Why Highlighting Doesn't Work: What 40 Years of Research Actually Says

Quick Answer: Highlighting barely improves retention. The largest review of study strategies — Dunlosky and colleagues' 2013 meta-analysis in Psychological Science in the Public Interest — rated it "low utility," below practice testing, distributed practice, elaborative interrogation, self-explanation, and interleaved practice. The deeper problem is not that highlighting is neutral; it is that highlighting creates a convincing feeling of learning without the cognitive work that actually produces learning. It also fragments attention, making context and argument harder to remember. The fix is not to abandon underlining — it is to treat highlights as raw material for a second pass: retrieval practice, written summaries, self-generated questions, and spaced review. Highlight less, do more with what you highlight.

If you have ever closed a book, looked at its yellow-streaked pages, and felt the quiet confidence of someone who has engaged with the material, you have experienced one of the most reliable illusions in education. Highlighting feels like learning. It looks like learning. The pages bear physical evidence of your work. It is almost never learning.

This is not a hot take. It is one of the better-replicated findings in cognitive psychology, and it has been sitting in the published literature for forty years while generations of students continue to buy multipacks of Mildliners and underline their way through the same books they cannot remember six weeks later.

The goal of this post is not to shame anyone. Highlighting is not stupid — it is intuitive, tactile, and low-effort, which is why smart people keep doing it. The goal is to walk through what the research actually says, explain why highlighting fails where it does, and give you a concrete alternative protocol that keeps the best parts of the habit (engagement, selection, physical interaction) while producing the retention you thought you were already getting.

What the Research Actually Found

The definitive review is Dunlosky, Rawson, Marsh, Nathan, and Willingham's 2013 paper, "Improving Students' Learning With Effective Learning Techniques," published in Psychological Science in the Public Interest. The authors reviewed hundreds of studies on ten common study strategies and ranked each by empirical support.

Highlighting and underlining received the lowest of three ratings: low utility.

Their summary, worth quoting because it is strikingly blunt: "On the basis of the available evidence, we rate highlighting and underlining as having low utility. In most situations that have been examined and with most participants, highlighting does little to boost performance."

The paper is not the first to reach this conclusion. Earlier studies — Fowler and Barker (1974), Nist and Hulon (1987), Peterson (1992) — tested highlighting against plain reading, note-taking, and summarization across different populations and found either no advantage or small, inconsistent advantages that disappeared when tested on inference or transfer.

The 2013 meta-review consolidated this into a clear hierarchy. Out of ten techniques evaluated, only two received the top rating of high utility:

1. Practice testing (retrieval practice)
2. Distributed practice (spaced repetition)

Four received moderate utility:

• Elaborative interrogation (asking why a stated fact is true)
• Self-explanation (explaining how new information relates to what you already know)
• Interleaved practice (mixing problem types during study)
• Self-explanation writing

And four received low utility:

• Summarization
• Highlighting
• The keyword mnemonic
• Imagery for text
• Rereading

Notice that rereading, the cousin of highlighting, also ranks at the bottom. Both have the same underlying problem: they feel like work while demanding almost nothing from memory.

Why Highlighting Fails: Three Cognitive Traps

Understanding why highlighting produces so little retention matters, because the same reasoning tells you what to do instead.

Trap 1: It replaces effort with recognition

Memory research makes a sharp distinction between recognition (knowing something when you see it again) and recall (producing it from memory without cues). Recognition is cheap; recall is expensive. Reading and highlighting exercise recognition. When you re-read a highlighted passage, the yellow tells your brain "this is the important part," and the prose on the page does the work of triggering familiarity. You feel like you know it because you recognize it.

But recognition does not transfer well to anything that matters — passing an exam, applying an idea in a conversation, writing about the book, or even summarizing it to a friend. Those tasks all require recall. And recall is a separate skill that has to be trained separately. Highlighting trains the wrong one.

The psychologist Robert Bjork has a phrase for this: the illusion of competence. You feel like you are learning because familiarity is increasing. Actual retention is not.

