Responding maintained by intermittent reinforcement: implications for the use of extinction with problem behavior in clinical settings.

The team worked with adults with intellectual disability who had long-standing problem behavior.

Each person first got continuous reinforcement (every response worked). Then they got intermittent reinforcement (only some responses worked).

After each schedule the staff used extinction. They timed how fast the behavior stopped under both conditions.

Extinction took the same time whether the behavior had been on continuous or intermittent reinforcement.

Giving a short run of continuous reinforcement right before extinction lowered the baseline level and made the first extinction burst smaller.

The old worry that intermittent reinforcement makes behavior harder to stop was not supported.

Lattal et al. (2020) saw the same thing in pigeons and rats: intermittent histories did not slow extinction and bursts were rare.

Woodford et al. (2024) now puts extinction last in sleep plans, matching the idea that you can wait and use milder steps first.

Nevin et al. (2016) adds a twist: lean signaled DRA plus extinction cuts later relapse, so pairing a thin schedule with a signal may help after all.

You no longer need to fear that thinning a schedule will lock problem behavior in place. Feel free to move to intermittent reinforcement when it helps the client or fits the setting. If you want an easier start, give a brief stretch of continuous reinforcement right before you begin extinction; it softens the first burst and keeps rates low.

Basic research established that behavior maintained on an intermittent schedule (INT) extinguishes more slowly than behavior maintained on a continuous schedule (CRF). This is the partial-reinforcement extinction effect: sparse, unpredictable reinforcement builds persistence.

A common explanation is discrimination. Under CRF, the switch to extinction is obvious and responding drops quickly. Under INT, non-reinforced responses were already routine, so the learner takes longer to detect that reinforcement has ended.

This applied study tested whether that laboratory finding complicates treatment. Three individuals with profound intellectual disability took part after functional analyses identified the reinforcers maintaining their self-injury, aggression, or disruption. Behavior was placed on CRF or INT before extinction, using reversal and multielement designs.

Contrary to the worry, problem behavior maintained on INT was not necessarily harder to treat with extinction than behavior on CRF. The results temper the assumption that a history of intermittent reinforcement dooms extinction, though schedule history still deserves attention in treatment planning.