Room 327

Dynamic Epistemic Logics for Games of Chance and Bluffing

Tim French

University of Western Australia

Many games, such as Clue or BlackJack, are useful examples of epistemic reasoning. Players know their own cards but are uncertain as to the cards of their opponent, and generally success in the game involves reducing your uncertainty without reducing the uncertainty of your opponent. Dynamic epistemic logic has been shown to be able to model reasoning in such games. However, when designing software agents to play these games, probabilistic agents using Bayesian reasoning perform much better than purely logical agents. In this talk we consider some preliminary ideas for probabilistic logics to represent the reasoning and learning capabilities for agents in these games, and present some basic concepts in the setting of a simple bluffing game.