On the evolution of cooperation in general games of common interest

• Autores: Fernando Vega-Redondo
• Localización: Working papers = Documentos de trabajo: Serie AD, Nº. 11, 1993, págs. 1-14
• Idioma: inglés
• Enlaces
  • Texto completo (pdf)
• Resumen

  • This paper investigates an evolutionary model of equilibrium selection in which agents are randomly paired every period to play some general symrnetric game of common interest (i.e., a game where some strategy profile Pareto- dominates al1 other configurations). Along the process, players tend to imitate the strategy (or strategies) which currently yield the highest average payoff. Occasionally, they also experiment with arbitrarily chosen strategies. In the long run, if the rate of experimentation is small, the population ends up playing the uniquely efficient strategy profile "most of the time". Moreover, the convergence to such state of affairs is relatively fast, independently of population size.