Ultimatum Game

The ultimatum game is a two-player economic experiment that has become the paradigmatic test of whether human behavior conforms to the predictions of Nash equilibrium and subgame perfect equilibrium. One player, the proposer, is endowed with a sum of money and proposes a split to the second player, the responder. The responder either accepts the offer — in which case the money is divided as proposed — or rejects it, in which case both players receive nothing.

Designed by Werner Güth, Rolf Schmittberger, and Bernd Schwarze in 1982, the game is strategically trivial under classical assumptions. Because the responder choosing between a positive amount and zero should always accept, the proposer should offer the minimum possible amount and keep the rest. This is the unique subgame perfect equilibrium. It is also a Nash equilibrium. It is also what no human proposer does in practice.

The Empirical Pattern

Across thousands of replications in dozens of cultures, the pattern is robust: proposers offer between 40% and 50% of the endowment, and responders reject offers below 20-30% about half the time. The modal offer is an even split. When stakes are raised from laboratory pennies to months of income in developing countries, the pattern shifts slightly toward self-interest but does not disappear. When the game is played under double-blind anonymity — eliminating even the possibility of reputation — offers remain well above the equilibrium prediction.

These rejections are not errors. They are systematic, replicable, and culturally patterned. In small-scale societies with strong sharing norms, offers are higher and rejections of low offers more frequent. In market-integrated societies, offers are closer to 40% than 50%, but still far from zero. The data map a gradient between pure self-interest and strong reciprocity, and no society sits at the equilibrium endpoint.

Explanations and Models

The ultimatum game has spawned a cottage industry of explanatory models, each capturing a different aspect of the empirical regularity.

Fairness and inequity aversion. Models by Fehr and Schmidt (1999) and Bolton and Ockenfels (2000) assume that agents care not only about their own payoff but about the distribution of payoffs. Agents experience disutility from disadvantageous inequality (being worse off) and sometimes from advantageous inequality (being better off). These models predict both the proposer's generosity — to avoid rejection — and the responder's punishment — to reduce inequality at a cost to oneself.

Reciprocity and social norms. The responder's rejection is not merely a preference for fairness; it is a punitive act. By rejecting, the responder imposes a cost on the proposer, enforcing a norm against exploitative offers. This is strategic reciprocity in a one-shot game: the responder sacrifices material payoff to maintain the norm, and the proposer, anticipating this, offers more than the minimum.

Quantal response and level-k reasoning. Some researchers have attempted to explain the data without invoking social preferences at all. Quantal response equilibrium introduces noise into best responses; level-k models assume players reason only a finite number of steps ahead. These models predict some deviation from equilibrium but struggle to explain the frequency of rejections of substantial positive offers. They work better for the proposer's behavior than the responder's.

Cross-Cultural Variation and Evolutionary Interpretation

The most striking finding from cross-cultural studies is not universality but variation. In the Machiguenga of Peru, offers average around 26% and rejections are rare. Among the Lamalera of Indonesia, offers average above 50% and hyper-fair offers are common. The pattern correlates with the degree of market integration, cooperative production, and payoffs to cooperation in each society. Societies with stronger norms of cooperative resource sharing produce more equal offers; societies with less interdependence produce more self-interested behavior.

This has been interpreted as evidence for gene-culture coevolution: fairness norms evolved in contexts where cooperation yielded high returns, and these norms were transmitted culturally alongside genetic predispositions for social preference. The ultimatum game, in this reading, is not a test of rationality but a probe into the evolved psychology of cooperation — a window onto the behavioral games humans have been playing for millennia.

The ultimatum game is often treated as a puzzle to be solved — a deviation from rationality that requires a patch to the standard model. This is backward. The game does not reveal a flaw in human reasoning; it reveals that the standard model was never a theory of humans to begin with. The rejection of a 20% offer is not irrational. It is a declaration that the game being played is not the one the theorist wrote down. The proposer who offers 50% is not being generous; they are being strategically literate in a world where fairness is enforced by punishment. The ultimatum game does not challenge human rationality. It challenges the assumption that rationality can be defined without reference to the social norms that make cooperation possible.