View article

We present a model of social interaction in which actors choose their partners and play the Chicken Game with them. In contrast to most previous models of the coevolution of games and networks, we assume that the actors can employ different actions against different partners. This allows us to derive two different solutions to the coordination and asymmetry problems in the game: egalitarian stationary conventions (based on indirect reciprocity at the network level) and alternating conventions (based on direct reciprocity in dyads). We derive predictions on the occurrence of the possible convention structures and test our theoretical findings in a computerized experiment. The experimental results indicate that the egalitarian conventions indeed emerge and persist and that the alternating conventions are much more likely than the stationary ones. However, although egalitarian in terms of payoffs, the convention …