How Much Money Would You Trust to a Stranger

The players don't see each other and play only once. The second player is under no obligation: she can keep everything, with no consequences. By a strict calculation of self-interest, the first player shouldn't give anything at all.

In practice, things turn out differently. On average, the first player gives about half, and the second returns roughly a third of what she received. This pattern appears in dozens of countries, although the shares vary noticeably.

The third bar is exactly why the game was invented. It shows the gap between the model of a person acting only for her own benefit and how people actually behave, using real money.

The strongest factor is the chance to see your partner or learn anything about her. Even a photograph or one small shared detail noticeably increases the amount passed on, although it has no effect on the payoff.

The second factor is the country and its social environment. In places where general trust is higher in surveys, people pass more in the game. The link is also noticeable between regions within a single country.

A separate story involves attempts to influence trust by inhaling oxytocin. Early studies produced a striking result that spread through popular articles, but large replications did not confirm it, and this line of research is now treated cautiously.

Another finding concerns expectations. The amount returned is best predicted not by the second player's character, but by how much she was given: generosity is perceived as a sign of trust, and people respond in kind.

There's also a version of the game without tripling, where giving produces no gain. People pass noticeably less there, and that's an important detail: trust in experiments rests not only on the hope of getting more back, but also on the possibility of mutual gain itself.