What the comparison actually shows
The standard experiment: everyone reads a passage. One group rereads it. Another group puts it away and writes down what they remember. Both groups are tested later. Immediately afterwards, the rereaders often do slightly better — which is exactly why the myth persists. After a delay of a few days to a week, the retrieval group wins clearly, and the gap widens the longer you wait.
| Rereading | Retrieval practice | |
|---|---|---|
| How it feels during | Smooth, productive, encouraging | Effortful, halting, discouraging |
| Confidence afterwards | High — often higher than justified | Low — often lower than justified |
| Recall after a week | Poor, and worse the longer you wait | Substantially better, and the gap grows |
| Diagnoses your gaps | No — you can't notice what you didn't notice | Yes, immediately and precisely |
| Transfers to new question wordings | Weakly | Well — you practised producing, not recognising |
The fluency illusion, and why it's so convincing
The second time you read a paragraph, it flows. Your brain interprets that ease as evidence of knowledge, because in ordinary life ease usually is. It's a reasonable heuristic that happens to be catastrophically wrong here: the ease comes from the text being on the page in front of you, and the exam will not have the text on the page in front of you.
This is why students walk out of exams genuinely shocked. They were not lying about feeling prepared. Their monitoring was accurate about familiarity and blind about retrievability, and nothing in the rereading experience distinguishes the two. Highlighting makes it worse by adding a physical act that feels like work — see why highlighting doesn't work.
Why retrieval physically does more
Retrieval is not just a measurement of memory, it modifies it. Dragging a fact out of storage strengthens the route back to it, and the harder the drag, the more it strengthens — up to the point where you fail entirely and learn nothing. That is the whole of the desirable difficulties principle: effort that you can just about complete is the mechanism, not the price.
Rereading provides no drag. The information is supplied, so no route is travelled and nothing is strengthened. You are, in a real sense, watching someone else know the material.
How to convert a rereading habit into a retrieval one
- 1
Read the chapter once — genuinely once
The first read is necessary; it's the second, third and fourth that are wasted. Give yourself permission to move on while feeling like you haven't got it, because you haven't, and rereading won't fix that either.
- 2
Close it and write what you remember
Blank page, ten minutes, no looking. It will feel humiliating the first few times. That feeling is the difference between what you recognise and what you know, and you were previously paying for it in the exam hall instead.
- 3
Check, and mark only what was missing
Crucially: don't reread the parts you got right. That's the moment the old habit reasserts itself. Only the gaps get attention.
- 4
Turn each gap into a question
Not a highlighted line — a question with a single answer. A highlighted line can only be reread. A question can be attempted, which means it can strengthen memory rather than describing it.
- 5
Re-attempt tomorrow, then in three days, then in a week
Retrieval plus spacing compounds; either alone is much weaker. Two spaced re-attempts turn a good afternoon into a memory that survives to the exam.
When rereading is the right call
The honest exceptions, because there are some:
- First encounter with genuinely difficult text. You cannot retrieve what never went in. One careful read of a dense proof or a legal judgment is not the enemy; the fourth is.
- Material you'll never be tested on but need to follow today — a paper for a seminar discussion in an hour.
- Immediately before a test, for a small set of facts. Massed rereading does produce short-lived recall, which is fine if the horizon is two hours.
- Text whose structure you're trying to learn, as opposed to its content — the shape of an argument, the layout of a mark scheme.
What unites the exceptions: short horizons or first exposure. For anything you need in a month, rereading is time transferred from a method that would have worked.
The uncomfortable implication
If the better method feels worse, then your sense of how well a study session went is systematically anti-correlated with how much you learned. A session that felt smooth and satisfying was probably rereading. A session that felt like failure was probably retrieval doing its job.
This is not a reason to seek suffering — struggling with material far beyond you is just failure, with no upside. It's a reason to stop using "that felt productive" as your quality signal and start using something measurable instead: how much of the blank page you filled this week compared to last.