seem to be quite mad to say that being an airfoil is causally inert. Airplanes fall down when you take their wings off; and sailboats come to a stop when you take down their sails. Everybody who isn't a philosopher agrees that these and other such facts are explained by the story about lift being generated by causal interactions between the airfoil and the medium. If that isn't the right explana­tion, what keeps the plane up? If that is the right explanation, how could it be that being an airfoil is causally inert?

Epiphobics primarily concerned with issues in the philosophy of mind might well stop here.

Many philosophers have the bad habit of thinking about only two sciences when they think about sciences at all; these being psychology and physics. When in the grip of this habit, they are likely to infer that if psychological theories have some property that physical theories don't, that must be because psychological states (qua psychologi­cal) are intentional and physical states (qua physical) are not. In the present case, if there's an argument that psychological properties are epiphenomenal, and no corre­sponding argument that physical properties are epiphenomenal, that must show that there is something funny about intentionality.

But we now see that it shows no such thing since, if the causal inertness of psycho­logical properties is maintained along anything like the lines of 1-3, there are likely to be parallel arguments that all properties are causally inert except those expressed by the vocabulary of physics. In which case, why should anybody care whether psychological properties are epiphenomenal? All that anybody could reasonably want for psychology is that its constructs should enjoy whatever sort of explanatory/causal role is proper to the constructs of the special sciences. If beliefs and desires are as well off ontologically as mountains, wings, spiral nebulas, trees, gears, levers and the like, then surely they're as well off as anyone could need them to be.

But, in fact, we shouldn't stop here. Because, though it's true that claims for the epiphenomenality of mountainhood and airfoilhood and, in general, of any non­physical-property-you-like-hood will follow from the same sorts of arguments that imply claims for the epiphenomenality of beliefhood and desirehood.

It seems to me, in light of the foregoing, that it ought to be a minimal condition upon a theory of what it is for something to be a causally responsible property that it does not entail the epiphenomenality of winghood, mountainhood, gearhood, leverhood, beliefhood, desirehood and the like. I'm about to propose a theory which meets this condition, and thereby commends itself as a tonic for epiphobics. It isn't, as you will see, very shocking, or surprising or anything; actually it's pretty dull. Still, I need a little stage setting before I can tell you about it. In particular, I need some caveats and some assumptions.

Caoeats First, curing epiphobia requires making it plausible that intentional properties can meet suflident conditions for causal responsibility; but one is not also required to show that they can meet necessary and sufficient conditions for causal responsibility. This is just as well, since necessary and suffident conditions for causal responsibility might be sort of hard to come by (necessary and suffident conditions for anything tend to be sort of hard to come by) and I, for one, don't claim to have any.

Second, the question "What makes a property causally responsible?" needs to be distinguished from the probably much harder question. "What determines which prop­erty is responsible in a given case when one event causes another?" Suppose that el causes e2; then, trivially, it must do so in virtue of some or other of its causally responsible properties; i.e., in virtue of some or other property in virtue of which it is able to be a cause. But it may be that el has many—perhaps many, many—such properties; so it must not be assumed that if el is capable of being a cause in virtue of having a certain property P, then P is ipso facto the property in virtue of which el is the cause of e2. Indeed, it must not even be assumed that if el is capable of being a cause of e2 in virtue of its having P, then P is ipso facto the property in virtue of which el causes e2. For again it may be that el has many—even many, many—properties in virtue of which it is capable of being the cause of e2, and it need not be obvious which one of these properties is the one in virtue of which it actually is the cause of e2. At least, I can assure you, it need not be obvious to me.

It is, to put all this a little less pedantically, one sort of success to show that it was in virtue of its intentional content that your desire to raise your hand made something happen. It is another, and lesser, sort of success to show that being a desire to raise your hand is the kind of property in virtue of which things can be made to happen. Curing epiphobia requires only a success of the latter, lesser sort.

Assumptions I assume that singular causal statements need to be covered by causal laws. That means something like:

4. Covering Principle: If an event el causes an event e2, then there are properties F, G such that;

4.1. el instantiates F;

4.2. e2 instantiates G;

and

4.3. 'T instantiations are suffident for G instantiations" is a causal law.⁴

When a pair of events bears this relation to a law, ΓU say that the individuals are each covered or subsumed by that law and I'll say that the law projects the properties in virtue of which the individuals are subsumed by it. Notice that when an individual is covered by a law, it will always have some property in virtue of which the law subsumes it. If, for example, the covering law is that Fs cause Gs, then individuals that get covered by this law do so either in virtue of being Fs (in case they are subsumed by its antecedent) or in virtue of being Gs (in case they are subsumed by its consequent). This could all be made more predse, but I see no reason to bother.

OK, I can now tell you my suffident condition for a property to be causally responsible:

5. P is a causally responsible property if it's a property in virtue of which individ­uals are subsumed by causal laws; or, equivalently,

5.1. P is a causally responsible property if it's a property projeded by a causal law; or, equivalently (since the satisfaction of the antecedent of a law is ipso facto nomologically suffident for the satisfaction of its consequent),

5.2. P is a causally responsible property if it's a property in virtue of the instantia­tion of which the occurrence of one event is nomologically suffident for the occurrence of another.⁵

If this is right, then intentional properties are causally responsible in case there are intentional causal laws; aerodynamic properties are causally responsible in case there are aerodynamic causal laws; geological properties are causally responsible in case there are geological causal laws... and so forth. To all intents and purposes, on this view the question whether the property P is causally responsible reduces to the question whether there are causal laws about P. To settle the second question is to settle the first.