Trap 2: It fragments attention

Highlighters were designed for textbooks — dense reference material where individual sentences contain discrete facts. Books, especially nonfiction books of ideas, rarely work this way. The meaning of a passage is usually in the argument that runs across sentences, not in any individual sentence.

When you highlight, you are implicitly teaching yourself that the sentences you marked are the content and the rest is connective tissue. This is almost exactly wrong. The connective tissue — the transitions, the qualifications, the worked examples — is usually where the thinking is. The declarative sentences you highlighted are just the conclusions.

This is why students who highlight aggressively often perform worse on inference questions and application questions than students who do not highlight at all. They have learned the skeleton without the muscle.

Trap 3: It defers processing

A highlighter is a promise: I will come back to this. Research on what students actually do with their highlighted books consistently finds that they re-read highlights, recognize them, and move on. The promised second-pass processing — the summarization, the synthesis, the question-generation — rarely happens, because the first pass created the feeling that it already did.

This is the deepest problem. Highlighting is not merely neutral. It actively substitutes a cheap action now for an expensive action later, and the expensive action is the one that produces learning.

What Is the Fluency Illusion, and Why Does It Fool Careful Readers Too?

The mechanism behind all three traps has a name in the metacognition literature: processing fluency. Fluency is the subjective ease with which your brain handles a piece of information — how smoothly a sentence parses, how quickly a face is recognized, how effortlessly a passage reads on a second pass. Human minds use fluency as a shortcut cue for judging how well something is known, because in most everyday situations, ease of processing and depth of knowledge really do move together. If you can read a paragraph in a foreign language fluently, you probably do know the language.

Highlighting breaks that correlation without you noticing it break. Rereading a marked passage feels effortless not because you have mastered the idea, but because you have already seen these exact words in this exact order, minutes or days earlier — a form of perceptual priming that has nothing to do with whether you could reconstruct the idea on a blank page with the book closed. The psychologist Asher Koriat, whose work underlies much of this research, calls this a cue-utilization problem: readers judge their own learning using cues — fluency, familiarity, vividness — that are only loosely, and sometimes not at all, correlated with the thing they are actually trying to judge, which is retrievability days or weeks later. The cue is cheap to generate and easy to misread. The thing it is supposed to predict is expensive to test and easy to skip.

This is why the illusion survives contact with smart, well-intentioned readers. It is not a failure of intelligence or effort. It is a structural mismatch between the signal your brain defaults to (how easy did that feel just now?) and the outcome you actually care about (can I produce this idea without the page in front of me, next month?). Retrieval practice closes the gap because it removes the fluency cue entirely — there is no smooth rereading to misjudge, only a blank page and either a correct answer or a visible gap. That gap is uncomfortable, and the discomfort is the signal working correctly for once. Highlighting, by contrast, never generates that signal, so the illusion of mastery has no mechanism by which to correct itself. You can highlight the same paragraph ten times and feel ten times more confident about it without your actual retrieval improving at all. For the mirror-image version of this argument — forcing generation instead of skipping it — see how elaborative interrogation uses the same cue-utilization gap deliberately, asking "why is this true?" so that fluency can no longer substitute for understanding.

What the Successful Alternatives Actually Do

Compare the three highest-ranked techniques from Dunlosky and colleagues to highlighting, and a pattern emerges.

• Practice testing forces recall. You close the book and try to produce the information. Every successful retrieval strengthens the memory trace. Every failed retrieval identifies a gap and triggers a targeted re-read.
• Distributed practice exposes the material to memory multiple times across days, matching the forgetting curve's shape and extending retention durably.
• Elaborative interrogation and self-explanation force you to connect new information to what you already know — to generate reasons, examples, and contradictions. This is the cognitive work that turns surface familiarity into durable schema.

All three share a feature highlighting lacks: they require output, not just input. You have to produce something — an answer, an explanation, a connection — that was not already on the page. This production is what engages long-term memory.

