Dual-system free-operant avoidance: extension of a theory

Our theory of positively reinforced free-operant behavior (Perez & Dickinson, 2020) assumes that responding is controlled by 2 systems. One system is sensitive to the correlation between response and reinforcement rates and controls goal-directed behavior, whereas a habitual system learns by reward prediction error. We present an extension of this theory to the aversive domain that explains why free-operant avoidance responding increases with both the experienced rate of negative reinforcement and the difference between this rate and that programmed by the avoidance schedule. The theory also assumes that the habitual component is reinforced by the acquisition of aversive inhibitory properties by the feedback stimuli generated by responding, which then act as safety signals that reinforce habit performance. Our analysis suggests that the distinction between habitual and goal-directed control of rewarded behavior can also be applied to the aversive domain.