The Development of Cooperation in Genetic

In the Prisoner's Dilemma, two players decide, without communicating, whether to cooperate or defect. The players engage in repeated rounds of the Prisoner's Dilemma in the Iterated Prisoner's Dilemma (IPD), a model of competition and cooperation in social interactions. A genetic algorithm is a search procedure inspired by the mechanisms of biological evolution. In it, a population of solutions to a problem evolves, producing better solutions. Researchers have investigated the development of cooperation using genetic algorithms whose populations consist of IPD-playing strategies. This paper presents such an inquiry, in three parts. In the first, the average level of cooperation among strategies increases with the length of their IPD contests. In the second and third, the population is divided into clans that apply different strategies in contests with members of the same and different clans. Cooperation in "foreign" contests rises with the length of the contests, but falls with the number of clans.