Variation Makes Practice Do Different Work

The question here is not simply whether learners should repeat a problem or see a new one each time. It is how repetition and variation change the kind of representation that problem solving, retrieval practice, or worked examples can build. The practical judgment is to vary cases when learners must recognize a shared structure, preserve enough common structure for comparison, and expect that useful variation may slow practice before it improves transfer.

This matters because the research traditions surrounding retrieval practice and worked examples have often used different kinds of practice. Worked-example studies commonly give learners a different problem on each trial. Retrieval-practice studies often repeat the same item. If a varied worked-example condition is compared with a repeated retrieval condition, the apparent advantage may belong partly to the practice material rather than to the instructional method. Separating those factors reveals what variation contributes on its own.

Repetition, Variation, and the Scope of What Is Remembered

Exact repetition gives a learner repeated contact with the same item. Each encounter can strengthen access to that item, its answer, and the cues associated with it. If the goal is to remember a particular fact or become able to retrieve one defined response, repetition may be well matched to the task. Repetition also makes practice look smoother: the learner becomes increasingly familiar with the prompt and may respond with less search.

But a familiar item is a narrow test of knowledge. A learner can remember that one problem has one answer without representing the relation that would solve a different problem. Repeating the same example can therefore strengthen a specific exemplar while leaving the learner uncertain about which features are essential and which are accidental. The learner may know what to do when the problem looks exactly the same, but not recognize the rule when the numbers, wording, or arrangement changes.

Variation changes the representational problem. Instead of encountering one item again and again, the learner meets several instances that share an underlying structure. The surface details may change while the relation that matters remains constant. To succeed, the learner must compare cases, notice what remains invariant, and separate the rule from the incidental features of any one example. That comparison creates an opportunity for abstraction: the learner builds a representation that covers more than a single prompt.

Suppose a learner is studying the distributive property. Seeing 2(x + 3) become 2x + 6 repeatedly may strengthen the connection between that expression and its result. Seeing 2(x + 3), 4(x + 5), and 3(x + 1) transformed in related ways creates a different demand. The learner has to notice that the outside factor applies to both terms even though the numbers change. The cases are useful because they vary in a controlled way while preserving the mathematical relation.

The same principle applies beyond mathematics.