Forgetting curve simulator
The forgetting curve shows how memory fades after you learn something. It falls fast at first and then more slowly. This simulator shows how a few reviews change that, using FSRS, the model Winda uses to schedule reviews. Pick a number of reviews under the chart to see its curve.
- Uses the FSRS-6 model
- No account needed
- Runs in your browser
96%chance you still remember something you learn today a year from now if you review it 4 times.
Each review comes as your chance of recall falls to 90%, on day 2, day 13, day 59, and day 222.
The curves come from FSRS-6 with its default parameters. The simulator assumes you remember it at every review. github.com/open-spaced-repetition/awesome-fsrs/wiki/The-Algorithm
Winda keeps this schedule for every card.
You make flashcards or questions from what you read. Winda brings each one back when your chance of recalling it falls to 90%. Each answer moves the next review further out.
The original forgetting curve
Hermann Ebbinghaus published the first study of forgetting in 1885. He learned lists of made-up syllables, waited, and then learned them again. He measured how much time the second learning saved.
| 0 learned | 100% |
|---|---|
| 20 minutes | 58% |
| 1 hour | 44% |
| 8.8 hours | 36% |
| 1 day | 34% |
| 2 days | 28% |
| 6 days | 25% |
| 31 days | 21% |
Savings is the share of the first learning time that he no longer needed. It shows how much of a list was still there. It does not show how much of the list he could recite.
Ebbinghaus, H. (1885). Über das Gedächtnis. Values as given by Murre, J. M. J., & Dros, J. (2015). Replication and analysis of Ebbinghaus' forgetting curve. PLOS ONE, 10(7). doi.org/10.1371/journal.pone.0120644
Questions
The forgetting curve is a graph of how memory fades after you learn something. Hermann Ebbinghaus published the first one in 1885. The curve falls fastest in the first hours and days. After that it flattens out.
There are several. A common textbook form is R = e^(−t/S), where R is the chance of recall, t is the time since you learned it and S is how stable the memory is. FSRS-6 uses a power curve instead, which falls more slowly over long gaps. In FSRS, stability is the number of days until your chance of recall falls to 90%.
The two curves measure different things. Ebbinghaus learned syllables with no meaning, and he timed how long relearning took. FSRS predicts whether you can answer a flashcard that you once got right. FSRS has been tested on about 727 million reviews from 10,000 Anki users. Material that you understand fades more slowly than nonsense syllables.
The best time to review is just before you would forget. With a 90% target and the default FSRS parameters, a fact that you got right on the first try needs reviews on about days 2, 13, 59 and 222. Those four reviews keep it near 90% for a year. Each gap is several times longer than the one before. One year after learning, the chance of recall is about 96% with those reviews and about 46% with none.
Cramming works for a test the next day. Reviews on the same day add little to each other, though. In the FSRS model, four reviews in one evening count about as much as one. What you crammed fades soon after the test.
FSRS stands for Free Spaced Repetition Scheduler. It is an open-source model that predicts how likely you are to remember each card today. Anki added it in version 23.10. Winda uses it to schedule every card.
Related
A reader that helps you remember.
Bring what you are studying. Read it with AI beside you, and let the questions come back to you on a schedule.
Free to start · No credit card needed