why we forget what we read

Why You Forget Everything You Read (And How to Fix It)

We forget 50% of what we read within an hour and 70% within a day. The reason isn't laziness — it's how memory encodes. Here's what to do about it.

edit

Oliver Grant

verifiedLast updated & fact-checked: May 24, 2026 · 19 min read

I once spent four months working through a serious six-hundred-page book on economic history. I finished it on a flight home from a wedding, closed the cover with the small private satisfaction of a long project finished, and felt — accurately, in the moment — that I understood it. Three months later a friend asked what I'd thought of the book. I could remember the cover. I could remember roughly where I was when I read certain passages. I could not produce a single specific argument the author had made.

This is not a story about that book. It is a story about almost every serious reader I know, and almost certainly about you too if you are honest about the last ten books you read. In a 2015 replication of Hermann Ebbinghaus's nineteenth-century forgetting-curve experiments, Dutch psychologists Jaap Murre and Joeri Dros confirmed that without active engagement, people lose approximately 50% of newly learned material within an hour and over 70% within 24 hours (Jaap M. J. Murre & Joeri Dros, Replication and Analysis of Ebbinghaus' Forgetting Curve, PLOS ONE, July 2015). For longer time horizons — the kind that matters when you ask yourself what you remember from a book three months ago — the numbers are worse.

The good news, if you want to call it that, is that the reason you forget what you read is not a personal failing. It is the documented default behavior of the human memory system. The better news is that fifty years of cognitive psychology has produced an unusually clear consensus on what actually fights the forgetting. The unfortunate news is that almost none of the things readers feel are helping them — highlighting, underlining, re-reading, audiobooks on 1.5x — appear on that list. This article is the why behind the forgetting, and the short list of what reliably reverses it.

Key Takeaways

  • The forgetting curve is real and well-replicated: in Murre & Dros's 2015 PLOS ONE study, average retention dropped to roughly 44% at one hour and 33% at one day without active retrieval (PLOS ONE, 2015)
  • Most book forgetting is encoding failure, not storage failure — the information never made it into a durable memory trace in the first place, because reading was passive
  • The single most replicated fix in cognitive psychology is retrieval practice: trying to recall material from memory before re-checking the page, which produces roughly 50% more retention than re-reading (Roediger & Karpicke, 2006)
  • Highlighting and re-reading were rated low-utility strategies in a comprehensive 2013 review of learning techniques — they produce confidence without comparable retention (Dunlosky et al., 2013)

Why is forgetting the default state of the brain?

In Murre and Dros's 2015 replication of the 1885 Ebbinghaus experiments, average retention of newly learned material dropped from about 58% at twenty minutes to 44% at one hour, 36% at nine hours, and 33% at one day (Murre & Dros, Replication and Analysis of Ebbinghaus' Forgetting Curve, PLOS ONE, 2015). The decay is not a quirk of nonsense syllables; it has been broadly replicated across meaningful material too, with the curve becoming somewhat shallower but the basic shape intact.

Cognitive psychologists generally split memory failures into three categories, which is useful because the fix for each is different. The first is encoding failure — the information never made it into a memory trace at all, because attention was divided or the material was processed only at a shallow level. The second is storage failure — the trace formed, then decayed because it was never reactivated. The third is retrieval failure — the trace is there, but you can't reach it without the right cue.

What is remarkable about reading is how much of the forgetting is the first kind. You feel like you read the book. You did read the book. But because the reading was passive — eyes moving across pages, occasional highlights, no effortful retrieval — most of the material never crossed the threshold into a durable trace. The book did not so much slip out of your memory as never fully arrive in it. This is why re-reading produces less learning than people expect: you are repeating an act that was already a poor encoding event.

Unique insight: The folk model of forgetting imagines memories as items in a leaky bucket. The cognitive-science model is closer to a pottery wheel: most of what you read never becomes a pot at all. It stays as wet clay that dries into nothing identifiable. The work of remembering is not in preventing leaks; it is in forming the pot in the first place.

What is the fluency illusion, and why does it lie to you?

