Many game theorists are turning to evolutionary simulations to model the behavior of boundedly rational agents. This new methodology allows researchers to observe purely adaptive behaviors in games, to observe differences of behavior due to changes in the games' parameters, to discover equilibria in games that are too complex to calculate analytically, and to discover new strategies for playing the games. I extend this methodology to a more complex class of games than had previously been attempted. I create a coevolutionary environment in which three agents, represented by classifier systems, play a characteristic function game. Although the agents have no computational capabilities, they learn to adapt reasonably intelligent behavior.
展开▼