Forgetting Curve

Hermann Ebbinghaus made himself the lab animal in 1885 and found that memory leaks fastest right after learning. He memorized lists of nonsense syllables like WID and ZOF, then measured how much work it took to relearn them later. The curve was not a straight line. The first hours did the most damage.

How it works

The forgetting curve says recall falls as time passes, with the steepest loss soon after the first exposure. Ebbinghaus measured "savings": if a list took 20 repetitions to learn the first time and 12 repetitions later, the saved effort was 40%. That gave memory a number, not just a feeling.

A simple version looks like this:

Recall = e^(-t/S)

t = time since learning
S = memory strength

The useful trick is not the equation. It is the shape. A second review too early wastes effort because the memory is still easy. A second review too late becomes relearning. Spaced repetition lives in that narrow middle.

Where it shows up

Case	What the curve changes	Concrete example
Language learning	Review timing	Anki schedules a card later when recall is easy
Medical exams	Retention planning	Facts reviewed across weeks beat cramming in one night
Training manuals	Refresh cycles	A safety rule forgotten after 30 days is a design failure
Personal notes	Retrieval design	A note that never returns is not knowledge

The curve turns memory from storage into scheduling. That links it to concept queueing theory: the scarce resource is not the note, it is the next review slot.

What's contested

The broad finding is settled: spacing beats massed practice across many tasks. The contested part is the exact curve. Ebbinghaus used one subject, himself, with artificial syllables, so his curve is not a law of nature.

Cepeda, Pashler, Vul, Wixted, and Rohrer reviewed 254 spacing-effect studies in 2006 and found the spacing effect across verbal recall tasks, motor learning, and classroom material. The live question is dose: the best gap depends on the target retention date, the difficulty of the material, and whether the learner is recognizing or producing the answer.

Why this crosses realms

The forgetting curve is a cousin of concept information theory. Surprise is expensive, repetition compresses it, and retrieval proves compression happened. A fact is not "known" because it exists in a notebook; it is known when the brain can reconstruct it under time pressure.

It also changes how I read mission voyager 1. A spacecraft can send a faint signal for 24 billion kilometers because engineers budget decay. A mind has to do the same thing with facts.

An open question

If an AI tutor can estimate your memory strength from every hesitation, typo, and wrong answer, does the forgetting curve become a private clock for each idea in your head?

Abhishek's take

The part that grabs me is not that people forget. Everyone knows that. The useful claim is that forgetting has shape, and shape can be scheduled against. I treat good notes less like a library and more like an orbit: the important objects need timed returns.

Key Sources

Hermann Ebbinghaus, Memory: A Contribution to Experimental Psychology (1885; English translation 1913), https://psychclassics.yorku.ca/Ebbinghaus/ - the original self-experiment behind the curve.
Cepeda, Pashler, Vul, Wixted, and Rohrer (2006), "Distributed practice in verbal recall tasks: A review and quantitative synthesis," Psychological Bulletin, https://doi.org/10.1037/0033-2909.132.3.354 - 254 studies on spacing.
Murre and Dros (2015), "Replication and Analysis of Ebbinghaus' Forgetting Curve," PLOS ONE, https://doi.org/10.1371/journal.pone.0120644 - modern replication using online participants.
Pashler, Rohrer, Cepeda, and Carpenter (2007), "Enhancing learning and retarding forgetting," Psychonomic Bulletin & Review, https://doi.org/10.3758/BF03194066 - practical review of spacing and retrieval.

Abhishek