Joint control and generalized nonidentity matching: Saying when something is not.

Four kids with typical development played a matching game.

Each trial showed a picture. The kids said what it was.

Then they saw three new pictures. One picture let them repeat their own word. The other two did not.

Kids got praise for picking the picture that did NOT match their word. This is called nonidentity matching.

After a few training rounds, all four kids picked the non-matching picture with new samples they had never seen.

They could say "not there" without extra teaching.

The study showed that hearing yourself talk can guide new choices. This is called joint control.

Lin et al. (2020) used the same conditional-discrimination idea with Chinese preschoolers with autism. They skipped easy drills and went straight to hard ones. Their kids also learned fast, showing the idea works in special ed rooms.

DeQuinzio et al. (2018) taught kids with autism to copy only correct models and say "I don’t know" when the model was wrong. Both studies train kids to notice what is missing or wrong, but one uses modeling and the other uses joint control.

Keintz et al. (2011) used stimulus-equivalence drills to teach coin values. Like the target study, kids got new skills without direct teaching. The 1992 paper adds evidence that equivalence-based methods create broad generalization.

You can use joint control to teach abstract concepts like "wrong," "missing," or "not the same." Start by letting the learner tact the sample, then reinforce picking the comparison that blocks the tact. Once the pattern is learned, new items generalize for free. This saves hours of trial time and builds the verbal skill of noticing absence—a key step for problem solving and safety.