Evolution of Indirect Reciprocity

Summary of: Evolution of Indirect Reciprocity

Author(s) / Editor(s)

Cooperation through indirect reciprocity, captured by the phrase "I help you, someone else helps me", requires the evolution of reputations and communication of those reputations among the larger group (as in the human instinct to gossip), cognitive abilities beyond being able to identify relatives (required for kin selection) or the individuals who have cooperated with you in the past (required for direct reciprocity).

Publication Reference

Published in/by
Nature 437, 1291-1298
October 27, 2005


  • "The hypothesis that more information leads to more cooperation has been confirmed in experiments, which compare three information conditions. In one condition, players have no information about their co-players; in the second they are told about what their co-players have decided when last in the role of a donor; and in the third they also know about the score of the recipient of the co-player. We note that this is not always enough to decide whether a previous defection was justified or not. However, the additional knowledge did enhance cooperation."
  • "Indirect reciprocity is situated somewhere between direct reciprocity and public goods. On the one hand it is a game between two players only, the donor and the recipient, but on the other hand it has to be played within a larger group. Richard Alexander claimed that indirect reciprocity originates from direct reciprocity in the presence of interested audiences."
  • "It is easy to conceive that an organism experiences as 'good' or 'bad' anything that affects the organism's own reproductive fitness in a positive or negative sense. The step from there to judging, as 'good' or 'bad', actions between third parties, is not obvious. The same terms 'good' and 'bad' that are applied to pleasure and pain are also used for moral judgements: this linguistic quirk reveals an astonishing degree of empathy, and reflects highly developed faculties for cognition and abstraction."
  • Even a group of players with discriminating strategies can be sidetracked by imperfect transfer of reputation information, as in unfounded rumors or exaggeration: "if players have different views about the reputation of others, then errors in perception can undermine cooperation."
  • In empirical studies, discriminating players are sensitive to their own score: "players who justifiably refuse to donate to a defector show an increased tendency to provide donations in the following round, as if to make up for that refusal. This indicates that they expect their refusal to lower their score in the co-players' eyes and that they do not rely on the community's understanding."

Cooperation through indirect reciprocity, can be captured by the phrase "I help you, someone else helps me". Indirect reciprocity helps explain how cooperation is possible at all when economic transactions move beyond small villages where one can easily keep track of one's interactions with everyone else. The success of strategies of indirect reciprocity in empirical studies might be attributable to the fact that humans care so deeply not only about how they are treated, but about the results of interactions between third parties. This concern and the desire to communicate concerns, or gossip, might in turn be explained by evolutionary psychology and the benefits of cooperation in large groups, surpluses resulting from division of labor. To test strategies of indirect reciprocity no two players can interact more than once and the scores of players (the portion of times they have cooperated with others) must be visible. A player choosing a simple version of indirect reciprocity will only cooperate with those whose score is above a certain threshold. However, this player might be punishing another player using indirect reciprocity who has only interacted with defectors. "Effectively, discriminating players pay a cost for punishing bad co-players. Such a form of altruistic punishment can promote cooperation in the community, but at a cost to the punisher, and thus can be viewed as a social dilemma." A more sophisticated strategy would have a player discriminate between justified defection (defecting to punish someone who always defects) and unjustified defection (defecting regardless of the recipients reputation). This strategy avoids the case where a group of players who always cooperate is invaded by a group of players who always defect, but it requires the cognitive abilities to keep track of interactions that are far removed from one's own.