In a series of experiments running from the late 1980s through the 2010s, UCLA psychologist Robert Bjork and colleagues documented a now well-known phenomenon: people are systematically bad at judging their own learning. When material is fluent — easy to read, familiar, presented in a polished format — readers consistently overestimate how well they will be able to recall it later (Robert A. Bjork, Elizabeth L. Bjork & Nate Kornell, Self-Regulated Learning: Beliefs, Techniques, and Illusions, Annual Review of Psychology, 2013).

This is the fluency illusion, and it is the silent killer of reading retention. When you finish a chapter, the ideas are still warm in working memory. Asked whether you understood the chapter, you would honestly answer yes. Asked, three months later, to summarize the chapter, you find that almost nothing comes. The error wasn't in your understanding at the time. The error was in mistaking the warmth of working memory for the formation of a long-term trace.

Highlighting is so dangerous because it is essentially a fluency-illusion-generating machine. The yellow ink registers as evidence of engagement. The brain sees the visual selection and concludes that the idea is now somehow more durable. The selection produces zero retrieval effort. You closed the book confident that the highlights would do the work; the highlights, on their own, almost never do.

[CHART 1: Ebbinghaus forgetting curve]

Memory retention without retrieval, by time elapsedMurre & Dros, PLOS ONE 2015 replication of Ebbinghaus 188580%60%40%20%0%20 min1 hr9 hrs1 day6 days31 days58%44%36%33%25%21%Time elapsed since learning, without any retrieval attempt

This is the dropoff that most readers fight against, usually badly. By the one-month mark — the time at which a friend is likely to ask what you thought of the book — roughly four-fifths of the material is gone if nothing else intervened.

How does passive reading sabotage memory?

In a 1972 paper that has shaped almost every learning-science result since, psychologists Fergus Craik and Robert Lockhart proposed that memory is a byproduct of the depth at which material is processed: shallow processing (recognizing letters, sounding out words) produces fragile traces, while deeper processing (extracting meaning, relating the idea to personal experience, generating implications) produces durable ones (Fergus I. M. Craik & Robert S. Lockhart, Levels of Processing: A Framework for Memory Research, Journal of Verbal Learning and Verbal Behavior, 1972). Their framework is now over fifty years old; it has held up extraordinarily well.

The implication for reading is uncomfortable. Most pleasure reading, even of serious books, is processed at a shallow-to-moderate level. You follow the argument. You enjoy the prose. You notice when a passage is striking. None of this constitutes deep processing in the Craik-Lockhart sense. Deep processing requires that you do something with the idea — restate it, compare it to a prior idea, ask whether it's wrong, apply it to a real situation. Without that something, the trace stays shallow and the curve does what the curve does.

This is also the cognitive-science explanation for why audiobooks at 1.5x or 2x speed retain so little. Speed compresses the input but does not compress the brain's encoding window. A 2010 study by Daniel Willingham and others, summarized in his book Why Don't Students Like School, argued that comprehension and retention are tightly linked to time spent thinking about meaning, not time spent receiving input. Faster input often produces less thinking, not more. The same applies to skim-reading and to reading while half-watching the news.

A stack of softcover books on a wooden surface in soft daylight — the visible accumulation of reading hours that may or may not have produced any retained understanding, depending on what happened after the books were closed.

In a comprehensive 2013 review of the research on learning techniques, John Dunlosky and four co-authors classified ten common study strategies as high, moderate, or low utility. Practice testing and distributed practice were rated high. Re-reading, highlighting, summarization by inexperienced readers, and keyword mnemonics were rated low (John Dunlosky, Katherine A. Rawson, Elizabeth J. Marsh, Mitchell J. Nathan & Daniel T. Willingham, Improving Students' Learning With Effective Learning Techniques, Psychological Science in the Public Interest, 2013). The two techniques the average reader uses most — highlighting and re-reading — are the ones the field has flagged as least effective. That is the central irony of how most of us read.

Does reading more actually make you remember more?

