When does perspective-taking predict coalition success? The interplay of propensity, accuracy, and social predictability

Theories of coalition formation assume that coalition outcomes are determined by static features of the game (e.g., players' resources), because these features are thought to elicit universal allocation norms that allow players to anticipate each other's demands. Yet the same structural features can carry different social meanings depending on what players want, expect, and consider fair. Perspective-taking helps players navigate this social information, but its importance may depend on how predictable the social context is. We examined whether perspective-taking predicts coalition success more strongly when allocation norms are difficult to anticipate, than when norms are clear. We tested this hypothesis in two experiments using a real-time three-person coalition paradigm, in which we manipulated the predictability of allocation norms ( High-Predictable vs. Low-Predictable ). In Experiment 1 ( N = 851), perspective-taking predicted greater inclusion and higher payoffs only in the Low-Predictable condition. In Experiment 2 ( N = 1206, preregistered), we replicated the main effect of perspective-taking across conditions, but not its moderation by norm predictability. Exploratory analyses distinguishing the propensity to take perspective from the accuracy of doing so found that in the Low-Predictable condition propensity translated into coalition success only when paired with accurate inference of partners' norms; in the High-Predictable condition, propensity predicted success independently of accuracy. These findings showed that socio-cognitive processes play a relevant role in coalition formation, and that accuracy adds value beyond propensity only when allocation norms are less predictable.