A Logic for Modeling Decision Making with Dynamic Preferences

We present a framework for decision making with the possibility to express circumstance-dependent preferences among different alternatives for a decision. This new formalism, Ordered Choice Logic Programs (OCLP), builds upon choice logic programs to define a preference/specialization relation on sets of choice rules. We show that our paradigm is an intuitive extension of both ordered logic and choice logic programming such that decisions can comprise more than two alternatives which become only available when a choice is actually forced. The semantics for OCL programs is based on stable models for which we supply a characterization in terms of assumption sets and a fixpoint algorithm. Furthermore we demonstrate that OCLPs allow an elegant translation of finite extensive games with perfect information such that the stable models of the program correspond, depending on the transformation, to either the Nash equilibria or the subgame perfect equilibria of the game.