Mixing rules: When to cooperate in a multiple-team competition

In the study of social dilemmas, there has been a strong divide between empirical studies of how people behave and simulation experiments focusing on how people ought to behave. This study bridges that gap by deducing rules-in-use by participants in a human experiment in a multiple-team social dilemma. It then uses a computer simulation to evaluate which of the identified rules-in-use generates the best performance. The results suggest that participants use either a Win-Stay/Lose-Revert strategy or a Win-Cooperate/Lose-Defect strategy. While a mean-Win-Stay/Lose-Revert strategy dominates most often, the Win-Cooperate/Lose-Defect strategy performs better in a two-team than in a single-team setting.