A computational approach to the evolution of competitive strategy*

Abstract

The weakness of the theoretical link between existing micro and macro approaches to the study of organizational behavior is particularly evident when one tries to follow the indications of contemporary theories of organization and analyze competition as an emergent social process. In this paper we present a computational model of how individual competitive behavior evolves into aggregates with complex collective properties which determine the structural context of future individual action. According to the model, aggregate regularities emerge from the local interaction among agents playing a spatial version of the repeated prisoner's dilemma. At each round the local strategy which yields the maximum individual benefit will survive and replace the strategy having inferior local fitness. Through this evolutionary mechanism, the strategy with the highest reproductive value at the strictly local level, propagates to more distant sites by influencing the choice of units in partially overlapping neighborhoods and generates emergent collective consequences. The model is simulated under a variety of evolutionary conditions to explore the effects of the range of local interaction and of frequency‐dependent selection on the evolution of competitive strategy.