I see that Alex Byrne and Michael Tye have just posted their reply, "Qualia Ain't in the Head", to Adam Pautz's paper "Sensory Awareness is not a Wide Physical Relation". Both papers are forthcoming in Nous. Pautz argues against externalist representationalism about sensory experience, and Byrne and Tye defend it. The centerpiece of Pautz's case is the scenario of Twin Maxwell. Maxwell is an ordinary perceiver of orange who represents it as a mixed hue, via activation of dual opponent-processing channels. Twin Maxwell is a counterfactual perceiver in a different environment, who normally responds to (what we call) orange things with activation of a single channel, the sort of activation that normally goes along with representing a color as a unique hue. Pautz argues that (1) Maxwell and Twin Maxwell have different experiences when looking at an orange in typical circumstances: Maxwell has a "mixed hue" experience, Twin Maxwell has a "unique hue" experience. He also argues that (2) wide representationalism (at least of the Dretske/Tye variety, where states represent those properties that they causally covary with under normal conditions) is committed to saying that Maxwell and Twin Maxwell have the same sort of experience: both are in an internal state that is caused by the same external physical property P in optimal conditions, so both states will represent P, so both states will have the same phenomenal character. So externalist representationalism is false.
Byrne and Tye say various things in reply, questioning (1) in some cases, and also noting that externalist representationalism is not committed to Dretske/Tye representationalism. But the core of their reply, toward the end of their paper, is to deny (2), holding that even Dretske/Tye representationalism is consistent with Maxwell and Twin Maxwell having different experiences. (Or at least, that it is consistent with there being different experiences in all the cases where (1) is plausible.) Their key point is that when Twin Maxwell looks at an orange thing, and assuming he has an experience as of unique red or some other unique hue here, then he is not perceiving under optimal conditions, precisely because he is perceiving the orange object as red (or as having some other unique hue), and such an experience will be nonveridical. If so, the state does not causally covary with P under optimal conditions, so it does not represent P, so it need not have the same phenomenal character as Maxwell's experience.
I think the state of play favors Pautz here. It seems illegitimate to appeal to nonveridicality in explaining why conditions are not optimal. For an experience to be nonveridical is for it to have a false content; and on the Dretske/Tye account, the content of an experience is to be explained partly in terms of the notion of optimality. If optimality is then explained partly in terms of veridicality, this account will be circular. So a noncircular account requires that optimality be explained without invoking notions such as veridicality and content, perhaps instead using notions such as normality, fitness, and so on. So to make their case, Byrne and Tye need to show that Twin Maxwell's conditions are suboptimal in some such independent sense. But they have not done this, and it is not easy to see how this could be done, since Maxwell and Twin Maxwell's circumstances seem to be symmetrical with respect to the natural candidates for the relevant independent features. (A version of this point is made by Pautz in his reply to the second objection [pp. 27-30] in the long version of his paper on the web, and, I gather, in a footnote in the abridged version that will appear in Nous.) Perhaps there is some independent grounding for suboptimality that could be found, but this is far from obvious.
Of course there are other replies available. If the externalist abandons the Dretske/Tye account of content, other options will be available. But it looks like the symmetry considerations generalize to many other accounts, so it would at least be interesting to see some other options spelled out. Personally I think the best reply for the externalist representationalist is Pautz's "third objection" [pp. 30-31 of the web version]: the appeal to compositional representation of distinct but necessarily coextensive complex properties. Maxwell might represent the property <R to degree 0.5 and Y to degree .5>, while Twin Maxwell might represent the coextensive property <R' to degree 1 and Y' to degree 0>, where R' and Y' are the properties tracked by Twin Maxwell's opponent-process channels corresponding to our R and Y channels. Pautz suggests in response that even the states of the single channels in Twin Maxwell (activation 1 on the first channel, activation 0 on the second channel) will track <R to degree 0.5> and <Y to degree 0.5> respectively. But the externalist can easily handle this by holding that the different states of a single channel are constrained to represent different degrees of a single quantifiable property R', which must differ from R, and that <R' to degree 1> and <R to degree 0.5> are distinct properties. As Pautz notes, this reply doesn't generalize to other inversion cases involving noncompositional representation in cases involving pain and taste. But the externalist might reply that the real power of Maxwell case comes from the compositionality, and that in these other cases it is easier for them to deny that the relevant subjects have different experiences.
Of course there's a lot more to be said. It will be interesting to see where things go from here. It's interesting in any case to see the recent groundswell of support for internalist versions of representationalism, in the work of people like Tim Crane, Terry Horgan and John Tienson, Joe Levine, Georges Rey, Sydney Shoemaker, Charles Siewert, and Brad Thompson, as well as Pautz and yours truly. Clearly this is the wave of the future!