The Fehr–Gächter public-goods experiment examined why people contribute to shared benefits when they could do better individually by withholding their resources. Its central finding was that cooperation weakened when participants saw others free ride, but could be restored when the rules made free riding visible and allowed it to be punished. The experiment presents cooperation not as automatic generosity, but as a conditional practice shaped by what people observe and by whether they believe unfairness will have consequences.

The game and its incentives

The basic experiment placed four people in a group for four rounds. Each participant began with twenty tokens and could either keep them or contribute them to a public pot. A contribution cost the person who made it one token, but gave 0.4 tokens to each member of the group. Thus, an individual who contributed bore a private cost while the group as a whole gained 1.6 tokens. If everyone kept all their tokens, each ended with twenty; if everyone contributed everything, each ended with thirty-two.

The structure separated individual advantage from collective advantage. A participant could benefit from other people’s contributions while keeping their own tokens, making free riding the individually attractive strategy in the ordinary economic account. But if all four chose that strategy, the public pot received nothing. The game therefore made the cooperation problem concrete: the group could achieve a better shared result, yet each member had an incentive to rely on others to pay its cost.

Cooperation erodes under free riding

Across similar public-goods experiments in the developed world, most participants did not begin by acting selfishly. They generally contributed about half their tokens to the common pot. The pattern changed over successive rounds, however, as players saw others free ride: contributions declined, and by the end, 70 to 80 percent of players were free riding. The group consequently ended much poorer than it would have if everyone had continued contributing.

Fehr and Gächter described participants in three broad categories. Roughly a quarter were consistently selfish, always choosing to free ride. A small minority were altruists, contributing heavily from the outset and continuing even when others withheld their tokens. The largest group consisted of conditional consenters: they began by contributing at least some of their wealth but became less willing to do so as they observed free riding. By the final rounds, almost all of these conditional contributors had stopped cooperating. Their behavior made the experiment’s central vulnerability plain: cooperation could unravel not only because some people never contributed, but because the conduct of free riders drove the larger, initially cooperative group away from contribution.

Visibility, punishment, and renewed contributions

Fehr and Gächter changed the game by revealing, at the end of each round, what every participant had contributed. This made free riders visible to the rest of the group. Participants could then punish them: for the cost of a third of a token, a person could take one token away from a free rider. Punishment was costly to the person imposing it and did not fit the narrow calculation of maximizing that individual’s token total.

Participants nevertheless spent money to punish free riders. The effect extended beyond the immediate penalty: the free riders began contributing their fair share. Moreover, participants continued to contribute in the final rounds, even though punishment could no longer be inflicted then. The outcome suggests that punishment did more than alter the material payoff of an individual decision. Its availability helped establish that free riding would be exposed and answered, and cooperation persisted after the threat of further punishment had been removed.

Conditional consent and the durability of cooperation

The experiment’s interpretation turns on the conditional consenters. Consistently selfish players were not the majority, and persistent altruists were only a small minority; the prospects for the group depended most on people whose willingness to contribute was responsive to others’ behavior. When free riding went unchecked, their contributions fell away. When participants could identify and punish free riders, cooperation became sustainable enough to continue even in the last rounds.

This is why the experiment matters beyond the token game. It offers a model for the fragility of collective provision: people may be willing to bear individual costs for a shared benefit, but that willingness depends partly on whether they believe others are doing their part. In the book’s discussion of taxation, the same logic connects voluntary compliance to trust in neighbors and confidence that the state will find and punish those who evade their obligations without punishing the innocent. The experiment does not make punishment costless or cooperation inevitable; it shows how accountability can keep conditional contributors from feeling that they alone are paying for a common good.