Serial alternative response training as intervention for target response resurgence.

Three adults with developmental disabilities took part. All had problem behavior that returned when reinforcement stopped.

The team compared two DRA setups. One group earned rewards for a single good response. The other group earned rewards for two different good responses that alternated each session.

An alternating-treatments design flipped the two setups across days. The researchers watched how much the old problem behavior bounced back when reinforcement paused.

Serial DRA won. When the two-response rotation was in place, problem behavior stayed low during the later test.

With classic single-response DRA, the old behavior quickly returned. The simple switch of reinforcing two actions in turn cut resurgence for every participant.

Nist et al. (2023) saw the opposite. Their rats got five alternating responses, yet resurgence stayed just as high as with one response. The clash is useful: rats are not people, and five choices may overload humans who need only two.

Fleck et al. (2023) extend the idea. They showed that two responses running side-by-side (concurrent DRA) also block resurgence, but only when problem behavior still earns rewards. The 2015 study adds the serial option for cases where you do use extinction.

Austin et al. (2015) had already sketched the theory: teach multiple alternatives so any resurgence lands on good, not bad, behavior. Perez et al. (2015) give the first human data that the plan works.

If you run DRA and worry about relapse, rotate two functional alternatives instead of drilling just one. Pick responses the client already enjoys and switch the reinforced option each session. The small change can spare you the big headache of resurgence when reinforcement later thins or stops.