ROBERT VAN GULICK
ARE BELIEFS BRAIN-STATES? AND IF THEY ARE WHAT
MIGHT THAT EXPLAIN? 1
(Received 9 March 1994)
Although I find myself more or less in sympathy with Lynne Baker's general views on explanation, reduction and supervenience, I find her position less than clear on some issues, 2 and I am less than convinced by some of the arguments she offers.
Let me first address the two main issues on which I find her paper less than clear. Her critical target is the view "that in order to explain behavior causally, beliefs must be brain states." Let's call this the brain- explain assumption. She takes this to be a common assumption of those who hold what she calls the Standard View, i.e. the thesis that beliefs are brain-states in any of several familiar senses - token identity according to which beliefs are identical to brain-state tokens, nonreductive materi- alism according to which beliefs are constituted by brain-states... ; and functionalism, according to which beliefs are functional roles occupied by brain-states. Prof. Baker may be correct that the "must be brain to explain" thesis is commonly assumed by those who hold the Standard View (though as an empirical matter I don't know how often holders of the Standard View in fact make such an assumption.) But as a logical matter the assumption is not entailed by the Standard View nor does Prof. Baker claim that it is.
It is here that the first unclarity arises; just what does Prof. Baker take herself to be rejecting and arguing against~? Is her target merely the common assumption or also the Standard View itself? I assume she wants to reject both since she goes on to propose an alternative to the Standard View, but her arguments are directed at the assumption. Since the brain explain assumption is not entailed by the Standard View, demonstrating its falsity would not suffice to falsify the Standard View;
Philosophical Studies 76:205-215, 1994. 1994 KluwerAcademic Publishers. Printed in the Netherlands.
206 ROBERT VAN GULICK
there's no immediate modus tokens to be had. One way to interpret her would be as making the following argument:
The brain/explain assumption provides the main reason for holding the standard view.
But that assumption is false.
Thus we are left with no good reason for the standard view and we should reject it.
This general approach is suggested by her talk of "undercutting the cen- tral motivation for the Standard View" But I am reluctant to attribute such a line of reasoning to her since it shows so many gaps when put in such blunt argument form:
1. Is the brain/explain assumption in fact the central motiva- tion for the Standard View? What analysis of the literature supports that claim?
2. Just because a reason for a given view fails it doesn't follow that there are no reasons for it.
3. Even if we had at present no reasons for accepting a view, it of course wouldn't follow that the view is false and should be rejected?
Thus I think it makes more sense to interpret her as intending to forge a more modest link, that of demonstrating the falsity of the brain/explain assumption in order to cast doubt on the Standard View without actually refuting it. I hope that she will tell us whether that is a fair reading of her project in her present paper)
However a second unclarity calls even this more modest interpre- tation into question. Though Prof. Baker pretty clear wishes to reject the Standard View, at one point in her paper she says some things might seem to imply acceptance of at least one version of it. I have in mind the remarks she makes regarding constitution and supervenience (pp. 182-183). She suggests that philosophers have gotten into trouble by conflating the two relations, which coincide in the case of nonre-
ARE BELIEFS BRAIN-STATES? 207
lational properties of physical objects but come apart in the case of intentional properties. Applying this distinction to her example of the S&L's bad investment. She refers to "the fatal equivocation: the S&L's investments are constituted by the U-state, but the property of being an investment does not supervene on the property of being a U-state" (p. 183 italics original). But since the whole thrust of the first half of Prof. Baker's paper is to assimilate belief-explanations to nonpyschological intentional ones (such as those about S&L investments and failures) the passage above seems to imply that we should apply the same distinction to beliefs and brain-states. That is, beliefs do not supervene on brain- states, though they are constituted by brain-states. However, that would be to accept one of the versions of the Standard View listed at the outset: "nonreductive materialism, according to which beliefs are constituted by brain states" (p. 175). I don't think Prof. Baker wants to do that since she clearly want to reject the Standard View tout course, but I don't see how to reconcile that with what she says about supervenience versus constitution. I hope she can clear up my confusion. I might add that since I consider myself a nonreductive materialist and have labeled myself as such in print, I would like to know more about why she finds that particular position unacceptable. From my perspective it accords well with much else that I find in her position. Where does she see a conflict?
Having made my two requests for clarification, let me turn then to Prof. Baker's arguments. Part 1 of her paper focuses on an argument against the brain/explain assumption; it is in the form of a conditional argument with one assumption and four premises 4
Beliefs have brain states as constituents.
