Effect of amount of training on rate and duration of responding during extinction.

KELLEHER et al. (1963) asked a simple question: does more continuous reinforcement change what happens when you stop reinforcing?

They first trained rats and preschool kids on CRF. Every correct response earned a reward.

Then they ran extinction. They counted how many extra responses popped up right after reinforcement stopped.

The group that got the longest CRF history showed the biggest extinction burst.

More training made the burst taller, longer, and more variable.

The same dose-response pattern appeared in both rats and children.

Capio et al. (2013) seems to disagree. They saw that CRF-made behavior persisted longer during extinction.

Look closer: M et al. measured how long the behavior kept going; T et al. measured the first spike. Both agree—dense reinforcement history makes extinction tougher.

Shahan et al. (2025) updated the finding. Bigger reinforcers, not just more trials, also grow the burst. They then showed you can shrink the burst by adding a rich alternative reinforcer during extinction.

Neisworth et al. (1985) took the lab lesson to a clinic. Two adults with developmental disability received CRF for brief self-stimulation, then extinction. Self-stim dropped fast, but only one client kept the gain at two weeks.

Before you place a behavior on extinction, check its reinforcement history. If the client has received continuous, large, or long-lasting rewards, plan for a bigger initial burst.

Have alternative reinforcers ready, as Shahan et al. (2025) suggest, and consider thinning the schedule first, as Capio et al. (2013) recommend.