Evaluation of resurgence following differential reinforcement of alternative behavior with and without extinction in a human operant model

DeWitt et al. (2024) asked: does resurgence still happen if you do DRA without extinction? They used a lab setup with neurotypical adults. Each person first earned points for pressing one button. Then that button stopped paying off and a new button paid instead. Finally, neither button paid. The team compared two groups: one had total extinction for the first button, the other still gave an occasional point for it.

Both groups showed resurgence. People went back to pressing the first button when reinforcement stopped. The size of the bounce was slightly bigger in the no-extinction group, but the numbers were not statistically different. Bottom line: skipping extinction does not protect you from resurgence.

The finding lines up with Wunderlich et al. (2017). They saw that any weakening of extinction (50% response blocking) let problem behavior return. Both papers show that halfway or no-extinction versions still leave the door open for relapse.

Gomes-Ng et al. (2023) used the same lab and adult volunteers to test differential versus nondifferential reinforcement. Like DeWitt, they found that letting ‘wrong’ responses earn rewards changes later behavior. Together these studies warn: if alternative responses still get fed, old ones stay alive.

Older pigeon work (A et al. 2001; J et al. 1981) shows signaled versus unsignaled delays shift preference. The theme matches here—what happens around reinforcement (extinction or not) alters future responding, even when the procedures look different on the surface.

You may avoid extinction in the clinic for safety or consent reasons. This study says plan for resurgence anyway. Build extra practice of the alternative response, add signals for when reinforcement is unavailable, and track data after you thin the schedule. Expect the old behavior to pop up and have a ready response.