Persistence of discrete-trial-instrumental behaviour despite outcome devaluation

This research investigates whether reinforcer encoding affects performance of an instrumentally-conditioned response or if such responses are automatic habits insensitive to current reinforcer value (S-R theory). Outcome devaluation was adopted to test whether responding was controlled by a response-reinforcer association. Unlike recent studies reporting outcome devaluation effects, the present experiments employed a discrete trial rather than a free operant procedure. Rats were initially trained to perform two responses (left and right levers) to obtain two reinforcers (food pellets and sucrose). One reinforcer was then devalued by inducing gastric malaise to condition an aversion to the outcome. Following devaluation rats were tested in extinction. Results revealed that devaluation had no effect on subsequent instrumental performance, with rats responding on both levers during test, despite the clear acquisition of an aversion to the devalued outcome. In contrast, when the discrete trial procedure was modified to approximate a free operant procedure (i.e., short intertrial interval and long lever insertion time), animals showed a reduction in responding for the devalued outcome following both a motivational shift and specific satiety. These results suggest that discrete trial instrumental behaviour is less sensitive to reinforcer value than free operant behaviour and thus is more like stimulus-bound habitual responding.