Contextual control of stimulus generalization and stimulus equivalence in hierarchical categorization.

Griffee et al. (2002) taught adults to sort nonsense syllables into groups. The twist: a colored background told them which rule to use. Red might mean 'pick the small animal,' while green meant 'pick the large one.'

After learning the rules, the team tested if new syllables slid into the right groups without extra teaching. They also checked if the colored cues alone could make the whole class of syllables pop up.

The setup worked. Participants built tidy, layered categories. Once the color cue appeared, they treated all linked syllables the same way—even syllables they had never seen before.

Functions and equivalence relations transferred across the board. In plain words, the context controlled both the group and what the group meant.

Reichow et al. (2011) ran a similar test but swapped the color cue for different-looking pictures. Only some adults passed, showing that arbitrary cues can be safer than look-alike ones. The two studies line up: context matters.

Brown et al. (1994) proved pigeons can do transitive matching. Karen et al. push the same logic further, proving humans can layer extra rules on top of transitivity.

Howard (1979) showed that a verbal label alone can bend generalization. Karen et al. add a concrete tool—conditional color cues—to make that bending reliable and teachable.

If you want flexible concept learning, tape a context cue onto the task. Use a colored border, a corner icon, or a spoken word to tell the learner which rule is ‘on.’ After a few trials, the cue alone can summon the whole category and every skill tied to it. Great for teaching conditional safety rules, multiple schedules, or category games where the same item can live in different buckets.