Thanks for your continued interest in this. This has forced me to get a lot clearer in my own mind on these issues. Hopefully this results in me getting clearer in print. I’d be curious as to your reaction to the following.

The multiple realizability of the mental by the neural does not alone entail the non-introspectibility of neural realizers. This is can be shown by constructing an argument by analogy I hinted at concerning bottles and perception. The mere fact that bottles might be multiply realized does not entail that their realizing properties are imperceptible: consider the perceptible differences between glass and metal bottles. By analogy, then, it remains open that while the mental is multiply realized by the neural, the neural realizers are nonetheless introspectible as such. (That is, as neural. Hereafter I’ll drop “as such” for brevity.) Something more needs to be added, then, if multiple realizability is to be wielded against neurointrospection.

Here are some promising candidates for things that need to be added. First off, add the assumption that the only properties of personal states that can be introspected are mental properties. This seems like a pretty harmless assumption to me: I suppose it’s definitional of introspection that only the mental is introspectible. Second, add the assumption that all, not just some, mental properties have multiple possible neural realizers. It follows pretty straightforwardly, it seems, that neurointrospection is false. One way to point out why neurointropection fails is that the only way, given the first added assumption, the neural can be introspected is if mental properties are identical to neural properties. However, given the second added assumption, mental properties can’t be identical to neural properties.

So, if neurointrospection is going to be true, it really does need my first chapter, and the arguments contained therein for the identification of mental properties with neural properties, to succeed.

Been meaning to reply for a while. Glad to see you’re still interested in looking at the MR/neurointrospection issue.

So I agree that the considerations you mention fail to raise any special problems for a theory that claims that bottles are perceptible. But now I’m a little confused. Because unless I’ve missed something, bottleness is the realized property, right? Well my issue isn’t with the knowability of the realized property (unless I’m really, really confused the realized property in the case of neurointrospection is surely just going to be a mental property). My issue is with the knowability of the realizer properties (this is going to be the neural properties, right?). And as you show with respect to bottles, there can be perceptually indistinguishable realizations of the same bottle type. To me, that looks problematic for neurointrospection.

So suppose, analogously, that some mental state type, M, can be realized by many different neural state types, N1, N2, N3, etc. Suppose also that it is qualitatively indistinguishable when M is realized by N1, or N2, or N3, etc. I go off and get my neuroscientific training I come to know all these nomologically possible realizers of M. I come to know that being in M means one is either in N1, or N2, or N3, etc. My question is: how will *that* knowledge suffice for me to know which of all those nomologically possible neural states I am actually in when I am in M? There are a range of candidates here and, as far as I can see, no obvious way to know which one is operative for any token instance of M.