Spaced repetition schedule: when should you review?

There is no one spaced repetition schedule that fits everyone. In the research, the best gap depended on how long you need to remember. To remember for a week, a review at 20 to 40% of that week worked best; to remember for a year, 5 to 10% of the year. And a review worked far better as recall, with the book closed, than as rereading.

Where do the 1, 3, 7, 14 day schedules come from?

No study we could find set those particular days. Search for a spaced repetition schedule and you get a list of days: a day later, three days, a week, two weeks. Another version runs ten minutes, a day, a week, a month. They often carry Ebbinghaus's name, and the days change from one guide to the next.

That does not make the lists wrong. Starting short and spacing out further is the direction the research points. But "a month later" is not the right gap for everyone, and the studies below show why.

Did Ebbinghaus find that we forget two thirds in a day?

The number is his, but it does not mean what it sounds like. It means relearning took 66% of the time the first learning had, not that two thirds of the list could not be recalled.

Ebbinghaus experimented on himself in the 1880s. He learned lists of nonsense syllables and, after a set time from about twenty minutes to 31 days, learned the same list again. What he measured was not how much he could recall but how much less time the relearning took. This is called savings. He wrote that this saving was the measure of what remained at the end of the interval, and after a day it was 33.7%.

In the same table Ebbinghaus also wrote down what was forgotten: 66.3% after a day, 78.9% after 31 days. That is where "we forget two thirds in a day" comes from: the rest of the savings score, so it too is measured in relearning time. Relearning faster also means something is still there, even when you cannot bring it to mind.

In 2015 Murre and Dros repeated the experiment the same way. One person spent 70 hours learning lists and relearned them after 20 minutes, 1 hour, 9 hours, 1 day, 2 days, 6 days and 31 days, and the results were close to Ebbinghaus's curve. The curve was not entirely smooth: it jumped upward at 24 hours. Earlier researchers had put this down to sleep, but Murre and Dros did not conclude that sleep was the cause.

Both experiments had a single subject learning nonsense syllables. Take the shape of the curve, but do not carry its numbers over to your own study as they are.

What decides the best gap between reviews?

How long you need to remember. The longer that is, the longer the best gap, but the smaller its share of that time.

In 2006 Cepeda and colleagues pooled the research on spaced learning: 317 experiments in 184 articles. In their analysis, the gap that produced the best retention grew as the time you had to remember grew.

Their 2008 study measured this directly. More than 1,350 people learned a set of facts, reviewed them once after a gap of up to 3.5 months, and took a final test up to a year after that. Three things came out.

• As the gap grew, scores first rose and then, past the best point, gradually fell.
• The further away the test, the longer the best gap.
• As a share of the time until the test, the best gap shrank: 20 to 40% for a test a week away, 5 to 10% for a test a year away.

Read this way, "review after a month" suits someone who wants to remember for a year. For a test next week it comes too late. That is why one list cannot fit everyone.

One thing to keep in mind: the study gave a single review. It does not say how to stretch the gaps across several reviews. The direction still holds: the longer you want to remember, the further apart you space.

Rereading or recalling: what should a review be?

Recalling, with the book closed. How you review matters as much as when, and rereading came out ahead only five minutes later. Roediger and Karpicke (2006) had students read a passage, then either read it again or write down what they could recall without looking. On a test five minutes later, rereading did better, 81% to 75%. Two days later it turned around, 54% to 68%, and a week later 42% to 56%. The students who recalled remembered slightly more after a week than those who reread did after two days. Yet the students who reread grew more confident that they would remember.

Karpicke and Roediger (2008) drew the line more sharply with foreign language vocabulary. Once a student had produced a word correctly, studying it again had no effect on later recall, while testing it again had a large one. The students' predictions of their own scores were unrelated to how they actually did.

So on a review day, close the notes and recall before you reread. The same goes for an answer from ChatGPT. Rereading a well written answer raises your confidence, and that confidence says nothing about what you will remember next week.

How to set your review days today

In one line: set the review day by how long you need to remember, and on that day close the book and recall first.

A page in a notebook is enough.

1. Decide how long you need to remember it. A test next week, or something you want to keep a year from now.
2. Set the review to fit that time. In the study above, a single review worked best a day or two later for a one-week goal, and three to five weeks later for a one-year goal. For anything in between, pick a share in between; the longer the time, the smaller the share.
3. On the day, recall first. Write what you remember on a blank page, or ask yourself one question and answer it. Then open the source and fix only what was wrong or missing.
4. If you are a day or two late, review anyway. Past the best gap, scores fell gradually.

Say it is Monday and the test is next Monday. Close the book on Wednesday and see what comes back.

In Aloom

In Aloom you ask AI, and keep only the answers you want in your own knowledge graph. The answers you keep come back each day, one, three or five as you choose, and instead of showing you the answer again, Aloom asks a question so you recall it first. Both ideas in this note, spacing and recall, are in that daily review. The more you recall a node, the darker its ink, so you can see at a glance what has settled. See Aloom

Frequently asked questions

Can I use a spaced repetition schedule template?

Yes. Something that remembers the days for you is useful on its own. Just change the days to fit how long you have until the test or goal, and on each review day recall before you reread.

What is the equation of the forgetting curve?

There is no single one. According to Murre and Dros (2015), Ebbinghaus himself fitted his curve with a power function in 1880 and with a logarithmic function in 1885. The exponential curve in many textbooks is one model among several.

What is active recall?

Bringing what you learned back from memory without looking at it. Writing on a blank page, asking yourself questions and doing practice problems are all active recall. In both studies above, it was the thing that lasted longer than rereading.

References

• Ebbinghaus, H. (1913). Memory: A Contribution to Experimental Psychology, Chapter 7. Trans. H. A. Ruger & C. E. Bussenius. Originally published 1885.
• Murre, J. M. J., & Dros, J. (2015). Replication and analysis of Ebbinghaus' forgetting curve. PLOS ONE, 10(7), e0120644.
• Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.
• Cepeda, N. J., Vul, E., Rohrer, D., Wixted, J. T., & Pashler, H. (2008). Spacing effects in learning: A temporal ridgeline of optimal retention. Psychological Science, 19(11), 1095–1102.
• Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.
• Karpicke, J. D., & Roediger, H. L. (2008). The critical importance of retrieval for learning. Science, 319(5865), 966–968.