Is Highlighting Actually Helping You Learn?

Highlighting is not useless. It is also not a reliable learning method merely because the page looks more studied afterward.

That awkward middle position is a good description of the research.

A highlighter can direct attention, mark information as important, and make a later return faster. Under some conditions, highlighted material is remembered better. But experiments have repeatedly failed to show that the ordinary act of highlighting consistently improves deeper comprehension or transfer by itself.

The practical mistake is treating three different things as if they were the same:

• selection: deciding which information matters
• storage: preserving a route back to that information
• learning: becoming able to recall, explain, or use it later

Highlighting clearly does the first two. The third depends on what happens during and after the marking.

The older evidence was mixed from the beginning

One of the often-cited early experiments is Robert Fowler and Anne Barker's 1974 paper, “Effectiveness of Highlighting for Retention of Text Material.”

In their first experiment, groups of college students read material under different highlighting conditions and were tested a week later. The researchers did not find an overall difference in total retention scores among the groups. More detailed analyses suggested that selected material could benefit from highlighting and that actively choosing material was better than merely reading someone else's highlights.

That is already more complicated than “highlighting works” or “highlighting does nothing.” The effect depended partly on what was selected and how the highlighting was done.

Decades later, a broad review by John Dunlosky and colleagues reached a cautious verdict. Their 2013 monograph, “Improving Students' Learning With Effective Learning Techniques,” rated highlighting and underlining as low-utility study techniques. The problem was not evidence that highlighting always harms learning. It was that benefits were inconsistent and much less generalizable than techniques such as practice testing and distributed practice.

Highlighting seems better at selection than integration

A useful modern experiment makes the distinction unusually clear.

Hector Ponce and Richard Mayer randomly assigned 130 college students to five conditions while learning from expository text. Some read ordinary text, some saw key words already highlighted, some actively highlighted, and others worked with graphic organizers. Their 2014 paper, “An eye movement analysis of highlighting and graphic organizer study aids for learning from expository text,” also tracked where students looked.

Both active highlighting and reading pre-highlighted text improved performance on a rote-memory test relative to the control condition. They did not produce the same improvement on the summary-based comprehension test. The graphic-organizer conditions improved both.

The eye-tracking results help explain why. Highlighting directed attention toward selected information. The graphic-organizer task required more organization and integration across the text.

That is a useful way to think about highlighting: it can help choose what receives attention without necessarily doing the intellectual work of connecting the chosen information.

A recent randomized study found no overall learning advantage

A 2024 digital-reading experiment by Lucia Mason and colleagues gives an even more current test.

In “Highlighting and highlighted information in text comprehension and learning from digital reading,” university students were randomly assigned to actively highlight a digital text, read a version in which important material was already highlighted, or read without highlighting. The study measured literal comprehension, inferential comprehension, transfer, and calibration both immediately and later.

There was no overall main effect of reading condition on those outcomes. Active highlighting, by itself, did not outperform ordinary reading.

But one result matters: within the group that generated its own highlights, the quality of the information selected significantly contributed to literal comprehension and transfer performance. Students who selected better material tended to learn better.

That does not prove that choosing good highlights caused the better outcomes. Better readers may simply be better at recognizing what matters. But it does shift the practical question away from the number of highlighted sentences and toward the judgment used to select them.

The same study also found that active highlighters spent substantially longer reading than the other groups. More interaction with the page is not automatically more learning. A study technique consumes time, and that time needs to produce something.

What you highlight matters more than the existence of yellow text

A 2023 study by Alexandra List and Chang-Jen Lin examined 278 undergraduates reading multiple digital texts. Students were prompted to highlight and annotate with different strategic goals, including relevance, cross-text connections, evaluation, and metacognitive monitoring.

The paper, “Content and quantity of highlights and annotations predict learning from multiple digital texts,” found no overall differences in comprehension and integration across the assigned strategy conditions. However, students' highlighting of source information and statistical evidence, along with other evaluative behaviors, predicted better multiple-text comprehension and integration.

This was a quasi-experimental study, and the predictive relationships should not be turned into causal claims. Still, it agrees with the broader pattern: highlighting is more promising when it reflects good judgments about information than when it is treated as a generic physical action.

The problem with highlighting everything

Selection only works if something is not selected.

If half a chapter is marked, the visual signal becomes weak. More importantly, broad highlighting can avoid the question that makes the technique cognitively interesting in the first place: What is actually important here?

A useful highlight often represents a decision about structure:

• This sentence states the argument.
• This paragraph contains the evidence.
• This definition controls the rest of the chapter.
• This exception changes the rule.
• This source detail matters to whether I should trust the claim.

Those judgments can support careful reading. But once the judgment has been made, the reader still has to remember or use the idea.

If the goal is memory, retrieval has stronger evidence

The strongest reason not to make highlighting the center of a study system is opportunity cost. Time spent rereading marked passages is time that could be spent trying to retrieve the ideas without looking.

Henry Roediger and Jeffrey Karpicke demonstrated this with educationally relevant prose in their 2006 experiments, “Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention.” Students either repeatedly studied passages or took free-recall tests. Repeated study produced better performance after five minutes, but prior testing produced substantially better retention after two days and one week.

This is not a study of highlighting, so it should not be cited as evidence that a particular highlighting workflow improves memory. It supports a narrower recommendation: when the objective is durable recall, do not stop at selecting the information. Try to retrieve it.

A better way to use highlights for learning

Highlighting works best as the front end of a learning process rather than the whole process.

1. Read first enough to understand the local context. Marking every sentence on first contact makes importance difficult to judge.
2. Highlight selectively. Prefer claims, definitions, evidence, distinctions, and exceptions that structure the material.
3. Later, hide the passage. Try to state the idea, answer a question about it, or explain how it connects to the larger argument.
4. Check the original. Retrieval is useful partly because failure becomes visible.
5. Keep only the highlights that have a future job. Some are for exam review, some for writing, some merely for finding a passage again.

This does not require converting an entire book into flashcards. It requires being clear about what you want to be able to do after reading.

Highlighting is useful when you ask it to do the right job

The evidence does not support the satisfying fiction that marking a sentence is equivalent to learning it.

It supports something more modest. Highlighting can guide attention. It can preserve important locations. The quality of the selection matters. In some experiments, highlighted material has shown memory benefits. But deeper comprehension generally requires more than emphasis.

So if a highlight is there because you want to quote the passage later, it may already be doing its job.

If it is there because you want to know the material, use it as a cue for something harder: explanation, connection, or retrieval.

Studies cited

• Fowler, R. L., & Barker, A. S. (1974). Effectiveness of highlighting for retention of text material. Journal of Applied Psychology, 59(3), 358–364.
• Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving Students' Learning With Effective Learning Techniques. Psychological Science in the Public Interest, 14(1), 4–58.
• Ponce, H. R., & Mayer, R. E. (2014). An eye movement analysis of highlighting and graphic organizer study aids for learning from expository text. Computers in Human Behavior, 41, 21–32.
• Mason, L., Ronconi, A., Carretti, B., Nardin, S., & Tarchi, C. (2024). Highlighting and highlighted information in text comprehension and learning from digital reading. Journal of Computer Assisted Learning, 40(2), 637–653.
• List, A., & Lin, C.-J. (2023). Content and quantity of highlights and annotations predict learning from multiple digital texts. Computers & Education, 199, 104791.
• Roediger, H. L., III, & Karpicke, J. D. (2006). Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention. Psychological Science, 17(3), 249–255.