Acquisition of matching to sample via mediated transfer.

Two adults with Down syndrome joined a matching game. The trainer first taught B-A matches, then C-B matches. No one ever practiced C-A matches.

The team wanted to see if the third link would appear on its own. They called this hidden jump 'mediated transfer.'

Both learners quickly chose the correct C-A pairs without training. The untaught relation popped out like a bonus skill.

The result showed that two taught links can weave a three-part network in the mind.

Preston (1994) ran the same logic with neuro-typical adults and a child. Emergence was even faster, proving the effect holds across populations.

LeFrancois et al. (1993) added a 5-second delay before the sample. The tweak cut errors in half for adults with ID, giving you a smoother way to build equivalence.

Hansen et al. (1989) seems to disagree. They had to drill simple parts first or matching failed. The gap is about method: M et al. chained two relations, J et al. broke the task into tiny pieces. Both paths work, but mediated transfer skips the drill when learners can handle bigger steps.

You can grow reading sets, sign sets, or social networks without teaching every last pair. Train A-B and B-C, then test for A-C and celebrate the free skill. Start tomorrow: pick six pictures, six printed words, and six spoken words. Teach picture-to-word and word-to-sound, then watch the picture-to-sound matches appear without extra trials.