Why Re-Reading Doesn't Work (And What Does)

• Re-reading and highlighting were rated low utility in a major review of ten study techniques.
• The culprit is the fluency illusion — ease of processing feels like knowing, and is not.
• Retrieval practice beats re-reading substantially on delayed tests, and the effect grows with delay.
• Roediger and Karpicke found repeated re-reading won at five minutes and lost badly at one week.
• The fix costs no extra time. It replaces re-reading rather than adding to it.

What the review actually found

In 2013, John Dunlosky, Katherine Rawson, Elizabeth Marsh, Mitchell Nathan and Daniel Willingham published a monograph in Psychological Science in the Public Interest assessing ten common study techniques against the evidence. They graded each on how well it generalises across materials, learners, conditions and test types.

The results were not evenly distributed.

Utility rating	Techniques
High	Practice testing, distributed practice
Moderate	Elaborative interrogation, self-explanation, interleaved practice
Low	Summarisation, highlighting/underlining, keyword mnemonic, imagery for text, re-reading

Two things stand out.

Only two of ten techniques earned a high rating, and both are effortful. And the low-utility column contains precisely the strategies students use most. A survey by Karpicke, Butler and Roediger published in Memory in 2009 found 84% of students reported re-reading as a study strategy and 55% named it their primary one, while only around 11% reported self-testing as a main approach.

The most popular method and the most effective method are not the same method. That gap is the whole story.

The fluency illusion

Why does something so ineffective feel so productive?

Every time you process a text, you process it more easily. Sentences parse faster, terminology stops snagging, structure feels familiar. This increasing ease is called processing fluency, and it is real — you genuinely are reading better.

The problem is that your brain uses fluency as a proxy for knowledge. Easy to read gets interpreted as "I know this." And on a page that is open in front of you, the two are indistinguishable.

They come apart the moment the page is closed. Recognising information — knowing it looks right when you see it — is a much weaker memory state than being able to retrieve it from nothing. Re-reading trains recognition efficiently and retrieval barely at all. Then an exam asks for retrieval.

This is why the classic experience exists: students who felt completely prepared and then went blank. They were not lying to themselves out of laziness. Their confidence was calibrated to a signal that was measuring the wrong thing.

Highlighting makes it worse, because it adds the feeling of activity to the illusion. Dragging a pen over a line requires almost no processing of meaning, and it produces a page that looks like work was done. Research has found highlighting can actually impair performance on questions requiring inference across a text — the marks focus attention on isolated facts and away from the relationships between them.

The experiment that makes the case

The most instructive single study here is Roediger and Karpicke (2006), published in Psychological Science, because it shows exactly how the illusion is generated.

Students read a prose passage and were then assigned to conditions: some re-studied it repeatedly, some studied then took free-recall tests without access to the text. Everyone was later tested at either five minutes or one week.

At five minutes, the repeated-study group performed best. At one week, the ordering had reversed, and decisively:

Condition	5-minute test	1-week test
Repeated study	~83%	~40%
Study then repeated testing	~71%	~61%

The repeated-study group lost more than half of what they had. The tested group lost comparatively little.

Note what this means. On a short-delay test — which is what a student running through material feels like they are simulating — re-reading genuinely does look better. Every signal available in the moment points at the method that fails a week later. Roediger and Karpicke also found students predicted they would remember more from re-reading. The illusion holds right up until it is tested at a realistic interval.

This is the testing effect, and its defining property is that the advantage grows with the retention interval. Which means it is largest at exactly the point that matters.

What retrieval practice actually is

Retrieval practice means getting information out of your head without looking at the source. That is the entire definition, and its simplicity is the point — the mechanism is the effort of retrieval, not any particular format.

Practical forms, roughly in order of convenience:

The blank page. Close the book and write down everything you remember about the topic. Then open it and check what you missed. This is close to free — it requires no preparation and no materials — and it is among the most effective things on this list. The gaps you find are precisely the material worth re-reading, which turns re-reading into a targeted tool instead of a default.

Self-quizzing. Turn headings and key terms into questions before you study, then answer them from memory afterwards.

Flashcards, used properly. The value is in attempting the answer before flipping. Flipping the card and thinking "yes, I knew that" is re-reading with extra steps.

Explaining it aloud. Teaching a concept to someone else, or to an empty room, forces retrieval and exposes gaps immediately — the sentence you cannot finish is the thing you do not actually know.

Practice problems and past papers, done closed-book first. Working a problem with the solution visible is not retrieval.

Two conditions make it work considerably better. Feedback matters: check your answers afterwards, since an unchecked wrong answer can consolidate the error. And space it, because retrieval practice and the spacing effect compound — the same act of recall does more work when it happens near the edge of forgetting.

The others in the low column

Summarisation rated low largely because it depends so heavily on skill. Good summarising is genuinely useful, but it requires identifying main ideas and restating them in your own words, and students who have not been trained tend to produce near-verbatim copies. As typically practised, it approximates transcription.

The keyword mnemonic — inventing a memorable image linking a foreign word to a similar-sounding native one — works, but narrowly. It suits paired-associate material, takes considerable time to construct, and the mnemonics themselves fade. It is a specialist tool oversold as a general one.

Imagery for text (forming mental pictures while reading) has thin evidence outside easily visualised material.

None of these are worthless. They were rated low on generalisability — how reliably they help across subjects, learners and test formats — not declared harmful. Re-reading a difficult passage once more to understand it is sensible. The finding is that these should not be where most of your time goes.

What to do instead

The change costs nothing extra, which is the most persuasive thing about it. You are not being asked to study more — you are being asked to spend the same hours differently.

Read once, properly. Then close the book and retrieve. That single reallocation is most of the benefit available.

Then layer on the other high- and moderate-utility techniques:

• Space it out. Four one-hour sessions across four weeks beats four hours the night before, at identical cost.
• Interleave. Mix problem types and topics rather than blocking them. It feels worse and performs better, particularly for maths and motor skills.
• Ask why. Elaborative interrogation — asking why a fact is true and answering it — connects new material to what you already know.
• Explain to yourself. Self-explanation while working through a problem outperforms passively following a worked example.

Expect it to feel worse. All of these are desirable difficulties in Robert Bjork's phrase: they slow down performance during practice and improve retention afterwards. The strategies that feel smooth are the ones producing memories that will not survive the week.

If you take one thing: the feeling of knowing is not evidence of knowing. The only way to find out whether you can retrieve something is to try to retrieve it — which is also, conveniently, the thing that makes you able to.

Further reading: The full review is Dunlosky, Rawson, Marsh, Nathan and Willingham, Improving Students' Learning With Effective Learning Techniques, Psychological Science in the Public Interest (2013) — there is also a shorter, freely available version in American Educator. For a book-length treatment, Make It Stick by Brown, Roediger and McDaniel (2014) covers the same evidence for a general audience.