I don't see offhand why the IBSTs can't get around Fodor's problem about representing uses in much the way that Fodor answers Dretske's objection on p. 514, viz., by exploiting recursive compositional structure.
    Suppose that some IBST is correct for labelling uses.<1>    Then we can determine a symbol's content in that context: "Platypus" means platypai because a labelling use of "platypus" is adverbially caused by a platypus.  Now that the symbol has that content, it has it, even when being used representationally.  The whole containing representation gets its content in the usual compositional way.
    Fodor might worry that my crucial appeal to the notion of a labelling use introduces circularity or some other illegal thing.  Certainly the notion isn't a purely causal one.  Maybe Fodor would object that it is itself at bottom a semantical notion, or that it presupposes something about intentional content that a psychosemantic theory is not allowed to presuppose.  If so, I don't yet see that.  Anyone?
 

Footnote

1. Fodor says he coined the term "labelling" for want of something better to call such uses.  Linguists call them "presentational" uses.  Presentational sentences such as "Lo, a rabbit" and "There's a cow" have characteristic syntactic properties.  But I'll stick with "labelling" here.