Inducing select/reject control in a matching‐to‐sample procedure with observing response: Effects on stimulus equivalence

Grisante et al. (2024) asked adults without disabilities to play a three-choice matching game. First, the adults had to press a small key to peek at the sample. This peek counted as an observing response. Then they picked the card that went with the sample.

The team wanted to know if this peek step would push people to use select control (picking the right card) or reject control (ruling out wrong cards). They tracked which style each adult used and checked if full equivalence classes still formed.

Every adult built new equivalence classes, no matter which control style they leaned on. Most adults showed strong select control, but clear reject control also worked. The peek requirement nudged both styles without extra training.

In short, you can grow brand-new stimulus classes even when the learner mainly learns by rejecting errors first.

Plazas (2019) already showed that baseline reject control can spread to new items in transitivity tests. Grisante et al. go one step further by purposely inducing that reject control during training, not just watching it transfer later.

Frame et al. (1984) warned that you must test separate S+ and S– control before claiming true equivalence. The new study follows that rule by using the observing response to separate select from reject moments, giving cleaner proof of equivalence.

Cullinan et al. (2001) used same/different cues to bias equivalence; Grisante replaces those cues with the peek response. Both tricks show that small procedural tweaks can shape how people come to treat stimuli as equal.

If a client keeps picking by elimination instead of pointing right to the target, you no longer need to panic. Let them look longer or require an observing response; equivalence can still bloom. Next time you run an MTS task, add a quick peek step and watch whether the learner leans on select or reject control. Either path can end in full stimulus classes, so you can press on with teaching derived relations without re-training the whole protocol.