Students can become highly accurate on familiar exercises and still struggle when a problem is presented differently. This is not necessarily evidence that the earlier learning was useless. It often shows that practice stabilised execution without sufficiently training recognition and transfer.
Surface features are powerful cues
Numbers, diagrams, wording and context help students identify a problem. If those features remain similar across practice, learners can associate the method with the surface rather than the underlying relationship.
Change one feature at a time
After routine practice, vary the wording while preserving the structure. Then change the representation. Change the context. Add irrelevant information. Each variation tests whether the learner still recognises what matters.
Compare problems
Ask why two different-looking problems can use the same principle. Then compare two similar-looking problems that require different methods. These contrasts teach students to discriminate structural clues from decorative ones.
Require a diagnosis before a solution
Before calculating or writing, ask what kind of relationship is present and what evidence supports that judgement. This makes method selection visible and gives teachers a chance to repair recognition errors separately from execution errors.
Use productive uncertainty
Unfamiliar problems should not be so distant that students lack all prerequisites. The challenge should force adaptation while leaving a plausible route through existing knowledge.
Return after a delay
Immediate variation can still be cued by the lesson. Revisit the principle later among competing topics. Delayed mixed practice provides stronger evidence of flexible access.
The repair
Once a method is understood, stop protecting it with an unchanged surface. Vary context and representation, compare structures, require diagnosis and mix the knowledge with alternatives. Students become better at unfamiliar problems by learning what remains invariant when everything else changes.