I don't mind if you find this proposal dull, but I would be distressed if you found it drcular. How, you might ask, can one possibly make progress by defining “causally responsible property" iη terms of "covering causal law"? And yet it's undear that we can just drop the requirement that the covering law be causal because there are non­causal laws (e.g., the gas law about pressure and volume varying inversely) and perhaps an event's being covered by those sorts of laws isn't suffident for its having a causally responsible property.

I can think of two fairly plausible ways out of this. First, it may be that any property in virtue of which some law covers an individual will be a property in virtue of which some causal law covers an individual;⁶ i.e., that no property figures only in noncausal laws. This is, I think, an interesting metaphysical possibility; if it is true, then we can just identify the causally responsible properties with the properties in virtue of which indi­viduals are covered by laws.

And, even if it's not true, it may be that what makes a law causal can itself be specified in noncausal terms. Perhaps it involves such properties as covering temporal successions, being asymmetric, and the like. In that case it would be okay to construe "causally responsible" in terms of "causal law" since the latter could be independently defined. Barring arguments to the contrary, I'm prepared to suppose that this will work.

We're now in a position to do a little diagnosis. According to the present view, the properties projeded in the laws of basic sdence are causally responsible, and so too are the properties projected in the laws of the special sciences. This is truistic since the present view just is that being projected is sufficient for being causally responsible. Notice, in particular, that even if the properties that the special sciences talk about are supervenient upon the properties that the basic sciences talk about, that does not argue that the properties that the special sciences talk about are epiphenomenal. Not, at least, if there are causal laws of the special sciences. The causal laws of the special sciences and causal laws of basic sciences have in common that they both license ascriptions of causal responsibility. Or so, at least, the present view would have it.

This is not, however, to deny that there are metaphysically interesting differences between special-science laws and basic science laws. Let me introduce here a point that I propose to make a fuss of later.

Roughly, the satisfaction of the antecedent of a law is nomologically sufficient for the satisfaction of its consequent.⁷ (I'll sometimes say that the truth of the antecedent of a law nomologically necessitates the truth of its consequent.) But a metaphysically interest­ing difference between basic and nonbasic laws is that, in the case of the latter but not the former, there always has to be a mechanism in virtue of which the satisfaction of its antecedent brings about the satisfaction of its consequent. If Ts cause Gs' is basic, then there is no answer to the question how do Fs cause Gs; they just do, and that they do is among the not-to-be-further-explained facts about the way the world is put together. Whereas, if Ts cause Gs' is mmbasic, then there is always a story about what goes on when—and in virtue of which—Fs cause Gs.

Sometimes it's a microstructure story: Meandering rivers erode their outside banks; facts about the abrasive effects of particles suspended in moving water explain why there is erosion; and the Bemouli effect explains why it's the outside banks that get eroded most. Sometimes there's a story about chains of macrolevel events that inter­vene between F-instantiations and G-instantiations. Changes in CO₂ levels in the atmo­sphere cause changes in fauna. There's a story about how CO₂ blocks radiation from the earth's surface; and there's a story about how the blocked radiation changes the air temperature; and there's a story about how changes in the air temperature cause climac­tic changes; and there's a (Darwinian) story about how climactic changes have zoologi­cal impacts. (I try to be as topical as I can.)

Or, to get closer home, consider the case in computational psychology: There are— so I fondly suppose—intentional laws that connect, for example, states of believing that P & (P → Q) to states of believing that Q. (Ceteris paribus, of course. More of that later.) Because there are events covered by such laws, it follows (trivially) that inten­tional properties (like believing that P & (P → Q) are causally responsible. And because nobody (except, maybe, panpsychists; who I am prepared not to take seriously for present purposes) thinks that intentional laws are basic, it follows that there must be a mechanism in virtue of which believing that P & (P → Q) brings it about that one believes Q.

There are, as it happens, some reasonably persuasive theories about the nature of such mechanisms currently on offer. The one I like best says that the mechanisms that implement intentional laws are computational. Roughly, the story goes: believing (etc.) is a relation between an organism and a mental representation. Mental representations have (inter alia) syntactic properties; and the mechanisms of belief-change are defined over the syntactic properties of mental representations. Let's not worry, for the mo­ment, about whether this story is right; let's just worry about whether it's epiphobic.

Various philosophers have supposed that it is. Steven Stich, for example, has done some public handwringing about how anybody (a fortiori, how I) could hold both that intentional properties are causally responsible and the ("methodologically solipsistic")

Making Mind Matter More 15 7 view that mental processes are entirely computational (/syntactic). And Norbert Horn- stein⁸ has recently ascribed to me the view that "the generalizations of psychology, the laws and the theories, are stated over syntactic objects, i.e., it is over syntactic represen­tations that computations proceed." (p. 18). But: THE CLAIM THAT MENTAL PRO­CESSES ARE SYNTACTIC DOES NOT ENTAIL THE CLAIM THAT THE LAWS OF PSYCHOLOGY ARE SYNTACTIC. On the contrary THE LAWS OF PSYCHOLOGY ARE INTENTIONAL THROUGH AND THROUGH. This is a point to the reiteration of which my declining years seems somehow to have become devoted. What's syntac­tic is not the laws of psychology but the mechanisms by which the laws of psychology are implemented. Cf: The mechanisms of geological processes are—as it might be— chemical and molecular; it does not follow that chemical or molecular properties are projected by geological laws (on the contrary, it's geological properties that are pro­jected by geological laws); and it does not follow that geological properties are causally inert (on the contrary, it's because Mt. Everest is such a very damned big mountain that it's so very damned cold on top).

