What Is Active Recall? Measured Across 299,120 Retrievals
Active recall means pulling the answer from memory before checking it. Using 299,120 retrieval records, we put a number on the gap between looking familiar and actually knowing.
What Is Active Recall? Measured Across 299,120 Retrievals
Reading a word list three times, or covering the translation and saying the word yourself — which one works?
The difference is not how long you spend. It is whether the memory actually gets retrieved. This article explains what active recall is, along with one concrete number from 299,120 retrieval records: one in every 6.6 retrievals lands in the state of "this looks familiar, but I cannot produce it."
What active recall is
Active recall means pulling the answer out of memory first, then checking it against the correct one. Its opposite is passive review — reading the textbook, the notes, or the word list again.
The two differ by the order of a single action:
| Passive re-reading | Active recall | |
|---|---|---|
| Order | You see the answer immediately | You retrieve first, then check |
| What the brain does | Recognition (I have seen this) | Retrieval (this means…) |
| How it feels | Smooth, easy | Effortful, sometimes blank |
| Do you learn whether you know it | No | Yes, immediately |
Common forms of active recall include flashcards, self-testing, covering the answer and reciting, and writing from memory with the book closed. What they share is that the answer does not appear until you have produced it.
Why re-reading misleads you
Seeing a word, finding it familiar, feeling that you know what it means — that feeling is familiarity.
Familiarity and retrieval ability are two different things. The more often you see something, the stronger the familiarity, but rising familiarity does not mean rising retrieval ability. That is why you can go through a vocabulary book five times and still draw a blank in the exam: five passes build recognition, and the exam asks for retrieval.
Active recall does two things at once. The act of retrieval strengthens the memory, and it tells you on the spot whether you actually know the item.
The data: 299,120 retrievals, four outcomes
On Tomaru, every card is answered the same way: try to recall, reveal the answer, then pick one of four options describing what just happened.
| Item | Value |
|---|---|
| Period | 13 April – 23 August 2026 |
| Retrievals | 299,120 |
| Users | 764 |
| Cards | 32,121 |
All figures are anonymised aggregates and contain no personal data.
How the four outcomes are distributed:
| What the user selected | Count | Share |
|---|---|---|
| Forgot (completely blank) | 56,547 | 18.9% |
| Fuzzy (came up, but unsure) | 45,359 | 15.2% |
| Clear (retrieved it) | 106,053 | 35.5% |
| Know it (barely had to think) | 91,161 | 30.5% |
34.1% of retrievals were not clean successes. Of those, 18.9% were completely blank, and another 15.2% landed on Fuzzy.
Fuzzy is the point of this article. Under passive re-reading, that entire 15.2% would have been counted as "I know this" — you see the answer, it looks right, you turn the page. Only by attempting retrieval first do you find out that you could not actually produce the word.
What Fuzzy predicts
Joining each retrieval to the next retrieval of that same card shows what the current state predicts:
| State this time | Sample | Next review after | Next time recalled clearly | Next time completely blank |
|---|---|---|---|---|
| Forgot | 52,917 | 0.17 days | 50.1% | 34.0% |
| Fuzzy | 37,124 | 1.48 days | 56.5% | 11.7% |
| Clear | 80,450 | 2.77 days | 80.4% | 9.7% |
| Know it | 39,925 | 7.38 days | 82.0% | 11.2% |
Read the third and fourth columns together.
The scheduler brings Fuzzy cards back quickly (1.48 days on average) and makes Know it cards wait a long time (7.38 days on average). In other words, Know it faces a test roughly five times harder — and yet it comes back clearly 82.0% of the time, while Fuzzy manages only 56.5%.
Even when a Fuzzy card is retested after just a day and a half, 43.5% of the time it does not come back cleanly. Familiarity does not survive two days.
Turning re-reading into active recall
1. The answer has to be genuinely hidden. A bilingual word list cannot support active recall, because both sides are in view at once. Cover one side, or use cards.
2. Produce first, then check. A vague sense that "it is something like give up" is not enough — you need the full meaning. Saying it aloud or writing it down is far stricter than thinking it silently.
3. When it does not come, wait three seconds before revealing. A failed retrieval attempt still has value, provided you genuinely tried. Flipping straight to the answer puts you back in re-reading.
4. Mark Fuzzy honestly as Fuzzy. This is where it is easiest to let yourself off. Mark it Clear and the card gets pushed a week out; by then it will most likely be blank, and the whole loop was wasted.
FAQ
What is active recall?
Active recall is retrieving the answer from memory first and only then checking it against the correct one, as opposed to passive review where you simply re-read the textbook or word list. The difference is whether the answer appears before you have produced it. Common forms include flashcards, self-testing, covering the answer and reciting, and writing from memory with the book closed.
How is active recall different from passive review?
Passive review exercises recognition — the content looks familiar. Active recall exercises retrieval — you produce the content with no prompt. Exams, conversation and writing all draw on retrieval. Our data shows 15.2% of retrievals land in the Fuzzy state: the answer looks right when revealed, but the learner could not produce it. Passive review never surfaces that category.
Why can I read something many times and still not remember it?
Because re-reading accumulates familiarity, not retrieval ability. The more passes you make, the stronger the sense of knowing when you see the word — but that feeling does not mean you can produce the meaning unprompted. In our data 34.1% of retrievals are not clean successes, and 15.2% of them are exactly this familiar-but-cannot-produce state.
Is active recall really more effective?
Yes. It does two things at once: the act of retrieval strengthens the memory, and it exposes on the spot which items you do not actually know. Our data shows cards marked Fuzzy still fail to come back cleanly 43.5% of the time even when retested after only 1.48 days, while cards marked Know it come back clearly 82.0% of the time after a 7.38 day gap.
How do I practise active recall?
Cover the answer, produce a complete answer yourself, then check. Four points: the answer must genuinely be out of sight; say it aloud or write it rather than thinking it silently; pause a few seconds before revealing when nothing comes; and mark your state honestly — if it was fuzzy, mark it fuzzy.
How is active recall different from spaced repetition?
Active recall decides what you do during a review; spaced repetition decides when the review happens. They are complementary: active recall on a badly timed schedule wastes effort on material you already know, while a perfect schedule filled with re-reading means doing the ineffective thing at exactly the right moment.
Does recalling silently in my head count?
It counts, but it is weaker. A vague mental impression easily gets graded generously, because a fuzzy trace auto-completes the moment it meets the correct answer. Saying it aloud or writing it forces a complete answer and leaves less room to overestimate yourself.
Should I reveal the answer as soon as I am stuck?
Pause a few seconds first. Revealing after a failed retrieval attempt is still more effective than revealing straight away, because you have confirmed the gap. But if you flip within a second every time, what you are actually doing is re-reading.
How much active recall should I do per day?
There is no fixed number. What matters is that each item gets retrieved shortly before you would have forgotten it, not that you do many at once. Testing 30 words a day at the right moments beats testing 200 words once and then not touching them for three weeks.
The data in this article comes from anonymised retrieval records on Tomaru. Further reading: What is the testing effect, What is the forgetting curve.