Prisoner’s Dilemma

The Prisoner’s Dilemma gives a precise form to a recurring conflict: each participant has reason to defect, yet both can do better if they cooperate. In its one-round version, individually rational choices lead to a collectively inferior result. In repeated encounters, however, the past can shape expectations and make cooperation sustainable. The book uses this contrast to connect game theory with reciprocal altruism, animal behavior, and human arrangements whose rules can turn apparent competition into shared advantage.

The one-round dilemma

The book presents the Prisoner’s Dilemma as a simple gambling game that has attracted the attention of political scientists, economists, mathematicians, and psychologists. A banker adjudicates and pays the two players. Each has two cards, marked cooperate and defect; both choose one and lay it face down, so their choices are simultaneous and neither knows the other’s move until the banker reveals the cards. The game’s structure matters more than the exact sums involved.

The outcomes are ranked so that the temptation to defect is better than the reward for mutual cooperation, which is better than the punishment for mutual defection, which in turn is better than the sucker’s payoff for cooperating while the other defects. The account gives mutual cooperation a payment of $300 each and unilateral defection a $500 temptation payoff for the defector and a $100 fine for the cooperator. The ranking creates the dilemma: whichever card the other player chooses, a player can reason that defecting would produce the better individual outcome.

The same logic explains why two rational players, each reasoning in their own interest, both defect. Their choices leave each with a poor result, although both know that mutual cooperation would have paid them more. The tension is not simply between good and bad intentions. It arises from the way the rules translate each player’s choice into a payoff, making defection individually advantageous even when its joint outcome is inferior.

The book also illustrates the structure through two prisoners, Peterson and Moriarty, held separately and unable to make a pact. Each can betray the other by turning King’s Evidence, or cooperate by remaining silent. If one betrays while the other stays silent, the betrayer goes free and the silent prisoner receives a heavy sentence; if both betray, both are convicted but receive some credit for giving evidence. If both remain silent, the evidence is insufficient to convict them of the main crime, and they receive a small sentence for a lesser offence. Because neither can ensure that the other will stay silent, each has reason to betray, and both risk the harsher shared result.

Repetition, trust, and reciprocal response

The book’s route out of the one-round dilemma is not a change in the payoffs but a change in the circumstances of play. A single encounter offers no reliable way to reach or police an agreement. The repeated version brings the same players back together for an indefinite number of rounds. After each simultaneous choice, the cards are collected and play begins again. The prospect of future encounters gives participants an opportunity to build trust or mistrust, reciprocate help, placate, forgive, or retaliate.

In a ten-round illustration, a player could collect as much as $5,000 by exploiting an opponent who cooperates every time, but mutual cooperation on all ten rounds would allow each to collect $3,000 from the banker. Players can use one another’s previous actions to judge whether continued cooperation is safe; they can, in effect, police each other’s behavior. Persistent mutual defection remains possible, as does a mixed sequence of cooperation and defection. Repetition does not guarantee trust, but it makes present conduct consequential for later encounters.

This logic supplies a framework for reciprocal altruism, in which a benefit is given now with the possibility of repayment later. The book links that framework to the problem of mutual grooming: a bird can remove a parasite from a friend’s head, a place the recipient cannot reach with its own bill, and may later receive the same service in return. Such help has a cost in time and energy, so indiscriminate helpers can be exploited by individuals who accept assistance but do not repay it. The model therefore directs attention to whether individuals can recognize one another and meet repeatedly, not just to whether cooperation seems intuitively sensible.

The book connects this approach to work on reciprocal altruism and the Prisoner’s Dilemma, including Robert Trivers’s discussion of delayed repayment and later work by Robert Axelrod, partly in collaboration with W. D. Hamilton. It also recalls the strategy of Grudgers: they help one another but withhold help from individuals who previously refused to help them. In the account, Grudgers came to dominate Suckers, who help indiscriminately and are exploited, and Cheats, who exploit others but also do one another down. The comparison shows why reciprocal cooperation differs both from unconditional giving and from unrestrained exploitation.

Cooperation beyond a single species

The model is also used to make sense of cooperation between different species. The book names cleaner fish and ants that tend aphids as examples of relationships in which reciprocal altruism is not confined to members of one species. The point is not that participants consciously calculate the game. The book describes strategies as unconscious behavior programs laid down by genes; game-theoretic language is a way to analyze the consequences of those behavior patterns.