It is, I should add, not in the least unusual to find that the vocabulary that's appropri­ate to articulate a special-science law is systematically different from the vocabulary that's appropriate to articulate its implementing mechanism(s). Rather, shift of vocabu­lary as one goes from the law to the mechanism is the general case. If you want to talk laws of inheritance, you talk recessive traits and dominant traits and homozygotes and heterozygotes; if you want to talk mechanisms of inheritance, you talk chromosomes and genes and how the DNA folds. If you want to talk psychological law, you talk intentional vocabulary; if you want to talk psychological mechanism, you talk syntactic (or maybe neurological) vocabulary. If you want to talk geological law, you talk moun­tains and glaciers; if you want to talk geological mechanism, you talk abrasion coeffi­cients and cleavage planes. If you want to talk aerodynamic law, you talk airfoils and lift forces; if you want to talk aerodynamic mechanism, you talk gas pressure and laminar flows. It doesn't follow that the property of being a belief or an airfoil or a recessive trait is causally inert; all that follows is that specifying the causally responsible macroproperty isn't the same as specifying the implementing micromechanism.

It's a confusion to suppose that, if there's a law, then there needn't be an implement­ing mechanism; and it's a confusion to suppose that, if there's a mechanism that imple­ments a law, then the properties that the law projects must be causally inert. If you take great care to avoid both these confusions, you will be delighted to see how rapidly your epiphobia disappears. You really will. Trust me. ^[11]

There seems to be some tension between the following three principles, each of which I take to be prima fade sort of plausible:

6. Strict covering: Just like 4 except with the following in place of 4.3; 'Tl instantiations are causally sufficient for P2 instantiations" is a strict causal law.

7. Anomia of the mental: The only strict laws are laws of physics. Spedfically, there are no strict 'psychophysical' laws relating types of brain states to types of intentional states; and there are no strid 'psychological' laws relating types of mental events to one another or to types of behavioral outcomes.

8. Causal responsibility of the mental: Intentional properties aren't epiphenomenal.

6means something like: Causal transactions must be covered by exceptionless laws; the satisfaction of the antecedent of a covering law has to provide literally nomologi­cally suffident conditions for the satisfaction of its consequent so that its consequent is satisfied in every nomologically possible situation in which its antecedent is satisfied.

7 means something like this: The laws of physics differ in a characteristic way from the laws of the spedal sdences (notably inducting psychology). Spedal sdence laws are typically hedged with 'ceteris paribus' dauses, so that whereas physical laws say what has to happen come what may, spedal-sdence laws only say what has to happen all else being equal.⁹

How we should construe 8 has, of course, been a main concern throughout; but, according to the account of causal responsibility that I've been trying to sell you, it effectively reduces to the requirement that mental causes be covered by intentional laws. So now we can see where the tension between the three prindples (6-8) arises. The responsibility of the mental requires covering by intentional laws. But given the revised notion of covering, according to which causes have to be covered by strict laws, it must be physical laws, and not intentional ones, that cover mental causes. So it turns out that the intentional properties are causally inert even according to the count of causal responsibility commended in part I.¹⁰

Something has to be done, and I assume it has to be done to 6 or 8 (or both) since 7 would seem to be okay. It is quite generally true about spedal-sdence laws that they hold only barring breakdowns', or 'under appropriately idealized conditions', or 'when the effects of interacting variables are ignored'. If even geological laws have to be hedged—as indeed they do—then it's more than plausible that the 'all else equal' proviso in psychological laws will prove not to be eliminable. On balance, we had best assume that 7 stays.

VVhat about 8 then? Surely we want 8 to come out true on some reasonable construal. I've opted for a robust reading: mental properties are causally responsible because they are the properties in virtue of which mental causes are subsumed by covering laws; which is to say that mental properties are causally responsible because there are inten­tional generalizations which specify nomologically suffident conditions for behavioral outcomes. But this reading of 8 looks to be incompatible with 7. 7 suggests that there aren't intentionally specifiable Suffident conditions for behavioral outcomes since, at best, intentional laws hold only ceteris paribus. So, maybe the notion of causal responsi­bility I've been selling is too strong. Maybe we could Ieam to make do with less.¹¹

Ibis is, more or less explidtly, the course that LePore and Loewer recommend in "Mind Matters": If the causal responsibility of the intentional can somehow be detached from its causal sufficiency for behavioral outcomes, we could then maybe recondie causal responsibility with anomicness. In effect, L&L's idea is to hold on to 6 and 7 at the cost of not adopting a nomological subsumption reading of 8. Prima fade, this strategy is

Making Mind Matter More 159 plausible in light of a point that L&L emphasize in their discussion of Sosa: The very fact that psychological laws are hedged would seem to rule out any construal of causal responsibility that requires mental causes qua mental to be nomologically sufficient for behavior. If it's only true ceteris paribus that someone who wants a drink reaches for the locally salient glass of water, then it's epiphobic to hold that desiring is causally respon­sible for reaching only if literally everyone who desires would thereupon reach. After all, quite aside from what you think of 6, it's simply not coherent to require the antecedents of hedged laws to provide literally nomologically sufficient conditions for the satisfaction of their consequents.

That's the stick; but Loewer and LePore also have a carrot on offer. They concede that, if the only strict laws are physical, then instantiations of intentional properties are not strictly sufficient for determining behavioral outcomes. But they observe that grant­ing 6 and 7 doesn't concede that the physical properties of mental events are necessary for their behavioral effects. To see this, assume an event m which instantiates the mental property M and the physical property P. Assume that m has the behavioral outcome b, an event with the behavioral property B, and that it does so in virtue of a physical law which strictly connects the instantiation of P with the instantiation of B. LePore and Loewer point out that all this is fully compatible with the truth of the Counterfactual:— Pm & Mm → Bb (i.e., with it being the case that m would have caused Bb even if it hadn't been P.) Think of the case where M events are ''multiply realized," e.g., not just by P instantiations but also by P* instantiations. And suppose that there's a strict law connecting P* events with B events. Then Mm → Bb will be true not only when m is a P instantiation, but also when æ is a P* instantiation. The point is that one way that—Pm & Mm → Bb can be true is if there are strict psychological laws; i.e., if being an M instantiation is strictly sufficient for being a B instantiation. But the Counterfactual could also be true on the assumption that B instantiations have disjoint physically sufficient conditions. And that assumption can be allowed by someone who claims that only physical laws can ground mental causes (e.g., because he claims that only physical laws articulate strictly sufficient conditions for behavioral outcomes).

