Researchers and policymakers often need to anticipate how changes in incentives, rules, or feasible actions reshape perceptions of socially appropriate behavior. However, existing approaches elicit such perceptions separately in each setting without developing a theory of how individuals form these judgments. This paper proposes a theory in which social appropriateness rankings emerge endogenously from universalization reasoning (i.e., evaluating an action by the payoff it would yield if everyone chose it). The resulting norm depends only on the game's payoffs, without requiring data on beliefs, preferences, or choices. I validate the framework by comparing its predictions against norms elicited in a pre-registered lab experiment and existing experimental evidence. The theory performs well in several settings, including the prisoner's dilemma, the stag hunt, and the dictator game with taking options. However, it departs systematically in settings where individuals approve of actions that benefit others even when those actions are socially inefficient. To account for these departures, I extend the framework to incorporate social preferences together with universalization reasoning, providing a tractable approach to studying norm heterogeneity. Together, the two frameworks provide a portable and tractable tool for predicting how perceptions of social appropriateness vary across environments.