International Workshop on
Regulated Agent-Based Social Systems:
Theories and Applications (RASTA'02)
Abstract for
- 16:30 - 17:30 (third presentation)
- The Evolution of Specialization in Groups
David Hales
In a previous paper (Hales 2002a) we presented simulation results that demonstrated the evolution of tag based groups composed of cooperative (in-group altruistic) individual agents performing specialised functions. We showed how teams of individual maximisers (who copy the behaviours of those who outperform them) come to form internally specialised and cooperative groups that efficiently exploit their environment. Also in a previous paper (Hales 2002b) we demonstrated that the efficiency of the specialisation process was highly dependent on the searching strategy employed by agents to locate in-group members with required skills. Specifically we showed that populations of agents with smart searching strategies outperformed populations of dumb (random) search strategies even when the costs of smart searching were much higher. We hypothesised that in mixed populations smart strategies would out-evolve dumb ones. In this paper we test this hypothesis. Our results show that smart strategies do indeed outperform dumb strategies for significant periods of time but that dumb strategies persist also. The time series of individual runs show cycles of smart and dumb strategies in the population over generations. We argue that the study of such phenomena at this given level of abstraction offers a possible minimal way towards understanding the evolution of institutional roles and internal specialisation without positing actions that originate at the supra-individual level (though we do not discount such actions).
Daniel Moldt
Last modified: Fri Jun 21 15:55:45 CEST 2002