Abstract:
Moral character judgments inform highly consequential decisions about whom to trust, reward, punish, praise, blame, and forgive. We posit that humans represent and predict ethical choices by determining how much the observed person values themselves (S), values others (O), and cares about outcome disparities (D). Observers are modeled as revising their preconceptions about these parameters after watching the social choices of other people, which they understand in terms of the three corresponding incentives within each situation. A Morality Game was developed, where across a series of trials each participant/observer predicts peoples’ next choices before and after watching their previous choices in game theoretic dilemmas with systematically varied payoff structures that include helpful, selfish, win-win, and malicious options. We find that participants represent their beliefs about other peoples' moral decision-making based on relative weightings of the above three independent dimensions and they revise these beliefs with evidence.