Trust in generosity: an experiment of the repeated Yes–No game

Compared to ultimatum games allowing for altruistic sanctioning via monitoring, Yes–No games only allow for unmonitored altruistic sanctioning. Game theoretically, the sequentially rational outcomes (for non-positive conflict payoffs of the responder) coincide, but the large multiplicity of equilibrium outcomes of ultimatum games is avoided by Yes–No games. As Avrahami et al. (J Socio-Econ 47: 47–54, 2013) for the ultimatum game, we experimentally implemented 100-period plays of Yes–No games for newly and randomly rematched player pairs to test whether their observation of fast and nearly universal convergence to equal sharing depends on its equilibrium property. Will there be significant altruistic sanctioning, convergence to solution play, or cyclicity in behavior? There are two possible pie sizes which either only proposers or both players know. The dynamics of play differ fundamentally from the quick convergence to equal sharing for ultimatum games. There is neither convergence to equal sharing nor to equilibrium play, but persistent heterogeneity in offers and (non)acceptance. Some participants engage in game theoretically unpredicted sanctioning even in later periods.