Over the past decade, experiments devised by Ernst Fehr of the University of Zurich in Switzerland, among others, have shown that many people will cooperate with others even when it is absolutely clear they have nothing to gain. A capacity for true altruism seems to be a part of human nature. It is a heartening discovery, yet one that has also touched off a firestorm of debate.

The experiments at the centre of the controversy are as simple as they are illuminating. They ignore theory-based preconceptions about how individuals ought to behave and focus instead on finding out what they actually do when playing games in which there is real money at stake.

One of the most basic of these games is the "ultimatum game". An experimenter gives one of two players some cash, say $20, and asks that person, called the "proposer", to offer a fraction of it to the second player, called the "receiver", whose identity is hidden from the other player. The proposer can offer any amount they choose, from nothing up to the entire $20. The receiver then has the choice of accepting or rejecting the offer. If he or she accepts, the cash is shared according to the original offer. A rejection means that no one gets anything. The game is played just once.

For the receiver, self-interest would seem to dictate accepting the offer no matter how small it is, since getting something is better than getting nothing. Knowing this, a similarly self-interested proposer should offer as little as possible. But over the past decade or so, research on student volunteers has shown that proposers in such experiments typically offer anything from 25 to 50 per cent, while receivers tend to reject offers of less than 25 per cent.

"People reject low offers," says anthropologist Joseph Henrich of Emory University in Atlanta, Georgia, "because they view them as unfair." And through their rejection, they show a willingness to punish the unfair offers even at a cost to themselves.

A vast number of other experiments illustrate the same point. Last year, for example, Fehr and his colleagues had students play a version of the famous prisoner's dilemma game, in which two people can prosper through cooperation but are also given strong incentives to cheat on one another. In this game, if the participants cooperate, each receives a worthwhile monetary pay-off. But either player can get an even higher pay-off by cheating while their opponent cooperates (see Diagram).

In this particular version of the game, the researchers got people to play sequentially: one would go and then the other, fully aware of what the first had done. In theory, anyone thinking only of their own personal gain would always cheat, as this pays more than cooperating. But in the experiments, although many of players who went first did cheat, others cooperated, despite knowing that the second player could sucker them by cheating. What's more, roughly half those who went second rewarded cooperation by treating their opponent fairly, even though that meant forgoing an easy pay-off for themselves (Human Nature, vol 13, p 1). "The facts are clear," Fehr says. "Many people are willing to cooperate and to punish those who don't, even when no gain is possible."
"Many people will cooperate, and punish those who don't, even when no gain is possible"

This tendency - which researchers call "strong reciprocity" - throws into question the assumption that apparently selfless behaviour must have some selfish explanation. Across disciplines, researchers now agree that people often act against their own self-interest. "This is the most important empirical work on the human sense of justice in many years," says evolutionary biologist Robert Trivers of Rutgers University in New Jersey.