Real world dilemmas rarely involved just two choices and perfect interactions without mistakes. In extending the realism of the Iterated Prisoner's Dilemma (IPD) game, prior evolutionary approaches included intermediate choices or mistakes (noise). This study takes a step further using a co-evolving population of neural networks playing the IPD game with both intermediate choices and noise. Several issues will be addressed, which include the evolution of cooperation and the evolutionary stability in the presence of noise and more choices. Our experimental study shows that noise has a negative impact on the evolution of cooperation, but could improve, surprisingly, the evolutionary stability.
|Title of host publication
|Proceedings of the 2004 Congress on Evolutionary Computation, CEC2004
|Number of pages
|Published - 2004