Correspondence training: an example of rule-governed behavior?

The team worked with four adults who had mild intellectual disability. Each adult sat alone in a room with a box of small objects.

Two adults got say-do training: they first said which object they would touch, then touched it, then earned a token only if the two matched. The other two simply earned tokens for touching the correct object with no verbal step.

The study asked: does adding the verbal promise help people learn to match their words and actions?

Both groups learned to pick the right object almost every time. Accuracy rose from about a large share to over a large share for everyone.

More important, both groups also started matching their words and actions in new tasks later, even when no one asked them to talk first. The verbal training gave no extra benefit.

Byra et al. (2018) later used the same correspondence idea with preschoolers with autism. They kept the verbal step because parents wanted to hear the child say the plan. The kids still generalized bathroom hygiene skills at home for six months.

Ortiz et al. (2022) tested awareness training for public-speaking tics. Like R et al., they found the full package was often overkill—just describing the problem cut tics for half the speakers.

These studies line up: the extra verbal part often does not speed learning. You can start with the simpler, do-only route and add words only if the learner or caregiver needs them.

If you want clients to follow through on promises, reinforce the follow-through first. Skip the verbal contract unless the setting demands it—like when parents or teachers need to hear the plan. This saves session time and reduces response effort for the learner. Try a quick do-only phase first; add say-do only if generalization stalls.