Stop Highlighting and Rereading: What Actually Works Is Called Active Recall
Think back to how you studied in school.
Open the textbook, run a highlighter over the key sentences, then read the chapter again. And again. Close the book feeling reassured — you studied.
If that's you, the next part stings a little. Over the past century, cognitive science has put every mainstream study technique through the lab, and rereading and highlighting land reliably in the bottom tier.
A study-technique ranking that pulls no punches
In 2013, John Dunlosky of Kent State University and several colleagues published a review in Psychological Science in the Public Interest. They took on a thankless job: gather the ten most common study techniques — highlighting, rereading, summarizing, self-explanation, practice testing, and so on — weigh each one against the actual experimental evidence, and sort them into three tiers.
The results were awkward. The two most effective techniques: practice testing and distributed practice. The bottom tier: highlighting, rereading, and summarizing — precisely the three with the most users.
That top-ranked "practice testing" sounds fancier than it is: retrieve the material from memory without looking at it. Getting a quiz question right, answering a flashcard, explaining a concept to a friend — all of it counts. Its plainer name is active recall.
Why rereading fails: it's designed to fool you
Rereading's core problem is that it's comfortable.
By the third pass through a paragraph, of course it feels like "I know this." The familiarity is real, but it only proves you recognize the text — not that you can produce it. Then the exam comes, or real life does, the page isn't in front of you, and your mind goes blank. Recognizing a face and recalling a name have always been different skills.
Active recall runs the opposite way. It's unpleasant, effortful, and easy to avoid. The seconds you spend stuck, trying to restate a concept with the book closed, are exactly when your brain is rebuilding the memory. Robert and Elizabeth Bjork coined a term for this family of effects: "desirable difficulties." The more effortful the retrieval, the more firmly the memory gets embedded.
Put bluntly: in studying, comfort and actual learning tend to move in opposite directions.
How to use it in everyday reading
The principle is simple; the discipline is the hard part. A few things we do ourselves:
Close it first, then restate. Ask yourself, "what was that about?" Whatever you can restate is yours; for whatever you can't, go back and fill the gaps. Don't settle for "I got the gist."
Rewrite key points as questions, and answer them a day later. "What are FSRS's three parameters?" beats rereading "FSRS has three parameters" ten more times. Questions are hooks for memory; declarative sentences are not.
Explain it to someone. No one around? Talk to an empty room. Teaching is the most brutal self-test there is — wherever you stumble is precisely where you didn't understand.
Use flashcards, but don't handcraft them. Cards are a ready-made active-recall tool, but the tedium of card-making is where traditional flashcard users drop out. We went into that in the previous post on spaced repetition.
It pairs naturally with spaced repetition
Active recall answers how to review; spaced repetition answers when. Conveniently, those are the two techniques at the top of Dunlosky's ranking.
Put them together: at the moment you're about to forget (spaced repetition does the math), pull the memory out with a question (active recall does the work). Anki works because a flashcard is, by nature, active recall and spaced repetition fused into one system.
Which leaves exactly one question: where do the cards come from?
Inling's answer: let AI write the questions
In the previous post we described how Inling Browser treats a bookmark as the starting point of learning. What happens after the bookmark is this post's subject.
The AI reads the article and takes over the two most tedious steps: judging what's worth keeping, and rewriting knowledge points as questions. Every card and quiz question it generates is a ready-made round of active recall. Scheduling stays with FSRS, which resurfaces each card right before you forget it. Your part is one tap on the bookmark button.
Here's the counterintuitive bit: everything this workflow removes is preparation. The part that actually produces learning — the effortful retrieval — isn't removed at all, and can't be. That's a feature.
If you take one sentence away from this post, take this one: next time you review, don't open the book and start from the top. Close it first, and try to recall. The effort isn't inefficiency — it's the memory setting in.
Inling Browser is in iOS beta. If you'd like AI to turn the articles you've read into cards and quizzes, request an invite on our homepage.
Further reading: Dunlosky et al. (2013), "Improving Students' Learning With Effective Learning Techniques," is the classic evidence review; "desirable difficulties" comes from Robert and Elizabeth Bjork's research program.