In 2013 Dunlosky and colleagues reviewed ten common study techniques and sorted them by how well the evidence supported them. Highlighting landed in the bottom group, alongside rereading and summarisation, rated low utility. Some studies found it produced no benefit at all, and at least one found it actively impaired the ability to draw inferences across a text.
It is also the single most popular study behaviour in the world. That gap is worth understanding, because the problem is not the marker.
Why it fails
Highlighting is a decision you make before you understand. You mark a sentence while reading it, which is the exact moment you are least equipped to judge whether it matters. What gets highlighted is what sounds important: confident phrasing, definitions, the first sentence of a paragraph. Importance is a property of the whole argument and is not visible yet.
It feels like work and is not. Marking a page produces the sensation of engagement without any of the processing that produces memory. You have handled the sentence, not thought about it. This is the illusion of competence in its purest form, and it is why heavily highlighted textbooks so reliably belong to students who did badly.
It defers the work forever. The implicit promise is "I will come back to this". Almost nobody does, and when they do, they reread the yellow parts, which is rereading with extra steps.
Over-highlighting destroys the signal. Once a third of the page is yellow, the marks carry no information. The commonest pattern in student books is a first chapter almost entirely marked and a last chapter untouched, which is a record of enthusiasm rather than of importance.
The underlying issue: highlighting is recognition, and learning requires production. Marking a sentence guarantees you will recognise it later, which is a much weaker state than being able to say it. See active recall.
The six rules
Highlighting is salvageable, but only as a first pass that creates work, never as the study itself.
1. Read the section first, mark second. Read a whole subsection with the marker down. Then go back and mark. This single change fixes most of it, because you are now marking with knowledge of where the argument ended up rather than guessing sentence by sentence.
2. Cap yourself. No more than two or three marks per page, or roughly five per cent of the text. An artificial limit forces selection, and selection is the cognitive work that highlighting normally lets you skip. If you have to choose between two sentences, you have to compare their importance, which is the whole point.
3. Mark claims, not topics. Highlight the sentence that makes an argument, not the one that announces a subject. "Retrieval strengthens the path more than review does" is worth marking. "This chapter discusses memory" is not, and headings are already headings.
4. Write in the margin whenever you mark. Three words minimum: why, or disagree, or contradicts Ahrens. This converts a passive mark into a small act of production, and it is the difference between a highlight and a note. See marginalia, and Adler's classic case for writing in books in How to Read a Book.
5. Use two marks, not one. A distinction gives the marks meaning. The pairing that works best in practice is important versus I do not understand this yet, because the second category is a genuine to-do list and it is the one people never make. Colour-coding beyond two or three categories collapses under its own overhead.
6. Process within a week, or do not bother. Highlights are a capture format, not a storage format. Turn them into something of your own: a few atomic notes, a summary written from memory, or answers to questions you posed while reading. Until that happens the reading has produced nothing durable.
Turning highlights into knowledge
Rule six is the one that matters and it deserves its own procedure. The move is from marked text to your words, and the cheapest reliable version is this:
- Close the book. Write down what it argued, from memory. This is the diagnostic step, and it is the blurting method applied to reading.
- Open to the highlights and add what you missed, in a different colour.
- Convert anything that survives into one note per idea, phrased as a claim you could defend, with the source attached. See literature notes and how to take notes from a book.
- Delete the rest. Most highlights do not deserve to become notes, and pretending otherwise is how archives fill with quotations nobody rereads. See the collector's fallacy.
Tiago Forte's progressive summarization formalises this as successive passes that narrow what is marked. It works, with one caveat worth stating plainly: layered highlighting can become an elaborate way to keep not writing anything. The pass that generates your own sentences is the one doing the work.
E-books and PDFs
Digital highlighting is easier, which makes over-highlighting worse and the export better.
The advantage is that your marks are extractable as text, so step one of processing is free. The danger is that a tidy list of exported highlights looks like a finished artefact. It is not; it is a pile of someone else's sentences, and an archive of them is the collector's fallacy at industrial scale.
Use the export as the input to writing, and delete the export afterwards. Practical mechanics in how to take notes from a book and the best notes app for PDF annotation.
The short version
Read first, mark after. Two marks a page. Write why in the margin. Convert to your own words inside a week, from memory, then throw the highlights away.
A highlight is a bookmark for future work. If the work never comes, the yellow does nothing at all.
More in how to take notes, how to remember what you read, and how to study from notes.