« Kim on General Mental Properties
Is Categorical Phenomenalism Physicalist? »

Kim on the Functionalizability of Qualia Differences and Similarities

By the end of Physicalism, Or Something Near Enough, Kim arrives at the view that all mental properties except qualia are reducible. Qualia, because they cannot be functionalized, resist reduction. Kim’s view is meant to be “near enough” to physicalism in that the cognitive and the intentional (but not the qualitative) are physically reducible. [As noted over at fragments of consciousness, the resulting view looks to be a form of epiphenomenalist dualism with regard to qualia.] Kim is particularly interested in arriving at a view of the mental that saves most of our intuitions about causal efficacy. To this end, he claims that in addition to the cognitive/intentional, the differences and similarities among qualia (but not their intrinsic characters) are functionalizable and “can enjoy causal powers as full members of the physical world”. But I am not convinced that this is a tenable position, nor that it is near enough to physicalism to satisfy our central intuitions of causal efficacy.

Let me start with the “near enough” bit. A favorite example of the causal efficacy of the mental is the experience of pain, and its causing a subject to engage in pain-related behavior. Pain has a very distinctive, and displeasing, phenomenal character. It seems to me that part of the causal explanation for my wanting to avoid sharp pointy things is precisely because I have found in the past that they can cause me to have very unpleasant painful experiences. I want to avoid having experiences with that intrinsic phenomenal character. The “painy” feel of pain strikes me as intrinsically aversive. Perhaps after doing some philosophy we decide we have to give up on this intuitive idea. But that would be a significant rather than minor departure from our conception of the causal efficacy of the mental.

What about the strategy of functionalizing qualitative similarities and differences? Note that even if this can be done, it does little to salvage the intuition of causal efficacy in the case of pain versus pleasure. However, it does seem more promising in the case of experiences of color. But I am not sure that, insofar as qualitative similarities and differences can be functionalized, these relations provide an element of causal efficacy for qualitative states.

To begin, notice how odd it would be for a relation between two properties to be physically reducible, but not the relata. This at least strikes me as very odd (perhaps someone can disabuse me of this feeling). Certainly there are relations that we think of as physical relations, such as temporal relations, that might conceivably be shared by non-physical relata. But if there really are non-physical relata that can be temporally related, then perhaps temporal relations are not strictly physical. Or at least the instances of temporal relations that relate non-physical relata are not physical. At any rate, the case at hand is importantly different, in that relations of phenomenal similarity are relations that, assuming qualia are not physical, could not possibly relate physical properties. If intrinsic phenomenal properties are physically irreducible, then I don’t think it makes much sense to suppose that a relation such as x being more phenomenally similar to y than z could possibly be physically reducible.

Kim supports the idea that phenomenal similarity is functionalizable by pointing out that differences in a subject’s color experiences are manifestable in behavior. For example, the subject’s ability to discriminate objects based on their color will indicate whether two or more color experiences had by the subject are the same or different. From their behavioral detectability, Kim concludes that phenomenal similarities and differences are functionalizable, and therefore also physically reducible and causally efficacious. But these moves, I think, are made much too quickly. To put my worry briefly, behavioral detectability will only support functionalizability (and thus physical reduction) if it known that the relevant differences in behavior are caused by, and not merely correlated with, differences in color qualia. But surely we cannot help ourselves to the idea that the so-called “behavioral manifestations” of differences among qualia are literally caused by differences among qualia–the latter is part of what Kim is attempting to establish. And having conceded that the intrinsic aspects of qualia are themselves physically irreducible and causally impotent, Kim is not in a position to insist that any law-like relationship between qualia differences and behavior must be a causal one. For presumably he is willing to suppose that qualia nomically supervene on the physical, despite being physically irreducible. It would follow from nomic supervenience alone that differences in behavior within an individual can indicate differences in color qualia. So behavioral detectability is not enough to support the physical reduction of similarities and differences among qualia.

Posted by Brad Thompson on Monday, October 30th, 2006 at 11:37 pm.

