Testing for transitive class containment as a feature of hierarchical classification.

Five adults without disabilities tried to learn three-stimulus classes.

Some classes used shapes that looked alike. Others used random shapes.

The trainer gave many matching-to-sample trials until each person passed.

Then the team tested if the adults could pick the "bigger group" inside each class.

Only two people passed the bigger-group test with look-alike shapes.

After more training, zero of three people passed.

When the trainer switched to random shapes, all three people passed.

Arbitrary shapes made the hidden rule pop out.

Brown et al. (1994) saw perfect transitivity in pigeons years earlier. Birds got it; most humans here did not. The bird task was simpler—no "bigger group" test—so success was easier.

Zhirnova et al. (2025) later added naming games: kids both said and heard the class name. Every child then showed transitivity plus analogy. Adding talk may fill the gap Brian et al. left.

Bruns et al. (2004) also saw mixed results with toddlers using random shapes. They proved that listener-only drills are too weak; you need naming or extra tact trials. Brian’s adults may have needed the same boost.

If your learner is stuck on equivalence, swap to arbitrary stimuli first. Then add bidirectional naming—have the learner both say and point to the class name. This two-step tweak can turn failed probes into solid emergent relations.