Abstract:
Tag-based ethnocentric cooperation is a highly robust behavior which can evolve and prevail under a wide variety of conditions. Recent studies have demonstrated, however, that ethnocentrism can temporarily be suppressed by other competing strategies, especially in its early evolutionary stages. In a series of computational experiments, conducted with an agent-based evolutionary model of tag-mediated cooperation, we addressed the question of whether a stochastically established and once dominant non-ethnocentric strategy such as indiscriminate altruism can stably persist and permanently outweigh ethnocentrism. Our model, simulated on various complex network topologies, employs simple haploid genetics and asexual reproduction of computational agents equipped with memory and heritable phenotypic traits. We find that in combination with an implemented memory mechanism and tags, random bias acting in favor of altruists can lead to their long-lasting victory over all other types of strategists. The difference in density between altruistic and ethnocentric cooperators increases with greater rewiring of the underlying network, but decreases with growing population size. These findings suggest that randomness plays an important role in promoting non-ethnocentric cooperation and contributes to our understanding of how other than adaptive mechanisms can initiate the design of novel behavioral phenotypes, thereby shaping surprisingly new evolutionary pathways.