Stimulus class formation and concept learning: establishment of within- and between-set generalization and transitive relationships via conditional discrimination procedures.

Three adults with intellectual disability sat at a table. They learned to pick the correct picture when given a sample. Each set had ten pictures. Some pictures earned snacks. Others did not. The team tested if the learners could still match new pairs they had never practiced.

The study used eight different ten-picture sets. Every session used a two-choice format. Correct picks that had earned food before kept their snack. The rest stayed neutral.

Only pictures with a snack history gave clean transitive relations. Learners showed within-set and between-set generalization for those items. Items that entered the class only through generalization never produced transitive control.

In plain words, reinforcement history made the difference. Generalization alone was not enough.

Dube et al. (1989) ran almost the same lab setup the same year and got the same positive result. The close match feels like a direct replication.

Dube et al. (1987) showed that reinforcers themselves join stimulus classes. Fetterman et al. (1989) extend that idea by proving the reinforcer link is required for transitivity to appear.

Ayres‐Pereira et al. (2025) now supersedes this work. Their multiple-baseline design across neurotypical adults gives stronger evidence and quantified large effects for forming equivalence classes with highly similar stimuli.

When you build stimulus classes, tie every member to a reinforcer at least once. Do not rely on pure generalization to create transitive relations. A quick praise or token for the untaught links can save you from silent failures during equivalence probes.