Logical Foundations for Belief Representation

This essay presents a phi l osophi cal ond computotionol theory of the represento-tion of de re, de dlcto, nested, and quasi-indexical belief reports expressed i n natural language. The propositional Semantic Network Processing System (SNePS) is used for representing ond reasoning about these reports. In particular, quasi-indicators (indexical expressions occurring i n intentional contexts and representing uses of indicators by another speaker) pose problems far natural-language representation and reasoning systems, because-unl i ke pure indicators-they cannot be repl aced by careferential NPs without changing the meaning of the embedding sentence. Therefore, the referent of the quasi-indicator must be represented i n such a way that no invalid careferential claims are entailed. The importance of quasi-indicators is discussed, and it is shown that all four of the above categories of belief reports can be handl ed by a single representational technique using belief spaces containing intensional entities. Inference rules ond belief-revision techniques for the system ore also examined. This essay presents a computational analysis of a referential mechanism-quasi-indexicality-first examined in philosophy some 20 years ago, but not hitherto employed in artificial intelligence (AI) studies of belief systems. In turn, a philosophical claim about the relations of de re, de ditto, and de se beliefs is made as a by-product of the computational analysis. I thus hope to illustrate the importance of philosophy for research in AI and the correla-tive importance of a knowledge of AI for philosophical research, in the spirit of Dennett's (1978) recommendations: