have grave difficulties accounting for the possibility of representational error (Fodor, 1990; Levine & Bickhard, 1999; Loewer & Rey, 1991). And, therefore, accounting for
representational content per se.
Such models can address complex representations as being constructed as combinations of more basic representations, and the more basic representations might be
themselves constructed out of still more basic representations, but this composition and decomposition process must have a base, a ground or foundation, of representations that are not constructed out of nor defined in terms of any other representations. It is this foundation that cannot be accounted for. The models only permit the creation of new representations in terms of already available representations. There is no account of newly emergent representation (even if a new correspondence — a new differentiation — were created, perhaps in a connectionist net, there is no account of how the system can get any
content about what those correspondences are supposed to represent). One move at this point is to posit that all basic representations must be innate, and
all representations that, say, adult human beings are capable of are created as complexes out
of those innate atoms (Fodor, 1981). But the problem here is a logical one: there is no
model of how correspondence representational content can come into existence, and evolution is just as helpless in the face of that problem as is individual level learning and development. Correspondence models, in other words, presuppose a foundational level of representation that they are incapable of modeling, that they themselves render impossible. They involve an internal contradiction in their assumptions, an incoherence (Bickhard, 1993).
Correspondence models render the emergence of new representation impossible. Neither evolution nor development nor learning can generate emergent representation, if
this approach is correct. But there were no representations at the moment of the Big Bang and there are representations now. Therefore, representation has emerged. Therefore,
representational emergence is possible. Therefore, any model that renders representational
emergence impossible is false. Therefore, correspondence models are false.
Note that the interactive model renders representational emergence almost trivial: any new construction of system process organization that happens to involve the right kinds of indications of interactive potentiality will constitute an emergent representation.