Reinforcer magnitude and resurgence

Oliver et al. (2018) worked with pigeons in a lab. First the birds pecked a key for food. Then the food stopped and the birds learned to peck a new key for food. Later the team cut or boosted the food amount for the new key while stopping it again.

The test asked: does changing the size of the new reinforcer bring the old pecking back? The old response returning is called resurgence.

When the new key paid less food, the birds went back to the old key. When the new key paid more food, the old key stayed quiet.

The result says resurgence is not just about change. It is about worsening conditions. Making the new deal worse drives relapse.

Weinsztok et al. (2022) ran a similar test with people and got the same big-picture answer. They showed that bigger reinforcement for the alternative behavior protects the treatment even when staff slip up. Both papers say larger magnitudes guard against relapse.

King et al. (2025) looked at kids with autism and asked a new question: does the type of reinforcer matter, not just size? They found giving the original reinforcer non-contingently cut resurgence a bit more than giving the alternative one. Oliver’s size rule still holds; King adds that the reinforcer’s history also counts.

Cashon et al. (2013) seems to clash at first. They saw that bigger reinforcers made pigeons less flexible, not more. The studies differ on what they measure. Oliver looks at relapse of an old response; H looks at variety within the current response. Bigger pay keeps the current response strong, which can both lower variety and block resurgence. No true conflict—just different outputs.

When you thin or shift reinforcement, keep the alternative payoff strong or even boost it. If you must reduce quality or quantity, expect a spike in the old problem behavior and plan extra supports. Watch for worsening conditions, not just change.