Intentional explanations generally are irreplaceable by phys- ical explanations of the constituents of the intentional phenomena.
Belief-explanations are intentional explanations.
Belief-explanations are causal explanations.
The causal-explanatoriness of beliefs requires beliefs to be
208 ROBERT VAN GULICK
brain-states only if belief-explanations are replaceable by brain-state explanations of the same phenomena.
She makes three inferences, reasoning as follows
From (P1) (P2) and (CP) to
(C1) Belief-explanations explanations.
From (P3) and (P4) and (C1) to
can not be replaced by brain-state
(C2) Beliefs need not be brain explanatory.
And finally discharging (CP) to
states in order to be causally
(C3) Even if beliefs are brain states, beliefs need not be brain states in order to be causally explanatory.
Prof. Baker pronounces the argument valid and concentrates her efforts on defending premises (P1) and (P4), leaving (P2) and (P3) to be defended elsewhere.
I find no problem with the logic of the present argument, 5 so let us consider what might be said about the truth of its premises, especially the two premises Prof. Baker undertakes to defend? I have no quarrel with (P1); indeed I have defended a similar position myself 6 and regard it as a central tenet of nonreductive materialism and of what is sometimes referred to as the "autonomy of the special sciences." If one is serious about the pragmatic dimensions of explanation, i.e. about such factors as the context of use, the goals and interest they are meant to serve, and the limits on the mental powers and memories of actual agents, then the practical indispensibility of special science explanations including economic and other intentional explanations seems obvious.
Premise (P4), however, seems more problematic. It makes a rather strong claim since it asserts that there is only one way in which the causal explanatoriness of beliefs could depend on their being brain states, namely if belief-explanation were replaceable by brain-state explana-
ARE BELIEFS BRAIN-STATES? 209
tions of the same phenomena. (P4) is in effect a negative existential asserting that there are no other possible ways in which such depen- dence could arise, and like all negative existentials it faces a tough burden of proof.
Prof. Baker believes (P4) follows from a general principle which she finds intuitive but admits she is unable to prove. Since the principle strikes me as less than obvious, 7 we need to consider the other reasons she gives in support of (P4)? She asks us to imagine that the universe had turned out to be Aristotelian. If so, she argues beliefs would have tumed out not to be brain-states, but our explanations in terms of beliefs would not have been affected at all. As she puts it, "if the sensible and social world were the same as in our world, we would not have had the same range of explanations, deployed in the same ways and with the same degree of success, that we actually have" (p. 187). She thus concludes that the explanatoriness of beliefs does not require them to be belief-states since they would have been just as explanatory in the Aristotelian world in which they are not brain-states.
I am unpersuaded by this argument, though it may be because I fall to appreciate all that is packed into the hypothetical assumption of an Aristotelean world. Though it seems possible to imagine a world in which beliefs are not brain-states but we still successfully use belief explanations, it's not clear what follows from that about what beliefs are in our world. Consider a parallel argument about genes. We can imagine a world in which genes did not tum out to be identical with sequences of DNA; perhaps they tumed out to be proteins. In such a world we might still be able to give and use all the sorts of explanations associated with classical genetics quite successfully. But nothing would seem to follow about genes not being DNA sequences in the actual world. And of more immediate relevance it would not follow that the causal explanatoriness of genes in our world is not dependent on their being DNA sequences. Why then should the situation be any different with respect to beliefs?
Those who have supported the Standard View by appeal to the causal explanatory role of beliefs have sometimes argued roughly as follows
. In both our folk and scientific psychologizing we need a notion of event-causation as well as a notion of dispositional
210 ROBERT VAN GULICK
causation, i.e. we need a notion of one mental event caus- ing another (as in reasoning when one thought gives rise to another whose content is a logical consequence of the content of the first) and a notion of mental events causing behavior (as when my realization that I had forgotten my wallet causes me to hit the brakes and pull back into the driveway).
2. In the actual world the only things that are available for filling the relevant internal causal roles are brain-states (and brain-events).
3. Thus actual beliefs (i.e. beliefs in the actual world) must brain-states.
Note that this conclusion is quite compatible with beliefs not being brain-states in some other possible world - Aristotelean or otherwise. If being a belief involves being the occupant of a certain causal role, there is no reason why the occupants of those roles couldn't be brain-states in our world but something quite different in other possible worlds. It's just a modal or cross-world version of multiple realizability.
