Provide contingent support: more help when needed, less when not

Calibrate help to the learner’s actual performance moment by moment rather than delivering a fixed dose.

Why it works

Wood, Bruner, and Ross found that effective tutors constantly adjust the specificity of their interventions in response to the learner’s immediate success or failure. When the learner succeeds, the tutor withdraws one level of support; when they fail, the tutor provides one level more. This "contingent tutoring" keeps the learner in the zone of proximal development throughout the session, whereas non-contingent support leaves learners either bored or lost.

How to do it

1. Define a five-level support hierarchy before a learning session: from most to least specific hint.
2. Start with minimal support; only escalate if the learner fails to progress.
3. Drop back one level of support immediately after each success.

Evidence

Wood, Bruner, and Ross’s 1976 tutoring study directly demonstrated the contingent-support principle, finding that mothers who adjusted their help to the child’s moment-to-moment performance produced faster acquisition than those who did not. Follow-up work has repeatedly confirmed contingency as a key ingredient of effective tutoring. (observational)

Original evidence is developmental; application to adult learning contexts is widely assumed and consistent with coaching research, but less directly studied.

Sources

• Wood, Bruner & Ross (1976), "The role of tutoring in problem solving", Journal of Child Psychology and Psychiatry

Common mistake

Offering maximum support at the start regardless of need, which short-circuits the effortful attempts that drive learning and creates dependency on the scaffold.