Derived stimulus relations, semantic priming, and event-related potentials: testing a behavioral theory of semantic networks.

The team taught adults two sets of four made-up words. Each set formed an equivalence class through match-to-sample training.

After training, they flashed word pairs on a screen. Some pairs came from the same class, others did not. They measured reaction time and brain waves (ERP) to see if related words sped up responses.

Words inside the same class primed each other. Reaction times dropped and brain waves showed a clear N400 effect, a sign of semantic processing.

Words from different classes did not prime. The results treat equivalence classes like mini language networks in the brain.

Dias et al. (2021) later showed that priming only appears if the words can be spoken. Abstract shapes alone will not do it. Together, the two papers tell us both equivalence and pronounceability matter for ERP signatures.

Barnes-Holmes et al. (2018) fold these ERP findings into a broad review. They use the priming data as hard evidence that derived relations are the building blocks of meaning.

Cohen-Almeida et al. (2000) stretch the idea further. They formed equivalence classes in people with minimal language, proving the model can work outside college labs.

You now have an objective, brain-based way to check if your client truly treats two stimuli as 'the same.' Run a quick priming test: same-class pairs should speed up responses. If they do, your equivalence class is solid and ready for generalization programming.