Picking the next review

Judgement

Pick the next review with the intended retention period in view. Do not treat one fixed interval or multiplier as correct for every learner and subject. Pavlik and Anderson’s model treats review timing as a trade-off: waiting creates more spacing, but waiting too long can produce failures that consume additional practice time.¹

A large single experiment demonstrates why the target date matters. More than 1,350 participants learned facts, received one review after gaps extending as long as three and a half months, and took final tests after further delays extending as long as one year. For each final-test delay, increasing the review gap first improved and then reduced performance. The best absolute gap became longer as the intended retention period became longer, but it did not follow one constant proportion.²

Single studies also show why adapting to responses is plausible. Undergraduates learning German-English pairs in one intensive session did better on a one-week test when a model selected items from their response histories than when ordering was random or learner-controlled.³ In a semester-long study, 179 middle-school Spanish students received equal numbers of review trials under massed, generic-spaced and personalised schedules. Personalised scheduling produced better performance on a proctored cumulative Spanish exam 28 days after the semester ended.⁴ These results concern particular materials and populations.

With a deck and twenty minutes, begin by naming when the knowledge must remain available. A schedule for an approaching exam can reasonably differ from one intended to maintain vocabulary well beyond a course. Use recent response history to decide which items deserve attention. In Atkinson’s roughly two-hour session with undergraduates learning 84 German-English pairs, training performance ranked in the reverse order from recall on the one-week delayed test.³ When the target period, material or error pattern changes, reconsider the timing rather than preserving a familiar interval automatically.

Boundary

Spacing is a strong finding. Selecting the next interval is a weaker practical decision built on that finding. Target-aware and response-sensitive scheduling are reasonable procedures, not universal prescriptions. The evidence here does not establish a fixed gap, multiplier or scheduling objective that remains best across subjects, learners and retention periods.