WikiSummary, the Social Science Summary Database

Fehr, Fischbacher, and Kosfeld: Neuroeconomic foundations of trust and social preferences

From WikiSummary, the Free Social Science Summary Database


Fehr, Fischbacher, and Kosfeld. Neuroeconomic foundations of trust and social preferences. American Economic Review.

In Brief

A review article. We know that the prisoner's dilemma does not accurately predict the way people behave in experiments. Why? Using PET scanning, the authors seek to shed light on this.

First, they expect to find that people enjoy punishing unfair behavior (i.e. people who defect against a cooperating player). They use three experimental conditions in which punishment is possible. In the first condition, punishment is costly (it costs both the cooperator and the defector); in the second, punishment is free (but costs the bad guy money); in the third/control, punishment is only symbolic (i.e. it incurs no financial costs). By comparing the costly and free cases to the symbolic case, they see that the reward areas of the brain are activated when we get to give an actual punishment.

Second, they expect to find that people enjoy cooperating for its own sake. They test this by using PET scanning when subjects play a prisoner's dilemma against either a real person or a computer. The monetary payoffs are the same in either case, so we can cancel out those areas of the brain. But in the human case, you might also get an additional payoff from the act of mutual cooperation. Turns out, you do; the reward centers of the brain are activated by mutual cooperation. There is an additional payoff from cooperating with a human.


They discuss recent neuroeconomic evidence that people derive non pecuniary utility from both mutual cooperation and from punishing unfair behavior in these games. Thus the mutual cooperation that takes place despite the free riding incentives and the punishment of people who do free ride is not irrational, but better understood as a rational response of people with corresponding social preferences.

The authors examine the effect of Neuropeptide Oxytocin (OT) on trusting and trustworthy behavior. OT is shown to elicit more trusting behavior when administered to subjects even when their beliefs do not change. These results in concert show not only that people exhibit prosocial behavior but that other people believe they do and trust them.


Subjects play a trust game where they send money to another player and hope the other player sends money back to them. The first player can keep his money, or he can send the money to the second player; the second player can keep this money, or transfer it back to the first. Any money that is sent back to the first player is automatically doubled. Thus, both players can be better off if both transfers occur than if neither transfer occurs, but the first player must trust the second player not to keep all the money.

The experimental conditon: The first player is given the option to punish the second player for failure to transfer money back.


Brain scan imaging shows that punishing untrustworthy people activates the reward response areas of the brain, suggesting that people get pleasure out of punishing untrustworthy people. Evidence further shows that stronger stimulation comes from punishing economically rather than randomly. That is, only when you are wronged instead of blindly punishing everyone.

Evidence also shows more stimulation of the reward part of the brain is produced through cooperative gains. That is, the authors scan people receiving equal monetary rewards, but one receives the reward as a result of cooperation with another human and the other just gets it (or gets it by playing a computer). There is additional activity in the brain's reward areas when the money comes from cooperation.

Adjusting OT levels in players also affects their ability to trust. Trust is variable, which means economists' assumptions about fixed preferences is wrong.