Re-emergence of under-selected stimuli, after the extinction of over-selected stimuli in an automated match to samples procedure.

Crane et al. (2008) ran two small lab experiments with neurotypical adults. The task was an automated match-to-sample game on a computer. First, people learned to pick one picture that always paid off. That picture became the over-selected stimulus. Next, the researchers turned off the payoff for that picture. They watched to see if any of the ignored pictures suddenly guided choices again.

When the money stopped for the dominant picture, adults started picking the old ignored pictures again. The previously under-selected stimuli regained control. The learning had not vanished; it was simply masked by the stronger cue.

Hawkes et al. (1974) showed the same pattern in autistic children. When they stopped reinforcement for hand-flapping, spontaneous toy play jumped up. Falligant et al. (2020) later repeated the idea with adolescents, using RIRD to block motor stereotypy and seeing leisure skills rise. All three studies tell the same story: remove the top response and the next one in line shows up.

Weisman et al. (1975) looked at the flip side. Pigeons kept pecking a key that turned the lights off during extinction. Once the blackout payoff was gone, the pecking stopped. Crane et al. (2008) echo this: extinguish the consequence that keeps the dominant behavior alive and other behavior re-emerges.

Matson et al. (2013) seems to disagree. They added extra social praise for stereotypy and later tried extinction. The stereotypy did not increase, but it also did not extinguish faster. The key difference is procedure. Laura’s team removed the payoff for the top stimulus only. L et al. added payoff first, then removed it. Adding reinforcement first muddies the later extinction effect.

You can use this in your next session. If a client keeps picking the same card, toy, or answer, try briefly removing the payoff for that choice. Watch if older, ignored options pop back. This gives you a quick way to see what the learner actually knows, not just what they repeat. It also reminds us that extinction is not just about killing behavior; it is about revealing the next layer of learning that was already there.