Evolution of the Prisoner’s Dilemma
Source: https://news.upenn.edu/news/penn-team-s-game-theory-analysis-shows-how-evolution-favors-cooperation-s-collapse
This article discusses what happens to a populations choices to the Prisoner’s Dilemma over the course of many repeated runs. To recap, if both cooperate, they both receive a payoff, if one defects, the receive all the payoff, and if both defect, the get a payoff lower than if they had cooperated. In this tests, they allowed successful players “reproduce”, or carry on their winning strategy to future games. The study found that over time, those who co-operated with each other would win more, as defectors couldn’t win against other defectors and would eventually die out.
They also ran a test where one could switch their choice of action, as well as change the value for cooperating. Initially, the test was very similar, cooperative behavior overtook defective behavior. However, as the population began to adopt more and more cooperative behavior, there became more and more incentive to defect to receive a more optimal payoff. The study found that after a certain point, defecting became the optimal solution once again, causing the tendency towards cooperation to reverse.
We learned in class that the dominant strategy for the prisoners dilemma is to defect, but this only takes into account the test being run once. However, as we see in the article, the dominant strategy can evolve when repeated tests are run, as people who defect will die out as they cannot win against each other. Even more fascinating is that after evolving, the strategy with then change again, as the potential payoff is very high for a population with many cooperators. In this sense, the dominant strategy is always evolving: creating a strange oscillating equilibrium between high populations of cooperators and defectors. This article is a great example showing how introducing a small change (such as repeated test runs) can result in evolutionary game theory.