Recency, repeatability, and reinforcer retrenchment: an experimental analysis of resurgence.
Old problem behavior can resurge even when the replacement skill still earns occasional reinforcement, so keep monitoring after you thin the schedule.
01Research in Context
What this study did
The team ran a lab test of resurgence. First, they paid people for one button press. Then they stopped that pay and paid for a new button instead. Finally they stopped all pay and watched the old button press come back.
They asked three questions. Does it matter how long ago the first button was paid? Will the old response return again in later tests? What happens if we keep the new button on a very thin pay schedule instead of turning pay off completely?
What they found
Resurgence showed up every time. The size of the bounce-back stayed the same no matter how many tests they ran. Cutting the new button’s pay rate to almost zero still let the old response return, though the bounce was smaller.
Free snacks delivered for doing nothing wiped the resurgence out completely. The old button press stayed quiet when food arrived no matter what the person did.
How this fits with other research
Fujimaki et al. (2024) used the same lab set-up and added a twist. They let people eat the old snack until it lost value. When the snack was no longer wanted, resurgence shrank. Together the two studies show you can weaken relapse either by removing the snack value or by giving free snacks.
P (2019) ran the same test with typical adults. People with lower depression and higher anxiety showed bigger resurgence. The 2003 paper did not look at traits, so the later study tells us resurgence size can also rest on who the person is.
Bernheim et al. (1967) looked at rich-to-lean schedule drops decades earlier. Their birds also dipped in responding when pay suddenly got thin. The 2003 work turns that dip into a full return of an old response, updating the older contrast idea.
Why it matters
You now know resurgence can appear even when the new skill still earns a tiny bit of reinforcement. Do not assume thinning the schedule keeps the old problem away. Watch for bounce-back whenever you fade extra reinforcement. If you must block it, give brief free reinforcement or first devalue the old reinforcer. Plan extra probes after any lean schedule change.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add a short extinction probe after you thin a DRA schedule and record if the old response pops up.
02At a glance
03Original abstract
Four experiments were conducted with pigeons to assess the experimental conditions necessary for the occurrence of resurgence. The general procedure consisted of the following conditions: Condition 1--reinforcement of key pecking; Condition 2--reinforcement of treadle pressing and concurrent extinction of key pecking; and Condition 3--the resurgence condition wherein resurgence was defined as the recovery of key pecking. In Experiments 1 and 2, the resurgence condition was conventional extinction. The effect of recency on resurgence magnitude was examined in Experiment 1 by manipulating the number of sessions of Condition 2, above. Resurgence was not a function of recency with the parameters used. Repeating the three conditions revealed resurgence to be a repeatable effect in Experiment 2. In Experiment 3, a variable-time schedule was in effect for the resurgence condition. Resurgence was not produced by response-independent food delivery. In Experiment 4, the resurgence condition was a variable-interval schedule for treadle pressing that arranged a lower reinforcement rate than in Condition 2 (92% reduction in reinforcers per minute). Resurgence was lower in magnitude relative to conventional extinction, although resurgence was obtained with 2 out of 3 pigeons. The results are discussed in terms of the variables controlling resurgence and the relations between behavioral history, resurgence, and other forms of response recovery.
Journal of the experimental analysis of behavior, 2003 · doi:10.1901/jeab.2003.80-217