Effects of reinforcement magnitude on spontaneous recovery.

Hatton et al. (1999) looked at how the size of reinforcers during treatment changes later problem-behavior bounce-back.

They used an A-B-A design. First, disruptive behavior earned big rewards. Then it was put on extinction while an alternative task earned either large or small rewards. Later they tested if the old behavior returned.

Kids who got larger alternative rewards showed less spontaneous recovery of the old problem behavior.

In plain words, richer reinforcement during treatment weakened the later rebound once extinction started.

Craig et al. (2017) and Ritchey et al. (2023) saw the same pattern in humans: bigger alternative rewards knock behavior out faster but rebound harder when you later thin them.

Shahan et al. (2020) updated the story with a math model. They show resurgence grows exponentially as you reduce reinforcement, giving you a formula to predict the bounce.

Arroyo Antúnez et al. (2026) repeated the effect in mice, proving the link is not just a human quirk.

When you plan an extinction plus differential-reinforcement program, start with rich rewards for the replacement behavior. Expect a bigger resurgence later if you must thin those rewards. Use the exponential model from Shahan et al. (2020) to pace the fade and schedule extra probes. This front-loads suppression now while preparing you for the bounce you know is coming.