As far as important papers in the philosophy of mind go, Frank Jackson’s Epiphenomenal Qualia and What Mary Didn’t Know are pretty high up on the list of must-reads. They’ve spawned a mass of literature devoted to picking apart just what Mary did or didn’t know made all the more intriguing because Jackson himself later distanced himself from the argument. Every possible response to the argument has been (seemingly) given, and there’s ample reason to regard anyone writing on it the same as someone writing on substance dualism – interesting, no doubt, but somewhat well-worn.
At any rate, I’m going to do just that, so deal with it. The basic outline and conclusion of the argument is will known and linked to above, so I won’t rehash too much of that. I do think, however, that David Lewis is correct when he says that if the hypothesis of phenomenal information is true, then the argument basically works, and refutes materialism. I’m not sure how I feel about his argument that a consequence of the knowledge argument working is that it proves too much (see the previous link), though.
Lewis himself endorses the ability account, where Mary gains a new ability to describe, imagine, remember or recognize an experience that she didn’t have before. This keeps Mary from learning any new information while still allowing her to ‘know what it’s like.’
James Madden, in Mind, Matter and Nature points out that even if this account is true, it leaves untouched the issue of just what exactly is imagining or describing, which is a first-person experience of what something is like:
The ability hypothesis does not address the issue of what exactly Mary is imagining or describing, that is, her first person experience of red. The materialist might attempt to account for imagination or description in terms of behavioral dispositions, various psycho-physical identities (type or token), but this only pushes the problem back one remove; in each case the proponent of the knowledge argument would just as likely ask whether these accounts actually capture what Mary imagines or describes, that is, her awareness of what it is like to see red. Thus, the ability hypothesis doesn’t pose a great threat to the knowledge argument. (p. 143)
Not a devastating objection but an interesting one nonetheless. Madden also interacts with Andrew Melnyk’s objection, which boils down to the idea that ‘there is nothing philosophically suspect about two conceptually distinct descriptions referring to one and the same object.’ (p. 139). Madden gets to what I think is the heart of the issue of the knowledge argument:
The point of the so-called knowledge argument is not that a descriptive (conceptual) difference implies different objects of knowledge, but that we can account for the difference in Mary’s newly acquired knowledge of what it is like to see red only in terms of a qualitative experience she has for the first time after leaving the room. (p. 140)
The so-called knowledge argument doesn’t trade on an inference from the fact that psychological states can only be known by two different descriptions to the conclusion that they must involve both nonphysical and physical properties. Rather, the point is that we can know that certain descriptions apply to a particular psychological state only by having the requisite experience. We need to posit a qualitative experience as a real feature of the world to account for the difference…Descriptive differences do not tell us much about what exists, but if we know that a certain description is true of some object, and the only account to be given of that knowledge is in terms of some experience, then we must grant the reality of the experience. (p. 141)
Here the knowledge argument has moved from the merely conceptual to the objective, while still recognizing the importance and reality of conceptualization of experiences. This to me is a substantial point, since most replies to the knowledge argument seem to take Melnyk’s route.
Madden effectively neutralizes most standard replies to the knowledge argument, it seems to me. As far as I can tell, the most promising attack for the materialist would be that of Lewis, who denies the hypothesis of phenomenal information (or rather, points towards why it should be considered suspect, since h explicitly states that he can’t refute it). Lewis makes another interesting argument to the effect that phenomenal information is epiphenomenal. In essence, he argues that Mary would not have done anything different had the phenomenal world been different than her initial leaving the room experience – if the phenomenal world had been Way 1, she would have done X, and if it had been Way 2, she would have also done X. Keeping in mind that Lewis is operating with phenomenal information being totally different the physical information, he comes to the conclusion that if he had acted differently, it would mean that the phenomenal – non-physical – would have caused the physical to do something, which Lewis argues is against the laws of physics. So, in a nutshell, if the phenomenal world is real (for the knowledge argument proponent) then it is qualitatively different from the physical, and thus epiphenomenal. If it’s not epiphenomenal, we have a case of the non-physical causing a change in the physical. And this, given current physics, simply cannot be. However, surely there are arguments one could give for why Lewis’s conclusion might be false – substance dualists give these kinds of arguments, for example (also see here). As Lewis states it, it’s an interesting objection but a rather seeming-question-begging one.
While none of this is really conclusive, I do think that Madden’s reply to the ability and description objections clear a substantial amount of ground. Lewis’s objections are interesting and seems to open up a potentially fruitful problem space, but as it stands his objection is question-begging, but should definitely be explored further if the materialist wants a solid rebuttal to the knowledge argument.