This morning I stumbled upon David Rosenthal’s forthcoming Philosophical Issues paper “How to Think About Mental Qualities” (available on his website). In this paper he argues that inverted qualia are inconceivable. He says,

Quality spaces cannot be symmetrical around any axis without making it impossible to distinguish qualities on one side of the axis from qualities on the other. And [our higher-order awareness] makes one aware of mental qualities in respect of their relative position in the relevant quality space. So that lack of symmetry carries over to the way the [higher order states] make us aware of mental qualities. Undetectable quality inversion is accordingly no more possible for conscious than for non-conscious mental qualities.

The first part of the argument is supposed to show that the usual move, that if saying that the quality space is conceivably symmetrical, for principled reasons about the way perception works. Any quality space complicated enough to do justice to human discriminatory capabilities will have to be asymmetric. This means that any inversion will be detectible from the third person and so quality inversion without detection is inconceivable. The intuition that it is conceivable is a reflection of one’s already accepting what Rosenthal calls a “consciousness-based” theory of the mental qualities. I am deeply sympathetic to this line of argument but I think that there is a problem with Rosenthal’s line of argument in the above passage.

Suppose that we grant the point about the asymmetric quality space, and that the content of the higher-order awareness is also asymmetric. Suppose that in both you and I when we are presented with an orange stimulus the same first-order sensory quality is produced and that this quality is more like the quality that represents yellow and red than it is like the one that represents green and blue, etc. But now suppose that in you there are higher-order thoughts that accurately reflect these similarity and differences. So in you, when the orange quality is consciousness, one has a higher-order state that picks it out as more similar to the yellow* and red* qualities than it is to the green* and blue* qualities. All along, though, we may suppose, in me the ‘opposite’ is happening. When I have a first-order quality that represents orange my higher-order awareness picks it out as the one that is more like green* and blue* than it is like yellow* and red*. But suppose also that I have the same perceptual thoughts that you do. So, I believe that there is something orange near me when I have orange* qualities and when that thought is conscious I consciously believe that something orange is near me. But what it is ike for me in this situation will be like it is for you when you see blue. So we have different conscious experience that is not detectable from a third-person standpoint. We may suppose even that when I introspect my third-order thoughts “get it right”. So, when I introspect my higher-order awareness that the quality is more like green than it is like orange, I am aware of it as one that has the quality more like yellow than it is like green. Thus my introspective reports will exactly match yours. I will also match you in all of my discriminatory abilities.

Granted that this is detectable in principle since we are stipulated to have different higher-order states, which could ultimately be detected in some very advanced neuroscience, but it is nearly undetectable and this is all we need t ground the intuition. In short, it takes a lot of work, even on a purely physicalist conception of the mind like quality space and higher-order theory, to show that these scenarios are not ideally conceivable. Whether they are or not it is important to see that they seem conceivable and that this allows us to get a common sense handle on what we are talking about.

On Friday I attended Pete Mandik’s Cogsci talk at CUNY. The talk was excellent; entertaining and with a spirited discussion of many interesting topics. This paper represents one of a two part project that Pete is pursuing that involves Swamp Mary. One project employs Swamp Mary against dualists while the other employs her in an in-house dispute between what Pete call ‘gappy’ and ‘non-gappy’ physicalism. Friday’s talk was the latter and is based on his paper Swamp Mary Semantics: A Case for Physicalism without Gaps.

Pete’s distinction is recognizable as meant to capture type-a and type-b physicalism. More generally it is the distinction between a priori and a posteriori phsyicalism. Or at least one may get that idea from the way Pete characterizes the distinction. He says that the issue is over whether or not what we can “classic Mary” in her pre-release condition is in a position to know what it is like to see red. Pete claims that the gappy physicalist will deny that she does while the non-gappy physicalist does not. Swamp Mary is then presented as a challenge to the gappy physicalist. In short the challenge goes as follows. Swamp Mary is a physical duplicate of post-release Mary, the one who has seen red and so knows what it is like to see red, but Swamp Mary has never herself seen red since she is a swamp being who, we can stipulate, has had no experiences herself at all. She is a duplicate of a being that has had the experience of seeing red but hasn’t herself had the experience. Nonetheless, Pete urges, it is natural to say that Swamp Mary knows what it is like to see red. This is natural since post-release Mary knows what it is like to see red and Swamp Mary is a physical duplicate of post-release Mary. But since it is also natural to say that Swamp Mary has never seen red, or had an occurrent red quale, it is now mysterious why it is supposed to be the case the pre-release Mary doesn’t know what it is like to see red. If Swamp Mary can know what it is like to see red without ever having a red quale then it ought to be the case that pre-release Mary can also know what it is like to see red without having seen it. Thus, Pete concludes, physicalist should be non-gappy physicalist and hold that Mary can know what it is like to see red in her pre-release state.

