Why you forget what you read: the forgetting curve, replicated, and the two things that bend it

Last updated: October 2026 · By Thomas Lovas Lokoy, founder & builder of BrainLoop. AI-assisted draft checked against the sources at the end. General information about memory research, not medical advice.

You read a chapter on Monday, feel you know it, and by Friday it is a vague shape. People search for "why do I forget everything I read" as if it were a personal failing. It is not. It is a curve that was first measured in 1880, redrawn in a modern laboratory in 2015, and found to be much the same. The good news is that the same body of research tells you which two habits bend the curve upward. Here is the honest version.

The curve, and the man who starved himself of sleep to draw it

Hermann Ebbinghaus was his own only subject. He memorised lists of nonsense syllables, waited, and then measured how much effort it took to relearn them, a method he called savings: the less relearning needed, the more had been retained. The retention intervals he used were 20 minutes, 1 hour, 9 hours, 1 day, 2 days, 6 days and 31 days.¹ The resulting curve, steep at first and flattening later, is the forgetting curve that has appeared in every study-skills book since.

For a long time it rested on one Victorian's self-experiment. Then in 2015 Jaap Murre and Joeri Dros published "Replication and Analysis of Ebbinghaus' Forgetting Curve" in PLOS ONE. One subject spent 70 hours learning lists and relearning them after 20 minutes, 1 hour, 9 hours, 1 day, 2 days or 31 days. Their verdict: "The results are similar to Ebbinghaus' original data", and "the Ebbinghaus forgetting curve has indeed been replicated".¹

So the shape is real. Most of what you lose, you lose early, within the first hours and the first day. What survives the first day decays much more slowly.

The bump nobody expects

The replication found something that textbook versions of the curve smooth away. In the authors' words, the curve "is not completely smooth but most probably shows a jump upwards starting at the 24 hour data point".¹ Retention after a day was slightly better than a smooth decline would predict, and the same small anomaly is visible in Ebbinghaus's own numbers. Murre and Dros interpret the discrepancy as an effect of sleep: a model that includes memory consolidation during sleep "would predict such a boost after one or two nights".¹

Hold that thought, because it is the first practical lever. Material you encounter and then sleep on is not simply decaying; the night does some work on it. Reading something at 11 p.m. and testing yourself the next morning is not the same as reading it at 9 a.m. and testing yourself at 10 p.m., even though the gap is similar.

The first thing that bends the curve: retrieve, don't reread

Ebbinghaus measured forgetting after a single learning episode. What happens if, instead of rereading, you test yourself? Roediger and Karpicke's 2006 study in Psychological Science is the clean comparison. Students learned prose passages either by restudying them or by taking memory tests. Immediately afterwards, the restudy group looked better. Two days and a week later, the tested group remembered substantially more, despite the restudy group having felt more confident.²

That reversal is the forgetting curve in action. Rereading props up the early, steep part of the curve and your confidence with it. Retrieval changes the slope of what comes after. The reason you forget what you read is partly that reading is the weakest way to encounter information twice.

The second thing: space the retrievals, and let the gap grow

If one retrieval helps, when should the next one be? Cepeda and colleagues answered that at scale in 2008, also in Psychological Science. More than 1,350 people learned facts, had a review after a gap of up to three and a half months, and were tested up to a year later. Retention rose as the gap lengthened, then slowly fell. The optimal gap grew with the time until the final test; as a fraction of that time it fell from roughly 20–40 per cent for a test one week away to 5–10 per cent for a test a year away.³

Read that against the curve. The best moment to retrieve something is not immediately, when it is still fresh and the retrieval is effortless, and not after it has vanished. It is somewhere along the slope, where recall is effortful but possible. A short gap for a short horizon; a long gap for a long one.

Putting the three findings together (my reading)

1. You will forget most of a single reading within a day. That is the curve, replicated. Plan for it rather than being surprised by it.
2. Sleep on new material before you judge whether it stuck. The 24-hour bump suggests the first night is doing something useful.
3. Make the second encounter a retrieval, not a reread. Close the book and ask yourself the questions. Confidence after rereading is the curve lying to you.
4. Space the retrievals, widening the gap as you succeed. Tomorrow, then a few days, then a week or more, depending on how long you need to keep it.

None of this requires an app. A piece of paper with questions on one side and answers on the other does it. But the questions have to exist, and the honesty of the self-test has to be enforced, which is where a tool earns its place.

How BrainLoop Flashcards leans on this

The Flashcards app does three small things that line up with the research. It makes every card a retrieval: you see the question, commit to an answer, click to reveal, and must press Got It! or Didn't Know before moving on. It gives you an honest score for the pass, so a deck you are "sure" about can be checked against what you actually produced. And it lets you turn a chapter into questions in seconds by typing a topic and pressing Generate, which asks the AI for six question-and-answer cards.

Two honest limits. Generated decks are not saved between visits, so the spacing schedule is yours to keep (write down the topic, regenerate when due, and expect slightly different cards). And the app does not know the forgetting curve; it does not schedule reviews for you. It gives you a clean retrieval loop and a score. The gaps are your job, and the research above tells you how to set them.

The short version

Forgetting is steep early and shallow later, a curve first drawn in 1880 and replicated in 2015, with a small upward bump after the first night that is probably sleep. Rereading mostly props up confidence. Retrieval and spaced, widening review gaps are the two habits with evidence behind them. Read once, sleep, test yourself tomorrow, then test again later, and the curve bends.

---

Sources

Footnotes

1. Murre, J. M. J., & Dros, J. (2015). Replication and Analysis of Ebbinghaus' Forgetting Curve. PLOS ONE 10(7): e0120644. Open access: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0120644 ↩ ↩² ↩³ ↩⁴

2. Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: taking memory tests improves long-term retention. Psychological Science. Abstract via Europe PMC: https://europepmc.org/article/MED/16507066 ↩

3. Cepeda, N. J., Vul, E., Rohrer, D., Wixted, J. T., & Pashler, H. (2008). Spacing effects in learning: a temporal ridgeline of optimal retention. Psychological Science. Abstract via Europe PMC: https://europepmc.org/article/MED/19076480 ↩