Evolutionary Game Theory

Evolutionary game theory applies the logic of strategic interaction to natural selection, showing how the success of a behavior depends not only on what it does but also on what other members of the population do. Its importance in the book is twofold: it offers a way to analyze conflict as a balance of costs and benefits, and it explains how cooperation can emerge within a framework of individual and gene-level competition.

From behavioral examples to strategic explanation

The book first distinguishes behavioral altruism and selfishness from claims about subjective motives. The relevant question is whether an act apparently raises or lowers the survival prospects of the actor and its recipient; apparent effects may prove misleading when examined more closely. Animal examples illustrate the distinction, but the author explicitly warns that stories are not serious evidence for generalizations. The larger explanation, he argues, must come from the fundamental logic of gene selfishness rather than from an assumption that animals sacrifice themselves for the good of a species or group.

That warning sets the methodological ground for evolutionary game theory. The theory does not infer a strategy merely because an outcome looks beneficial to a group, nor does it require an animal to calculate consciously. Instead, it asks what behavioral policies would persist under selection, given the other policies present and the consequences of encounters among individuals.

Strategies, payoffs, and stability

The framework begins with a cost-benefit analysis of whether to fight and, once a fight begins, whether to escalate or retreat. An elephant seal considering a rival’s harem provides an illustrative case: the potential prize must be weighed against the chance of injury or death, the rival’s likely fighting ability, and the time and energy consumed by combat. Waiting to grow stronger may improve the eventual prospect of success. This imagined calculation is not a claim about conscious deliberation; it expresses how behavior can be analyzed as if costs and benefits were being weighed.

John Maynard Smith, working with G. R. Price and G. A. Parker, gave this approach a formal expression through game theory. His central concept is the evolutionarily stable strategy, which he traced to W. D. Hamilton and R. H. MacArthur. A strategy is a pre-programmed behavioral policy, described as a set of instructions for analytical convenience rather than as a plan consciously formed by an animal. A strategy is evolutionarily stable when, once adopted by most of a population, no alternative strategy can do better by invading it. Because each individual’s best response depends on the behavior of the others, stability is a property of the population’s strategic composition, not a ranking of behaviors in isolation. After environmental change, the population may pass through a period of instability or oscillation; once a stable strategy is reached, selection penalizes deviations from it.

The book makes the logic concrete with an abstract contest between Hawk and Dove strategies. Hawks fight persistently and retreat only when seriously injured; Doves threaten but do not inflict harm, and eventually yield after prolonged posturing. Against a Dove, a Hawk wins without being injured; Hawk-versus-Hawk contests risk severe injury or death; Dove-versus-Dove contests consume time before one backs down. To compare the consequences, the model assigns arbitrary payoffs: 50 points for winning, zero for losing, minus 100 for serious injury, and minus 10 for wasting time. These points stand for effects on gene survival, not for an independently important numerical scale. The question is therefore not simply whether a Hawk can beat a Dove, but whether either strategy can remain stable when its frequency in the population changes. The book concludes that neither pure strategy is evolutionarily stable on its own in this model.

Asymmetry and territorial behavior

Game-theoretic reasoning also explains why an encounter need not be symmetrical. A resident may know the terrain better and be less tired than an intruder. The conditional policy of resident wins and intruder retreats can be stable because individuals benefit from holding a particular place and remaining there. Its reverse, in which the intruder wins and the resident retreats, has a self-undermining consequence: individuals would be selected to avoid being residents, producing continual movement and tending to erase the resident category itself.

Niko Tinbergen demonstrated the role of location with two male sticklebacks that had built nests at opposite ends of a tank. He placed each fish in a glass test-tube and held the tubes together, allowing them to display toward one another without direct contact. Near male A’s nest, A adopted an attacking posture and B attempted to retreat; near male B’s nest, their roles switched. Moving the tubes from one end of the tank to the other changed which fish behaved as resident and which as intruder. The observed pattern was consistent with the conditional strategy: attack when resident, retreat when intruding. On this account, territorial defense need not require a separate explanation in terms of a direct benefit to the species; it can arise as an evolutionarily stable response to the asymmetry created by time of arrival at a patch of ground.

Cooperation, rules, and repeated interaction

The same strategic perspective extends beyond contests. In the book’s account of the Prisoner’s Dilemma, two players each choose between cooperation and defection, placing their cards face down so neither can respond to the other’s move before choosing. A banker reveals the choices, and each player’s winnings depend on both cards. The structure matters because an apparent clash of interests can sometimes be transformed into a situation in which both parties benefit from cooperation.

Divorce provides an institutional example. A couple may have reasons to cooperate even after a marriage breaks down, including the welfare of children and the preservation of family finances. Yet, as the book describes the rules then operating in England and until recently in all fifty states of the United States, professional restrictions prevented one lawyer from representing both members of the couple. Each person had to find separate legal representation, and the case could be framed as an adversarial contest even when the couple wanted an amicable outcome. The account notes that lawyers might benefit from the resulting zero-sum framing, while also acknowledging that some lawyers persuade clients to settle cooperatively out of court. The example shows that outcomes depend not only on individual dispositions but also on rules that shape who can cooperate and what choices are available.

The Bristol–Coventry football match of May 18, 1977, supplies a sharp illustration of how new information can change the game. With relegation still at stake for Sunderland, Bristol, and Coventry, a draw between Bristol and Coventry would save both if Sunderland lost. The Bristol–Coventry match was running five minutes late, so the Sunderland result became known before it ended. Two minutes before the final whistle, news that Sunderland had lost was displayed to the players. A draw now served both teams, and scoring further goals could expose either side to the risk of defeat and relegation. The players shifted from vigorous competition to protecting the draw; spectators who had been rivals joined in a shared celebration. A contest that had functioned as zero-sum became nonzero-sum when outside information altered the teams’ incentives.

Cooperation in repeated games depends on the expectation of future encounters. The book calls this expectation the shadow of the future. If players know precisely that a game will end after a fixed number of rounds, the last round resembles a one-off Prisoner’s Dilemma, encouraging defection; reasoning backward then makes defection appear advantageous in each preceding round. For that reason, repeated-game analysis assumes that the ending is unpredictable or known only to the banker. The possibility of future interaction can sustain cooperation, but estimates that the encounter is about to end may instead make defection tempting.

Scope and evidential limits

The book presents game theory as a way to connect strategic models with evolutionary explanation, not as proof that every helpful-looking behavior serves a group. Its author repeatedly distinguishes an illustrative example from evidence capable of establishing a general claim. In discussing Wynne-Edwards’s proposal that mass gatherings help animals gauge population density and restrain reproduction for the group, he finds the proposal plausible in outline but says its evidence consists of examples open to alternative interpretations, including explanations in selfish-gene terms. This contrast underscores the need to test models against evidence rather than treat an appealing account of collective benefit as established.

The same caution applies to adaptive explanations more generally. For the caddisfly house, the argument is that selection could favor ancestral genes that led their bearers to build effective structures, because adaptation requires hereditary variation on which selection can act. But the book also notes that the genetics of caddis houses had not actually been studied and that breeding caddisflies in captivity is difficult. The theoretical interpretation rests on the house being a Darwinian adaptation, not on a completed genetic investigation. Evolutionary game theory thus supplies a disciplined way to ask which strategies and outcomes selection could maintain, while the strength of any particular application still depends on the evidence for its assumptions and mechanisms.