A computational model of affect theory: simulations of reducer/augmenter learned helplessness phenomena

摘要

This paper outlines the initial results of an ongoing research effort to build and test a computational model that simulates key components of a theory of mind proposed by psychologist Silvan S. Tomkins in his four-volume series Affect, Imagery, Consciousness (1962, 1963, 1991, 1992). Tomkins' model of the mind is able to account for an impressive range of motivational phenomena, including the crucial developmental link between cognition and emotion, how conflicting motives are resolved and decisions about moment-to-moment behavior are made, and how individuals can be motivated simultaneously by highly idiosyncratic goals that emerge from experience, as well as the complete array of innate, instinctual drives. To begin testing the explanatory power of Tomkin's model, key components of his theory have been implemented in computational terms and tested in experimental simulations. In particular, Tomkins' affect theory, a sub-theory of his overall view of the mind, has been targeted for primary research purposes. Affect theory has been selected as the focus of study because it is the most unique component of Tomkins' model, and many authors, e.g., Nathanson (1992) and Schore (1994), consider it to be his primary contribution to the understanding of the mind. Affect theory also has the potential to explain how "primary" emotions, as described by Damasio (1994), are generated and can evolve into cognitively generated "secondary" emotions over time. Reducer/augmenter (Larsen & Zarate, 1991) and learned helplessness phenomena (Peterson, Maier, & Seligman, 1993) were selected for simulation purposes. These affective/behavioral phenomena have been repeatedly documented to occur in studies involving living subjects. Moreover, these phenomena are at the opposite ends of the nature/nurture continuum. For reducer/augmenter phenomena, differences in affectively motivated behavior are attributed to innate, constitutional factors. In contrast, learned helplessness effects are attributed to differences in environmental experience. The simulation of these causally diverse affective/behavioral tendencies, therefore, should provide a useful initial test of the range of phenomena for which Tomkins' theory can account.