You remember what you read by deciding in advance what the text is for, capturing the parts that matter in a form you can be tested on, and reviewing them on a spaced schedule. Rereading feels productive and does little. The comprehensive review by Dunlosky and colleagues rated practice testing and distributed practice as high-utility techniques, and rereading as low-utility.
Creator disclosure: Heni Hazbay creates Summarise Visually and has a direct interest in the app. The learning guidance below comes from published research, not from the app.
Publication and evidence note: This guide was prepared from the cited sources and reviewed project evidence. Its four-stage method, capture block, and review schedule are editorial tools, not a validated study protocol or a claim about your results.
Why you forget, and what that tells you
Forgetting is not a failure of attention alone. Three ordinary things cause most of it.
You read without a question. Text you have no use for has nothing to attach to, and it decays quickly.
You read for coverage instead of retrieval. Finishing a chapter records that you were present; it does not record that you can produce the content later.
You review by rereading. Familiarity rises, so confidence rises, while the ability to recall the material without the page in front of you does not. Dunlosky’s review found exactly this pattern: rereading produces a sense of fluency that overestimates learning.
One more factor applies to screens specifically: Nielsen Norman Group’s research found that people scan web pages rather than read them word by word. If you are reading on a phone or laptop, assume scanning is your default and design against it.
This guide covers the reading-to-capture side. Once you have checked material, the retrieval mechanics — building prompts, answering them cold, and handling feedback — are covered in key points and Q&A for study.
Stage 1: set the question before you read
Write the question the text is supposed to answer, in one line, before reading. “Read chapter six” is coverage. “What conditions does the author say make the model fail?” is a question, and it changes what your attention does with every paragraph.
If you cannot form a question, skim the headings, abstract, or conclusion first and form one. That is not cheating; it is the reason experienced readers of research start at the end, as reading a research paper efficiently sets out.
Stage 2: capture in a testable form
Capture as you read, in a form that can be turned into a question later. Cornell’s note-taking system is built for this: a narrow cue column for questions beside a wider column for notes, with a summary line underneath, so the notes can be covered and the cues used as prompts. Other shapes of material suit other formats, compared in note-taking methods.
A capture block that works for reading of any kind:
Question this text answers:
Claim (in my words):
Why it holds (evidence, mechanism, or argument):
Conditions or exceptions:
Where it lives (page, section, timestamp):
What I could be asked about this:The last line is what converts notes into review material. If you cannot write a question about a note, you probably recorded a phrase rather than an idea.
Stage 3: close the book and retrieve
Within a day, answer your own questions without the text visible. Write or say the answer in full, then check it against the source and mark what was wrong or missing.
Two rules keep this honest. Answer before you look, because reading the answer first converts practice testing back into rereading. And check against the source rather than your notes, because an error in the notes otherwise gets rehearsed rather than corrected.
Stage 4: space the reviews
Distributed practice — spreading study over time rather than massing it — was the other high-utility technique in Dunlosky’s review. A workable default for reading you intend to keep:
| Review | When | What you do |
|---|---|---|
| 1 | Same day | Answer your capture questions cold |
| 2 | 2–3 days later | Answer only the ones you got wrong or hesitated on |
| 3 | About a week | Answer all of them, then summarize the whole text from memory |
| 4 | Before it is needed | Answer the weakest third |
The intervals are a starting point, not a measured optimum. Adjust from your own error rate: material you keep missing needs shorter gaps.
What to do with a summary
A summary is useful at two moments and useless at a third.
Before reading, a summary orients you and helps you form the question.
After reading and checking, a summary is a compact review object — the thing you reread quickly before you attempt retrieval.
Instead of reading, a summary gives you the feeling of knowing without the material to retrieve. This is the substitution that produces confident, empty recall in an exam.
Using Summarise Visually in this method
Current project review supports six summary modes and on-device summary history, so a checked summary stays available for later review. Two placements fit this method: an orientation summary before you read a long source, and a checked review object afterwards.
What the app cannot do is the part that produces retention. Generating a summary is not retrieval, reading it again is not practice testing, and no mode establishes that you have learned anything. Keep the questions, the cold attempts, and the corrections yourself — using AI to make study notes sets out the prepare, generate, check, retrieve sequence in full.
Frequently asked questions
Why do I forget what I read almost immediately?
Usually because the reading had no question attached and was never retrieved. Material that is read once for coverage and reviewed by rereading is not practised in the way that supports later recall.
Does rereading help at all?
A single rereading can help comprehension of difficult text, but as a study strategy Dunlosky’s review rated it low-utility compared with practice testing and distributed practice.
How soon should I review after reading?
Attempt retrieval the same day, then space the next reviews out over days and weeks, shortening the gaps for material you keep missing.
Do I need to take notes to remember what I read?
You need something you can be tested on. Notes are the common form; questions written in the margin or a cue column work too. Highlighting alone gives you nothing to attempt.
Is reading a summary a substitute for reading the source?
No. A summary supports orientation and review. Used instead of the source, it produces familiarity without the detail, conditions, and evidence that questions actually test.
Source and evidence note
The relative utility of practice testing, distributed practice, and rereading comes from Dunlosky and colleagues’ review in Psychological Science in the Public Interest. The cue-column and summary structure comes from Cornell University’s note-taking system. The finding that readers scan on screens comes from Nielsen Norman Group. All three were accessed on July 25, 2026.
The four-stage method, capture block, and review schedule were created for this guide by applying that published guidance to reading. The intervals are editorial defaults, not measured optima, and no learning outcome is claimed for them. Product statements reflect current project review; no retention study was conducted here.
Limits and verification
Retention depends on the material, your prior knowledge, and the demands of the assessment, so no schedule transfers unchanged between subjects. Treat the tables here as starting points and adjust from your own error record.
AI assistance can produce notes and questions quickly, but a question generated from a summary can test a detail the source does not emphasise, or one it does not contain. Check generated study material against the source with the AI summary accuracy checks, and follow the AI-use rules that apply to your course.
Publication was authorized by Heni Hazbay after review of the cited sources and current project evidence.