We view learning as a social as well as a cognitive process, mediated through complex interactions among intelligent agents that coevolve with dynamic environments. As each agent's interactions and decisions bring it new information about the world, the agent interprets that information, refines old strategies of action and adopts new strategies that gradually influence the world. The agent's internal representation, or model, of the world changes both as the agent accumulates greater knowledge and experience, and as the world itself changes. We seek to elucidate the relationships between internal changes in an agent's model of the world, and external changes or events in which the agent participates.
Current work in this area centers on the SimEd project, which aims to model the U.S. educational system using elements of game theory, learning theory and multi agent simulation. We cast learning as a cooperative interaction among individuls playing out a meta-game of learning, loosely based on the Iterated Prisoner's Dilemma paradigm, with multiple players and multiple, sometimes conflicting, goals.
Relevant papers: