Social decision-making is increasingly studied with neurocomputational modeling. Here we discuss how this approach allows researchers to better understand and predict behavior in social settings. Using examples from the study of resource distributions and social learning, we illustrate how this methodology provides a flexible way to quantify social values and beliefs, identify specific motives and cognitive processes underlying social choice and learning, and arbitrate between competing theories of social behavior. We also critically discuss open questions and potential problems associated with this methodology.