A Critical Examination of the Cohen-Levesque Theory of Intentions

. We examine the formal theory of intentions recently proposed by Cohen & Levesque [2]. We evaluate the assumptions made by this theory and their consequences relative to fairly general intuitions about intentions, especially as they are applied in AI domains. We show that the theory is conceptually problematic and that certain technical claims made by the authors are false or counterintuitive in most natural scenarios. Keywords: intentions, logic in AI. 1 Introduction It is of great use to a scientific discipline when its theories are critically examined and their unstated assumptions, and the unexpected consequences of their explicit assumptions, are brought out. Unfortunately, there is not as much evaluation of claims and theories in AI as in other, more classical, disciplines. This is changing now, as AI becomes established itself. This paper contributes to this trend by critically examining a theory of intentions proposed by Cohen & Levesque (henceforth, C&L) [2]. The importance..