A more detailed example comes from G. S. Wilkinson’s observations of vampire bats. A bat that finds a blood meal may return to its roost with a surplus, while another returns empty; on another night their fortunes may reverse. Wilkinson observed bats regurgitating blood for less fortunate roostmates. Of 110 regurgitations, 77 could readily be understood as mothers feeding their young, and many other instances involved genetic relatives. Some donations, however, were between unrelated bats, especially frequent roostmates with opportunities to meet repeatedly.

The book examines whether this sharing fits the incentives of the game. Wilkinson estimated how long bats at different levels of starvation would take to die, translating blood into hours of prolonged life. The same quantity of blood could extend the life of a highly starved recipient more than that of a less starved one. Although a donor’s act increased her own risk, the recipient’s survival benefit could be much larger than the donor’s cost. On an unlucky night, a bat could benefit greatly from receiving blood; on a lucky night, it might gain slightly by withholding a donation, if it could avoid a later response.

Recognition is crucial to that possibility. In Wilkinson’s captive-bat experiment, one bat was removed for a night and starved while the others were fed, then returned to the roost. The population included bats from two caves many miles apart, allowing the experiment to distinguish familiar roostmates from newcomers. Across thirteen donations to experimentally starved bats, twelve came from an old friend from the same cave and one from a new friend. The result supports the relevance of individual recognition and repeated association to reciprocal sharing, while the book presents the case as a fit with the model rather than as proof that the bats consciously follow a strategy.

Rules, institutions, and shared gains

The Prisoner’s Dilemma also helps the book distinguish situations that look competitive from those that can yield mutual benefit. Divorce is used as an example: even after a marriage ends, a couple may have reasons to cooperate, including the welfare of children and the cost of paying two lawyers. Yet the legal arrangements described in the book require a lawyer to accept only one member of a couple as a client. The other must seek separate advice or go without it, and the case is framed as one side against the other even when the couple would prefer an amicable settlement. The rules can thus reinforce an adversarial game; the book notes that lawyers may benefit, while also acknowledging that some persuade clients to settle out of court.

A football match in the English Football League shows how a change in circumstances can alter the game’s structure. On May 18, 1977, Bristol and Coventry played each other while Sunderland’s result affected which club would be relegated from the First Division. Bristol and Coventry’s match ran five minutes late, so news from Sunderland reached the ground before it ended. Sunderland had lost, and both Bristol and Coventry needed only a draw to avoid relegation. With that news displayed on the electronic message board, scoring ceased to serve either team’s interest: pursuing a goal could leave its defense exposed and risk defeat. The teams shifted from a fierce contest to securing the draw, and supporters who had been rivals joined in celebration. A game that had seemed zero-sum became one in which both clubs could benefit from the same result.

The book contrasts that match with the general appeal of spectator sports, which are usually exciting because teams strive against each other, and with ordinary life, which is not organized for spectators. In human, animal, and plant life, an external circumstance can make it possible for participants to benefit from one another’s success rather than having to defeat a rival. In this sense, cooperation remains compatible with the book’s account of selfish genes: mutually beneficial behavior need not depend on universal selflessness.

Repeated play, however, depends on uncertainty about when the interaction will end. If two players know that a game lasts exactly 100 rounds, the final round becomes a one-off dilemma, making defection the individually rational choice. The same reasoning then applies to round 99, and works backward through every earlier round. For cooperation to remain viable, the end must be unpredictable or known only to the banker. The expected duration of the relationship—the book’s “shadow of the future”—therefore affects whether players have reason to maintain it. Even an uncertain ending can become a strategic concern if players infer from circumstances that the game is nearly over.

Scope and significance

The Prisoner’s Dilemma is presented as more than a puzzle about two people choosing cards. The book uses it to investigate how incentives, information, repetition, recognition, and rules shape the possibility of cooperation. The same basic contrast recurs in the prisoners’ mutual distrust, the vampire bats’ repeated relationships, divorce procedures that encourage adversarial framing, and a football match whose stakes change when news arrives from another ground.

The model does not claim that every generous act is disguised self-interest, or that animals consciously calculate payoffs. The book explicitly leaves room for genuine disinterested altruism in discussing unpaid blood donors, while using the vampire-bat case to show how reciprocal sharing might be favored under particular conditions. It also stresses that gene-level explanations do not prescribe human morality: the possibility of cooperation within an evolutionary framework is not an instruction to treat people selfishly.

Its broader significance lies in showing how a basically competitive account of evolution can accommodate cooperation without treating it as a contradiction. A one-off encounter may reward defection, while an ongoing relationship can make reciprocity viable; a rule can lock participants into conflict, while changed circumstances can reveal shared gains. The decisive question is therefore not simply whether people or animals are selfish, but what structure of interaction makes cooperation stable, vulnerable, or worth maintaining.