In short, LePore and Loewer show us that we can get quite a lot of what we want from the causal responsibility of the mental without assuming that intentional events are nomologically sufficient for behavioral outcomes; i.e., without assuming that inten­tional laws nomologically necessitate their consequents; i.e., without denying that the mental is anomic. Specifically, we can get that the particular constellation of physical properties that a mental cause exhibits needn't be necessary for its behavioral out­comes. I take LePore and Loewer's advice to be that we should settle for this; that we should construe the causal responsibility of the mental in some way that doesn't require mental events to be nomologically sufficient for their behavioral consequences. In effect, given a conflict between 6 and a covering law construal of 8, LePore and Loewer opt for 6; keep the idea that causes have to be strictly covered, and give up on the idea that the causal responsibility of the mental is the nomological necessitation of the behavioral by the intentional.

Now, this may be good advice, but I seem to detect a not-very-hidden agenda. Suppose, just for the sake of argument, that there is some way of providing intentional­ly sufficient conditions for behavioral outcomes. Then this would not only allow for an intuitively satisfying construal of the causal responsibility of the mental (viz., mental properties are causally responsible if mental causes are covered by intentional laws, as per part I), it would also undermine the idea that mental causes have to be covered by physical laws. If the laws of psychology have in common with the laws of physics that both strictly necessitate their consequents, then presumably either would do equally well to satisfy the constraints that 6 imposes on the laws that cover mental causes. But the idea that mental causes have to be covered by physical laws is the key step in the famous Davidsonian argument from the anomia of the mental to physicalism. It may be that LePore and Loewer would like to hang onto the Davidsonian argument; it's pretty clear that Davidson would.

I take Davidson's argument to go something like this:

9.1. Mental causes have to be covered by some strict law (strict covering);

9.2. but not by intentional laws because intentional laws aren't strict; the satisfac­tion of their antecedents isn't nomologically sufficient for the satisfaction of their consequents (anomia of the mental);

9.3. so mental causes must be covered by physical laws;

9.4. so they must have physical properties. Q.E.D.

But if there are intentionally sufficient conditions for behavioral outcomes you lose step 9.2; and if you lose step 9.2, you lose the argument. It appears that the cost of an intuitively adequate construal of mental responsibility is that there's no argument from mental causation to physicalism.

Well, so much for laying out the geography. Here's what happens next: First, I'll try to convince you that your intuitions really do cry out for some sort of causal sufficiency account of causal responsibility; something like that if it's m's being M that's causally responsible for b's being B, then b is B in all nearby worlds where m is M. (This is, to repeat, a consequence of defining causal responsibility in terms of strict covering laws, since it is a defining property of such laws that the satisfaction of their antecedents necessitates the satisfaction of their consequents.) I'll then suggest that, appearances to the contrary, it really isn't very hard to square such an account with the admission that even the best psychological laws are very likely to be hedged. In effect, Γm claiming that, given a conflict between 6 and 8, there's a natural replacement for 8. At this point the question about physicalism becomes moot since it will no longer be clear why hedged psychological laws can't ground mental causes; and, presumably, if hedged psychological laws can, then strict physical laws needn't. It still might turn out, how­ever, that you can get a physicalist conclusion from considerations about mental causation, though by a slightly different route from the one that Davidson follows—a route that doesn't require the subsumption of causes by strict laws as a lemma.

My first point, then, is that, Loewer and LePore to the contrary notwithstanding, the notion of the causal responsibility of the mental that your intuitions demand is that Ms should be a nomologically sufficient condition for Bs. Accept no substitutes, is what I say. Γm not, however, exactly sure how to convince you that this is indeed what your intuitions cry out for; perhaps the following considerations will seem persuasive.

There aren't, of course, any reliable procedures for scientific discovery. But one might think of the procedures that have sometimes been proposed as, in effect, codi­fying our intuitions about causal responsibility. For example, it's right to say that Pasteur used the 'method of differences' to discover that contact with stuff in the air—and not spontaneous generation in the nutrient—is responsible for the breeding of maggots. This is not, however, a comment on how Pasteur went about thinking up his hypotheses or his experiments. The method of differences doesn't tell you how to find out what is causally responsible. Rather, it tells you what to find out to find out what's causally responsible. It says: thrash about in the nearby nomologically possible worlds and find a property such that you get the maggots just when you get that property instantiated. That will be the property whose instantiation is causally respon­sible for the maggots.

Γm claiming that Pasteur had in mind to assign causal responsibility for the maggots, and that, in doing so, it was preeminantly reasonable of him to have argued according to the method of differences: viz., that if the infestation is airborne, then fitting a gauze top to the bottle should get rid of the maggots, and taking the gauze top off the bottle should bring the maggots back again. Assigning causal responsibility to contact with stuff in the air involved showing that such contact is necessary and sufficient for getting the maggots; that was what the method of differences required, and that was what Pasteur figured out how to do. If those intuitions about causal responsibility were good enough for Pasteur, I guess they ought to be good enough for you and me.

So then, I assume that the method of differences codifies our intuitions about causal responsibility. But this implies that assigning causal responsibility to the mental re­quires the truth of more Counterfactuals than L&L are prepared to allow. Intuitively, what we need is that m's being M is what makes the difference in determining whether b is B, hence that ^rBb whenever Mm' is true in all nearby worlds. If the method of differences tells us what causal responsibility is, then what it tells us is that causal responsibility requires nomological sufficiency.¹² So the causal responsibility of the mental must be the nomological sufficiency of intentional states for producing be­havioral outcomes.