3 Responses to “Kim on the Functionalizability of Qualia Differences and Similarities”

Hi Brad,
Very interesting post. Let me check if I understood right the first part as I haven’t read Kim’s book… Kim considers the model where the “intristic character” of qualia is not physically reducible and not causally efficient. So, you point that according to our intuition it is the intristic phenomenal character of pain that we want to avoid, which would also mean that this intristic character of qualia is causally efficient.

I think similar case can be made for colors. People tend to like more some specific colors and dislike others. Some people have favorite color(s). And as in the case of the pain, the intuitive account is that they do that based on the intristic phenomenal character of the color. So again it doesn’t seem that intristic character of qualia is causally impotent, as people can say “my favorite color is blue”, or it might affect what they wear, and so on.

Also, how can this sort of epiphenomenalism about intristic character of qualia account for our talking about that intristic character of qualia? (It seems clear that our talk about intristic character of qualia is because qualia has that intristic character, hence the intristic character is not causally impotent). Of course this goes for any epiphenomenalism. I’m not sure how can one get over this?

Left by Tanasije Gjorgoski on October 31st, 2006

Hi Brad,

Thanks for sending me your blog entry on some things I’ve said in my recent book. You make your points clearly and compellingly, and they obviously raise substantive issues. I am not sure I can respond in a satisfactory way, but let me give you my initial reactions.

Let me start with the “near enough” bit. A favorite example of the causal efficacy of the mental is the experience of pain, and its causing a subject to engage in pain-related behavior. Pain has a very distinctive, and displeasing, phenomenal character. It seems to me that part of the causal explanation for my wanting to avoid sharp pointy things is precisely because I have found in the past that they can cause me to have very unpleasant painful experiences. I want to avoid having experiences with that intrinsic phenomenal character. The “painy” feel of pain strikes me as intrinsically aversive. Perhaps after doing some philosophy we decide we have to give up on this intuitive idea. But that would be a significant rather than minor departure from our conception of the causal efficacy of the mental.

That is of course what’s so counterintuitive, and perhaps disturbing and unsettling, about epiphenomenalism. I think that’s true of any philosophical thesis that involves an error theory–a claim that a whole range of statements we ordinarily take to be true, and obviously so, are in fact false. I wonder whether epiphenomenalism about qualia is worse off in this regard than, say, moral antirealism or the thesis that there is no free choice/action. Perhaps, epiphenomenalism is like the claim that there is no freedom in that we cannot live by it, that is, we cannot live as though it is true. My feeling is that in this respect the thesis of no freedom is worse off than qualia epiphenomenalism. But can’t we see that as philosophers we might very well defend, as many have, the position that there in fact is no free action? One further question that arises for me is why we shouldn’t go for an error theory about the whole bag of qualia in general (I suppose, like Dennett) instead of a restricted error theory about qualia causation. I have sometihng to say about this but will save it perhaps for another occasion.

To begin, notice how odd it would be for a relation between two properties to be physically reducible, but not the relata. This at least strikes me as very odd (perhaps someone can disabuse me of this feeling). Certainly there are relations that we think of as physical relations, such as temporal relations, that might conceivably be shared by non-physical relata. But if there really are non-physical relata that can be temporally related, then perhaps temporal relations are not strictly physical. Or at least the instances of temporal relations that relate non-physical relata are not physical. At any rate, the case at hand is importantly different, in that relations of phenomenal similarity are relations that, assuming qualia are not physical, could not possibly relate physical properties. If intrinsic phenomenal properties are physically irreducible, then I don’t think it makes much sense to suppose that a relation such as x being more phenomenally similar to y than z could possibly be physically reducible.

I gave a talk several years ago on this material and someone (I think it was Jim Pryor) asked the following question. I am saying that something X and something Y are epiphenomenal. But I am also claiming that X standing in a certain relation R to Y has causal powers. Isn’t this strange? How can this happen? The point you raise seems to be in the same, or at least nearby, ballpark. I don’t have an answer to these general metaphysical questions. But I don’t see any obvious inconsistency in the claim that although X and Y are each physically irreducible, that X stands in R to Y is physically reducible. Here, the focus is on R, and I don’t see that the irreducibility of X and Y must affect the reducibility of R one way or another. A similar comment would apply to Jim Pryor’s point.

