Playing Games and Making Decisions with Complex State Spaces, Resource- Bounded Agents, and Unforeseen Contingencies.

摘要

The goal of this project was to construct models for decision making and game theory that can deal with complex state spaces, resource-bounded and cognitively-limited agents, and nnforeseen contingencies in a robust way. A number of advances were made in three areas: 1. A general game-theoretic framework for reasoning about strategic agents performing possibly costly computation was developed. The framework provides psychologically appealing explanations for behavior in well-studied games, such as (such as finitely repeated prisoner's dilemma and rock-paper-scissors). 2. A logic that can capture awareness and knowledge oflack of awareness was developed. The logic subsumes a number of logics introduced earlier. A sounds and complete axiomatization was provided for the logic. 3. A new solution concept, iterated regret mininmization, was defined and characterized. well-studied games.