The first—and crudal—step in getting what a robust construal of the causal respon­sibility of the mental requires is to square the idea that Ms are nomologically sufficient for Bs with the fact that psychological laws are hedged. How can you have it both that special laws only necessitate their consequents ceteris paribus and that we must get Bs whenever we get Ms. Answer: you can't. But what you can have is just as good: viz., that if it's a law that M → B ceteris paribus, then it follows that you get Bs whenever you get Ms and the ceteris paribus conditions are satisfied.¹³ Ibis shows us how ceteris paribus laws can do serious scientific business, since it captures the difference between the (substantive) claim that Fs cause Gs ceteris paribus, and the (empty) claim that Fs cause Gs except when they don't.

So, it's sufficient for M to be a causally responsible property if it's a property in virtue of which Ms cause Bs. And here's what it is for M to be a property in virtue of which Ms cause Bs:

10.1. MscauseBs;

10.2. ^,M → B ceteris paribus' is a law;¹⁴ and

10.3. the ceteris paribus conditions are satisfied in respect of some Ms.

I must say, the idea that hedged (including intentional) laws necessitate their conse­quents when their ceteris paribus clauses are discharged seems to me to be so obviously the pertinent proposal that Γm hard put to see how anybody could seriously object to it. But no doubt somebody will.

One might, I suppose, take the line that there's no fact of the matter about whether, in a given case, the ceteris paribus conditions on a special science law are satisfied. Or that, even if there is a fact of the matter, still one can't ever know what the fact of the matter is. But, surely that would be mad. After all, Pasteur did demonstrate, to the satisfaction of all reasonable men, that ceteris paribus you get maggots when and only when the nutrients are in contact with stuff in the air. And presumably he did it by investigating experimental environments in which the ceteris paribus condition was satisfied and known to be so. Whatever is actual is possible; what Pasteur could do in fact, even you and I can do in principle.

I remark, in passing, that determining that ceteris paribus stuff in the air causes maggots did not require that Pasteur be able to enumerate the ceteris paribus conditions, only that he be able to recognize some cases in which they were in fact satisfied. Sufficient conditions for the satisfaction of ceteris paribus clauses may be determinate and epistemically accessible even when necessary and sufficient conditions for their satis­faction aren't. A fortiori, hedged laws whose ceteris paribus conditions cannot be enumerated may nevertheless be satisfied in particular cases. Perhaps we should say that M is causally responsible only if Ms cause Bs in any world in which the ceteris paribus clause of ^zM → B all else equal' is discharged. This would leave it open, and not very important, whether 'all and only the worlds in which the ceteris paribus conditions are discharged' is actually well-defined. It's not very important because what deter­mines whether a given law can cover a given event is whether the law is determinately satisfied by the event. It is not also required that it be determinate whether the law would be satisfied by arbitrary other events (or by that same event in arbitrary other worlds). It seems to me that the plausibility of Davidson's assumption that hedged laws can't ground causes may depend on overlooking this point.

Finally, it might be argued that, although the ceteris paribus conditions on other special-science laws are sometimes known to be satisfied, there is nevertheless some­thing peculiar about intentional laws, so that their ceteris paribus conditions can't be. I take it that Davidson thinks that something of this sort is true; but I have never been able to follow the arguments that are supposed to show that it is. And I notice (with approval) that LePore and Loewer are apparently not committed to any such claim.

Where does all this leave us with respect to the classical Davidsonian argument that infers physicalism from the anomalousness of the mental? It seems to me that we are now lacking any convincing argument for accepting principle 6. Suppose it's true that causes need to be covered by laws that necessitate their consequents; it doesn't follow that they need to be covered by strict laws. Hedged laws necessitate their consequents in worlds where their ceteris paribus conditions are satisfied. Why, then, should mental causes that are covered by hedged intentional laws with satisfied antecedents and satisfied ceteris paribus conditions require further covering by a strict law of physics?

The point till now has been that if strict laws will do to cover causes, so too will hedged laws in worlds where the hedges are discharged. I digress to remark that hedged laws can play the same role as strict ones in covering law explanations, so long as it's part of the explanation that the ceteris paribus conditions are satisfied.

When the antecedent of a strict law is satisfied, you are guaranteed the satisfaction of its consequent, and the operation of strict laws in covering law explanations depends on this. What's typically in want of a covering law explanation is some such fact as that an event m caused an event b (and not, nb, that an event m caused an event b ceteris paribus).¹³ Indeed, it's not clear to me that there are facts of this latter sort. Hedged generalizations are one thing; hedged singularly causal statements would be quite an­other. Well, the point is that strict laws can explain nt's causing b precisely because if it's strict that Ms cause Bs and it's true that there is an M, then it follows that there is an M-caused b. "You got a B because you had an M, and it's a law that you get a B whenever you get an M." But if that sort of explanation is satisfying, then so too ought to be: "You got a B in world w because you had an M in world w, and it's a law that ceteris paribus you get a B whenever you have an M, and the ceteris paribus conditions were satisfied in world w." The long and short is: One reason you might think that causes have to be covered by strict laws is that covering law explanations depend on this being so. But they don't. Strict laws and hedged laws with satisfied ceteris paribus conditions operate alike in respect of their roles in covering causal relations and in respect of their roles in covering law explanations. Surely this is as it should be: Strict

Making Mind Matter More 163 laws are just the special case of hedged laws where the ceteris paribus clauses are discharged vacuously; they're the hedged laws for which 'all else' is always equal.

