Game of chicken
Also known as: Hawk-dove game, Snowdrift game
Two players race toward collision; whoever swerves first loses face, but crashing is worst.
What it means
An anti-coordination game in which each player wants to stand firm while the other yields, but mutual aggression yields the worst outcome for both. Its two pure equilibria are asymmetric — one swerves, the other doesn't — so the strategic problem is which player gets to be tough, often resolved by commitment, reputation, or signaling resolve. Paradoxically, visibly removing your own option to swerve (tearing off the steering wheel) can win by forcing the other to yield. It matters as a model of brinkmanship, from nuclear standoffs and strikes to traffic merges, where credibility and the threat of catastrophe shape the outcome.
How it works
The payoff structure defines the game. Two players each choose between yielding (swerving) and standing firm (driving straight). Both yielding gives a modest, shared outcome; one yielding while the other stands firm gives the firm player the best payoff and the yielder only a small loss of face; both standing firm produces the worst outcome for both, the crash. What makes it an anti-coordination game is that the players want to choose differently: each does best when its own move is the opposite of the other's. Formally there are two pure-strategy Nash equilibria, and they are asymmetric, since in each one player stands firm and the other swerves. The theory therefore identifies stable outcomes but does not say which player occupies which role. A third, mixed-strategy equilibrium has each player standing firm with some probability, which leaves a positive chance of the mutually ruinous crash. Chicken is usually placed beside two neighbours. In the prisoner's dilemma, mutual defection is itself an equilibrium, so the tension runs between individual and collective interest. In chicken the disaster of mutual aggression is nobody's equilibrium, and the problem is distributional: who gets to be the tough one. The stag hunt, by contrast, rewards matching rather than mismatching, and its danger is timidity rather than collision.
Why commitment can win
The counter-intuitive lever, developed by Thomas Schelling, is that reducing your own freedom of action can improve your position. If a driver visibly tears out the steering wheel and throws it from the window, the rival can see that swerving is no longer possible and, being rational, yields. The move works only because it is both observable and irreversible: a commitment that can be quietly undone, or that the other side cannot verify, carries no weight. This is why credibility, not mere willingness, is the currency of brinkmanship. Many real tactics are ways of tying one's own hands. Reputation built over repeated encounters, sunk costs that make retreat expensive, delegation to an agent who is not authorised to compromise, and public promises that would be humiliating to break all serve to remove the option of backing down. The same logic exposes the danger. If both sides tear off their wheels, catastrophe becomes certain; the manoeuvre that wins when one side does it destroys both when both attempt it. Schelling framed brinkmanship not as certain resolve but as the deliberate manipulation of a shared risk that neither party fully controls, a contest in generating danger rather than in absolute courage.
What the evidence shows
The empirical anchor in the behavioural literature is the repeated-play experiment. Rapoport and Chammah (1966) had pairs play chicken many times in the laboratory and traced how the payoff parameters and the availability of communication shaped the rate of mutual swerving. They found behaviour sensitive to the size of the crash loss relative to the gain from winning: the steeper the catastrophe, the more players moved toward yielding. This work is descriptive, small by modern standards, and drawn from student subjects, so its quantitative rates should not be treated as fixed constants. The same payoff structure reappears in biology as the hawk-dove game. Maynard Smith and Price (1973) and Maynard Smith (1974) recast animal contests over a resource in these terms and introduced the evolutionarily stable strategy, showing that a population does not settle at all-aggression but at a stable mix of aggressive and yielding types, because universal aggression is self-defeating. That is a formal echo of the mixed equilibrium, and it has held up as a foundation of evolutionary game theory. The honest summary of the human evidence is narrower. Outcomes depend heavily on framing, on whether pre-play communication is allowed, and on whether the game is one-shot or repeated, and much of the literature rests on modest samples. The qualitative pattern is robust, in that mutual aggression is comparatively rare and commitment or signalling shifts who yields, but precise magnitudes travel poorly between settings.
