Source
Trivers (1971), “The Evolution of Reciprocal Altruism”, LIT-510 (The Quarterly Review of Biology 46: 35–57), read in full in NOTE-402.
What was actually shown
The model (pp. 35–39). Altruism dispensed at random is eliminated. Altruism dispensed by reference to the recipient’s own altruism spreads if altruists who are cut off by cheaters gain more than non-altruists do. Conditions (p. 37): many altruistic situations in a lifetime, repeated interaction with a small set of individuals, and symmetric exposure. Hamilton’s sketch: over 2n situations two altruists each get nR, and an altruist who switches after being cheated gets S + (n − 1)P. A rare non-altruist cannot spread unless T ≫ R. A rare altruist faces a barrier that weakens as n grows.
One case that fits. Cleaning symbioses: hosts do not eat their cleaners, signal before leaving and chase off threats. The evidence is natural history for each precondition. NOTE-402 rates it moderate.
One case that does not. Bird warning calls are explained by the caller’s own interest, not by reciprocity: a predator that eats a neighbour is more likely to learn and stay near the caller’s species and place. The abstract lists warning calls among the cases the model explains; the section rejects that (NOTE-402, corrections).
The human system (pp. 47–54). Friendship and liking motivate giving. Moralistic aggression, provoked by “injustice, unfairness, and lack of reciprocity”, protects the giver against cheaters. Gratitude and sympathy are sensitive to the cost and benefit of an act. Guilt motivates reparation once a breach is found, and more after a public transgression than a private one. Then: “once moralistic aggression has been selected for to protect against cheating, selection favors sham moralistic aggression as a new form of cheating” (p. 54), and likewise for the others, with suspicion to catch them. With language, “selection may favor the elaboration of norms of reciprocal conduct” (p. 52).
What could have come out otherwise. The social-psychology literature could have shown these emotions indifferent to cost, benefit and reciprocation. The findings Trivers cites fit, but they were gathered before the account and for other purposes.
What this does not say
- That the emotions were tested as adaptations. The account is fitted to existing findings. No study was designed for it (NOTE-402, C4, weak–moderate).
- That sham emotions are detected. Asserted, with no studies (C5).
- That the model is stable against invasion. That was shown later, in the form and with the limits of THEORY-672.
- That warning calls are reciprocal altruism. The paper argues they are not.
- Anything about fair division. The “unfairness” here is cheating in an exchange. Morality-as-cooperation’s division domain is not grounded here.
- That the dispositions are heritable. Assumed (p. 48).
Connections
- Axelrod & Hamilton (LIT-498, THEORY-672) formalize the model with a continuation probability w. Hamilton’s Part I (LIT-497, THEORY-408) is the foil: this model is built for cases where kin selection “can be ruled out”.
- The multiparty passage anticipates indirect reciprocity, which Alexander (LIT-495) made the basis of moral systems and Nowak (LIT-501, THEORY-567) gave a rule.
- Tomasello (LIT-125, NOTE-185; THEORY-495) argues that tit-for-tat reciprocity yields gratitude, not obligation. Here guilt is the regulating emotion of exactly that reciprocity. The two disagree on whether reciprocity can be a source of moral emotion beyond liking and gratitude.
- Morality-as-cooperation takes its exchange domain from here (LIT-503). Curry’s list for that domain (trust, gratitude, punishment, revenge, forgiveness) is close to Trivers’s.
- Moral disciplining (LIT-101, NOTE-132, A1) rests on the premise that reciprocal cooperation requires delaying gratification, which is the time lag between giving and return here.