Still, I think that there is something to be said for the intuition that strict physical laws play a special role in respect of the metaphysical under-pinnings of causal relations, and I think there may after all be a route from considerations about mental causation to physicalism. I'll close by saying a little about this.

In my view, the metaphysically interesting fact about special-science laws isn't that they're hedged; it's that they're not basic. Correspondingly, the metaphysically interest­ing contrast isn't between physical laws and special science laws; it's between basic laws and the rest. For present purposes, I need to remind you of a difference between special laws and basic laws that I remarked on in part I: If it's nonbasically lawful that Ms cause Bs, there's always a story to tell about how (typically, by what transformations of microstructures) instantiating M brings about the instantiation of B. Nonbasic laws want implementing mechanisms; basic laws don't. (That, I imagine, is what makes them basic.)

It is therefore surely no accident that hedged laws are typically—maybe always—not basic. On the one hand, it's intrinsic to a law being hedged that it is nomologically possible for its ceteris paribus conditions not to be satisfied. And, on the other hand, a standard way to account for the failure of a ceteris paribus condition is to point to the breakdown of an intervening mechanism. Thus, meandering rivers erode their outside banks ceteris paribus, but not when the speed of the river is artificially controlled (no Bernoulli effect); and not when the river is chemically pure (no suspended particles); and not when somebody has built a wall on the outside bank (not enough abrasion to overcome adhesion). In such cases, the ceteris paribus clause fails to be satisfied because an intervening mechanism fails to operate. By contrast, this strategy is unavailable in the case of nonbasic laws; basic laws don't rely on mechanisms of implementation, so if they have exceptions that must be because they're nondeterministic.

We see here one way in which ceteris paribus clauses do their work Nonbasic laws rely on mediating mechanisms which they do not, however, articulate (sometimes be­cause the mechanisms aren't known; sometimes because As can cause Bs in many different ways, so that the same law has a variety of implementations). Ceteris paribus clauses can have the effect of existentially quantifying over these mechanisms, so that 'As cause Bs ceteris paribus' can mean something like There exists an intervening mechanism such that As cause Bs when it's intact.' I expect that the ceteris paribus clauses in special science laws can do other useful things as well. It is a scandal of the philosophy of science that we haven't got a good taxonomy of their functions.

However, I digress. The present point is that:

11. non-basic laws require mediation by intervening mechanisms; and

12. there are surely no basic laws of psychology.

Let us now make the following bold assumption: all the mechanisms that mediate the operation of nonbasic laws are eventually physical.¹⁶ I don't, I confess, know exactly what this bold assumption means (because I don't know exactly what it is for a mecha­nism to be physical as opposed, say, to spiritual); and I confess that I don't know exactly why it seems to me to be a reasonable bold assumption to make. But I do suspect that if it could be stated clearly, it would be seen to be a sort of bold assumption for which the past successes of our physicalistic world view render substantial inductive support.

Well, if all the mechanisms that nonbasic laws rely on are eventually physical, then the mechanisms of mental causation must be eventually physical, too. For, on the

current assumptions, mental causes have their effects in virtue of being subsumed by psychological laws and, since psychological laws aren't basic, they require mediation by intervening mechanisms. However, it seems to me that to admit that mental causes must be related to their effects (including, notice, their mental effects) by physical mechanisms just is to admit that mental causes are physical. Or, if it's not, then it's to admit something so close that I can't see why the difference matters.

So, then, perhaps there's a route to physicalism from stuff about mental causation that doesn't require the claim that ceteris paribus laws can't ground mental causes. If so, then my story gives us both physicalism and a reasonable account of the causal respon­sibility of the mental; whereas Davidson's story gives us at most the former.¹⁷ But if we cant get both the causal responsibility of the mental and an argument for physicalism, then it seems to me that we ought to give up the argument for physicalism. Γm not really convinced that it matters very much whether the mental is physical; still less that it matters very much whether we can prove that it is. Whereas, if it isn't literally true that my wanting is causally responsible for my reaching, and my itching is causally responsible for my scratching, and my believing is causally responsible for my say­ing.... if none of that is literally true, then practically everything I believe about anything is false and it's the end of the world.

Notes

This paper is a revised and extended version of some remarks presented at an APA symposium on December 30, 1987, in reply to Emest LePore and Barry Loewer's "Mind Matters," Journal of Philosophy 84.11 (Nov. 1987): 630-642. I am grateful to them and to Brian McLaughlia for much stimulating conversation on these and related issues.

1. I shall more or less assume, in what follows, that events are the individuals that causal laws subsume and to which causal powers are ascribed. Nothing will turn on this; it's just a bore to always be having to say "events, or situations, or things or whatever...."

2. It facilitates the discussion not to worry about which of their properties events have essentially. In particular, I shall assume that we can make sense of Counterfactuals in which a certain mental event is supposed to have no intentional content, or an intentional content or a physical constituency different from its actual content or constituency. Nothing geπnane to the present issues hangs on this since, as far as I can tell, the same sorts of points I'll be making about Counterfactual properties of events could just as well be made about relations between events and their counterparts.

3. What follows is a very crude approximation of the aerodynamic facts. Enthusiasts will find a serious exposition in W. Ross, Sail Power (New Yoric Alfted A Knopf, 1975).

4. The Covering Principle is generally in the spirit of the proposals of Donald Davidsoa except that, unlike Davidsoa I'm prepared to be shameless about properties.

5. 5.2 is in the text to emphasize that the nomological subsumption account of the causal responsibility of the mental is closely connected to the idea that mental events are nomologically sufficient for behavioral outcomes. We will thus have to consider how to square the nomological subsumption story with the fact that the antecedents of psychological laws generally do not specify nomologically sufficient conditions for the satisfaction of their consequents (because, like the laws of the other special sciences, the laws of psychology typically have essential ceteris paribus causes). See part ∏.

