In Chapter 2 of The Subjective Brain, I defend the Neuro-introspection thesis whereby brain states are introspectible as such. An objection I owe to Dan Cavedon-Taylor concerns whether the alleged multiple-realizability of the mental by the neural (MR) would be inconsistent with Neuro-introspection. The way Dan puts it is available here and I reproduce my response below.
Regarding MR, due to arguments set forth in ch. 1, I don’t take it particularly seriously. But considering my introspection thesis in isolation from ch. 1, I can grant, for the sake of discussion, relatively strong versions of MR. Since I think perception and introspection are analogous in many significant ways, it is useful to consider the MR issue by constructing an analogy to perception. Suppose there is some object type that is not only multiply realizable but multiply realized. Suppose further that the object type bottle is one such example. So there are lots of distinct physical realizers of bottles, e.g. glass ones and aluminum ones. But this supposed fact (the multiple realizability of bottles) is not all by itself a problem for standard accounts of object perception. There’s not an obvious problem of how one can perceive not only that a bottle is present but that a glass bottle is present, is there?
Perhaps you think my example concerning glass vs. aluminum is a poor one since glass and aluminum are readily perceptually distinguishable. Suppose then that we switch examples to perceptually undetectable realization differences, e.g. two kinds of glass that can only be distinguished with special instruments. Now we have an example in which some properties of bottles are imperceptible. But this doesn’t raise any special problems for a theory that claims that bottles are perceptible. It’s pretty obvious that even though perceptible objects must have perceptible properties, they may nonetheless have imperceptible properties as well. I offer, then, that an analogous thing is true of introspection: neural states have neural properties that are introspectible, but perhaps they also have some neural properties that are introspectively undetectable. How does that show the failure of neuro-introspection?