A proponent of the Standard View might give such an argument to show that unless beliefs are brain-states we can't account for the full range of their causal explanatoriness in the real world in particular we wouldn't be able to account for our successful real world appeal to beliefs in explanations invoking event-causation. The counterfactual possibility of there being beliefs in some other world that are not brain- states would not really affect what must be true in the real world to account for their explanatory role here. Something must occupy the relevant causal roles here, and brain-states are the only real world can- didates. Or so at least a supporter of the Standard View might argue. I don't see how Prof. Baker's imaginary example of an Aristotelean world would have any force against such reasoning. I suspect she would challenge the argument's first premise and deny that we need a notion of event-causation covering mental states, but that's then where the real fight must be fought and it's not an issue that can be settled by appeal to her imagined Aristotelean world.
Let me summarize my response to the Aristotelean world example. It's largely about the scope of the word 'must'. When the supporter of
ARE BELIEFS BRAIN-STATES? 211
the Standard View claims that beliefs must be brain states, that is a claim made relative to the actual world and its actual causal structure, it is not intended as modal claim about all words in which belief explanations can be used. Counterfactuai Aristotelean words have no bearing on such claims solely about what must be true in the actual word to account for our actual explanatory practice.
Thus I don't find that Prof. Baker has made a convincing case for (P4) nor shown that her master argument against the brain/explain assumption is sound as she claims. And in so far as she has not refuted the assumption, she has not thereby succeeded in casting doubt on the Standard View.
There may of course be other ways to undermine the Standard View, such as offering an attractive alternative account of what beliefs are and how they explain, and that is just what Prof. Baker sets out to do in the final section of her paper. I will devote the balance of my comments to assessing her alternative.
According to Prof. Baker, beliefs are not brain-states; they are "global states of whole persons" (p. 192). She summarizes her posi- tive view into two theses:
A. S has a belief that p at t if and only if 'S believes that p' is true at t.
B. 'S believes that p' is true at t if and only if there is an associated set of counterfactuals true of S at t.
In support of her claim that beliefs are global states of whole persons, she invokes the following analogy: "Horses win races; legs have states." I agree. Leg states are needed to win a horse race, but legs don't win races though a horse may win a race "by a leg." Sometimes wholes have properties that are not properly described as states of their parts. People eat with their mouths but mouths don't eat, nor is eating prop- erly described as a state of one's mouth. However, some properties of wholes are aptly described as states of their parts. Pancreases don't have diabetes; people have diabetes. But diabetes is properly described as a state of one's pancreas. People, not lungs, have emphysema, but emphysema is a state of one's lungs. Similarly it is people not brains
212 ROBERT VAN GULICK
that are comatose, but to be comatose is surely to be in a certain type of brain-state. What then of belief? Is it more like being comatose or like eating and winning a race? Adding the modifier "global" to "state of a whole person" may unfairly bias the answer. Although Prof. Baker talks of "global states of persons" and in an earlier version of her paper spoke of "discrete brain-states" the proponent of the Standard View need not be committed to discreteness; that's a separate issue. There is no reason why one couldn't hold that beliefs are relatively global brain- states. Some specific versions of the Standard View, such as the inner sentence theory, might be committed to discreteness and local represen- tation but the Standard View per se makes no commitment on the issue. It asserts only that beliefs are brain-states - local and discrete or global and distributed. Although Prof. Baker sometimes invokes connec- tionism against the Standard View, there is again no real conflict. One can be a connectionist and still hold that beliefs are brain-states, as long as one gets the right sort of brain-states, i.e. global and distributed ones. Thus if one leaves out the distinct issue of global versus local, it is far from obvious that beliefs are not better treated on a par with being comatose than with winning a race. One can admit that it is people not brains that have beliefs while still maintaining that to have a given belief is to be in a relevant sort of brain state.
