ilXt~OS~QCtiVQ bQliQf

Autoepistemic (AE) logic is a formal system character-izing agents that have complete introspective access to their own beliefs. AE logic relies on a fixed point defini-tion that has two significant parts. The first part is a set of assumptions or hypotheses about the contents of the fixed point. The second part is a set of reflection prin-ciples that link sentences with statements about their provability. We characterize a family of ideal AE rea-soners in terms of the minimal hypotheses that they can make, and the weakest and strongest reflection prin-ciples that they can have, while still maintaining the interpretation of AE logic as self-belief. These results can help in analyzing metatheoretic systems in logic programming