Where it shows up
The model is invoked wherever two parties would both prefer to avoid a shared catastrophe yet each wants the other to concede. Nuclear standoffs are the canonical case, and much Cold War strategic thinking was built on exactly this structure, with the threat of mutual ruin doing the bargaining work. Labour disputes fit the pattern when a union and an employer each threaten a shutdown that neither truly wants; so do litigation-versus-settlement standoffs, legislative confrontations over deadlines, and market-entry races into a niche that can support only one survivor. Everyday traffic supplies the miniature version: a narrow bridge, an unmarked junction, or a lane merge where two drivers must sort out who goes first. In each case the analytic value of the model is the same. It locates the two asymmetric equilibria, explains why the mutually destructive outcome is unstable yet not impossible, and identifies commitment, reputation, and credible signalling as the tools that decide which side ends up yielding.
Limits and caveats
The clean model assumes both players know they are in a game of chicken, know the payoffs, and act on them. Real encounters rarely satisfy this. Payoffs are often private and uncertain, and a player who misreads the situation, believing the game is a prisoner's dilemma or that the opponent is irrational or suicidal, will behave very differently. The tearing-off-the-wheel result is a metaphor for irrevocable commitment, and genuinely irreversible commitment is both hard to achieve and dangerous to hold, because it forfeits the flexibility to de-escalate if the other side also commits. The static two-by-two game is also silent on which equilibrium is reached; without added structure such as a focal point, a visible asymmetry, or a first-mover advantage, the theory cannot predict who swerves. Continuous-time cousins such as the war of attrition, where the cost accrues gradually until one side quits, give richer and often more realistic predictions than the single-shot collision. Finally, the framing can be self-fulfilling: describing a negotiation as a game of chicken encourages the very brinkmanship and escalation the label names, so the model is best used as an analytic lens on incentives rather than as a script for action.
Examples
In a labor standoff, each side refuses to budge hoping the other caves first, knowing a strike that wrecks the firm hurts them both.
Two drivers approach a single-lane bridge from opposite ends; each would rather cross first, but if neither brakes they collide, so the driver who accelerates hard and looks away, visibly committing, pressures the other to yield.
A union and an employer in a contract standoff each brandish a strike or lockout; both prefer a settlement to a shutdown, yet each wants the other to concede terms, and the side that credibly signals it will not settle first gains leverage.
Two firms eye a market that can profitably support only one; each would rather the rival withdraw, but simultaneous full-scale entry burns both, so a public, sunk-cost commitment to stay can drive the competitor out.
Neighbouring powers in a border dispute mobilise forces near the line; mutual escalation risks a war neither wants, so each manoeuvres to appear immovable and to shift the burden of backing down onto the other.
First described in Game theory; analyzed by Schelling, Russell.
Key references
- Rapoport, A., & Chammah, A. M. (1966). The game of chicken. American Behavioral Scientist, 10(3), 10-28. doi.org/10.1177/000276426601000303
- Maynard Smith, J., & Price, G. R. (1973). The logic of animal conflict. Nature, 246(5427), 15-18. doi.org/10.1038/246015a0
- Maynard Smith, J. (1974). The theory of games and the evolution of animal conflicts. Journal of Theoretical Biology, 47(1), 209-221. doi.org/10.1016/0022-5193(74)90110-6
- Schelling, T. C. (1960). The strategy of conflict. Cambridge, MA: Harvard University Press.
- Kahn, H. (1965). On escalation: Metaphors and scenarios. New York: Praeger.
- Wang, Y., Lin, Y., Fu, C., Huang, Z., Xiao, S., & Yu, R. (2021). Effortless retaliation: The neural dynamics of interpersonal intentions in the Chicken Game using brain-computer interface. Social Cognitive and Affective Neuroscience, 16(11), 1138-1149. doi.org/10.1093/scan/nsab064