Can we neutralize social preference in experimental games?

We propose an experimental method whose purpose is to remove social concerns in games. The core idea is to adapt the binary-lottery incentive scheme, so that an individual payoff is a probability to see one's preferred social allocation implemented. For a large class of social preference models, the method induces payoffs in the game that are in line with subjects' (social) preferences. We test the method in several popular experimental games, contrasting behaviors with and without our methodology. Our results suggest that a substantial part of the difference between predictions based on selfishness and observed behaviors seems driven by such preferences, since our method does induce more “selfish” behaviors. But they also indicate that a considerable share is left unexplained, perhaps giving weight to alternative explanations or other types of social concerns.