It's not that science can't remind us of, or call attention to, modal truths that would be accessible without science. The surest proof of a possibility is an actuality. It would have been a lot harder to come up with a dream argument for skepticism if people didn't dream, or an argument from the possibility of hallucination if people didn't have non-veridical experience. But it would only have been more difficult. The possibilities upon which the arguments trade are real whether or not they ever materialize. The possibility that the neural activity associated with mental states differs from creature to creature, and even person to person, is (and was always obviously) real whether or not the possibilities materialize.
Indeed, after all of the empirical data are in, we might ask again: What distinctively philosophical questions will be, or even could be, answered by science? What distinctively philosophical controversies will be advanced, let alone settled? Suppose that we have obtained exhaustive correlations of the sort described above. Are we any closer to answering any of the following? Can you even imagine any empirical research that would shed light on any of the following?
(1) Are knowledge arguments for substance/property/fact/event dualism sound?
(2) Are there important distinctions to be made between substance/property/fact/event dualism? Is there such a thing as substance? What is the connection between differences between kinds of substances and kinds of properties exemplified by substance? What is the difference between a mental property and a physical property? What is the distinction between an event and a fact?
(3) Is functionalism a plausible account of the nature of mental states?
(4) On a functionalist account of mind, should we identify mental states with that which "realizes" the functional state (i.e. plays the functional role) or should we identify mental states with the exemplification of the second-order functional property?
(5) Do we "fix the reference" of predicate expressions picking out kinds of mental states with definite descriptions that nevertheless do not capture the meaning of the predicate expressions, or shall we view terms for kinds of mental states as having the meaning of definite descriptions? If either, what are the relevant definite descriptions?
(6) Should we be internalists or externalists about the identity conditions for mental states? If brain states are intentional (representational) what makes them intentional? Is it some feature intrinsic to the states, or has it more to do with the causal origin of the states. [sic] If the latter, why were we trusting first-person reports of mental states in trying to correlate neural activity with, say, desires or fears? Why would anyone have privileged introspective access to the character of a mental state whose identity conditions involve facts that are clearly the purview of empirical science?
Do any of the philosophers partnering with cognitive science think that they'll get any useful information from empirical science that will help them answer any of the above questions?
Don't misunderstand me. Many of the answers to these philosophical questions will clarify important empirical questions that someone might be interested in answering. If you decide that functionalism is true, you will probably need to turn to empirical science to discover what takes the value of the variable used in the specification of the functional state. If you become really interested in that question, then by all means stop doing philosophy for a while and do (or consult) the relevant empirical research. Alternatively, you could view your task qua philosopher as finished once you've come up with your functionalist analysis, content then to let the chips fall where they may with respect to the hard-wiring of the brain or the nature of a Cartesian ego.
If we decide that either mental predicate expressions, physical predicate expressions, or both have their reference fixed, or meaning given, by definite descriptions, and we can isolate the relevant definite descriptions, it may turn out that only the cognitive scientist can tell you whether or not the predicate expressions have an extension, and if they do what that extension is. I've already implied that I think it wildly implausible to suppose that our grasp of pain is somehow indirect (is "by description"), at least as we actually have an experience of pain of which we are consciously aware -- we'll talk about this issue in more detail shortly. I cant see how it can possibly turn that "is in pain" fails to pick out a property. Any view that allows for such a possibility is for that reason an implausible view. It's hardly the case, for example, that I think of being in pain as the property I typically exemplify when I'm bleeding all over the floor, or the property exemplification of which causes me to grimace, complain, answer questions in certain ways, or what have you. If such a view were correct, then, to be sure, it would be a matter of empirical investigation as to what that property is, and it might turn out in humans to be a pattern of neurons firing. When I think of Jack the Ripper, I think of a person who is causally responsible for certain atrocities committed at the end of the nineteenth century and there are any number of people who might turn out to be the referent of "Jack the Ripper." But I already know what property I'm talking about when I talk about pain. I don't pick it out via some metaproperty it exemplifies.
Knowledge, Thought, and the Case for Dualism
(Cambridge Studies in Philosophy)