Fine behavior. The prospect of punishment can lead to cooperation.

Digital Vision/Getty Images

The Fear Factor

Nobody likes to pay fines for speeding or littering, but does the punishment really promote good behavior? The answer appears to be yes, according to research published today in Science. Even just the long-term prospect of a penalty seems to motivate people to cooperate for the common good.

Cooperation for the public welfare is one of the characteristics of human societies, yet it's hard to see how altruistic behavior evolves from individuals looking out for their own interests. Altruistic behavior can have a high individual price, after all--the chance of death in war or costly carbon taxes, for example--and that tempts people to sneak a free ride. Societies have developed a wide variety of threats, from physical punishment to fines, to encourage cooperation.

But previous experiments, in which people played a game where they were fined if they didn't work together, suggested that punishment is very costly (ScienceNOW, 6 April 2006). Simon Gächter of the University of Nottingham, U.K., lead author of the new study, suspected that "the time horizon in these experiments was just too short to allow the beneficial effects of punishment to show up."

To test that hypothesis, Gächter and his team conducted similar experiments involving 207 participants divided into 69 groups of three. The groups played a game in which each participant was initially given 20 tokens and asked to decide how many to invest in a common project. In each round, the three group members earned 0.5 money units for each token contributed to the project, regardless of the individual contribution. So keeping all tokens is individually beneficial. But if every group member contributes generously, the group will be collectively better off. For half the groups playing this game, participants were given the opportunity at the end of each round to "punish" uncooperative group members, by deducting tokens from their stash. The other groups couldn't punish anyone.

Gächter's team had the groups play this cooperation game for either 10 or 50 rounds, lasting up to 90 minutes. The shorter games with punishment opportunities generated less income for the participants than the nonpunishment runs--just as in the earlier experiments. But when the games lasted 50 rounds, the groups that could punish noncooperating members generated higher average round contributions and higher net earnings for the participants than nonpunishment groups. The researchers conclude that punishment encourages cooperation over the long term.

Somewhat surprisingly, the 50-round experiments also saw participants punish each other less overall, in terms of tokens, than those in the much shorter game. "The upshot is that punishment is beneficial if it works as a mere threat that has not to be used very often," says Gächter.

The present study shows that Gächter's previous conclusion that punishment was detrimental "was an experimental artifact generated by a relatively small number of interactions," says anthropologist Rob Boyd of the University of California, Los Angeles, who studies the mechanisms that shape human culture. "It takes a while for the punishing individuals to teach the 'would be' cheaters to cooperate," he says. During these initial stages, punishment costs are high, but once the cheaters learn to cooperate, the mere threat of punishment is apparently enough to keep them in line.