Spaced practice: why a missed item comes back after one, three, seven, fourteen and thirty days

What the forgetting curve measures

A name can slip away in the middle of a conversation and still look entirely familiar when you meet it on an invitation a week later. That difference between recognising something and recalling it helps in reading the forgetting curve associated with Hermann Ebbinghaus. He published in German in 1885. The English translation, Memory: A Contribution to Experimental Psychology, appeared in 1913. He learned lists of nonsense syllables, relearned them after a delay and measured how much work the earlier learning had saved, a method he devised himself. The only participant in those experiments was Ebbinghaus himself. The observations therefore say little about other people or about meaningful material.

Ebbinghaus found that savings fell steeply at first and then more slowly as time passed. His measure compared the work needed for relearning with the work of the original learning session, rather than counting facts he could freely recall. The familiar curve therefore describes a pattern in one set of experiments. It cannot tell you that a fixed share of yesterday's lesson has disappeared from your mind, and it offers no personal deadline for remembering a colleague's name.

For ordinary learning, the useful implication is to arrange another encounter while something remains to be recovered. If you wait until an important name has become inaccessible, you may face more relearning work than a timely return would have required. Returning while you can still retrieve the answer with some effort also gives you a different task from repeating it while it sits in immediate memory.

A gap can make a review worth more

Imagine learning the meaning of a term in a training manual, then covering the definition and explaining it aloud. Immediately after reading, you may still hear the wording in your head. Returning on another day forces you to reconstruct the meaning from the term and the situation where it belongs, and successful recall of that kind can support later retention, although the size of the benefit depends on the task and the length of the gap.

The phrase “just before forgetting” describes a practical aim rather than an instant that a learner can reliably observe. A well-timed review can save the work of rebuilding an answer that has become inaccessible, while still making recall more demanding than an immediate repetition. This is a rationale for economical practice, and no measured saving is attached to the schedule described on this page. If you cannot retrieve the answer, check it and rebuild your understanding before another attempt.

Nicholas Cepeda and colleagues pooled the evidence on distributed practice in a 2006 meta-analysis, Distributed Practice in Verbal Recall Tasks: A Review and Quantitative Synthesis, published in Psychological Bulletin. It covered 839 assessments from 317 experiments in 184 articles, all using verbal recall tasks, and it concluded that the gap between study sessions and the delay before the final test act together, since the gap giving the best retention grew as the delay grew. The consequence for a learner is that timing should serve a retention goal. Repeating everything during a single evening gives no practice at recovering it on a later occasion.

The time you need to remember changes the useful gap

Preparing for tomorrow's meeting and learning terminology for work over the coming months are different timing problems. Cepeda and colleagues tested that relationship in their 2008 paper, Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention, in Psychological Science. More than 1,350 people were taught a set of facts, given a review after a gap of up to three and a half months and tested up to a year later. At any given test delay, final recall first rose as the gap before review grew and then declined, and the best gap became longer as the final test moved further away.

Longer retention goals therefore supported longer gaps in that experiment, although the best gap did not grow in a fixed proportion to the test delay: it was roughly 20 to 40 per cent of a one-week delay but only 5 to 10 per cent of a one-year delay. That rules out a universal calendar multiplier. The study also involved a single review after initial learning, rather than a whole ladder of returns. When planning your own revision, put the date you need the knowledge beside the date you learned it, then reserve opportunities to recall it between those dates.

Expanding gaps have no clear claim to superiority

A widening sequence looks plausible because familiar material seems ready for a longer wait. Jeffrey Karpicke and Henry Roediger tested that reasoning in their 2007 paper, Expanding Retrieval Practice Promotes Short-Term Retention, but Equally Spaced Retrieval Enhances Long-Term Retention. In their first experiment, 48 undergraduates learned vocabulary pairs, and expanding retrieval gave a modest advantage on a test ten minutes later. Equally spaced retrieval produced higher recall after two days, which reversed the short-term comparison.

Later experiments in the paper separated the delay before the first retrieval from the gaps between subsequent attempts. Delaying that first retrieval helped long-term retention whichever later schedule was used, while expanding the gaps added little. These findings do not establish equal gaps as ideal for every material or retention goal. When choosing a review date, leave enough time that you must genuinely recover the answer, then check whether you can still retrieve it accurately.

The five returns are a practical ladder

A flashcard user will recognise the organising idea: keep returning to material that needs attention and allow longer waits as it becomes familiar. In the Leitner tradition, cards move among boxes associated with different review frequencies. Spaced-repetition flashcard software makes that organising work easier to maintain, although individual systems use different schedules. The shared purpose is to distribute encounters across time instead of repeatedly browsing an entire deck.

Funga Wega brings missed items back after one, three, seven, fourteen and thirty days, treating those public intervals as a practical ladder with no claim of optimal timing. Facts return as the same question, including facts first met in a short reading-room lesson. Generated puzzles return as fresh questions of the same kind. Revisiting a reasoning problem therefore asks you to apply an approach again, while a factual return asks you to recover knowledge encountered in context.

The figure below illustrates declining access and repeated restoration using invented values. Every rise represents a successful review in this drawing, whereas real attempts can fail, and neither the rises nor the gentler later slopes predict an individual learner's results.

A useful review begins with an attempt

For a book or training course, choose a question whose answer you will need away from the page. Explain it with the source closed, compare your explanation with the original and record the particular omission. Later, attempt the question again before looking at your notes. When reviewing a calculation method, change the quantities so that recognising yesterday's answer cannot do the work, then check the reasoning that led to your result.

When practising the idea of sample bias, for example, explain how a survey of people already attending a library could misrepresent the reading habits of everyone in the town. On a later attempt, use a survey taken at a sports ground instead, so you have to identify the sampling problem in another setting. Write down the clue that made you recognise the problem.

An unsuccessful review gives you something specific to repair. Revisit the missing step and arrange another attempt you can realistically keep. If an important deadline is approaching, bring that attempt forward rather than treating any published ladder as a reason to arrive unprepared.