If you ask most students how to study, the answer is some version of rereading notes or highlighting a textbook. It feels productive — the material looks familiar, which gets mistaken for actually knowing it. The research says otherwise: actively testing yourself on material, known as retrieval practice, beats rereading by a wide margin, and it's not close.
Why testing beats rereading
Rereading builds familiarity, not recall. When you reread a paragraph, your brain recognizes it — "yes, I've seen this" — and that recognition feels like knowledge. But recognizing something and being able to produce it from memory, unprompted, are different skills. Retrieval practice — closing the book and trying to answer a question about it — forces the second, harder skill, which is also the one you actually need in an exam or a real conversation.
The effect gets stronger over time
One of the more counterintuitive findings is that retrieval practice's advantage grows the longer you wait before being tested again. Cramming with rereading might look identical to cramming with quizzing in the short term, but a week later, the group that quizzed themselves remembers meaningfully more. If you're studying for something that matters weeks from now, not tomorrow, this is the technique that actually pays off.
Turning this into a habit
The practical version of this doesn't require flashcards or a study app you'll abandon in a week. After reading a chapter or sitting through a lecture, write down (or generate) five to ten questions covering the material, close your notes, and answer them from memory. Get one wrong, and you've just found exactly what to reread — which is a far more targeted use of time than rereading everything.
- Test yourself before you feel ready — that discomfort is the part that builds memory.
- Space it out: quizzing yourself once today and again in three days beats quizzing twice today.
- Mix topics instead of drilling one at a time; the difficulty is part of what makes it work.
- Don't just reread wrong answers — try the question again a few minutes later.
Why rereading feels like it is working
The uncomfortable part of this is not that testing works. It is that rereading feels better while working less well, and the feeling is the problem. Each pass over a page is easier than the last, and the brain reads that increasing ease as increasing knowledge. It is not. It is recognition, which is the sensation of having seen something before, and it disappears the moment the page is taken away.
Retrieval feels worse for the same reason it works. Pulling an answer out without the text in front of you is effortful and you notice the gaps. Noticing a gap is unpleasant, and it is also the entire mechanism: you cannot repair a hole you did not know was there.
Spacing: the cheapest improvement available
The same number of questions does more work when spread out than when stacked together. A quiz at the end of a topic and again a week later beats two quizzes on the same afternoon, because the second one comes after some forgetting has begun and forces a real retrieval rather than an echo.
The practical version for a teacher is small and repeatable: three or four questions at the start of each lesson, drawn from the lesson before and one from a few weeks back. It costs five minutes and does more than a revision session at the end of term, because the term-end session is one long stack with no gaps in it.
Low stakes or no stakes
This only works if being wrong is safe. A learner who fears the score will avoid the risk, and avoidance is the opposite of retrieval — they will answer only what they are certain of, which practises the things they already know. Ungraded quizzes, or ones where only the learner sees their own result, get more honest answers and therefore do more good.
Anonymity helps in a group setting for the same reason. The scoreboard is a motivator for the people at the top of it and a discouragement for everyone else, which is worth remembering before turning one on in a class where the range is wide.
Three ways this goes wrong
- Only quizzing at the end. A test used to measure learning is assessment. A test used to cause learning has to come before you would think to measure anything, and repeatedly.
- Only recognition questions. Multiple choice with an obvious answer is close to rereading, because the answer is on the screen. Free-text answers, or options that are genuinely plausible, force the retrieval that does the work.
- No feedback, or feedback too late. Getting a question wrong and learning the right answer a week later is two separate events. Learning it immediately is one, and it is the one that sticks.
This is also exactly why a lot of teachers build a short quiz into the last five minutes of class instead of a recap slide: the recap feels like review, but the quiz is what actually sticks.