Computer simulations showed that simple methods to obtain reciprocal cooperation based on spatial association or the sharing of similar features cannot work in the absence of memory of past interactions
Cooperative computerized agents that relied on such simple methods were regularly cheated by selfish agents that never returned cooperation. In evolving artificial populations of agents, cooperative agents always went extinct in a few generations. These studies show so called "symmetry-based reciprocity" cannot be evolutionary stable.
Computerized "agents" behave cooperatively
A set of computer simulations showed that artificial "agents" can reproduce several characteristics of animal cooperation if they can choose their partner based on cooperation received.
Recently, we developed agent-based models aimed at testing whether a strategy of pure partner choice could promote the evolution of cooperative behavior and reproduce significant aspects of cooperation in group living animals. Our models showed that agents equipped with a behavioral strategy of "choosing cooperation" can develop diffentiated social relationships and show a positive correlation between cooperation given and received. Furthermore, choosing cooperators won over selfish agents (that never cooperate) in an evolutionary model that included reproduction and selection along multiple generations. These computer simulations suggest partner choice is a significant aspect of cooperation in animals.
Researcher involved
