Choice of partners in multiple two-person prisoner's dilemma games: An experimental study

We examine the effect of unilateral and mutual partner selection in the context of prisoner's dilemmas experimentally. Subjects play simultaneously several finitely repeated two-person prisoner's dilemma games. We find that unilateral choice is the best system. It leads to low defection and fewer singles than with mutual choice. Furthermore, with the unilateral choice setup we are able to show that intendingdefectors are more likely to try to avoid a match than intending cooperators. We compare our results of multiple games with single game PD-experiments and find no difference in aggregate behavior. Hence the multiple game technique is robust and might therefore be an important tool in the future for testing the use of mixed strategies.