The most useful thing you can do with a chapter you just finished is close it and write five questions about it. Not highlight it, not summarize it: write the questions. Writing a question forces you to decide which idea carried the weight and what a real test of it would look like, and that decision is most of the work of understanding. Then you answer them a day or two later, cold, and whatever you miss tells you precisely what to restudy. The whole loop costs about ten minutes a session.
Why the writing half is the part that teaches you
Answering someone else's questions is useful. Writing your own is better, and most people skip straight past it to a premade question bank. To write a question you have to hold the idea in your head, pick the part of it that actually matters, and invent a wrong answer that would still tempt someone who half-understood. That is producing something rather than reviewing something, which is the difference the generation effect describes.
There's a second payoff. A question you wrote reveals what you thought was important, and comparing that against the material later catches misreadings that no amount of rereading would have surfaced.
Write them with the material closed
With the book open, your questions trace the page. They follow the order of the headings, they quiz the sentences sitting in front of you, and they come out shaped like the text instead of shaped like the subject. Closed book, they come out of what you retained, and the holes in your set are information all by themselves. If you can't write one question about the third section, you don't have the third section yet, and that's worth knowing before an exam rather than during one.
Give yourself four or five minutes right after reading, page closed, and write whatever comes. Then open the material and fix the questions that turned out to be about something you misremembered.
Four kinds of question worth writing
- Definition in your own words: "What is X, without using the phrasing the book used?" Cheap to write, and worth one or two per topic at most.
- Mechanism: "Why does X lead to Y?" This is the one that catches material you've memorized without understanding, because a definition can be recited with no idea what drives it.
- Discrimination: "What separates X from Y, and which one applies in this case?" Most exam errors are confusions between two neighboring ideas rather than blanks.
- Prediction: "What happens if I change this one input?" Flip a sign, drop a condition, double a value, remove an assumption, then ask what breaks and why.
A good set leans on the last three. If every question you wrote opens with "what is," you've written a vocabulary list, and vocabulary lists are the easiest thing in the world to pass while understanding nothing.
Two mistakes that make your questions useless
Questions you can answer by recognizing a word
If the phrasing of your question contains enough of the answer that the right response pops out on sight, you'll pass it every time and learn nothing from passing. Strip the giveaway terms out. "Why does spacing reviews out help retention more than repeating them back to back?" is a real question. "Why is spaced repetition, which spaces reviews out over time, effective?" is a sentence with a gap in it.
Questions that quiz the page instead of the idea
"What are the four stages listed in section 2.3?" tests whether you remember a layout. Nobody outside that book will ever ask it, and it'll be gone in a week anyway. Ask instead what the stages are for, what happens if one is skipped, or which stage a given example belongs to.
Answer them later, not now
Answering immediately after writing tests nothing, because the answer is still sitting in working memory where you just put it. Wait at least until the next day. Waiting long enough that recalling takes visible effort is the point, not a side effect: the effortful retrieval is what strengthens the memory, and an answer that comes back instantly did nothing for you.
Keep the questions and answers physically apart. Questions on one page, answers on another, or answers written only when you sit down to test yourself. If the answer is visible while you read the question, your brain will take the free ride every time.
The ten-minute routine
- Finish the chapter or the session and close the material.
- Write five questions from memory, no peeking, in about four minutes.
- Open the material and correct any question that's based on something you misremembered. That correction is free learning.
- File them somewhere you'll actually find them, with the answers kept separate.
- Answer them cold the next day, in writing, before looking anything up.
- Restudy only what you got wrong, and write one new question about that specific thing.
- Requeue the misses for a few days later, and retire a question once you've answered it correctly twice from cold.
Where this fits with a TopicLearn lesson
A TopicLearn course plans active recall and spaced review into the lesson sequence from the start, with exercise formats matched to the subject: a code runner for programming, drag and match for terminology, recall checks and written reflection for conceptual material. The recall checks deliberately don't hand you a grade, because the point is finding the gap rather than scoring the attempt. Writing your own questions on top of that is still worth doing, especially for material you brought yourself, because the act of writing them is a different exercise from answering them.
If you can't write a single question about a section, you don't know that section yet. The empty spot in your question list is the finding.