Abstract

I argue that a specific version of Theory theory is necessary and sufficient for attributions and predictions of others' emotions. Theory theory is the view that we attribute and predict others' mental states on the basis of a body of generalisations about mental states, their situational input, and behavioural output. Theory's antagonist, Simulation theory, is the view that we ascribe mental states to others by simulating - or running ' off-line ' - their doxastic, emotional, and contextual situations. My argument for Theory's necessity and sufficiency develops in three stages: First, I show that some version of Theory is necessary for predictions of all mental states on the basis of the ascriber's knowledge of the subject's other mental states. The linchpin of the arguments here consists of considerations from relevant similarity between the ascriber's and the subject's mental states. Simulation cannot provide criteria for such similarity, and so, I argue, predictions must advert to Theory. Second, I develop a sui generis model of emotions, according to which emoticns' necessary objects and typical causes are concern-based construals; and emotions qua attitudes are complex states embedded in a narrative structure, characterised in terms of their object, their expressive behaviour, and their phenomenology. Third, I show that, considering the nature of the objects of emotions, some Theory is necessary for emotion-predictions and -attributions. Moreover, I develop a version of Theory, based on my analysis of emotions and narrative structures, and argue that this version of Theory is both necessary and sufficient for emotion-predictions and -attributions