We present a framework for decision making with circumstance-dependent preferences and decisions. This formalism, called Ordered Choice Logic Programming, allows decisions that comprise multiple alternatives, which become available only when a choice between them is forced. The skeptical semantics is based on answer sets for which we provide a fixpoint characterization and a bottom-up algorithm. OCLPs can be used to represent and extend game theory concepts. We demonstrate that OCLPs allow an elegant translation of finite extensive games with perfect information such that the c-answer sets correspond to the Nash equilibria of the game. These equilibria are not player-deterministic, in the sense that a single player, given the other players' actions, could rationally leave an equilibrium state by changing her action profile. Therefor cautious Nash equilibria are introduced as the answer sets of the transformed game.
|Title of host publication
|Al 2002: Advances in Artificial Intelligence
|Number of pages
|Published - 2002
|Lecture Notes in Artificial Intelligence