How do I remember what I read?

Be asked about it without the page in front of you, at increasing intervals, starting before you have forgotten it. Re-reading and highlighting alone do not work because they are recognition, not retrieval — the page is doing the remembering for you.

Why do I forget what I read so quickly?

Because reading is recognition. The words on the page supply the answer, so your brain never practises producing it. Without retrieval, most of what you read decays within days — which is the ordinary forgetting curve, not a personal failing.

Is highlighting useless?

Highlighting is useless as a way of learning and useful as a way of marking. The mark records a decision about what mattered; the learning happens only if something later asks you about the marked passage with the book closed.

How many times do I need to review something?

Fewer times than you would expect, if the reviews are spaced correctly. A handful of well-timed retrievals over months holds material far better than many reviews crammed together, because each successful recall at the edge of forgetting lengthens the next interval.

What is spaced repetition?

Scheduling each review for the moment just before you would forget, then pushing the next one further out each time you succeed. Modern schedulers such as FSRS estimate that moment per item rather than applying one fixed ladder to everything.

How to remember what you read

You remember what you read by being asked about it without the page in front of you, at increasing intervals, starting before you have forgotten it. Everything else — highlighting, re-reading, summarising, notes — is preparation for that, and none of it is a substitute.

Why re-reading fails

Re-reading feels like learning because it produces fluency: the second pass is smoother than the first, the sentences arrive already familiar, and familiarity is very easily mistaken for knowledge. It is not knowledge. It is recognition, and recognition collapses the moment the page is taken away.

The test is simple and slightly unpleasant. Close the book and say what the section argued. Most people who have just re-read a chapter twice cannot, and the gap between how well they expected to do and how well they did is the whole problem — it is why the strategy survives despite not working.

The method, in four steps

1. Read once, properly. Do not take notes as you go. Notes taken during reading are usually transcription, and transcription is the thing that feels most like work while teaching least.

2. Mark what matters, and nothing else. A highlight is a decision, not a study act. If you are marking more than a few passages a page, you are not deciding — you are decorating.

3. Get asked about each mark, from memory. This step is the study, and it is the step people skip. Pulling the answer back before you look is what makes it stay - the testing effect, one of the best-replicated findings in learning research. Writing the question yourself helps a little more, if you actually do it; most people don't, so a prompt made from your own marked sentence beats a perfect one never written.

4. Get asked at increasing intervals. Each successful recall should push the next attempt further away. Reviewing something you still know well is wasted effort; reviewing it after you have forgotten it is re-learning it from scratch. The value is in the narrow window between.

Why the interval matters more than the effort

Forgetting is not random. It is regular enough to be modelled, which means it is regular enough to be interrupted on purpose. That is the entire idea behind spaced repetition: rather than reviewing everything on a fixed schedule, you estimate when each individual item is about to slip and ask about it then.

Modern schedulers such as FSRS do this per item, using your own history of right and wrong answers rather than one ladder applied to everything. The practical consequence is that a few correctly-timed retrievals hold material across months, while the same number of reviews crammed into a week does not survive the term.

Doing it without an app

You can run all of this on paper, and it is worth knowing that. Write each question on a card. Keep boxes for daily, every-three-days, weekly, monthly. A card you get right moves up a box; a card you get wrong goes back to the first. This is the Leitner system, it is decades old, and it works.

Two things go wrong in practice. The card-making gets skipped, because it is a separate session that happens after the reading. And the boxes drift, because hand-scheduling dozens of items across months is genuinely tedious. Software is worth using for exactly those two failures and not for much else.

How FlowRecall runs this loop

FlowRecall exists to close the first failure. You read your own PDF or EPUB inside it, and when something matters you write it as a card on the spot — two lines, a question and its answer, in your own words, with no separate session and no re-reading the chapter later to work out what was worth asking. Writing the question is part of the work: you chose what matters, and being asked about it is the part that works.

It closes the second failure by running FSRS-6 on your device, so the interval is computed per item and works offline. What you are left with is the method above, with the two pieces that people reliably abandon taken off your hands.

Common questions

How do I remember what I read?: Be asked about it without the page in front of you, at increasing intervals, starting before you have forgotten it. Re-reading and highlighting alone do not work because they are recognition, not retrieval — the page is doing the remembering for you.
Why do I forget what I read so quickly?: Because reading is recognition. The words on the page supply the answer, so your brain never practises producing it. Without retrieval, most of what you read decays within days — which is the ordinary forgetting curve, not a personal failing.
Is highlighting useless?: Highlighting is useless as a way of learning and useful as a way of marking. The mark records a decision about what mattered; the learning happens only if something later asks you about the marked passage with the book closed.
How many times do I need to review something?: Fewer times than you would expect, if the reviews are spaced correctly. A handful of well-timed retrievals over months holds material far better than many reviews crammed together, because each successful recall at the edge of forgetting lengthens the next interval.
What is spaced repetition?: Scheduling each review for the moment just before you would forget, then pushing the next one further out each time you succeed. Modern schedulers such as FSRS estimate that moment per item rather than applying one fixed ladder to everything.