Kim supports the idea that phenomenal similarity is functionalizable by pointing out that differences in a subject’s color experiences are manifestable in behavior. For example, the subject’s ability to discriminate objects based on their color will indicate whether two or more color experiences had by the subject are the same or different. From their behavioral detectability, Kim concludes that phenomenal similarities and differences are functionalizable, and therefore also physically reducible and causally efficacious. But these moves, I think, are made much too quickly. To put my worry briefly, behavioral detectability will only support functionalizability (and thus physical reduction) if it known that the relevant differences in behavior are caused by, and not merely correlated with, differences in color qualia. But surely we cannot help ourselves to the idea that the so-called “behavioral manifestations” of differences among qualia are literally caused by differences among qualia–the latter is part of what Kim is attempting to establish. And having conceded that the intrinsic aspects of qualia are themselves physically irreducible and causally impotent, Kim is not in a position to insist that any law-like relationship between qualia differences and behavior must be a causal one. For presumably he is willing to suppose that qualia nomically supervene on the physical, despite being physically irreducible. It would follow from nomic supervenience alone that differences in behavior within an individual can indicate differences in color qualia. So behavioral detectability is not enough to support the physical reduction of similarities and differences among qualia.

When I was working on the book, I had an idea about how functionalization of phenomenal similarities and differences might go but I never worked things out in detail. Now I can’t really remember what I had in mind then. So I have to approach this question from scratch. Let’s take as our target of functionalization the following: q1 and q2 are phenomenally indiscernible to S (where the q’s are qualia and S is a subject). The following would seem serviceable as a general form of functional characterization of this term.

q1 and q2 are phenomenally indiscernible to S = def. S is in some state T such that if systems like S are in state T, this causes them to exhibit behavior of kinds K1, …, Kn.

Being in state T would also typically cause S-like systems to go into various mental states as well, e.g., judging in certain ways, forming certain beliefs, etc., but we can skip that here. Anyway, what are these behaviors K1, …, K2? What I have in mind of course is various forms of discriminative behavior–in the present case, since the two qualia are indiscernible to S, the behavior kinds would mainly concern the absence of certain discriminative behavior. But discriminative behavior toward what objects? Here, I would say something like this: toward objects that present q1 and q2 to S. What does “present” mean? Your concern, transposed to this context, would be, I believe, that this “present” must mean “cause”. As you would have seen, I want to avoid that–for me, qualia are epiphenomenal in a stronger sense than T.E. Huxley’s, namely qualia have neither causes nor effects. My answer is that here nomological supervenience or correlation is all we have and that that’s enough for our purposes. It so happens that lemons nomologically correlate with yellow qualia, and in that sense lemons present yellow quale to sensory systems like us. So if lemon1 and lemon2 present q1 and q2 to S respectively, then, given that q1 and q2 are indiscernible to S, S presumably would be unable to discriminate between lemon1 and lemon2 on the basis of their color. A lot of details remains to be fixed and many caveats have to be entered, but that is the general idea. The main point in reply to your argument is that to functionalize phenomenal indiscernibility between qualia this way, we do not, as far as I can tell, need to presuppose that qualia are causally efficacious.

In any case, the above is what I am inclined to say now. If I keep thinking about it, I may want to say something different tomorrow, but probably nothing wholly different.

Left by Jaegwon Kim on November 1st, 2006

Hi Tanasije,

I am inclined to agree with you about the causal efficacy of color qualia, and I think your examples are good ones. I focussed on pain because it is perhaps the most obvious and intuitive case in which intrinsic phenomenal character seems to play a causal role. The color case is less obvious, as witnessed by the fact that behaviorally undetectable spectrum inversion is much easier to imagine than a subject for whom pain and pleasure are swapped.

Good question regarding the intrinsic character of qualia and our talk about qualia. Certainly one of the most disconcerting things about qualia epiphenomenalism is that, on this view, our talk of qualia is not caused in any way by our having qualia. This is related to what Chalmers discusses as the “paradox of phenomenal judgment” in chapter 5 of The Conscious Mind. I do think that this consequence of the view is one of the best reasons to hope for an alternative to epiphenomenalism.

Left by Brad Thompson on November 3rd, 2006

Something to say?