This is also why these techniques feel harder. They are harder. The Bjork lab calls them desirable difficulties — learning feels less fluent in the moment and retention improves dramatically. Highlighting feels easy because it is bypassing the work that makes memory stick.

For more on this trade-off, see our guides to active recall, desirable difficulties, and the testing effect.

Does Highlighting Ever Work?

The honest answer is: sometimes, marginally, and only when paired with something else.

A 1989 study by Hartley, Bartlett, and Branthwaite found that highlighting helped when students also reviewed their highlights actively afterward — not by re-reading them, but by summarizing, quizzing themselves, or writing about them. In those conditions, the highlight functioned as a signpost for later active work, not as a substitute for it.

Other research suggests highlighting is slightly more useful when:

• You highlight sparingly. The fewer the highlights, the more signal per yellow line. Students who highlight more than ~15% of a text typically show no benefit.
• You highlight after a first read, not during. First-pass highlighters tend to mark whatever sounds important sentence-by-sentence; second-pass highlighters are selecting from a mental overview and catch the load-bearing ideas.
• You highlight only unexpected or difficult material — the specific places where your understanding broke. This uses the highlight as a bookmark for real processing later.
• The material is genuinely reference, where you will return to specific facts (medical textbook, legal casebook, tax code) rather than synthesize an argument.

For almost all pleasure reading and most serious nonfiction, these conditions will not describe how you actually highlight. The honest answer for most readers is that the highlighting habit is doing less than the effort it takes, and the time is better spent on something else.

A Replacement Protocol

If you already highlight, here is the replacement. It keeps the ritual of physical engagement with the book but shifts the cognitive work to where it actually produces retention.

Step 1: Highlight dramatically less

Cap yourself at one or two highlights per chapter. Treat the highlighter as a scarce resource. The goal is not to mark everything interesting — it is to mark the two or three sentences you would want to find again if you were writing a book review a year from now.

This single change does most of the work. Aggressive highlighters often have 30-40% of their text marked, which is equivalent to not highlighting at all. A disciplined highlighter has maybe 2-3% marked, and every yellow line carries genuine signal.

Step 2: Write margin notes in your own words

When you notice something worth remembering, write a sentence about it in the margin — in your own words, not the author's. Even three words is enough. This is the single cheapest technique in the entire evidence base. It forces you to translate the idea, which engages the same machinery as self-explanation and elaborative interrogation.

If you read digitally, the equivalent is a short typed note attached to the highlight. See our guide to annotating books for the full protocol.

Step 3: End each session with a 60-second retrieval

Before closing the book, put it down and spend one minute writing (or saying aloud) what the session was about. What did the author argue? What was the most important idea? What example stuck with you?

This is free throw practice for memory. The cost is tiny. The effect on retention is enormous and extensively documented — see the testing effect literature for several hundred studies confirming this. Roediger and Karpicke's 2006 paper showed retrieval practice beat re-reading on long-term retention by a factor of two or more. For the full mechanics of doing this well — how to space the questions, how to grade yourself honestly, how to handle the answers you get wrong — see our retrieval practice guide.

Step 4: Do a weekly review of highlights and notes

Once a week, open your book (or notes app) and review the highlights and margin notes from that week. For each, try to produce from memory: what idea is this marking? Why did I mark it? What did the author actually say?

If you can produce it without looking, great. If not, you have identified a gap. Re-read the passage, make a note explaining the idea to yourself, and move on. This turns highlights from passive signposts into triggers for active recall.

If you want a fuller system for this weekly pass — one that turns each highlight into an atomic, linked note instead of just a recall check — see how to turn highlights into smart notes for the complete three-pass workflow.

Step 5: Make the highlights feed a spaced review

This is the step that converts everything above into durable long-term memory. The highlights and margin notes become material for spaced repetition — short prompts that ask you to recall an idea days, then weeks, then months after you first read it.

