We propose a new modeling framework to study affective decision-making, which produces many notorious ”irrationalities” in human behavior. Building on biologically-inspired mod- els of reinforcement learning, we provide a description of a boundedly-rational affective agent who holds mood-dependent beliefs and exhibits prospect-theory-like behaviors in situations with uncertainty. By construction, affective agents possess personal and social iden- tities and desire to achieve higher social status, which allows them to cooperate with other like-minded agents. We show how moral sentiments, indicative of adherence to identity- based norms, emerge to strengthen cooperation even further. The model of affective decision- making and moral sentiments complements our previous work on moral reasoning among rational agents (“A Theory of Injunctive Norms”). Together, the two models constitute an attempt at a comprehensive framework to study norm-driven human behavior.