G. Roberts
Human cooperation among non-kin has long been accepted as being based on reciprocity: the idea that we return help from individuals that we re-meet. However, this paradigm has been challenged by Efferson et al. (Efferson, C., et al., Super-additive cooperation. Nature, 2024. 626, 1034-1041) who found that repeated interaction was insufficient to support cooperation without selection between groups. The authors based their conclusion on a model in which cooperative partners used a novel function to determine their response. In their system, cooperation through repeated interaction broke down. I point out here that the collapse of cooperation was not the result of the novel strategy set. Instead, it reflects a standard modelling result in which a cooperative environment allows drift towards unstable, undiscriminating strategies. I show how modifying model assumptions such as changing the implementation of mutation or allowing persistent defection maintain the variation which selects for discrimination. This discrimination in turn stabilizes reciprocity. I conclude that Efferson et al.'s results can readily be explained by existing theory and that repeated interaction can support cooperation without requiring between-group competition.