--- In quickphilosophy@xxxxxxxxxxxxxxx, Ron Allen <wavelets@...> wrote: > > Hi Walter: >  > I've been out of town and unable to devote much time to Fodor. >  > Just an observation: I think you are reformulating Fodor's development and > combining parts of different arguments. That's OK, though. >  > I agree completely that he's attacking BCP as if it were committed to > sorting and inference as sufficient conditions for concept possession, when, > in all likelihood, they are just necessary conditons. Individually and > conjoined, then, they are still weaker than concept possession. >  > Would not a kind of Chinese Room Argument be relevant to this question? It seems so to me. >For example, suppose I had a person inside the CRA who was asked to sort and >make inferences upon objects that were presented to the window. The little >guy in the CRA has to sort dieseldowns and he has to make inferences about >them. But, he just follows a set of ad hoc rules that somehow, >coincidentally, just happen to work for the particular sequence of dieseldown >and non-dieseldown examples we hold up in the window and for the dieseldown >and non-dieseldown deductive problems we hold up to the window. Now, this ad >hoc, rule-driven luck is not sufficient to say that the Room has the concept >of a dieseldown, even though, sure enough, it made a whole handsome series of >sortations and a pretty decent pattern of inferences. >  Yes. > Also, another thing about your reformulated argument below. In your point (2) > you define compositionality as concepts being decomposed into smaller ones. > This is different from Fodor, who only asserts that they add together or > permute in some fashion, not that they decompose. (2) doesn't require that all concepts decompose. It just says what's supposed to follow from concepts that do. >  Also, the decomposition leads to an infinite regress. I either come to some > concepts that are not compositional (and not composed of smaller ones) or I > have compositional concepts all the way down (to minus infinity). But, if we > don't regress, then all of our concepts are built up from atomic concepts, > and a BCP advocate with an Atomic Propositions and Atomic Perceptions > foundationalism underneath would not at all be unsatisfied with that result. > If Fodor were to argue for the kind of compositionality that you seem to > suggest here, then his Dog as such concept would be just a collection of > atomic perceptions, better or worse known to a subject. I take from that article that he actually is some sort of atomist. >  > So, at least with your (2) below, I think that you have an infinite regress, No, there's no regress from (2), which, again, only says what's supposed to follow if a concept IS "decomposable." It doesn't require that every concept is. and if not that, then you dissolve concepts as such into a variant acceptable to BCP. >  I don't see that. Can you explain? I repost the argument below for easy reference. W > > > (1) For language/understanding to work (be generative), most concepts must be > compositional. > > (2) For any concept C, C is compositional iff (if C is composed of (littler > concepts) A + N, then for any person S, if S understands A + S understands N, > then S will also understand C). > > (3) For the BCPer, for all persons S and concepts C, if S undertands C, then > S can sort representative Cs in favorable conditions. > > (4) Therefore, for the BCPer, a concept C is compositional iff (for all Cs > and Ss, if C is composed of A and N, then if S can sort both As and Ns in > favorable conditions, then S can also sort representative ANs (i.e. > representative Cs) in favorable conditions. [from (2) and (3)] > > (5) But for many (perhaps most?) concepts of the AN type, there are many > (perhaps most?) Ss who can sort both representative As and representative Ns > in favorable conditions but CANNOT sort representative ANs in favorable > conditions. [Fodor gives his night bluebird as an example here]. > > (6) Language/understanding is generative and does work. > > (7) Therefore, BCP is wrong. >