# The forgetting curve: why you can’t remember what you were just told

> Half of what you learn today will be gone within the hour. That’s not your memory failing, it’s how memory normally works. Here’s what to do about it.

Category: Understand. Published 2026-08-28, updated 2026-09-05.
About 8 minutes to read. URL: https://memorypilot.fr/en/blog/forgetting-curve

Someone gives you the gate code. You repeat it once, you nod, you’re sure you’ll remember it. Two hours later you’re standing at the gate and it’s gone. That’s not absent-mindedness, and it’s not age. It’s a mechanism described a hundred and forty years ago, measured, and reproduced since with the same results.

## What Ebbinghaus measured, on himself, for five years

Hermann Ebbinghaus was a German psychologist. Between 1880 and 1885, he made himself his only test subject, and he deliberately picked the dullest material imaginable: meaningless syllables like “WID,” “ZOF” and “KAX.” The choice wasn’t a quirk. A sentence that means something latches on to what you already know and skews the measurement; an empty syllable doesn’t.

![Photographic portrait of Hermann Ebbinghaus, a middle-aged man with a long beard and glasses, in a dark suit.](https://memorypilot.fr/img/article/ebbinghaus.webp)

*Hermann Ebbinghaus. For five years, he learned and relearned lists of nonsense syllables, timing himself every time. (Unknown author, public domain, via Wikimedia Commons)*

His method was brutally simple. He learned a list until he could recite it without a single mistake. He waited. Then he relearned it, and timed himself. The difference between the two learning sessions showed how much was left: what he called **savings**, “Ersparnis” in German. It was all published in 1885 in [Über das Gedächtnis](https://archive.org/details/berdasgedcht00ebbi), which you can still read today; it came out in English in 1913 as “Memory: A Contribution to Experimental Psychology.”

![A figure from the 1885 book: a hand-drawn curve on a graduated chart, captioned “Fig. 2.”](https://memorypilot.fr/img/article/gedaechtnis.webp)

*A figure from “Über das Gedächtnis,” 1885. Ebbinghaus ran ninety-two series of trials and drew his curves himself. (Hermann Ebbinghaus, public domain, via Wikimedia Commons)*

The curve he drew from it drops sharply, then levels off. Roughly half of what has just been learned fades within the hour. After a day, about a third remains. After a week, a small fraction. It’s not linear at all: **almost all of the loss happens in the first few minutes**, and whatever survives the first day holds up fairly well after that.

![Graph: Ebbinghaus’s data points and the fitted curve. The vertical axis shows the proportion retained, the horizontal axis the delay in days, on a logarithmic scale.](https://memorypilot.fr/img/article/courbe.webp)

*Ebbinghaus’s measurements, replotted by Jaap Murre. The bottom axis is logarithmic: the first point sits at twenty minutes, the last at thirty-one days. (Jaap M. J. Murre, CC BY 4.0, via Wikimedia Commons)*

**This isn’t some old theory people keep repeating to each other**

In 2015, Jaap Murre and Joeri Dros redid the original experiment using Ebbinghaus’s protocol, and found his curve again. Their paper, [Replication and Analysis of Ebbinghaus’ Forgetting Curve](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0120644), is open access. It’s one of the few 19th-century psychology findings to have held up in a modern replication, which is why you can rely on it without reservation.

## Why your brain does this, and why it’s good news

Forgetting isn’t a malfunction, it’s a filter. A memory that kept everything would also keep your parking spot number from three years ago, the price of last Tuesday’s coffee and the color of the car that passed you this morning. What doesn’t get recalled fades, and that’s precisely what lets you find everything else.

So the problem isn’t forgetting. The problem is that the filter can’t tell the price of a coffee from the gate code. It doesn’t keep what’s important, **it keeps what comes back**. And something you hear only once never comes back.

> Your memory doesn’t sort by importance. It sorts by frequency. That’s the whole difference, and it’s where everything gets lost.

## Three ways around it, and what they cost

### 1. Repeat, at wider and wider intervals

This is the classic answer, and it works: reviewing a piece of information just before you forget it pushes the curve back up each time, a little higher and for a little longer. That’s the principle behind spaced repetition, flashcards and every memorization app out there.

It’s also the least realistic answer for everyday life. Nobody is going to schedule five reviews spread over three weeks for a bike lock combination. Spaced repetition is built for things you want to **know by heart**, not for things you just want to be able to find. Mixing up the two is the most common mistake on the subject.

### 2. Write it down, somewhere

That’s what everyone does, and it’s where the real problem starts. The code goes on a sticky note on your screen. Your niece’s shoe size ends up in a text thread. The birthday goes in the calendar, your sister-in-law’s allergy in a note, the parking spot in a photo.

![A wall of wooden card-catalog drawers, each with a handwritten label in a metal label holder.](https://memorypilot.fr/img/article/meuble-cartes.webp)

*A library card catalog. It worked for a century, on one condition: that someone filed every card in its place, every single day. (MarkBuckawicki, CC0, via Wikimedia Commons)*

> Writing things down doesn’t solve forgetting. It moves it: you no longer forget the information, you forget where you put it.

And search doesn’t make up for it, because you never search with the words you wrote. You noted “Léa, seafood.” Six months later, you search for “who can’t eat shellfish.” No words in common, so no results, so a perfectly written note that is perfectly useless.

That’s exactly the wall that card systems hit, from the library catalog to the [Zettelkasten](https://memorypilot.fr/en/blog/digital-twin-promise-vs-reality): they work for people willing to file things, and everyone else abandons them after three weeks.

### 3. Hand it to a place that understands what you say

That’s the third way, and it’s the one we built. One single place where you type or dictate the way you talk, without choosing a folder, making up a title, or deciding whether it’s a note, a reminder or a contact. It files itself, and above all you can find it again with the words you’ll have in mind that day, not the ones you used the day you wrote it.

The difference from a notebook isn’t storage capacity; it never was. It’s the **way back**. A phrase like “who can’t eat shellfish” has to bring back “Léa is allergic to seafood,” and that’s all that matters.

## What it changes in practice

- **You stop deciding where things go.** That decision, not a lack of features, is what makes people abandon most note-taking systems after three weeks.
- **You find things by meaning, not by the exact word.** “Sarah’s allergy” comes up when you ask what she can’t eat.
- **What needs to ring, rings.** A date without an alarm is a date you’ll read the day after.
- **What’s sensitive stays hidden.** A code doesn’t show up in a list where someone can read it over your shoulder: see [where to keep your codes](https://memorypilot.fr/en/blog/where-to-keep-codes-and-passwords).

You can’t beat the forgetting curve, but you can work around it. Your head is excellent at understanding, deciding and connecting; it’s bad at storing a four-digit number it heard once. You might as well take away the job it does badly.

**If you remember only one thing**

It’s not that your memory is bad. It’s that we ask it to do a hard drive’s job when it’s built for something else. Give it less to hold, and it’ll do better. And if you want to know what you can really expect from a digital memory, [we wrote about that too](https://memorypilot.fr/en/blog/digital-twin-promise-vs-reality).