On Chalmers' view, wherein the 'psychophysical laws' are contingent, it seems that across possible worlds most brains like ours will be zombies or at least have 'associated' qualia that don't 'match' the information processing in the brain. So sophisticated brains proceeding according to ordinary standards of rationality should zombie-conclude that they probably are not conscious (as they don't have access to any non-material qualia), despite their zombie-perceptions of being conscious (shared by both zombie and non-zombie brains). Yet Chalmers thinks that in our actual world the psychophysical laws lead to conscious experience mirroring the information processing in the brain. So, upon hearing the argument, shouldn't Chalmers' brain zombie-conclude that it is probably a zombie brain, and 'phenomenal Chalmers' consciously think the same?
No. Conclusions are drawn by people, not brains. Standards of rationality likewise apply to agents and their beliefs, not to their physical components (brains and neural states) in isolation.
On my view, beliefs are partly constituted by phenomenal properties -- that's what gives them their representational content. Zombies don't have beliefs like we do. They exhibit all the same behaviour, and make all the same noises, but there's no meaning in it. It's not really about anything.
One might define a 'z-belief' as the functional (physical, dispositional) component of a belief. It's not so clear how to assign pseudo-contents to these z-beliefs, but I guess a reductionist may offer a stipulation of some kind: S has a z-belief that P iff S has such-and-such physical dispositions [e.g. 'S behaves as though P were true', or 'S has a brain state which covaries with evidence of P', or some such. See my essay 'What Behaviour is About' for a more sophisticated empirical approach to attributing "content".]
Presumably we're to suppose that whenever I really have the belief that P, my brain has the z-belief that P. But I doubt whether any such reduction can be given that perfectly mirrors my actual belief contents. (If epiphenomenalism is true, and qualia are partly determinative of belief content, then the physical facts underdetermine what it is that I believe. My inverted-spectrum duplicate has the same brain -- hence z-beliefs -- as me, but our phenomenal beliefs are very different. My 'red' is his 'blue', or whatever.)
There's a more fundamental problem, even if we grant the reductionist his impossibly fine-grained z-content. Let's grant - per impossibile - that my brain (and zombie twin) "z-believes that P" iff I believe that P. However, my brain (understood as a purely physical system, i.e. excluding its phenomenal properties) is in possession of only a subset of my total evidence. Qualia - the contents of experience - are among my evidence if anything is. But these phenomenal properties are not causally accessible to my neural processes. So the conclusion 'I am conscious' follows from my evidence, but not from the "information" available to my brain. One can be a rational person, or have a "rational" brain, but not both.
Now, it's pretty obvious that being a rational person is better than having a "rational brain" (insofar as the latter attribution is even meaningful). Brains are parts of people, and like any body part we really only care about it for how it can serve the whole person. If quick feet didn't make for a quick person, we wouldn't much care for the former. Similarly, a rationally desirable brain is one that makes for a rational person, with justified beliefs.
One could imagine a brain that is instead built in such a way that it tends to produce "z-justified" z-beliefs. What this means is that it tends to end up in physical states such that a conscious person in that physical state would have beliefs in line with the physically accessible subset of their evidence. When put like that, it becomes clearer that what we've really described here is a defective brain. Let's call it "z-rational", and reserve the term 'rational' for brains that give rise to rational people -- people whose beliefs are in line with their total evidence.
Here are two implications:
(1) A z-rational brain can be expected to have more true z-beliefs (across all possible worlds).
(2) A rational brain can be expected to yield more true beliefs.
Fortunately, my brain is rational rather than z-rational. Hopefully yours is too (otherwise, you're a defective agent). One might try to argue that there's something "wrong" with a brain that isn't z-rational, but I don't think that'll work. For one thing, since you're really just describing a physical state it's not clear that brains or z-beliefs are even open to this sort of normative assessment. Norms apply primarily to people, and to our organs only derivatively. What a well-functioning agent really needs is a brain that will make them rational, not z-rational. As suggested above, a z-rational brain is defective from the standpoint of contributing to the functioning of the whole person (which is the relevant standpoint against which to assess brains). Further, when you stop to think about what it really means to have 'z-rational z-beliefs', you see that there's not really anything significant (worth caring about) there.