Let me try to motivate that view with a little experiment that requires only that you listen carefully to the next sentence I utter, which I promise you is true. My maternal grandmother, whose name was Agnes, lived to be 101 years old. As the result of hearing what I just said each person in this room 8 has acquired a new belief, since until a moment ago no one here knew anything about my grandmother Agnes. Were I to compare the state of any one of your brains just prior to and just after hearing and understanding my statement, there would have to be a physical difference in each case that corresponded to your acquisition of that belief. It need not be a local change; it might involve changes in many different parts of your brain. Nor need the same change - described in neuroscientific terms - have taken place in all of you; there can be lots of variation across individuals. But if I were now to ask any one of you the name of my maternal grandmother or her how old she lived to be, it would be that resultant brain state that would be responsible
ARE BELIEFS BRAIN-STATES? 213
for your answering as you would - assuming you are cooperative and trying to give what you took to be the correct reply. This seems to me to both a very common sensicai way to think about how beliefs cause behavior and one that treats beliefs as brain-states. It doesn't assume beliefs are local brain-states, or sentences in the head, or type-identical with any properties specified in the language of neuroscientific theories. It is agnostic on all those issues. But it does view beliefs as real causally efficacious brain-states, that qualify as beliefs in virtue of the causai functional roles they occupy, roles that in part involve making true just the sorts of counterfactuals to which Prof. Baker refers. As brain-states the occupants of those causal roles reside within the head, but the roles they occupy need not be similarly restricted. Nothing prevents one from categorizing such roles widely, bringing in social and environmental factors that reach well beyond the confines of the head and body. Wide functionalism is just as much an option as narrow functionalism, and which is the best way of specifying and individuating roles will likely vary with context and one's explanatory interests. And finally in speci- fying those roles we need not accept any reductive restrictions that limit us to specifying those roles solely within the linguistic and conceptual resources of nonintentional physical theory. The sorts of reasons that Prof. Baker has given against reductive replaceability and that I have endorsed in favor of the autonomy of the special sciences apply equally to how we should go about specifying functional roles in explicating belief and the other attitudes.
The upshot would seem to be that one can eat a good deal of the cake that Prof. Baker has baked while still keeping the view that beliefs are brain-states. In particular one could hold all of the following:
Belief explanations cannot be reduced to or replaced by phys- ical explanations without explanatory loss.
Beliefs are not local brain-states nor sentences in the head.
For many purposes beliefs should be individuated widely in a way that is sensitive to social and environmental factors.
It is people, not brains, that have beliefs.
214 ROBERT VAN GULICK
while also consistently holding that
5. Beliefs are brain-states.
Indeed I count myself 9 as holding that very conjunction of 1-5. Perhaps understood in this way, Prof. Baker will be willing to accept
the thesis that beliefs are brain-states though I suspect she won't. I do wonder why she won't since it seems a congenial match with so much else that she does accept. I f she won't accept my invitation to become a reformed brain-state theorist, I hope she will nonetheless explain why she finds the view mistaken.
i This paper is an abridged version of the commentary that I delivered at the t 993 Oberlin Philosophy Colloquium in response to Prof. Lynn Baker's paper entitled "Attitudes as Nonentities". I have deleted several sections of my original paper because they raised criticisms against parts of Prof. Baker's colloquium paper that have been dropped or significantly modified in her present published version. I have noted in the text some of the places where I have made deletions, but have otherwise added nothing to the commentary. I am pleased that my comments were of value of Prof. Baker in revising her paper and I am grateful for her generous acknowledgement. 2 Several of the unclarities in her original paper, especially regarding the form of her master argument have been remedied in the present version of her paper and remarks in those regards have been deleted. 3 At the Oberlin Colloquium, Prof. Baker responded that such a reading did indeed reflect her general intent. 4 The argument as presented at Oberlin did not have this form. It was not a conditional proof, a fact which generated some apparent problems of inconsistency. It also differed somewhat in its other premises and relied upon the vague notion of assimilating one type of explanation to another. The present argument avoids both those defects and I have deleted my critical comments accordingly. 5 I did have problems with the reasoning of the original argument especially with its reliance on the vague notion of assimilation, but again those defects have been remedied in the present version of her paper. 6 See for example my paper"Nonreductive Materialism and the Nature of Intertheoret- ical Constraint" in A. Beckermann, H. Flohr and J. Kim (eds.) Emergence or Reduction ? Essays on the Prospects of Nonreductive Physicalism Walter de Gruyter, Berlin and New York, 1992, pp. 157-79.
ARE BELIEFS BRAIN-STATES? 215
7 The principle as stated in the original Oberlin paper was open to clear counterexamples. The addition of a level restriction to the present version of the principle blocks those counterexamples, and I have deleted them accordingly. However, even in its present version the principle strikes me as far from obvious and thus not able to carry the weight of the argument without additional argument of the sort that Prof. Baker attempts to provide. s Or in the present case as the result of reading the above passage, you the reader have acquired the relevant belief. 9 See for example my "Nonreductive Materialism and the Nature of Inter-theoretical Constraint" cited above.
Department of Philosophy Syracuse University Syracuse, NY 13244 USA