WHAT SHOULD DEFAULT REASONING BE, BY DEFAULT

This is a position paper concerning the role of empirical studies of h uman default reasoning in the formalization of AI theories of default reasoning. We note that AI motivates its theoretical enterprise by ref erence to human skill at default reasoning, but that the actual resear ch does not make any use of this sort of information and instead relie s on intuitions of individual investigators. We discuss two reasons th eorists might not consider human performance relevant to formalizing d efault reasoning: (a) that intuitions are sufficient to describe a mod el, and (b) that human performance in this arena is irrelevant to a co mpetence model of the phenomenon. We provide arguments against both th ese reasons. We then bring forward three further considerations agains t the use of intuitions in this arena: (a) it leads to an unawareness of predicate ambiguity, (b) it presumes an understanding of ordinary l anguage statements of typicality, and (c) it is similar to discredited views in other fields. We advocate empirical investigation of the ran ge of human phenomena that intuitively embody default reasoning. Gathe ring such information would provide data with which to generate formal default theories and against which to test the claims of proposed the ories. Our position is that such data are the very phenomena that defa ult theories are supposed to explain.