In its 2021 reading survey, the Pew Research Center reported that the average American adult read about 14 books in the preceding 12 months, with a median of 5 (Pew Research Center, Who doesn't read books in America?, January 2022). The most committed readers — roughly the top quartile — read 20 or more books per year. The cultural assumption is that those readers are also the ones who know the most. That is not necessarily true.

If you read 20 books per year using passive methods, the forgetting curve grinds through 80% of nearly every one of them. By the end of the year, you have produced a great deal of motion and very little durable knowledge. Meanwhile, a reader who reads five books a year and aggressively engages with each — sets a reading question before starting, retrieves the main argument out loud after each chapter, writes a one-page summary after the book — keeps a far larger share of what they read, even though their book-count is a quarter of the heavy reader's.

This is the case that the cognitive-science research keeps making, and that most readers — including me, for years — keep refusing to hear. The number of books in your reading log is a measure of input, not of output. There is no straightforward relationship between input and output if the encoding is broken.

[CHART 2: Retention by study method]

One-week retention by study methodAfter studying the same factual passages, retention tested at 7 days80%60%40%20%42%Re-read only(no retrieval)56%Single retrieval(read + 1 test)61%Three retrievals(read + 3 tests)Source: Roediger & Karpicke, Psychological Science, 2006

In Henry Roediger and Jeffrey Karpicke's 2006 study, students who simply re-read a factual passage retained roughly 42% of it on a test seven days later. Students who read the passage and then took a single recall test retained about 56%. Students who took three recall tests retained about 61% (Roediger & Karpicke, Test-Enhanced Learning, Psychological Science, 2006). The retrieval-practice group, despite spending the same total time on the material, walked away with about 50% more retention than the re-reading group did. This is the testing effect, and for our purposes it is the single most important finding in this article.

A four-step protocol for not forgetting your next book

If you take only one thing from the science above, it is this: the trick is not to read more carefully, and it is not to highlight differently. It is to retrieve the material from memory, on a schedule, before checking the page. The protocol below is the shortest defensible version of what the literature recommends. The fuller method is in our companion piece on the simple three-step note-taking system.

  1. Set a question before you start. Before opening the book, write down — in one sentence — the question you are hoping the book will answer for you. This primes attention and gives the brain something specific to encode toward. The Bjork research on learning shows that having a clear retrieval target during encoding substantially improves later recall.
  2. Pause for sixty seconds after each reading session. Close the book. Without looking at it, write down the one to three ideas from this session that you most want to keep. This is the first retrieval attempt, performed while the memory trace is still warm enough to consolidate.
  3. Write a one-page synthesis within 24 hours of finishing. In your own words, summarize the book's central argument, three supporting ideas, and one way you might actually use it. No looking back at notes. This is the second retrieval attempt and the moment most readers' retention either solidifies or fails.
  4. Revisit the synthesis at one week and one month. Each revisit is another retrieval attempt, spaced to take advantage of the distributed-practice effect documented by Cepeda and colleagues (Distributed Practice in Verbal Recall Tasks, Psychological Bulletin, 2006). Two minutes each time is plenty.

The full method, with the small operational adjustments that make it survive a busy week, is in the note-taking article we published this same month.

A vintage hardcover book opens onto a worn page in low light — the kind of book whose contents are easy to admire and very easy to forget, unless something happens between the chapter and the next morning.

When forgetting isn't about reading at all

Sometimes the reason you can't remember the book has nothing to do with how you read it. Three under-discussed variables routinely overwhelm any reading strategy: sleep, divided attention, and chronic stress. A 2013 study published in Science found that during sleep the brain's glymphatic system clears metabolic waste from the interstitial space at substantially higher rates than during waking, and that memory consolidation is materially impaired by sleep deprivation (Lulu Xie et al., Sleep Drives Metabolite Clearance from the Adult Brain, Science, October 2013). If you are reading on five hours of sleep, no amount of retrieval practice will fully compensate.

The same is true of divided attention. Reading while a television plays in the background, while a phone notifies in your pocket, or while you wait for an email to arrive does not produce the same encoding as reading in a quiet room with no devices nearby. Cognitive psychologists have known this for decades, but the modern reading environment has made the problem newly acute. The book you forgot may have been a book you were never quite reading in the first place.

