Learning and the mind
Spaced repetition
Reviews spread out over time beat the same reviews crammed together, and the best gap grows with how long you need to remember.
Hermann Ebbinghaus noticed the effect in his 1885 memory experiments; the modern schedule work is by Nicholas Cepeda, Harold Pashler and colleagues (2006 and 2008).
If you have five chances to review something, spreading them over weeks beats spending them in one sitting. Psychologists call this the spacing effect, and it is one of the best-documented findings in the field. Cepeda, Pashler, Vul, Wixted and Rohrer pooled 839 measurements from 317 experiments in 2006 and found the same pattern almost everywhere: a gap between study sessions helps, and the gap that helps most gets longer as the time you need to remember for gets longer. Spaced repetition is the method built on this. Each item gets its own schedule, with reviews far enough apart that recall takes work, and the interval stretches each time you succeed.
The same group's 2008 study gives the numbers to plan with. Over 1,350 people learned 32 trivia facts, reviewed them once after a gap of up to 105 days, and were tested up to a year later. At every test delay, recall rose as the gap grew and then fell once the gap got too long. The best gap gave 64 percent more recall than reviewing straight away. Where the peak sat depended on the delay: about one day for a test a week out, 11 days for a test at five weeks, 21 days at ten weeks and at a year. As a rule the best gap is roughly 10 to 20 percent of the time you want to remember for, falling to about 5 percent at a year. The likely mechanism is that a review after some forgetting forces a fresh retrieval rather than a quick recognition, and that effort is what strengthens the memory.
The research is mostly on facts, word pairs and short texts, so the schedules are best trusted for that kind of material. You rarely know the retention interval in advance, which is why software uses an expanding schedule and adjusts it from your answers; that is a practical compromise, and the 2008 study tested a single gap, not a sequence. And spacing feels worse while you do it. Cepeda's subjects did best with the gap that would have felt least productive, which is why learners left to themselves cram, and why the schedule is better set by a system than by mood.
Example
In a study published in Psychological Science in 2008, Nicholas Cepeda, Edward Vul, Doug Rohrer, John Wixted and Harold Pashler had 1,354 people learn 32 obscure facts online, come back for one review after a gap of between zero and 105 days, and sit a final test 7, 35, 70 or 350 days after that. For the one-week test the best gap was one day; for the 35-day test it was 11 days; for the 70-day and one-year tests it was 21 days. Compared with reviewing immediately, the best gap raised recall by 10, 59, 111 and 77 percent at the four delays. The authors concluded that many educational practices are therefore "likely to be highly inefficient".
Where it applies
The flashcard decks at /flashcards run on a spaced-repetition scheduler, and the "Test yourself" feature at /models/review spaces these models the same way; both only work if you grade yourself honestly, since the schedule is set by your misses. If you build a scheduler, Cepeda's ridgeline is the lookup table: decide how long the memory must last, then set the first gap at 10 to 20 percent of it.
Think about it
Which fact or skill did you last cram in one sitting, and when is the next time you will actually need it?
Applied it to something real? Log it.
Test yourselfIf you need to remember something for about ten weeks, how far apart should two study sessions be, according to Cepeda and colleagues (2008)?
Roughly 10 to 20 percent of the retention interval, so about three weeks for a ten-week target. The best gap grows with the delay but shrinks as a fraction of it, down to about 5 percent at a year.
Related models
- The forgetting curveMemory for new material drops fast in the first hours and then slowly; what is left after a day is most of what will be left after a month.
- Retrieval practicePulling something out of memory strengthens it more than putting it in again, so a test is a better study tool than a reread.
- Desirable difficultiesConditions that make learning feel slower and harder often make it last longer; the ease you feel while studying is a poor guide to what you keep.
- Half-lifeA quantity that loses a fixed fraction per period has a half-life, and after a few of them most of it is gone, whether it is carbon-14 or a fact.
Sources and further reading
- Nicholas Cepeda, Edward Vul, Doug Rohrer, John Wixted and Harold Pashler, Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention, Psychological Science 19(11) (2008)
- Elizabeth Bjork and Robert Bjork, Making Things Hard on Yourself, But in a Good Way: Creating Desirable Difficulties to Enhance Learning, in Psychology and the Real World (2011)