Pete then goes through various responses that might be made by the type-b physicalist and tries to show that they all have problems. He does this by examining four different psychosemantic theories, which he also claims to be exhaustive, and arguing that they cannot provide the relevant explanation (i.e. of the difference between pre-release Mary and Swamp Mary’s knowledge of what it is like to see red). The four categories of psychosemantic theory are: 1. Quotation 2. Actual Cause 3. Nomological 4. Descriptive-homomorphic/isomorphic. Briefly, 1 is supposed to capture any kind of self-presentational view about phenomanl concepts like that of Chalmers or perhaps Block. 2 is supposed to capture any kind of teleological or causal theory of content while 3 is supposed to capture anything that resembles Fodor’s psychosemantics (think ‘asymmetric dependance’ here). Finally, 4 is supposed to capture any kind of conceptual role kind of theory. The problems for each of these, to make a long story short, are that 1 and 2 cannot explain how Swamp Mary does know what it is like to see red while 3 & 4 end up having problems explaining why pre-release Mary doesn’t know what it is like to see red.

There is a lot of detail that I am leaving out but in general Pete is trying to construct an argument against a central claim of the type-b physicalist. This is the intuition, shared by many, that one cannot really have the full concept RED without having the red quale present in ones consciousness. Or to put it more common sensically, one cannot know what it is like to see red unless one has (a) had a red experience and (b) has the ability to think, or otherwise identify, that the experience is red while one is having it. Pete has the intuition that Swamp Mary knows what it is like to see red and yet neither (a) nor (b) seem to be met and therefore concludes that the type-b physicalist is wrong. I suggested in discussion that Pete was running over the distinction between ‘knowing what it is like’ in the general sense, and ‘knowing what it is like FOR one’. To know what it is like for one too see red requires that one have, or be able to recall, a red experience and to be able to say, so to speak, to one’s self ‘this is what it is like to see red’. To know what it is like in the generic sense is to know what it is like to see red in the way that pre-release Mary is typically thought of by the type-b physicalist. She can know a lot about what it is like to see red. She can know that it is more like seeing something pink than it is like seeing blue, and all other kinds of facts. But intuitively she doesn’t know what it is like for her to see red. Once we have this distinction in mind we no longer have a problem with Swamp Mary. Swamp Mary knows what it is like in the generic sense, in just the same way as pre-release Mary, but she does no know what it is like for her to have the experience, again just like pre-release Mary. To put this in different terminology, both pre-release Mary and Swamp Mary lack the “pure” phenomenal concept, though they have many others. On the other hand, if we think that Swamp Mary does know what it is like in this way it will be because she has an ability to call up the red experience and identify it, whereas Classic Mary cannot do this.

Finally, does saying this really make one a gappy/type-b phsyicalist? I claim that it doesn’t. I agree that in the way the Mary thought experiment is usually set up we arrive at the conclusion that pre-release Mary cannot know what it is like for he to see red. This is because she lacks the appropriate concept, the pure phenomenal concept. In order to acquire this concept she needs to have the experience. But once she does there is no longer any gap between the physical and phenomenal. Once Mary has the pure phenomenal concept she is able to know what it is like to see red in the complete way necessary to make deductions from physical facts to phenomenal facts without relying on the pure phenomenal concept (or introspection) for justification of any step in the deduction. Thus in order to really count as a gappy physicalist in Pete’s sense we would have to deny that even once Mary had this concept she would be unable to know what it is like for her to see red just on the basis of the physical facts. This is much less plausible to me.

After the talk at the bar we started to talk about the role that intuitions should play in philosophy. This discussion was started by thinking about the experience principle. Why does anyone think that it is true that in order to fully have the concept of red one must have seen red, or had a red experience? I claimed that have prima facie evidence for this claim from the fact that it is intuitively obvious to many philosophers. I agree, basically, with Michael Devitt’s view that intuitions, especially those of experts, should count as defeasible evidence. Thus, we can have a priori justification for believing something but not a priori knowledge in the traditional sense. The mere fact that most philosophers think that Mary couldn’t know what it was like for her to see red from within her room should count as evidence, not for epiphenomenalism or dualism, but rather for the experience principle. That is what is really being tested there. Now, I agree that this is defeasible evidence. We could come to have reason to reject the idea. I *think* I can at least negatively conceive of a situation where pre-release Mary comes to know what it is like to see red without seeing it first. She will know that when people see red they are in a certain brain state. She will also know that people talk about knowing this state in a special first-person way and that they can only do this when in the relevant brain state. She might then conclude that to know what it is like to see red in this sense requires that she be in this brain state. Since she is not allowed to have the stimulus that will produce the brain state she must find some other way to produce the brain state. She might then realize that when one imagines seeing red one goes into something like the brain state that one is in when one actually sees red. Could she come to token this brain state without the stimulus? Well, obviously she could rig some kind of autobrain stimulation to stimulate the area and produce the experience. But is it absurd to think that she could do this without brain stimulation? That is, could she imagine what red looked liked without ever having seen it? I am not sure…my intuitions tend to go with the experience principle but I can’t see anything contradictory about the scenario just described…

There is more to say about all of this but this is getting too long already!