6. I'm leaving statistical laws out of Consideratioa If some laws are irremediably statistical then the proposal in the text should be changed to read: "any property in virtue of which some deterministic law covers an individual will be a property in virtue of which some causal law covers an individual."

7. But this will have to be hedged to deal with ceteris paribus laws. Part II is about what's the right way to hedge it.

8. N. Homstein, "The Heartbreak of Semantics," Mind and Language 3 (1988): 18.

9. Special science laws are unstrict not just de facto, but in principle. Specifically, they are characteristical­ly "heteronomic”: You can't convert them into strict laws by elaborating their antecedents. One reason why this is so is that special science laws typically fail in limiting conditions, or in conditions where the idealizations presupposed by the science aren't approximated; and, generally speaking, you have to go outside the vocabulary of the science to say what these conditions are. Old rivers meander, but not

Making Mind Matter More 165 when somebody builds a levee. Notice that "levee" is not a geological term. (Neither, for that matter, is "somebody.")

1 emphasize this point because it's sometimes supposed that heteronomidty is a proprietary feature of intentional laws qua intentional. Poppycock.

10. It could, no doubt, be said that accepting 6 doesn't really make the mental properties drop out of the picture because, even if mental causes have to be covered by physical laws, it can still be true that they are also covered by intentional laws (tnz., in the old 4.3 sense of "covering" which didn't require covering laws to be strict). As Brian McLaughlin (ms) has rightly pointed out, it's perfectly consistent to hold that covering by strict laws is necessary and sufficient for causal relations and also to hold that covering by loose laws is necessary, or even sufficient, for causal relations, so long as you are prepared to assume that every cause that is loosely covered is strictly covered, too.

However, it is not clear that this observation buys much relief from epiphobia. After all, if mental properties really are causally active, why isn't intentional covering all by itself sufficient to ground the causal relations of mental events? I've been urging that intentional properties are causally responsible if mental causes are covered by intentional laws. But that seems plausible only if mental events are causes in virtue of their being covered by intentional laws. But how could mental causes be causes qua intentionally covered if, in order to be causes, they are further required to be subsumed by noninten- Honal laws? Taken together, 6 and 7 make it look as though, even if mental events are covered qua intentional they're causes only qua physical. So again it looks like the intentional properties of mental events aren't doing any of the work.

11. I'm doing a little pussyfooting here, so perhaps I'd better put the point exactly: On the view that I will presently commend, there are circumstances in which instantiations of mental properties nomologi­cally necessitate behavioral outcomes. What isn't, however, quite the case is that these circumstances are fully specified by the antecedents of intentional laws. On my view, only basic laws have the property that their antecedents fully specify the circumstances that nomologically necessitate the satisfaction of their consequents (and then only if they're deterministic).

12. It will be noticed that I'm stressing the importance of causal sufficiency for causal responsibility, whereas it was causal necessity that Pasteur cared about most. Pasteur was out to show that contact with stuff in the air and only contact with stuff in the air is causally responsible for maggots; specifically that contact with stuff in the air accounts for all of the maggots, hence that spontaneous generation accounts for none. 1 take it that it is not among our intuitions that a certain mental property is causally responsible for a certain behavior only if that sort of behavior can have no other sort of cause.

13. So, what I said above—that a law is a hypothetical the satisfaction of whose antecedent nomologically necessitates the satisfaction of its consequent—wasn't quite true since it doesn't quite apply to hedged laws. What is true is that a law is a hypothetical the satisfaction of whose antecedent nomologically necessitates the SaHsfacfion of its consequent when its ceteris paribus conditions are satisfied.

14. If it's a strict law, then the ceteris paribus clause is vacuously satisfied.

15. To put it another way: Suppose you're feeling Hempelian about the role of covering laws in scientific explanations. Then you might worry that (i) ceteris paribus As cause Bs together with (ii) Aa yields something like (iii) ceteris paribus Bb which isn't strong enough to explain the datum (Bb). 'Ceteris paribus Bb' doesn't look to have the form of a possible data statement. I wonder in the text whether it even has the form of a possible truth.

16. "Eventually" means: either the law is implemented by a physical mechanism, or its implementation depends on a lower level law which is itself either implemented by a physical mechanism or is dependent on a still lower level law which is itself either implemented by a physical mechanism or..., etc. Since only finite chains of implementation are allowed, you have to get to a physical mechanism "eventually."

We need to put it this way because, as we've been using it, a "physical" mechanism is one whose means of operation is covered by a physical law (i,e., by a law articulated in the language of physics). And though, presumably, physical mechanisms implement every high-level law, they usually do so via lots of levels of intermediate laws and implementations. So, for example, intentional laws are imple­mented by syntactic mechanisms that are governed by syntactic laws that are implemented by neuro­logical mechanisms that are governed by neurological laws that are implemented by biochemical mechanisms that... and so on down to physics.

None of this really matters for present purposes, of course. A demonstration that mental events have neural properties would do to solve the mind/body problem since nobody doubts that neural events have physical properties.

17. On the other hand, 1 don't pretend to do what Davidson seems to think he can: viz., to get physicalism just from considerations about the constraints that causation places on covering laws together with the

truism that psychological laws aren't strict. That project was breathtaldngly ambitious but maybe not breathtaldngly well advised. My guess is: If you want to get a lot of physicalism out, you're going to have to put a lot of physicalism in; what I put in was the independent assumption that the mechanism of intentional causation is physical.

