Remember to keep clear on what kinds of mental states you are discussing when you focus on their relationship to behaviour. There are good reasons, for example, for thinking the connection between propositional attitudes and behaviour is of a very different sort to the connection between phenomenal states (such as pain) and behaviour.

“When you see someone else writhing on the ground, it is impossible for you to doubt that they are in pain. This is not merely because of your kind, sympathetic nature, but because it is part of the meaning of our concept of pain to be correctly applied in such situations.”

I’m not sure I quite agree with this. The writhing-behaviour is a criterion for our ascription of pain in the sense that it is evidence for the subject being in pain: it is not constitutive of the meaning of the concept. What is the pull towards the further step?

Also, some of the Wittgenstein quotes do not really support the further step. PI 303 just says that it is difficult to believe that people are not in pain when it appears as if they are: but our ability to doubt is to do with the epistemology of mental state ascriptions, not the semantics of them. The case is a bit more ambiguous with PI 281. Literally construed, it is false: I can, of course, say “a rock has sensations”. Wittgenstein must, I think, either mean that we cannot intelligibly say such a thing, or that we cannot truely say it. If the former, then this section does support your interpretation; if the latter, however, then not.

With PI 284 Wittgenstein seems more clearly to suggest that the issue is semantic. However, notice that he is on weak grounds here. He says that we “might as well” ascribe pain to a number as to a rock, but this directly begs the questions against the materialist. The materialist will say that pain is a configuration of matter, and so it makes sense to ascribe it to material things, but not to abstract notions like numbers.

Last thing:

“Perhaps that is a weak example – the person could be feigning pain, and then the application of the concept would be incorrect.”

This isn’t really a problem for behaviouristic philosophies of mind. The behaviourist asserts that mental states consist in *dispositions* to behaviour, and the person who is faking being hurt has different dispositions to someone who is really in pain (e.g. if you offer to give someone faking severe pain a fiver if they quit the pretense, they probably will, whereas someone who is really in severe pain won’t be able to).