Testosterone does not have a good reputation where social skills are concerned. Many associate high levels of the hormone with antisocial, egotistical and even aggressive behaviours. “We believed it ourselves,” admits Ernst Fehr, an economist at the University of Zurich in Switzerland. But when he set out to test the effect of testosterone on human social behaviour, he was surprised to find that he had been too quick to judge.

Fehr is interested in the role of biological substances on social interactions. In 2005, he reported that giving people the hormone oxytocin in a nasal spray makes them more likely to trust others (Nature 435, 673–676; 2005). “Most of our colleagues had predicted we would not be able to find anything because they thought a single biological substance could not have an effect on such complex behaviour,” says Fehr.

Encouraged by the oxytocin results, Fehr and his colleagues decided to tackle testosterone because studies in rodents had clearly shown that the hormone promotes aggression. They tested the effects of testosterone on social interactions using a game, played between two people, called the ultimatum game. The first player has ten banknotes and has to offer the other player five, three, two or none of those banknotes. The second player can accept or reject the offer. If the offer is accepted, the two split the money according to their agreement. If it is rejected, both players receive nothing.

The researchers then asked how testosterone would affect the bargaining choices made by the players. For example, would the hormone lead to more offers being fair — at five banknotes — or unfair, at three, two or none? Fehr and his co-workers predicted that the latter would be more likely. But they also e-mailed several researchers in the field to ask their views and two replied that “people with high levels of testosterone want to be leaders, so would want to induce cooperation because this would legitimize their leadership”, Fehr recalls. This alternative view predicted that the hormone would increase the number of fair offers.

Given that there was already evidence that a single, low dose of testosterone has cognitive and biological effects in women, Fehr and his colleagues put the predictions about the hormone to the test in a group of 121 women. Half the women received a placebo, the other half a testosterone pill. Four hours later, they got to play the game. The results, described on page 356, show that women who received the testosterone bargained more fairly.

And this wasn't the only unexpected result. After playing the game, participants were asked what kind of pill they thought they had been given. Women who believed they had received testosterone — regardless of whether they actually had — had behaved more unfairly than women who thought they had been given the placebo. Thus, ironically, the negative perception of testosterone as a substance linked to aggression may have affected women's behaviour in a negative way, even though the hormone itself had a positive effect on social interactions.

Will the findings be enough to clear testosterone's reputation? Fehr is uncertain. “I think it will take a long time to change the view of testosterone, because it is so widespread.”