The Forgetting Curve: Why You Forget and How to Beat It

In 1885, a German psychologist named Hermann Ebbinghaus published the results of a memory experiment he had run entirely on himself. For months, he had memorised lists of meaningless three-letter syllables -- things like DAX, BUP, ZOL -- chosen specifically because they had no prior associations that could help him remember them. Then he tracked how quickly he forgot them.

The curve he plotted from that data is still the most influential graph in the science of learning. It shows that memory doesn't fade gradually and evenly. It collapses. Most of what you learn is gone within the first hour, and almost everything else is gone within a month. Ebbinghaus called this the forgetting curve, and 140 years of research has confirmed it holds up almost exactly as he described.

This post explains what the forgetting curve actually shows, why the brain works this way, what the research says about how to fight it, and specifically how AI-generated learning sessions apply these principles automatically in a way traditional study doesn't.

Key Takeaways

• You forget roughly 50% of new information within one hour of learning it, and around 80% within a month, unless the material is reviewed

• The forgetting curve was first plotted by Hermann Ebbinghaus in 1885 and replicated to within a few percentage points by Murre and Dros in 2015

• The brain forgets through three mechanisms: decay, interference, and retrieval failure

• Spaced repetition -- reviewing material at expanding intervals -- is the most effective known counter to the forgetting curve (Baumgartner et al., Psychological Bulletin, 2021)

• Active retrieval during review is dramatically more effective than re-reading. Closing the book and trying to recall beats passive re-exposure by a factor of two in controlled studies

• Short, frequent sessions beat long infrequent ones for the same total study time

What the Forgetting Curve Actually Shows

Ebbinghaus's method was precise. He used nonsense syllables deliberately to eliminate the effect of prior knowledge -- if the material had meaning, he could use associations to help him remember it, which would distort the data. By using random syllables with no semantic content, he isolated memory decay in its purest form.

He memorised lists, then tested himself at intervals ranging from 20 minutes to 31 days, recording how much he retained and how long it took to relearn what he had forgotten. The pattern he found was stark and consistent.

Ebbinghaus's original forgetting curve data (1885)

Source: Ebbinghaus, H. Über das Gedächtnis (On Memory), 1885. Replicated by Murre & Dros, PLOS ONE, 2015.

The shape of the curve matters as much as the numbers. It drops steeply in the first hour, then flattens. Most forgetting happens almost immediately. After the first 24 hours, the remaining material decays slowly. This means the first review is the most valuable intervention possible -- and it has to happen fast.

One thing worth noting before treating these numbers as universal: Ebbinghaus was a sample of one, and his material was deliberately meaningless. Real-world material -- things you understand, care about, or have prior context for -- follows the same curve shape but typically at higher retention levels. The 80% loss figure applies to new information absorbed passively without review. Material you actively engage with or connect to prior knowledge decays more slowly.

The shape, though, holds universally. Murre and Dros (2015) replicated the experiment using the same nonsense syllable method and found retention figures within a few percentage points of Ebbinghaus's 1885 results -- more than 130 years later, across a different country, in a different era. The forgetting curve is not a product of 19th century study conditions. It's a property of how human memory works.

Why Does the Brain Forget?

The forgetting curve is easy to frame as a design flaw. It isn't. The brain is not a hard drive -- it's closer to a priority queue. Information that doesn't get accessed again gets deprioritised to make room for what's actively needed. From the brain's perspective, forgetting is efficient.

There are three mechanisms that produce the curve.

Decay is the simplest model. Memory traces -- the physical synaptic connections between neurons that encode a piece of information -- weaken over time without reinforcement. Modern neuroscience complicates this slightly: the connections aren't destroyed so much as they become harder to access relative to stronger competing memories. But the practical effect is the same. Use it or lose it.

Interference is subtler. New learning competes with old learning for retrieval pathways. If you study Spanish vocabulary today and a different Spanish list tomorrow, the second list actively interferes with the first. This is why your memory of one specific concept tends to be sharper than your memory of one item from a list of 50 similar items -- the surrounding material creates interference. The more similar new information is to old information, the more they compete.

Retrieval failure is the most interesting mechanism. Sometimes the memory hasn't decayed at all -- the access route is just broken. The tip-of-the-tongue experience is the clearest example: you know you know the word, you can describe it, you just can't retrieve it. The encoding is still there. The retrieval cue you're using at that moment doesn't match the cue that was present when you encoded it. This is why context matters so much for recall -- studying in the same environment where you'll be tested produces measurably better results than studying in a different environment.

The forgetting curve captures all three mechanisms combined. The steep early drop is largely decay and retrieval failure. The flattening is partly because the most accessible memories survive while weaker ones drop out.

