Tit for Tat

Tit for Tat is a strategy for repeated interactions: it begins by cooperating and thereafter copies the other player’s previous move. Its force lies in a compact rule that joins openness to cooperation with a response to defection. In the account of the Prisoner’s Dilemma, it is both a tournament-winning computer program and a model for understanding how reciprocal cooperation can arise without assuming that participants act for the good of a group.

Definition and Operation

In the original two-player game, each player chooses between cooperating and defecting without seeing the other’s choice. A banker reveals both choices, and each player’s payoff depends on the combination. Tit for Tat turns this one-round dilemma into a rule for play across successive rounds: cooperate on the first move, then do what the opponent did on the previous move. The strategy is not a consciously reasoned plan attributed to an animal or program; it is a pre-programmed policy whose operation can be described as though the player followed simple instructions.

The rule is conditional rather than unvaryingly benevolent. Its first move is cooperative, but it defects in response to a prior defection. In Axelrod’s terminology, it is “nice” because it is never the first to defect, and “forgiving” because retaliation answers the recent move rather than preserving a permanent grievance. It therefore distinguishes between an unprovoked defection and a partner’s return to cooperation: it responds promptly to the former, but can resume cooperation when the other player does.

Tournament and Outcomes

In the first computer tournament described in the book, fifteen preprogrammed strategies played against one another, including copies of themselves. The winning entry was Tit for Tat, submitted by the psychologist and game theorist Anatol Rapoport. Its rule was strikingly simple compared with the more elaborate programs in the competition. The tournament’s strategies were not human players; their authors supplied programs that served as proxies, in the book’s analogy, for genes programming bodies.

When Tit for Tat meets another copy of itself, both cooperate on the opening move. Each then copies the other’s cooperation, so the pattern continues and both reach the full benchmark score of 600 points. This result shows the strategy’s best-case outcome not as a one-sided victory but as sustained mutual benefit: neither needs to exploit the other for both to do well.

The contrast with Naive Prober shows how the rule responds when cooperation is interrupted. Naive Prober is a hypothetical teaching example, not an entry in the tournament. It follows Tit for Tat except that it defects occasionally without provocation. If it defects after a run of cooperation, Tit for Tat has cooperated on that move and receives the sucker’s payoff of zero, while Naive Prober gains five points. On the next move, however, Tit for Tat defects in imitation of its opponent’s last move, while Naive Prober copies Tit for Tat’s preceding cooperation. The players then alternate between five and zero points apiece, averaging 2.5 per move rather than the three points each could receive through mutual cooperation. A single opportunistic gain thus disrupts a more profitable joint pattern.

Across the tournament, the eight highest-scoring strategies were the same eight classified as nice: they never defected first. Tit for Tat averaged 5,045 points, or 84 per cent of the 6,000-point benchmark as reported in the passage; the other nice strategies scored slightly less. The strongest nasty strategy, Graaskamp, reached 66.8 per cent. The result supports the book’s claim that, under repeated play, a strategy need not be the first to exploit others in order to succeed.

Retaliation, Forgiveness, and Limits

Tit for Tat’s retaliation is immediate but not permanent. The book contrasts it with Grudger, a strategy that never forgets a defection. A formally identical strategy called Friedman appeared in the tournament; although technically nice because it did not defect first, it performed next to worst among the nice strategies. Its lasting grievance could trap both players in mutual recrimination, even if an opponent became remorseful and returned to cooperation. Tit for Tat’s shorter memory offers a route out of that cycle: after registering a defection, it can answer cooperation with cooperation again.

The book also notes a strategy that would be more forgiving still. Tit for Two Tats would tolerate two consecutive defections before retaliating. Axelrod’s calculations indicated that, had it been submitted, it would have won the tournament, because it would have been especially effective at avoiding extended retaliation. This comparison clarifies that Tit for Tat’s success does not mean its exact degree of forgiveness is always optimal; the strategic value of a response depends on how it shapes the next turns of the interaction.

Repetition and Real-World Applications

The strategy’s cooperative outcome depends on the possibility of meeting again. The book calls the expectation of future encounters the “shadow of the future.” If two players know that a repeated game ends after exactly one hundred rounds, the logic of the final round can work backward: each expects the other to defect at the end, and the same reasoning can carry defection through earlier rounds. For cooperation to remain viable, the ending must be unpredictable to the players or known only to the banker. Tit for Tat is therefore not a general instruction to cooperate in every situation; its relevance rests on repeated interaction and the incentives created by future responses.

The book uses Tit for Tat to interpret examples beyond computer games. In Eric Fischer’s observations of hermaphrodite sea bass, paired fish alternated between female and male reproductive roles. Playing the female role when it was one’s turn is compared to cooperation; trying instead to take the cheaper male role is compared to defection. A partner could respond by refusing to play the female role on a later turn or by ending the partnership, and pairs with uneven sharing of roles tended to break up. The observed alternation is presented as consistent with a Tit for Tat-like arrangement, not as proof that the fish consciously follow a rule.

The vampire-bat example likewise connects reciprocal behavior to recognition and repeated association. G. S. Wilkinson found that bats sometimes regurgitated blood for less fortunate roostmates. His captive-bat experiment separated bats from two caves, starved one individual overnight while feeding the others, and then returned the hungry bat to the roost. Of thirteen observed donations, twelve came from an old friend from the same cave and one from a bat taken from the other cave; the odds against that distribution arising by chance were calculated at less than one in 500. The evidence supports the importance of recognizing particular partners, a condition for reciprocal strategies to work, while the discussion also distinguishes such sharing from the many donations between genetic relatives.

Taken together, these examples make Tit for Tat more than a tournament entry: it is a way to analyze how cooperation can emerge from self-interested behavior when participants meet repeatedly, can respond to one another, and do not remain locked into grudges. Its opening cooperation makes mutual benefit possible; its retaliation discourages opportunistic defection; and its capacity to return to cooperation prevents one breach from dictating every later move.