Basic and applied research on extinction bursts

WFrazier et al. (2023) wrote a theory paper about extinction bursts. They asked, "Why do problem behaviors spike when we stop giving reinforcement?"

The team built a new rule called the temporally weighted matching law. It counts recent reinforcers more than old ones. The paper also lists ways clinicians can soften bursts before they start.

The new rule predicts when and how hard a burst will hit. If a child got lots of attention for screaming yesterday, the model says today's burst will be big.

The authors give three take-home moves: thin reinforcers first, shift some responses to new behaviors, and start extinction when the client is calm and busy.

Sayers et al. (1995) first argued that matching-law math could guide real-world ABA. WW et al. pick up that baton and aim it squarely at extinction bursts.

Stüttgen et al. (2024) push the matching law even further. Their trial-by-trial model shows how each past reinforcer tweaks the next choice, matching WW's "recent counts more" idea.

Sailor (1971) showed that simply skipping a scheduled reinforcer makes rats respond faster. WW's model explains the same jump in people by giving extra weight to those fresh omissions.

You can now look at baseline data and forecast if extinction will start with a bang. Before you withhold reinforcement, drop the rate, build alternate behaviors, and pick a busy work period. Less burst means safer, faster treatment for your client and less stress for you.