An interesting study.

Want cooperation? Rewarding the helpful can be more effective than punishing wrongdoers, a new experiment in game theory suggests.

In the public goods game, players choose whether or not to contribute money to a common pot. The pot is multiplied and redistributed equally, regardless of who contributes and who doesn't. When people play a pure version of the game, the temptation to freeload – reap the rewards without contributing anything – often leads to rapidly disintegrating cooperation.

Previous research found that cooperation is promoted by allowing players to punish freeloaders: cooperative players would pay a small cost that enables them to inflict a loss on the offender. This approach was more effective than reward, at least in games where players switch partners every round.
More carrot, less stick

David Rand and his colleagues at Harvard University modified the public goods game to reflect what they argue is a more natural scenario: people play with the same group for many rounds, establishing reputations with each other.

Players could choose to reward or punish others at a small cost to themselves. Rand found that rewarding or punishing were equally likely to lead to cooperation and higher earnings, but when players had the option to either punish or reward, but chose to reward, they received higher absolute payoffs. "It becomes in one's self-interest to help the group," says Rand.

"It's sort of a 'you scratch the group's back and I'll scratch yours'," he says.