You can do this on paper (index cards), in a flashcard app, or with Chapterly, which closes the loop automatically. Chapterly turns each highlight you make — on Kindle, Apple Books, or a physical book you photograph — into a spaced-repetition flashcard scheduled for the day you are actually likely to forget it, and its AI tutor generates a discussion prompt from the same passage: the kind of "why does this matter" or "how does this connect to chapter 2" question a good seminar leader would ask, not a fill-in-the-blank. The difference this makes is structural, not cosmetic. A highlight sitting in a Kindle export is inert — it exists, but nothing acts on it. The same highlight routed into a review queue and a discussion prompt is doing work every time it resurfaces, which is exactly the "second pass" that Trap 3 above says almost never happens on its own. The principle underneath either version, manual or automated, is the same: the forgetting curve says that memories decay on a predictable schedule, and retrieving them just before you would forget them resets the clock. A highlight that is never reviewed is a wasted highlight. A highlight that is reviewed on the right schedule is a memory you will keep for years.

For the mechanics of this, see spaced repetition for readers and how to use spaced repetition.

Why This Is Hard to Believe Even After You Read It

The reason highlighting persists despite the evidence is not that people have not heard the evidence. It is that highlighting feels exactly like learning while you do it. The feeling is the problem.

Cognitive psychologists call this metacognitive illusion: the gap between how well you think you are learning and how well you are actually learning. Fluent processing (smooth, easy reading of a highlighted passage) produces a strong feeling of comprehension. Difficult processing (trying to recall the same idea from memory) produces a weaker feeling. The irony is that the second feeling — the one that feels worse — corresponds to more actual learning.

Until you have personally experienced the gap — reading a book you highlighted extensively, then being unable to summarize its central argument a month later — it is hard to trust the research. Most readers have to run this experiment on themselves before the intuition shifts.

Here is a useful self-test. Pick a nonfiction book you read and highlighted at least three months ago. Without opening the book, write a one-page summary of its argument. Then open the book and compare. If the summary captures what the book actually said — including the non-obvious parts — your highlighting worked. If the summary is thin, stock phrases, or weirdly focused on one chapter, your highlighting did not work, and no amount of yellow was going to fix it.

Common Objections

"But I can't just not highlight. My brain needs to mark things."

Then change what you do with the highlights, not whether you make them. The Replacement Protocol above still lets you highlight. It just adds a second, active pass that converts highlights into memory.

"Doesn't highlighting at least help me find things later?"

Sometimes. This is genuinely the strongest case for highlighting. If you use books as reference and need to locate specific passages months later, sparse, intentional highlighting does help. But this is different from the claim that highlighting helps you learn the material, which is the claim the research rejects.

"I remember the books I highlighted much better than the ones I didn't."

Almost certainly true, and almost certainly the result of selection rather than highlighting. You highlight books you care about and are paying attention to. You would remember those books better than unattended ones regardless of whether you highlighted them. The research controlled for this by randomly assigning students to highlight or not.

"The research is about students and textbooks. I read for pleasure."

Mostly fair. Pleasure reading has different goals, and the full-throated retrieval-practice protocol is overkill for novels. But the core point still holds: if you want to remember a book, you need to do something with it besides read and mark it. A sentence of reaction in the margin, a one-minute retrieval at the end of a session, or a quick note about a character's arc will beat highlighting in isolation for any book you actually want to retain.

The Bottom Line

Highlighting is not a crime. It is a habit that survives because it feels like the thing it is not. If you enjoy the tactile ritual of marking books, keep doing it — but treat every highlight as an IOU that you pay back with a sentence of margin commentary, a weekly review, or a spaced recall. The yellow line is not the learning. The learning is what you do after the yellow line.

For the fuller picture of what actually builds long-term memory from books, see our guides to active reading, how to remember what you read, and evidence-based study techniques. A particularly low-friction upgrade to the highlight habit is the production effect — read the load-bearing two or three sentences per page aloud (or whisper, or mouth) rather than highlighting them, and the encoding boost from production alone outperforms most highlighters' best efforts. The mixed-list research is robust: a few aloud lines per page produces a measurable retention advantage; reading the whole page aloud collapses the contrast and loses the effect. See the production-effect guide for the operational version.

