In an experiment, two strangers are given the opportunity to share $100, subject to the following constraints: One person—the "proposer"—is to suggest how to divide the money and can make only one such proposal. The other person—the "responder"— must either accept or reject the offer without qualification. Both parties know that if the offer is accepted, the money will be split as agreed, but if the offer is rejected, neither will receive anything.

This scenario is called the Ultimatum Game. Researchers have conducted it numerous times with a wide variety of volunteers. Many participants in the role of the proposer seem instinctively to feel that they should offer 50 percent to the responder, because such a division is "fair" and therefore likely to be accepted. Two-thirds of proposers offer responders between 40 and 50 percent. Only 4 in 100 offer less than 20 percent. Offering such a small amount is quite risky; most responders reject such offers. This is a puzzle: Why would anyone reject an offer as too small? Responders who reject an offer receive nothing, so if one assumes—as theoretical economics traditionally has—that people make economic decisions primarily out of rational self-interest, one would expect that an individual would accept any offer.

Some theorists explain the insistence on fair divisions in the Ultimatum Game by citing our prehistoric ancestors' need for the support of a strong group. Small groups of hunter-gatherers depended for survival on their members' strengths. It is counterproductive to outcompete rivals within one's group to the point where one can no longer depend on them in contests with other groups. But this hypothesis at best explains why proposers offer large amounts, not why responders reject low offers.

A more compelling explanation is that our emotional apparatus has been shaped by millions of years of living in small groups, where it is hard to keep secrets. Our emotions are therefore not finely tuned to one-time, strictly anonymous interactions. In real life we expect our friends and neighbors to notice our decisions. If people know that someone is content with a small share, they are likely to make that person low offers. But if someone is known to angrily reject low offers, others have an incentive to make that person high offers. Consequently, evolution should have favored angry responses to low offers; if one regularly receives fair offers when food is divided, one is more likely to survive. Because one-shot interactions were rare during human evolution, our emotions do not discriminate between one-shot and repeated interactions. Therefore, we respond emotionally to low offers in the Ultimatum Game because we instinctively feel the need to reject dismal offers in order to keep our self-esteem. This self-esteem helps us to acquire a reputation that is beneficial in future encounters.