Resurgence as Choice in Context: Treatment duration and on/off alternative reinforcement

Shahan et al. (2020) tested how long alternative reinforcement should run before you stop it. They used pigeons that first earned food by pecking one key. Then that key stopped paying off. A second key began to pay. After different lengths of time on the second key, the payoff also stopped. The team watched how much the birds returned to the first, now-unpaid key.

The study cycled the second key's payoff on and off. This let the researchers see if short versus long periods of payoff changed later return to the first key.

Longer periods of payoff on the second key slightly reduced the birds' return to the first key. But the original math model had to be updated. It now had to include both the time spent on the alternative key and the local cues that signaled payoff was available.

In short, duration matters, but so do the little signals that say 'food here now.'

Shahan et al. (2020) directly replicate Shahan et al. (2020). Both 2020 papers tweak the Resurgence-as-Choice model and confirm it fits the data. One changes duration; the other changes rate. Together they show the model works across different ways of weakening alternative reinforcement.

Craig et al. (2017) earlier found that bigger food pellets sped up extinction but also caused bigger resurgence. The target paper now shows that longer payoff time, not just size, also tempers the bounce-back. The model needs both pieces.

Greer et al. (2024) later applied the duration idea to children with destructive behavior. Gradual thinning, not sudden drops, kept resurgence low. This extends the pigeon finding to real clients.

Shahan et al. (2024) seems to clash by saying higher-quality payoff causes more resurgence. But quality and duration are different levers. Better payoff boosts later return when removed; longer payoff slightly lowers it. The model now includes both factors.

When you fade token or edible reinforcement, run the alternative payoff a bit longer before each thinning step. Watch for signals like the bin still being in view or the timer beeping—these cues also drive response allocation. Plan longer treatment phases and fade gradually to keep problem behavior from roaring back.