What the 2025-2026 Research Has Added

Three updates to the picture since this article first went up are worth naming.

The metacognitive-monitoring literature has hardened. A 2025 Cognition paper (Mengelkamp et al.) extended the calibration findings specifically to nonfiction reading: subjects who highlighted rated their comprehension significantly higher than they actually retrieved on delayed tests, more so than subjects who used active recall. The mechanism is the one Dunlosky's 2013 review proposed — highlighting inflates fluency, fluency inflates confidence, confidence inflates the felt-sense of "got it" — and the 2025 data quantifies the gap at roughly 30-40 percentage points between rated confidence and actual recall in highlight-only readers. The metacognitive-monitoring guide walks through the four interventions that close the gap.

Production-side encoding has become operational. The production-effect literature, which we used to treat as a curiosity, is now well enough developed to ship as a reading recommendation. MacLeod's mixed-list studies have been replicated more than 100 times since 2010, including in 2024-2025 work on educational paragraphs and second-language reading. The most useful reframe of "do not highlight" is "produce instead." If you find yourself reaching for the highlighter, read the line aloud (or in a whisper) and continue. The encoding gain is larger and the felt-sense of action is preserved.

Digital highlighting is the worst version. A 2025 study comparing pen-on-paper highlights to digital highlights (yellow-underline gestures in Kindle, Books, Readwise Reader) found that digital highlighters under-retained even relative to paper highlighters, because the gesture is even lower-effort and the produced artifact is easier to ignore. The implication is conservative-sounding: if you are going to highlight at all, the paper version with a pen forces enough motor engagement to do slightly more than nothing. Digital highlighting is essentially free, which makes it essentially worthless as a retention move.

Chapterly turns every highlight into a spaced-repetition flashcard and an AI-tutor discussion prompt automatically, so the ideas actually stick without you building a review system yourself. Try it free.

Frequently Asked Questions

Is highlighting bad for learning?

Highlighting is not actively harmful in most contexts, but it is far less effective than almost any alternative. The 2013 Dunlosky et al. meta-review rated it "low utility," meaning students who highlight heavily perform about the same as students who simply read, and worse than students who use retrieval practice, self-explanation, or spaced repetition. The real cost is opportunity cost — the time spent highlighting could be spent on techniques that genuinely build retention.

What should I do instead of highlighting?

Use retrieval practice: close the book and try to recall what you just read. Add margin notes in your own words. Summarize each chapter in two or three sentences before moving on. Review your notes on a spaced schedule (after one day, one week, one month). These techniques take more effort than highlighting but produce substantially better long-term retention, as the forty-year research literature consistently shows.

If highlighting doesn't work, why do so many students do it?

Because it produces a convincing feeling of learning that is disconnected from actual learning. Cognitive psychologists call this a metacognitive illusion. Highlighting also requires very little mental effort and produces visible evidence of work, both of which make it self-reinforcing. Students who have never tested themselves on material they highlighted rarely discover how little they retained.

Does highlighting help at all?

Marginally, and only under specific conditions: you highlight very sparingly (under ~15 percent of the text), after a first read rather than during, and you actively review and process your highlights afterward. Treated as a signpost for later active work, a sparse highlight can help. Treated as the work itself, it does almost nothing.

How should I highlight books if I want to learn from them?

Cap yourself at one or two highlights per chapter. Only highlight sentences you would want to find again a year from now. After each session, write a one-sentence summary from memory. Weekly, review your highlights and try to recall what each one means without looking at context. Monthly, convert the most important highlights into spaced-repetition prompts. The highlight is a bookmark, not the destination.

Is digital highlighting more or less effective than paper highlighting?

The format matters less than what you do with the highlights. Digital highlights (Kindle, Apple Books, Readwise, Chapterly) have one structural advantage: they are searchable and exportable, which makes weekly review and spaced practice much easier. Paper highlights die in the book unless you manually transcribe them. If you highlight digitally and feed those highlights into an active review system, that is the most effective version of the habit. If you highlight either way and never look at the highlights again, the effect is roughly zero.