Fig. 4: Evolution of strategies under three scenarios.
a–f, Relative frequencies of initial transfers (a,c,e) and response functions (b,d,f). We bin initial transfer values and indicate relative frequencies by bin with a greyscale. We categorize response functions into discrete types. Because ambiguous strategies cover such a broad range, we separately identify extreme forms of ambiguous reciprocity (supporting information section 2.1.1726). Specifically, quasi-de-escalators are similar to de-escalating strategies in that they generate extremely low levels of cooperation in the long run. Similarly, quasi-escalators are similar to escalating strategies in that they generate extremely high levels of cooperation. For the results here, migration rates are high, group competitions have outcomes that are highly sensitive to differences between groups, and cancellation effects at the individual and group levels are as strong as possible (supporting information section 2.1.1726). Initial conditions favour cooperation in the repeated interactions (a,b) and group competition scenarios (c,d), but they disfavour cooperation in the joint scenario (e,f). Nonetheless, uncooperative forms of reciprocity prevail in the former two scenarios, whereas cooperative forms of reciprocity prevail in the latter scenario. For each scenario, results are based on five independently simulated populations.
