None

What is it for something in the mind to represent something? Robert Cummins looks at the familiar problems of representation theory (what information is represented in the mind, what form mental representation takes, how representational schemes are implemented in the brain, what it is for one thing to represent another) from an unprecedented angle. Instead of following the usual procedure of defending a version of "indicator" semantics, Cummins begins with a theory of representational error and uses this theory to constrain the account of representational content. Thus, the problem of misrepresentation, which plagues all other accounts, is avoided at the start. Cummins shows that representational error can be accommodated only if the content of a representation is intrinsic - independent of its use and causal role in the system that employs it. Cummins's theory of error is based on the teleological idea of a "target," an intentional concept but one that differs importantly from that of an ordinary intentional object. Using this notion he offers a schematic theory of representation and an account of propositional attitudes that takes exception with some popular positions, such as conceptual role semantics, Fodor's representational theory of the mind, and Putnam's twin-earth examples.