A Viewpoint Distinction in the Representation of Propositional Attitudes

Proceedings of the AAAI Conference on Artificial Intelligence, 5

Volume

Science

Knowledge Representation

A representation scheme can be used by a cognitive agent as a basis for its normal, inbuilt cognitive processes. Also, a representation scheme can serve as a means for describing cognitive agents, in particular their "mental" states. A scheme can serve this second function either when it is itself naturally used by a cognitive agent (that reasons about agents), or when it is merely an artificial, theoretical tool used by a researcher. In designing a representation scheme one must pay very careful attention to two related questions: the question of whether, for any given agent, the scheme is used by the agent or is used to describe the agent (or both); and the question of whether the scheme is being used as a theoretical tool as well as, perhaps, being used by agents). I show by example that representational pitfalls can be encountered when these questions are not clearly addressed. The examples revolve around Creary’s logic-based scheme and Maida and Shapiro’s semantic network scheme, both of which were designed primarily to facilitate the representation of propositional attitudes (beliefs, hopes, desires, etc.). However, the general points have wider application to schemes for propositional attitude representation. By appeal mainly to the Maida and Shapiro case I demonstrate also that it is possible to be misled by the ambiguity of whether "to represent" means "to denote" or "to be an ambassador/representative/abstraction of".