I shall outline some of the functions emotion has been supposed to serve in non-biological theories of narrative, examine competing hypotheses among biologically based theories, and propose a model that integrates elements from both. I shall argue that one main purpose of narrative is to engage the reader in an emotional process. Narrative stimulates a distinct train or sequence of emotions and brings this sequence to a distinct conclusion. As a subsidiary hypothesis, I shall argue that the production and reception of narrative is a positive social process. Among other functions, it serves as a form of emotional training and social bonding. The hypotheses in this model will be presented as alternatives to hypotheses implicit or explicit in the following models of narrative: (1) models that presuppose no emotional function but merely assume an autonomous drive toward representation (many realist models, including some sociobiological applications); (2) models that assume fantasy projection as an end in itself (some Freudian and sociobiological applications); (3) models that emphasize sexual display or other forms of status seeking (some sociobiological applications); (4) models that give exclusive attention to social manipulation (some Marxist and sociobiological applications); (5) models that assume the existence of emotional states but propose no specific psychological functions for these states (Frye's system of genres); and (6) postmodern models that postulate states of cognitive and emotional suspension parallel to the deconstructive conditions of linguistic indeterminacy (for example, Peter Brooks's).