Introduction

The selections from Thomas Aquinas, Thomas Hobbes, and Rene Descartes raise the question ofwhether thought is imagistic. For Aquinas the answer is yes (for people) and no (for noncorporeal beings). For Hobbes the answer is yes. For Descartes the answer is no. According to this trio of writers, imagery is something that only embodied creatures can have. This may seem surprising in the case of Descartes, for imagery seems to be a canonical mental phenomenon. But Descartes believes that if we were disembodied, we could no longer have images. Descartes makes a distinction between images and ideas, pointing out that you can have an idea of a chiliagon (a thousand­sided figure) but not an image of one.

Hobbes also articulates two important themes that recur in later philosophy: that mental imagery is decaying sense, and that the faculty of imagination is identical to the faculty of memory.

David Hume addresses the relation between perception, memory, and imagination. Like Hobbes, he argues that memories and mental images are merely sense perceptions that are less vivid. The crucial difference between memories and images, he argues, is that memories (though less vivid than perceptions) are still more vivid than images.

William James takes up the themes introduced by Hobbes and Hume. He argues for the physiological basis of mental images, maintaining that the processes underlying imagery are identical to the processes underlying perception. He cites tum-of-the- century neurophysiological data, -including a study that suggests that if the vision center of a sighted person is severely damaged, the person not only loses vision but will not even be aware of a deficit. The reason, according to James, is that the person can no longer have mental images and thus can have no idea of what is now unseen.

Following Hume, James argues that images are simply less vivid sense impressions. James suggests that there are common experiences that support this idea. For example, if a baby cries in a distant room, one may be unsure whether one is actually perceiving a baby or imagining it. That is because the perception is so faint that it is no more vivid than an auditory image.

James also addresses the question of whether thoughts (ideas) are identical to images. He denies both that general thoughts correspond to vague images and that thoughts about particular objects correspond to sharp images. So, for example, he approves of Berkeley's observation that the idea of a triangle does not correspond to a vague or confused image. There may rather be a sharp image of some prototypical triangle. The converse also holds. One may have a vague image of a particular individual.

Oswald Kiilpe (a continental psychologist at the turn of the century) also takes exception to the claim that all thought is imagistic. Kulpe notes that introspective experiments show that certain mental activities cannot be reduced to images. For exam­ple, the acts of attending, willing, and so forth, do not appear to be imagistic.

John Watson argues that mental images merely stand in the way of a proper scientific psychology (specifically, behaviorism) and suggests that they must be dispensed with altogether. According to Watson, mental images have no place in behavioral psychol­ogy for they are not publicly observable.

Gilbert Ryle challenges the idea of mental pictures as well, but on conceptual grounds rather than scientific grounds. He notes that if a child imagines a smile on a doll's face, the child does not actually see a smile floating in front of the doll. Rather the child is playing (behaving) as though the doll is smiling. Ryle goes on to note that the case for mental images is not so appealing when we consider the other sensory modalities. While we are quick to talk of the "mind's eye" seeing a mental picture, Ryle suggests that no one would speak of the "mind's nose" smelling a mental aroma.

Ryle also takes issue with the idea (from Hobbes, Hume, and James) that images can be thought of as less vivid sense impressions. Ryle notes that while dolls may be lifelike, we would never say that a real baby is lifelike. Likewise we may call an image vivid, but we would never consider an actual sense impression vivid. Moreover, an image of a loud noise is not loud and will not even drown out someone whispering. Ryle con­cludes that it is a simple confusion to try and compare the vividness of an image with that of a sense impression.

Daniel Dennett suggests several additional reasons for doubting that there can be anything like pictures in the head. First, he notes that most images share some physical property with the object they represent. So, for example, an image of an orange must be either round, or orange, or both. The question is, what physical properties could mental images possibly share with the things they are images oft

Second, Dennett notes that images, unlike pictures are incomplete. Close your eyes and imagine a tiger for a second or two. Now, how many stripes were visible in the image? Probably there is no answer to the question because the image was vague. On the other hand, there is a determinate answer to how many stripes are visible in a picture of a tiger.

In response to such doubts there is a great deal of work in cognitive psychology that attempts to establish the existence of visual mental images. Shepard and Metzler report an experiment in which subjects are given two pictures of geometrical figures and are asked to indicate whether the figures are identical. In some cases the second figure is identical to the first but rotated; in such cases the amount of time it takes the subject to indicate that the figures are identical is directly correlated with the degree to which the second figure has been rotated. The further the image has been rotated, the longer it takes the subject to respond. Shepard and Metzler conclude that the subject forms a mental image of the figure and rotates the mental image at a certain limiting rate.

Stephen Kosslyn reports the results of scanning experiments in which subjects must scan (with their "mind's eye") between points on an image. For example, a subject might be instructed to attend to a particular location on a mental image of a map and indicate whether there is a lake in another location on the mental map. The greater the relative distance between the two points, the longer it takes the subject to respond. Kosslyn concludes that this time difference can be accounted for if we suppose that the subject is scanning a mental picture at a certain limiting rate.

Zenon Pylyshyn remains unconvinced by these experiments. He argues that the experimental conditions are such that the subjects understand their task as imagining that they are actually scanning a map. Utilizing their real-world knowledge about scanning maps, they delay their reaction times to correspond to the time that they realize it would take to scan an actual map.

Kosslyn replies to Pylyshyn on this point citing a number of recent experiments designed to show that die subject has no expectation that it should take longer to scan greater distances.

FurtherReading

Block, N., ed. 1981. Imagery. Cambridge, MA: MTT Press.

Finke, R. 1989. Principles of Mental Imagery, Cambridge, MA: MTT Press.

Kosslyn S. 1980. Image and Mind. Cambridge, MA: Harvard University Press.

Pinker, S., ed. 1984. Visual Cognition. Cambridge, MAi MTT Press.

Shepard R., and L Cooper. 1982. Mental Images and their Transformations. Cambridge. MA: MTT Press. Tye, M., 1991. The Imagery Debate, Cambridge, MA: MIT Press.