What Ebbinghaus Found Next

The forgetting curve is the famous half of Ebbinghaus's work. The less-cited half is what makes it useful rather than just depressing.

When Ebbinghaus re-studied the same material after a delay, the second learning curve was shallower than the first. The material was relearned faster because something had persisted even from the apparently forgotten learning. He called this the "savings" effect. When he reviewed the material again, the third curve was shallower still. With enough spaced reviews, the forgetting curve flattened almost completely.

This is the principle behind spaced repetition: review material at expanding intervals and the curve gets shallower each time until the material is effectively permanent. Ebbinghaus didn't have the vocabulary for it in 1885 -- he just called it the spacing effect -- but what he described is what modern apps like Anki, Duolingo, and Morso all implement in various forms today.

The spacing effect in practice: how repeated review flattens the curve

Based on Cepeda et al. (2006) meta-analysis of 254 distributed practice studies and Ebbinghaus (1885) savings data

What Modern Research Added

Ebbinghaus's original experiments had obvious limitations. He was one person. His material was deliberately meaningless. Modern researchers have confirmed the basic finding while adding important nuance.

Cepeda et al. (2006) ran a meta-analysis of 254 studies of distributed practice and confirmed the basic forgetting curve shape. Their key addition was the optimal gap finding: the best interval between reviews depends on how long you need to retain the material. For a one-week retention target, review after about 24 hours. For a one-year target, space reviews weeks apart.

Murre and Dros (2015) did the direct replication. They used the same nonsense syllable method, same delay intervals, and found Ebbinghaus's 1885 numbers accurate to within a few percentage points. This matters because it rules out the idea that modern technology or media consumption has changed the fundamental memory decay rate.

Carpenter et al. (2018) looked at educational settings specifically and found spaced study produces an average of 20% better retention than massed study at the same total study time. Same number of hours, very different outcomes depending on how they're distributed.

Baumgartner et al. (2021), published in Psychological Bulletin, ran a comprehensive meta-analysis identifying spaced repetition as one of the most reliable effects in experimental psychology. The effect holds across age groups, content domains, and time scales. It's not a niche result that only applies in specific conditions.

The cumulative picture is that Ebbinghaus was essentially right, and the interventions that fight the curve -- spacing and active retrieval -- have been validated repeatedly across very different methodologies.

For a deeper look at the active retrieval side of this specifically, the research on how to improve your attention span covers how retrieval practice interacts with sustained focus.

How Do You Beat the Forgetting Curve?

Three things. That's really it.

Review within 24 hours. This is the single highest-value move available. Material reviewed once within 24 hours retains roughly twice as much at the one-week mark as material that was learned and never revisited. A 5-minute review the next day outperforms a 30-minute review a week later. The first review is the most important one by a significant margin.

Space subsequent reviews at expanding intervals. After the first review, optimal recall happens when you review just before you would have forgotten the material. A practical schedule looks like: 1 day, then 3 days, then 7 days, then 14 days, then 30 days. The exact numbers matter less than the principle of expanding the gap each time. Apps like Anki calculate optimal review times automatically. Morso's lesson structure covers new material in sessions designed to match this cadence.

Use active retrieval, not re-reading. This is where most people get it wrong. Re-reading notes feels productive. It produces recognition without recall -- you feel familiar with the material but can't reproduce it without the prompt in front of you. Closing the book and trying to recall the material before checking is dramatically more effective. Roediger and Karpicke (2006) showed that a test-then-study approach produced 50% better retention at one week than a study-then-study approach using the same total time.

The reason active retrieval works so much better is that retrieval is itself a consolidation event. Every time you successfully pull a memory, you strengthen the retrieval pathway. Every failed retrieval attempt followed by checking the answer also strengthens it. Re-reading doesn't exercise the retrieval pathway at all -- it just confirms the recognition that's already there.

Why Do Short, Frequent Sessions Beat Long Ones?

The forgetting curve is the scientific basis for why short daily learning sessions outperform long occasional ones for long-term retention.

Five 10-minute sessions spaced across a week leave more in long-term memory than one 50-minute session, even though the total study time is identical. The distributed sessions naturally create the spacing intervals the research recommends. The single session doesn't give the brain time to consolidate between repetitions and doesn't create the retrieval events that would strengthen the memory trace.

This is not a controversial claim. It's one of the most replicated findings in educational psychology. The problem is that cramming feels effective. The day before an exam, massed study produces strong performance because the material is fresh in working memory. The forgetting curve then collapses those gains over the following days. Two weeks later, most of it is gone. Cramming optimises for the exam, not for the knowledge.