There is also the question of when forgetting is a clinical issue rather than a reading-method issue. Persistent, significant memory loss that goes beyond the normal forgetting curve — losing track of recent conversations, struggling to recall familiar names, finding gaps in autobiographical memory — is a medical question and not a self-help one. If something about your forgetting is worrying you, or if a family member is worried about it, that conversation belongs with your doctor, not with a notebook. For the ordinary, healthy frustration of "I read a book and remember none of it," the methods in this article are the right tool. For anything more, please get a real opinion from a real clinician.

Read fewer books, remember more of them

If you have read this far, you have probably noticed an implicit recommendation underneath all the cognitive science: most of us would be better off reading less and engaging more. There is no shame in finishing fewer books a year. There is real cost to finishing a great many and remembering none of them — the cost is the years of attention you spent and the false confidence that you understand things you cannot, in fact, articulate. If you want a small library of books you actually use, a guide to reading when nothing seems to be working out or a guide to reading when you feel behind in life is a better place to start than another 50-book annual challenge.

If your problem is that you want to re-read books you've previously forgotten without buying them all again, a subscription service like Kindle Unlimited is an unusually good fit for retention work — it lets you re-borrow titles for the second-pass retrieval work without re-purchase. Our 2026 Kindle Unlimited review walks through whether the catalogue actually contains the kinds of books most readers want to re-read.

Frequently Asked Questions

Is it normal to forget what I read?

Yes. In Murre and Dros's 2015 replication of the Ebbinghaus forgetting curve, average retention dropped to about 44% within an hour and 33% within a day without any active retrieval (PLOS ONE, 2015). For richer book material the curve is somewhat shallower, but the basic decay is universal across healthy adults.

Does highlighting help me remember more?

Mostly no. A 2013 review by Dunlosky and colleagues, published in Psychological Science in the Public Interest, classified highlighting as a low-utility learning strategy, alongside re-reading (Dunlosky et al., 2013). Highlighting is fine as a bookmark for ideas, but it does not perform the retrieval work that actually consolidates a memory.

How long does it take to remember a book using retrieval practice?

Less time than most people expect. In the Roediger and Karpicke 2006 study, students who took a single recall test after reading a passage retained about 56% of it at one week, versus 42% for re-readers (Psychological Science, 2006). Three brief retrieval attempts, spaced across days, are typically enough for a non-academic book.

Why do audiobooks feel like I'm learning, but I can't summarize them later?

Most audiobook listening is processed at a shallow level — comprehended in the moment but not retrieved afterward. Without a deliberate retrieval step after each chapter, audiobooks decay along the same forgetting curve as silent reading. Listening faster than 1x typically worsens, not improves, retention.

Is reading slowly better for memory than reading fast?

Slower reading can improve encoding at the margins, but it doesn't substitute for retrieval. A passage read slowly without any post-reading recall attempt will still be largely forgotten within a few days. The single highest-leverage move is to add a retrieval step, not to change the speed at which you read.

The takeaway

Forgetting is the default; remembering is the deliberate exception. Almost everything about modern reading — the highlighters, the speed settings, the year-end book counts, the audiobook libraries — quietly assumes that input is what matters and that retention will follow. Fifty years of cognitive psychology suggests the opposite is closer to the truth. Input is cheap; retention is the entire game, and retention is won by retrieval, spaced over time, in the small uncomfortable moments after the book is closed and you have nothing to look at but the inside of your own head.

If you read one book this year using the four-step protocol above, you will remember more of it a year from now than you remember of any of the last twenty books you read without it. That is not a metaphor. It is, approximately, what the data says. The fix for forgetting is not more reading. It is a small change in what you do thirty seconds after you finish.

For the operational version of the method above, with the small adjustments that make it survive a busy week and a long book, see our simple three-step note-taking method for remembering every book you read.

Sources

Affiliate disclosure: BetterLifeReads earns a commission on Kindle Unlimited sign-ups initiated through links in this article. This does not change the price you pay or our editorial recommendations.