The extinction burst: Impact of reinforcement time and level of analysis on measured prevalence

Nist et al. (2021) tested how measurement choices change what we call an extinction burst. They used pigeons that earned food on fixed-ratio 1 or 5 schedules. Then they stopped all food and watched what happened.

The team scored the same sessions two ways: minute-by-minute and whole-session totals. They also tried two math tricks for the baseline rate: count every second or subtract the time birds spent eating.

Minute-by-minute scoring caught more bursts than whole-session totals. When they removed eating time from the baseline, even more birds looked like they had a burst. Same birds, same data, different answer.

In short, the stricter the lens, the more common the burst appeared.

Katz et al. (2020) saw the same thing a year earlier: bursts show up only when you slice the data thin and watch the first minutes closely. Both labs used pigeons and reached the same mixed-finding verdict.

Lerman et al. (1995) and Muething et al. (2024) each reported a 24% burst rate in big reviews—one with lab data, one with clinic cases. Nist's measurement tweak can push the count toward that 24% benchmark without any real change in behavior.

Shahan et al. (2025) added that larger reinforcers before extinction make bursts bigger. Nist teaches us to spot those bursts reliably; Shahan shows how to shrink them by offering richer alternative rewards.

If you only graph session totals, you may miss a brief spike that signals an extinction burst. Switch to minute-by-minute probes during the first three sessions of any new extinction plan. Also, subtract time the client spends consuming reinforcers from your baseline rate; it gives a cleaner picture of true responding. These small steps help you decide whether to stay the course or add supports before the burst escalates.