Let me begin with two of what I think of as an extremely simple, indeed grossly over-simplified, truism. Wittgenstein told us that meaning is use. He also told us that meaning should not be understood as primarily consisting in the relationship that holds between a name and its bearer. My truism is this: Wittgenstein’s two dicta are logically independent. It might be that meaning is use: that is, it might be that ultimately anything I mean has to be explicated in terms of something I use it for. For instance, it might be that every time I utter ‘cat’, I am doing something cat-related. It may nonetheless be true that the best way to understand the word ‘cat’ is as naming the concept CAT. Conversely, it may be that my utterances of ‘cat’ have to be understood in complex, situationally variable ways, so that the word cannot be explicated as naming anything. Nonetheless, it may be true that meaning is independent of use. In short, one of these dicta is about the communicative and pragmatic aspects of language, and the other about semantics, and though closely related in Wittgenstein’s thought, they are logically independent and they have to be argued for separately.
In a similar vein, the meaning-is-use claim is logically independent of Wittgenstein’s no-inner-mentality ideology, unless you follow a stolidly behaviourist line of thought. You might think that understanding the meaning of ‘cat’ is a matter of behaving in a certain way with regard to cats. But this has little to do with whether or not you change your inner state when you come to understand ‘cat’.