Discriminative effects of massed extincttion.
One long extinction session ending at 10 % of baseline rate sharpens stimulus control and steepens generalization curves.
01Research in Context
What this study did
Researchers worked with pigeons that pecked a colored key for food. They first trained the birds to tell two light wavelengths apart.
Then they ran one long extinction session. The session ended only when pecking dropped to 10 % of the baseline rate. After that they tested how well the birds told similar colors apart.
What they found
The extinction rule made the birds' color choices much sharper. Their generalization curve got steeper after just one long session.
The sharper curve showed up no matter how the birds were trained before extinction. Massed extinction alone tightened stimulus control.
How this fits with other research
Brinker et al. (1975) ran the same setup and got the same peak-shift. The 1974 finding replicated cleanly in pigeons.
Takashima et al. (1994) moved the idea to children. When they put one toy play action on extinction, kids started doing new play moves. The same extinction process that sharpens color control also creates response variety.
Craig et al. (2019) later showed a cost: each new extinction session weakens resistance to extinction. So the steep gradient comes at the price of faster response drop next time.
Why it matters
You can use a single, continuous extinction bout to clean up stimulus control. Set a clear criterion like 10 % of baseline rate to end the session. After that, probe for sharper discrimination. Just remember: later extinction probes may work faster, so re-check baseline often.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Run extinction in one unbroken block until the target response hits 10 % of last session's rate, then immediately test stimulus discrimination.
02At a glance
03Original abstract
Prior studies have reported that generalization gradients are not steepened if periods of non-reinforcement in S- follow and are not interspersed with periods of reinforcement in S+. Sharper gradients are produced by this massed-extinction procedure if it is preceded by prior discriminative training on a dimension orthogonal to the S+, S- dimension. The present study, using pigeons, found that generalization gradients along the wavelength dimension were steepened by massed-extinction sessions in 570 nm that had been preceded by: (1) discriminative training in which the S+ was a 550-nm light and the S- was a black vertical line superimposed on the 550-nm light; (2) non-differential reinforcement training with a 550-nm light and a black vertical line superimposed on the 550-nm light; (3) reinforcement training with only the 550-nm light. Massed-extinction sessions were administered until the response rate in the presence of the 570-nm stimulus was one-tenth of the mean response rate in the presence of the 550-nm stimulus during prior reinforcement training. Prior studies have used a time-dependent criterion, rather than a response-rate criterion of extinction, and this difference may be responsible for the differences in the effects of massed extinction on stimulus control.
Journal of the experimental analysis of behavior, 1974 · doi:10.1901/jeab.1974.22-161