Active recall: an evidence-based learning technique

Active recall, or retrieval practice, means producing an answer from memory before viewing the material. It can be a question, blank-page exercise, problem, or brief explanation. It need not be graded: the attempt itself can support later retrieval.

In Roediger and Karpicke's study, restudying prose helped more after a very short interval, but retrieval practice led to better recall after two days and one week (2006). This influential result does not mean every test always wins. Outcomes depend on the material, delay, feedback, and match between practice and the eventual task.

A reread paragraph feels familiar, which can produce an overly optimistic prediction of later performance. A closed-book attempt reveals what you can actually produce and guides the next study step. A broad review rated practice testing and distributed practice as high-utility strategies, within the normal limits of education research (Dunlosky et al., 2013).

Difficulty is not valuable by itself. An impossible or ambiguous question, or a repeatedly practiced error without correction, need not improve learning.

1. Blank page: write what you recall, then compare it with the source.
2. Headings as questions: hide the section and answer in your own words.
3. Flashcards: one prompt, one answer, then immediate checking.
4. Problems: solve a new case without watching an example, then analyze the error.

Schedule returns with flashcards and spaced repetition.

Suppose you read a chapter on supply and demand.

Minutes 0–3—closed book: draw axes and curves and label equilibrium. Write what shifts demand versus what produces movement along a curve.

Minutes 3–7—questions:

• What happens to equilibrium price and quantity when production costs rise?
• Why does a change in the product's own price not “shift demand”?
• Give one factor that shifts demand.

Minutes 7–11—answer: solve and graph without looking. Add a one-sentence “because” to each claim.

Minutes 11–15—feedback: compare with a verified example. Add missing assumptions in another color; keep the original error visible as diagnostic evidence.

Minutes 15–18—repair: if you confused a shift with movement along a curve, make a contrast pair: a change in the product's price versus a change in consumer income.

Minutes 18–20—next attempt: write two prompts for a later date. Do not copy the answers onto their fronts.

After a genuine attempt, check reasonably soon. Distinguish among no answer, partial answer, correct result with faulty reasoning, and correct but too-slow performance for your goal. Then answer again instead of merely reading the correction. Research suggests feedback can reduce persistence of errors, especially after an incorrect initial response (Butler and Roediger, 2008).

Use a cue when a topic is new, then remove it gradually. Space attempts instead of repeating five identical questions immediately. After consistent success, lengthen the interval or ask a transfer question. After repeated failure, simplify the prompt, re-explain the mechanism, or add an example.

For lists and arbitrary pairs, add a mnemonic, but make the final practice demand the answer needed in real use—not merely the mnemonic image.

• Roediger and Karpicke (2006), testing and later recall
• Dunlosky et al. (2013), review of learning techniques
• Butler and Roediger (2008), the role of feedback