Forgetting Curve

Every student has run the experiment without meaning to. You finish a lecture feeling like you own the material, skip it for ten days, and then open your notes to find what might as well be a stranger’s handwriting. The instinct is to blame yourself: bad memory, not enough focus, should have paid more attention.

The better explanation is that you forgot on schedule. Memory decay follows a predictable curve, first measured in 1885 and still replicating today, and once you know its shape you can study against it instead of being ambushed by it. That is the entire premise behind spaced repetition: place your reviews where the curve says forgetting is about to happen, and a small amount of work preserves a large amount of knowledge.

This article looks at where the curve comes from, what it actually shows (less than the internet claims, more than skeptics admit), and how to build a review schedule on top of it.

Ebbinghaus and his nonsense syllables

Hermann Ebbinghaus was a German psychologist who wanted to study memory in its purest form, uncontaminated by meaning, associations, or prior knowledge. His solution was to invent material with no meaning at all: nonsense syllables like zof, wid, and kep, consonant-vowel-consonant combinations that connected to nothing.

His method was heroic and slightly mad. Acting as his own only subject, he memorized list after list of these syllables to the point of perfect recitation, then waited: twenty minutes, an hour, a day, a week, a month. After each delay he relearned the list and recorded how much faster the relearning went compared with the first time.

That last detail matters more than it seems. Ebbinghaus did not measure how many syllables he could still recite. He measured savings: the reduction in effort needed to relearn. Even when he could not consciously recall a single syllable from a list, relearning it went faster than learning a fresh one. Some trace remained below the surface. Forgetting, in other words, is rarely total erasure; it is loss of access. That is why relearning old material always goes faster than the first pass, and why reviewing “forgotten” notes is a far better investment than it feels.

What the curve actually shows

Plot retention against time and you get the famous shape: a steep plunge in the first hours, a continuing slide through the first days, then a long, nearly flat tail. Whatever survives the first week tends to survive much longer.

Three honest caveats before you build a study plan on it:

• The precise numbers online are mostly decoration. You will see confident claims that students forget some exact percentage within 24 hours. Ebbinghaus’s own figures describe one man memorizing meaningless syllables, the hardest possible case. Meaningful, well-structured material decays more slowly. Trust the shape, not the decimals.
• The curve describes unreviewed memory. It is the trajectory of material you learn once and never touch again. The whole point of knowing it is to refuse those terms.
• Steeper for nonsense, gentler for understanding. The more connected material is to things you already know, the flatter your personal curve. This is why understanding a mechanism beats memorizing its description, and why techniques like active recall that force you to process meaning give you a better starting curve as well as better reviews.

The modern replication

A fair question about any 19th-century self-experiment: does it hold up? In 2015, Murre and Dros redid the whole ordeal, with one subject memorizing nonsense syllables under conditions matched as closely as possible to the original. Their curve tracked Ebbinghaus’s remarkably well. A 130-year-old finding from a sample size of one, reproduced.

The surrounding literature is even more reassuring. The spacing effect that falls out of the curve, the fact that distributed reviews beat massed ones, was confirmed at meta-analytic scale by Cepeda et al. (2006) across hundreds of comparisons. Forgetting curves have since been measured for vocabulary, medical knowledge, CPR skills, and lecture content. The decay rates differ; the shape keeps showing up.

How review timing flattens the curve

Here is the mechanism that turns a depressing graph into a study strategy. When you successfully retrieve a memory, two things happen: retention jumps back up, and the new forgetting curve from that point decays more slowly than the previous one. Each review buys you a longer-lasting memory than the review before it.

Run that logic forward and the optimal schedule writes itself:

1. First review after about a day, when the initial drop is well underway but the memory is still recoverable with effort.
2. Second review a few days later, riding the new, flatter curve.
3. Third review about a week after that.
4. Subsequent reviews at multiples of weeks, each one a quick confirmation rather than a rescue mission.

Notice the asymmetry of effort. Reviewing at the right time takes a minute per item, because you are catching the memory before it fully fades. Reviewing far too late means relearning, which costs nearly as much as the first study session did. The curve punishes lateness more than it rewards earliness, which is the polite scientific way of saying: cramming the night before is paying full price for knowledge you could have had at a discount.

Sleep deserves a line here too. A meaningful share of consolidation happens overnight, which is why the same fact reviewed across two days with sleep in between tends to outlast the same two reviews crammed into one evening. The curve is measured in days partly because the brain does its filing in nightly batches. An all-nighter attacks both ends at once: it skips the spaced reviews and sabotages the consolidation that would have rescued some of what you learned.

There is one wrinkle worth knowing: reviewing too early is also wasteful. Retrieval strengthens memory most when it requires some effort, Bjork’s “desirable difficulty.” Re-testing yourself an hour after learning feels great and accomplishes little. The sweet spot is just before the forgetting wins.

A practical schedule built on the curve

Suppose your physiology exam is four weeks out and the course is a stack of lecture slides and a 40-page packet of notes. A curve-aware plan looks like this:

Week 1: convert and encode. Turn the material into things you can test yourself on, question-and-answer pairs, not summaries. Doing this by hand is useful but slow; a flashcard generator that works from your PDFs and slides gets you to the reviewing stage the same day. Do the first retrieval of each batch within 24 hours of creating it.

Weeks 2–3: ride the intervals. Each item comes back at roughly day 3, day 7, then day 14 after you learned it, sooner if you missed it last time. Daily load: typically 15 to 30 minutes. Build this into your calendar like a class; the guide on making a study schedule covers how to slot it around everything else.

Final week: confirmation, not panic. By now most items are on the flat part of their curves. Your queue naturally shrinks to your stubborn weak spots, which is exactly where the remaining marks live. Add a timed practice test or two for format practice.

The bookkeeping, tracking which of 300 items is due on which day, is the part humans are bad at and software is built for. StudyDone runs the curve math for you: it spaces each question from your own notes and compresses the intervals to fit your exam date, so nothing is still “due for its second review” the morning of the test.

What to take from a 140-year-old graph

Forgetting is not a character flaw. It is the default behavior of a healthy brain triaging information, and it runs on a timetable regular enough to plan around. The students who seem to “just remember things” are usually not blessed with better hardware; they are touching the material more than once, at intervals that happen to land near the curve’s danger points.

You can do that on purpose. Learn it once properly, retrieve it tomorrow, again in a few days, again next week. The curve does the rest.

FAQ

What is the forgetting curve?

It is the shape of memory decay over time: retention drops steeply in the first hours and days after learning, then the decline flattens. Hermann Ebbinghaus first measured it in 1885 by memorizing nonsense syllables and testing himself at set intervals.

Is the forgetting curve real or a myth?

The shape is real and has been replicated, including a careful modern replication by Murre and Dros in 2015 that reproduced Ebbinghaus's curve closely. What is mythical are the suspiciously exact percentages often attached to it online; the actual rate of forgetting depends on the material and the learner.

How fast do students actually forget lecture material?

It varies with how meaningful the material is and how it was learned, but the steepest losses happen within the first day or two. Meaningful, well-understood material decays slower than disconnected facts, which is why understanding first and reviewing second is the right order.

How do I stop the forgetting curve?

You can't stop it, but you can flatten it. Each successful retrieval of a memory resets the curve and slows the subsequent decay, so a few reviews at expanding intervals, such as one day, three days, then a week after learning, preserve material with very little total time.

Did Ebbinghaus really test only himself?

Yes. He was both the experimenter and the only subject, memorizing thousands of nonsense syllables over months. That is a real limitation, but his core findings, including the curve's shape and the benefit of spaced practice, have held up in controlled studies with many participants.