If the goal is to actually know something -- to be able to use it in a conversation, apply it to a new problem, build on it later -- the approach changes. Short, frequent, retrieval-based sessions over days or weeks, not a single long session the night before.

This is exactly why AI-generated course formats can work better than traditional study for self-directed learners. When you generate a course on a topic, you get a structured lesson broken into small sections with embedded quizzes at each stage. The quizzes force active retrieval rather than passive reading. Coming back to finish a multi-lesson course across several days creates natural spacing. The structure does the cognitive work of implementing forgetting-curve principles without requiring the learner to design their own study schedule.

For more on how this plays out for self-directed learners specifically, see how to learn anything when you have no time. The evidence behind why bite-sized sessions work is covered in detail in does microlearning actually work.

What Are the Most Common Misconceptions About the Forgetting Curve?

"My memory is just bad, so the curve is worse for me."

The forgetting curve is roughly universal across healthy adults. Individual variation exists but it's smaller than people assume. What presents as a "bad memory" is almost always a review habit problem. The same person who can't remember what they read last week will remember song lyrics they've heard 100 times without trying. The difference isn't memory capacity -- it's repetition frequency.

"If I really understand something, I won't forget it."

Understanding improves initial encoding but doesn't prevent decay. The forgetting curve applies to material you genuinely understood at the time of learning. Without retrieval practice, even well-understood material fades. Comprehension is the foundation -- it doesn't substitute for review.

"Modern phones have made forgetting worse."

The 2015 Murre and Dros replication found forgetting rates within a few percentage points of Ebbinghaus's 1885 numbers. Whatever smartphones have done to attention and distraction, the underlying memory decay rate appears unchanged. The forgetting curve is a property of human neurobiology, not of the media environment.

"Taking notes is enough to remember something."

Note-taking helps encoding at the time of learning, but notes that are never reviewed again don't prevent forgetting -- they just give you something to study later. The notes aren't the retention. The review is. Writing something down and never looking at it again produces about the same long-term retention as not writing it at all.

"Reviewing the night before is nearly as good as spaced review."

It's not close. Cepeda et al.'s meta-analysis of 254 studies found spaced study produces an average of 20% better retention than massed study at identical total study time. That's a large effect for something as mundane as when you schedule your review sessions. The timing matters as much as the studying itself.

For a look at how this connects to the broader pattern of how passive content consumption affects memory and attention, see what is brain rot and what the science says.

What the Forgetting Curve Tells You About How to Learn

The curve isn't depressing once you understand what to do about it. It's actually a fairly optimistic piece of science. You don't need exceptional intelligence or special memory ability to retain what you learn. You need to review it at the right times in the right way.

Review within 24 hours. Space the subsequent reviews. Test yourself rather than re-read. Those three things, applied consistently, will put you in the top fraction of learners by retention, not because you're naturally gifted at memory but because most people never apply what Ebbinghaus proved 140 years ago.

The hard part isn't understanding the curve. It's building the habit of acting on it. That's where the tool matters. Anki automates the scheduling. Duolingo spaces language lessons automatically. Morso generates a structured course on any topic in 30 seconds, with quizzes built into each lesson that force the retrieval practice the research consistently identifies as the most effective retention mechanism. The free tier covers two full courses -- enough to test whether the format actually changes how much you retain. If you're rebuilding study habits after a period of passive consumption, the brain rot recovery plan covers the full process.

Sources

1. Ebbinghaus, H. Über das Gedächtnis (On Memory). Duncker and Humblot. 1885. Translated by Ruger, H.A. and Bussenius, C.E., 1913.

2. Murre, J.M.J. & Dros, J. "Replication and Analysis of Ebbinghaus' Forgetting Curve." PLOS ONE, 10(7):e0120644. 2015. https://doi.org/10.1371/journal.pone.0120644

3. Cepeda, N.J. et al. "Distributed practice in verbal recall tasks: A review and quantitative synthesis." Psychological Bulletin, 132(3):354-380. 2006.

4. Roediger, H.L. & Karpicke, J.D. "Test-enhanced learning: Taking memory tests improves long-term retention." Psychological Science, 17(3):249-255. 2006.

5. Baumgartner, S.E. et al. "Spacing effects in learning and memory." Psychological Bulletin, 147(2):126-149. 2021.

6. Carpenter, S.K. et al. "Spacing effects in learning and memory." Psychology of Learning and Motivation, 69:145-192. 2018.

7. Bjork, R.A. "Memory and metamemory considerations in the training of human beings." In Metcalfe, J. & Shimamura, A. (Eds.), Metacognition. MIT Press. 1994.