Multiple schedules, off‐baseline reinforcement shifts, and resistance to extinction
Extinction in an alternate schedule component can soften future extinction resistance even when that component never meets the target behavior.
01Research in Context
What this study did
Craig et al. (2018) asked a simple question: can we weaken extinction resistance by changing reinforcement somewhere else? They used a two-part schedule. In one part, pigeons pecked for food. In the other part, the birds also pecked, but the payoff rules changed behind the scenes.
The twist happened off-baseline. Sometimes the second part stopped paying altogether. Other times it paid more. The birds never felt these shifts in the target part. The team then extinguished the target pecks and counted how many extra responses survived.
What they found
Only the off-baseline extinction cut later resistance. Birds gave up faster when the alternate component had gone cold beforehand. Boosting the alternate payoff had no effect. Reinforcement history in another corner of the cage mattered more than current payoff size.
How this fits with other research
Craig et al. (2017) showed bigger pellets speed extinction but also spike resurgence. The 2018 study moves the lever in the opposite direction: removing pellets off-baseline softens extinction resistance. Same lab, same birds, complementary story.
Shahan et al. (2020) later found longer alternative reinforcement slightly lowers resurgence. Craig et al. (2018) add that the quality of that history, not just its length, sets the stage. Together they say both duration and deprivation in alternate contexts shape relapse.
Capio et al. (2013) saw rich alternate rates suppress fastest yet rebound hardest. Craig et al. (2018) agree richness alone does nothing unless you later remove it. The papers line up: value helps only if you take it away, and removal away from baseline still counts.
Why it matters
When you run DRA across settings, extinction history in one setting can guard or invite resurgence in another. Before you fade supports, check whether other therapists or teachers have already gone cold on reinforcement. If they have, you may see a smoother extinction curve. If they have not, plan for extra bursts even if the current payoff feels lean.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Ask your team where reinforcement has already stopped for other behaviors; use that context as your first extinction session for the current target.
02At a glance
03Original abstract
Resistance to extinction in a target multiple-schedule component varies inversely with the rate of reinforcement arranged in an alternative component during baseline. The present experiment asked whether changing the reinforcer rate in an alternative component would impact extinction of target component responding if those changes occurred in an off-baseline phase during which the target component was never experienced. Pigeons' key pecking was studied in three types of conditions, and each condition consisted of three phases. In Phase 1, pecking produced food in the target and alternative components of a multiple schedule according to variable-interval 60-s schedules. In Phase 2, the alternative-component stimulus was presented alone in a single schedule. Pecking during this phase produced the same reinforcer rate as in baseline in the Control condition, a higher rate of food (variable-interval 15 s) in the High-Rate condition, or was extinguished in the Extinction condition. Extinction of target- and alternative-component key pecking then was assessed in a multiple schedule during the final phase of each condition. Resistance to extinction of target-component key pecking was the same between the Control and High-Rate conditions but lower in the Extinction condition. These findings are discussed in terms of discrimination and generalization processes.
Journal of the Experimental Analysis of Behavior, 2018 · doi:10.1002/jeab.300