Response effort and resurgence

Walter et al. (2023) asked a simple question: Does the amount of work a behavior takes change how often it comes back after you stop reinforcing it?

They taught adults with and without autism two different ways to get tokens. One way took a single button press. The other took ten fast presses. Then they stopped giving tokens and watched which old response popped up more.

The easy, one-press response always returned stronger. When reinforcement ended, people used that low-effort move far more than the tiring ten-press move.

High-effort responses barely resurfaced at all. Effort during training acted like a shield against later relapse.

The finding lines up with Takashima et al. (1994) and Fingeret et al. (1985). All three used single-case lab designs and saw big, clear behavior changes. Each paper shows that how you set up the task matters as much as what you teach.

Rasing et al. (1992) looked at timing inside a session, not resurgence across sessions. Still, both studies remind us that small mechanical details—press rate, clock pattern—can steer big-picture results.

No direct clash here. Walter et al. simply add a new lever: response effort controls comeback strength.

When you plan extinction, pair it with a replacement skill that is easy to do. If the new skill is simpler than the old problem behavior, the client will choose it later when times get tough. Drop the effort bar and you drop the chance of resurgence.

Resurgence is the reappearance of a previously reinforced behavior when reinforcement for a more recently reinforced alternative behavior is discontinued or reduced. Three phases define it: first a response (R1) is reinforced, then R1 is placed on extinction while an alternative (R2) is reinforced, and finally R2 also stops producing reinforcement. When R2 stops working, R1 comes back, even though nobody is reinforcing it.

The classic clinical example is functional communication training. A learner's aggression is put on extinction while a mand like 'break, please' is reinforced. Treatment works, then the learner enters a busy classroom where staff cannot honor every mand. The mand contacts extinction, and aggression resurges. The behavior did not need to be re-reinforced to return; the treatment challenge alone was enough.

Resurgence is not the same as spontaneous recovery. Spontaneous recovery is the return of an extinguished response after time passes between extinction sessions, with no alternative behavior involved. Resurgence is specifically driven by worsening conditions for the alternative response, which is why it matters so much for predicting relapse after FCT and other differential reinforcement treatments.

Response effort is the physical or temporal work a response requires: how hard, long, or awkward it is to emit. It is one of the most manipulable dimensions in behavior analysis. Moving preferred items out of reach raises the effort of grabbing them; placing a communication card at arm's length lowers the effort of manding.

Effort functions like a price. All else being equal, behavior flows toward the response that costs less, which is why effort manipulations show up both as antecedent interventions (making problem behavior more effortful) and as design principles for replacement behaviors (making the mand, the card touch, or the device press as cheap as possible).

In this study, effort was manipulated by changing the size and speed of targets participants had to touch on a screen: large, slow targets made responding easy, and small, fast targets made it hard. That let the researchers ask a question with direct clinical stakes: does the effort a behavior took during its reinforcement history change how strongly it resurges later?

Across two experiments with typically developing adults and adolescents with autism, resurgence was greater when the original response was lower effort. Easy responses came back strongly during the final extinction test; effortful responses barely returned. A cheap behavior with a reinforcement history is apparently always near the surface.

Two practice implications follow. First, expect more resurgence when the target problem behavior is low effort, like dropping, yelling, or a simple swipe, and plan for relapse in those cases with denser schedule thinning and treatment-challenge probes. Second, protect your replacement behavior by keeping its effort low: if the mand is harder to emit than the problem behavior it replaces, extinction conditions will favor exactly the response you are trying to eliminate.