A Transactional Theory of Within-Group Conflict.

Transactional models of social evolution emphasize that dominant members of the society can be favored to donate parcels of reproduction to subordinate members in return for cooperation. I construct a formal theory of intragroup conflict within the framework of transactional models by determining the maximum extent to which colony members can be selfish without destabilizing the group. The difference between the maximum value of the subordinate's fraction of group reproduction that the dominant can tolerate before ejecting the subordinate and the minimum value required by the subordinate to stay and cooperate peacefully in the group defines the "window of selfishness," which in turn predicts the frequency of within-group conflict. The window of selfishness tends to increase with increasing group reproductive output, increasingly harsh ecological constraints on solitary breeding, and, counterintuitively, increasing relatedness between subordinate and dominant. Increasing fighting ability of the subordinate can either widen or narrow the window of selfishness, the latter being most likely when ecological constraints on group living are strong. Although increasing relatedness is predicted to increase the rate of within-group aggression, the mean intensity of an aggressive act should decline, as predicted by the general theory of honest signaling between relatives and the tug-of-war models of within-group selfishness. In the bidding game, in which multiple dominants bid for the services of a subordinate, the window of selfishness is predicted to have zero width. A zero-width window of selfishness and low conflict also are predicted for saturated N-person groups, that is, groups whose total output is a concave function of group size and in which the dominant is not favored to admit additional subordinates. The model's predictions are compared to empirical evidence and to predictions of alternative models of intragroup aggression, including the value-aggression model and the pure tug-of-war model.