Published on Sep 28, 2024
The objective:
The model of the Iterated Prisoner's Dilemma has been studied closely by many researchers in how it applies to the development of cooperation. In sociology this has been extended to how punishment - essentially hurting your opponent at a smaller cost to yourself - works to maintain cooperation.
This project takes a more concrete approach, evolving strategies in an effort to see whether retribution is a robust phenomenon like cooperation has been shown to be.
This is an entirely computer-based experiment. It uses a cellular-automata based simulation (coded in C++) to model interactions between hundreds of "organisms" playing the Prisoner's Dilemma over and over again.
Strategies evolved based on which were the most successful Data analysis was done in MATLAB, looking at the strategies of many different organisms over time in a number of sophisticated ways using tools developed by the author.
Punishment was not used in any clear strategies for any extended period of time in any of the runs under a variety of conditions.
Part of the reason punishment did not arise may have been because of the volatile environment that was created. However, we can conclude with almost complete certainty that retribution is not a robust phenomenon, and does not arise simply from a situation in which players attempt to maintain cooperation.
This project uses an evolutionary simulation to understand whether the general phenomenon of retribution can be explained easily through game theory.