Knowledge and the frame problem

1 Abstract The frame problem, which arises in any reasonably expressive theory of action, shows up in a strengthened form in integrated theories of knowledge and action. In this paper, we present two instances of this strengthened frame problem: the Third Agent Frame Problem and the Vicarious Agent Frame Problem. We show that these problems cannot be solved at all by using frame axioms, and that most non-monotonic temporal logics are likewise inadequate for handling these problems. Finally, we show how an existing non-monotonic temporal logic, Motivated Action Theory, can be extended to handle both problems. 2 Introduction It has long been recognized that agents who operate in complex environments must be able to reason about the interactions between knowledge and action Agents must reason about the knowledge they need in order to perform an action, how actions that they perform aaect their knowledge, and how they can plan to achieve their goals with only partial knowledge. The bulk of the research in this area has focussed upon the rst of these issues, more commonly known as the knowledge preconditions problem. Speciically, researchers have sought to formalize in 1