Formal and computational models of Reinforcement Sensitivity Theory

Abstract

Jeffrey Gray (Gray 1970, 1981, 1987) developed a ‘bottom-up’ approach to personality which began with the identification of large-scale brain-behavioural systems in animals (based on lesion and psychopharmacological evidence). The systems which he emphasized were: a reward system (later termed the Behavioural Activation System or BAS); a punishment system; an arousal system; and a Fight-Flight System. His view of personality was that inter-individual differences in the functioning of each discrete system should give rise, in human beings, to a major dimension (or trait) of personality. The theory also talked particularly about the sensitivity (i.e., reactivity) of each system to its characteristic inputs (i.e., the inputs which activated that system rather than any of the others). Inter-individual differences in the sensitivities of the systems were taken to be the causal basis of human personality traits. As the names ‘reward system’ and ‘punishment system’ implied, the characteristic inputs of the systems were reinforcing stimuli. To capture these aspects of the theory, I suggested the name Reinforcement Sensitivity Theory (RST) for the personality theory which he developed. Basic tenets of the original RST In the original version of the theory (which I shall call ‘old RST’ in this chapter), the personality trait corresponding to variations in functioning of the BAS was suggested to be impulsivity. The punishment system (so-called because it was thought that this system responded to conditioned stimuli signalling impending punishment) was referred to as the Behavioural Inhibition System (BIS): the personality trait corresponding to variations in functioning of the BIS was suggested to be anxiety.