mytetz › Study guides › When to review what you read
When to review what you read
Put a gap between your reviews, and make each gap longer than the last. A practical schedule is one day, one week, then one month. A gap beats a repetition in the same sitting: across 254 studies, spaced study gave 47.3% correct recall and massed study gave 36.7%.
Key points
- A gap between two reviews beats two reviews together.
- The best gap grows as the delay to your test grows. One fixed schedule fits nobody.
- Ebbinghaus measured savings in relearn time, and not recall. One person produced the data.
- Review only what you failed to recall. Do not re-read the whole section.
What is the forgetting curve, and what did Ebbinghaus measure?
- The forgetting curve
- A plot of memory against time after study. The loss is fast in the first hours. The curve then flattens.
Hermann Ebbinghaus collected the original data on himself in 1879 and 1880, and published it in 1885. Two details matter, and most articles get them wrong.
- He measured savings, not recall. He learnt a list of nonsense syllables, waited, then learnt the same list again. The savings score is the share of time that the second learning saved.
- One person produced the data. The subject was Ebbinghaus himself. The material was nonsense syllables, and not prose.
His savings score fell from 58% after 19 minutes to 34% after one day. It then fell slowly, and reached 21% after 31 days. The shape is the useful part: the steep loss happens early, so the first review belongs early.
Does the curve replicate?
Yes, in shape. Murre and Dros repeated the experiment and published the result in PLOS ONE in 2015. Their subject spent about 70 hours on the task. The authors call the resemblance between the curves remarkable.
| Delay | Ebbinghaus (1880) | Dros (2015) |
|---|---|---|
| 20 minutes | 0.582 | 0.472 |
| 1 hour | 0.442 | 0.373 |
| 9 hours | 0.358 | 0.276 |
| 1 day | 0.337 | 0.317 |
| 2 days | 0.278 | 0.230 |
| 6 days | 0.254 | 0.168 |
| 31 days | 0.211 | 0.041 |
Read the table with the same care. The 2015 replication also used one subject. Two earlier replications, by Mack and by Seitz, each used one subject too. The 31-day value of 0.041 sits far below the other three studies, and the authors give no explanation for it.
The authors also report one deviation from a smooth curve: an upward jump that starts near the 24-hour point. They link it to sleep, and they state that this link remains to be established.
What schedule do you use?
No paper gives one correct schedule. Cepeda and colleagues found that the best gap grows as the delay to the test grows. So match the gap to your goal, and make each gap longer than the last.
| Goal | First review | Second | Third |
|---|---|---|---|
| A test next week | Same day, in the evening | After 2 days | After 5 days |
| A test next month | After 1 day | After 1 week | After 3 weeks |
| Knowledge that you keep | After 1 day | After 1 week | After 1 month, then every few months |
A review is a test, and not a re-read. Cover the material. Produce the answer. Then check. The guide on how to test yourself while you read gives the method.
What do you review, and what do you skip?
- Open your marks from the last session. These are the points that you failed to recall.
- Answer each one from memory, before you open the source.
- Keep a mark on any point that you fail again. Remove a mark after two clean recalls.
- Skip every point that you recalled correctly twice. Time on those points buys little.
What does the research show?
Spaced study against massed study
Cepeda, Pashler, Vul, Wixted and Rohrer combined 254 studies with 14,811 participants in 2006. Spaced study produced 47.3% correct recall. Massed study produced 36.7%. That is a difference of about ten percentage points, from the same total study time.
The same review gives the limit that most articles drop: the optimal gap is not fixed. It grows with the interval between study and test. A gap that suits an examination in one week is too short for knowledge that you want in one year.
Why a gap is rated so highly
In 2013 a review of ten study techniques placed distributed practice in the top group, with practice testing. The two work well together: a spaced schedule decides when, and a retrieval decides what you do at each point.
Sources
- 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. doi:10.1037/0033-2909.132.3.354
- Murre, J. M. J., & Dros, J. (2015). Replication and Analysis of Ebbinghaus' Forgetting Curve. PLOS ONE, 10(7): e0120644. doi:10.1371/journal.pone.0120644
- Ebbinghaus, H. (1885). Über das Gedächtnis. Leipzig: Duncker & Humblot. English translation: Ruger & Bussenius (1913), Memory.
- Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Psychological Science in the Public Interest, 14(1), 4–58. doi:10.1177/1529100612453266
Common questions
Is the forgetting curve a solid group result?
No. One person produced the original data, and the 2015 replication also used one person. The shape repeats across those studies, so use the shape. Do not quote a percentage as a general human average.
Do I need a spaced-repetition application?
No. Software schedules the reviews for you, which helps with hundreds of items. A calendar entry and a list of your marked gaps do the same work for a few topics.
Is a longer gap always better?
No. A gap that is too long leaves you with nothing to retrieve. Set the gap so that recall feels difficult but possible.
Can I review by reading my notes again?
That is a re-read, and the 2013 review rated it in the lowest group. Cover the notes and produce the answer first. Read them afterwards, to check.
Where do you go next?
- How to test yourself while you read — what you do at each review.
- How to study on your own — the loop that this schedule completes.
Start with a topic
The memory research is itself a topic in the catalogue. Read it, then apply the schedule to it.
Open the catalogue