This paper, authored by Gary Charness and Matthew Rabin, explores the nature of social preferences in economic experiments. The authors design a series of simple games to test existing theories of social preferences more directly than previous studies. They find that subjects are more concerned with increasing social welfare—sacrificing to increase payoffs for all recipients, especially those with lower payoffs—than with reducing differences in payoffs. Subjects also exhibit reciprocity, withdrawing willingness to sacrifice when others are unwilling to sacrifice and sometimes punishing unfair behavior. The paper critiques the confounds in existing experimental games that have led to the overinterpretation of difference-aversion models. These models suggest that Pareto-damaging behavior (rejections in ultimatum games) is driven by a preference to minimize differences in payoffs. However, the authors argue that this explanation is confounded by the fact that such behavior only arises when a clear motivation for retaliation is aroused, and only when it involves reducing inequality. The experiments conducted by Charness and Rabin offer a broader range of choices to test different social motivations, eliminating these confounds. The authors develop a simple linear, two-person model of preferences that captures various types of social preferences, including difference aversion, social welfare, and reciprocity. They analyze the experimental data using this model and find that social-welfare preferences better explain the observed behavior than difference aversion. The data show that subjects are less likely to cause Pareto damage when it increases inequality than when it decreases it, suggesting that inequality reduction is not a strong explanation for Pareto-damaging behavior. The paper also examines the role of reciprocity through response games, where Player B's choice follows Player A's decision to forego an outside option. The results replicate recent evidence that positive reciprocity is not a strong force in experimental settings. However, subjects exhibit a form of concern withdrawal, where they withdraw their willingness to sacrifice when others are unwilling to sacrifice for fairness. The data also show that subjects significantly increase Pareto-damaging behavior following selfish actions by Player A. Overall, the findings suggest that social-welfare preferences explain the data better than difference aversion, and that subjects behave reciprocally. The authors conclude by discussing the implications of their results for the development of more accurate models of social preferences and suggest new directions for research.This paper, authored by Gary Charness and Matthew Rabin, explores the nature of social preferences in economic experiments. The authors design a series of simple games to test existing theories of social preferences more directly than previous studies. They find that subjects are more concerned with increasing social welfare—sacrificing to increase payoffs for all recipients, especially those with lower payoffs—than with reducing differences in payoffs. Subjects also exhibit reciprocity, withdrawing willingness to sacrifice when others are unwilling to sacrifice and sometimes punishing unfair behavior. The paper critiques the confounds in existing experimental games that have led to the overinterpretation of difference-aversion models. These models suggest that Pareto-damaging behavior (rejections in ultimatum games) is driven by a preference to minimize differences in payoffs. However, the authors argue that this explanation is confounded by the fact that such behavior only arises when a clear motivation for retaliation is aroused, and only when it involves reducing inequality. The experiments conducted by Charness and Rabin offer a broader range of choices to test different social motivations, eliminating these confounds. The authors develop a simple linear, two-person model of preferences that captures various types of social preferences, including difference aversion, social welfare, and reciprocity. They analyze the experimental data using this model and find that social-welfare preferences better explain the observed behavior than difference aversion. The data show that subjects are less likely to cause Pareto damage when it increases inequality than when it decreases it, suggesting that inequality reduction is not a strong explanation for Pareto-damaging behavior. The paper also examines the role of reciprocity through response games, where Player B's choice follows Player A's decision to forego an outside option. The results replicate recent evidence that positive reciprocity is not a strong force in experimental settings. However, subjects exhibit a form of concern withdrawal, where they withdraw their willingness to sacrifice when others are unwilling to sacrifice for fairness. The data also show that subjects significantly increase Pareto-damaging behavior following selfish actions by Player A. Overall, the findings suggest that social-welfare preferences explain the data better than difference aversion, and that subjects behave reciprocally. The authors conclude by discussing the implications of their results for the development of more accurate models of social preferences and suggest new directions for research.