The Extinction Burst: Effects of Alternative Reinforcement Rate

Discontinuing reinforcement for an operant response can sometimes produce a transient increase in the rate of that response (i.e., an extinction burst). Although differential reinforcement of alternative behavior (DRA) can attenuate the extinction burst, the effects of rate of alternative reinforcement on the extinction burst have never been examined. Using a within-session transition from a variable-interval (VI) 1.5-s schedule to extinction, we examined how arranging different rates of reinforcement for an alternative response during extinction affected target and alternative responding of rats. Consistent with prior findings, reinforcing an alternative behavior eliminated the extinction burst. Surprisingly, all arranged alternative reinforcement rates eliminated the extinction burst and similarly reduced target response rates, despite generating differential rates of alternative behavior. This outcome is contrary to predictions of the temporally weighted matching law, which predicts both graded suppression of target behavior and higher alternative response rates with higher alternative reinforcement rates. Early in extinction, alternative responding was elevated and initially nondifferentiated across groups, suggesting that rats required some time to discriminate the rate of alternative reinforcement. Incorporating a temporary generalization-of-value across target and alternative responses substantially improved model fits. These findings show that arranging even lean DRA schedules might prevent the extinction burst and suggest potential new avenues for studying response generalization during transitions to extinction.