The forgetting curve
Ebbinghaus’s curve · retention over time
The forgetting curve is what survives of a memory as time passes without review — steepest in the first day.
What it shows
Learn a list, then measure what survives an hour later, a day later, a week later, and you get a steep early fall that flattens out. That shape is the curve.
The flattening matters as much as the fall. Most of the loss happens in the first day or two, so a first review inside that window does a different job from one a week later.
What the famous percentages are not
It is not a table of fixed rates. Hermann Ebbinghaus measured himself, on material invented to carry no prior meaning, in the 1880s. Anything quoted as "you forget 90% within a day" is somebody’s rounding of that.
Meaningful material holds better. Something that connects to what you already know, or that carries an image or a story, decays far more slowly than nonsense syllables. The curve describes a tendency, not your exam.
How to use it
Read it as instruction about timing: review when the memory has weakened but not gone. A review that arrives after everything is lost is re-learning, and it costs what re-learning costs.
Each recall that succeeds flattens the curve for that item, so the gaps can lengthen. That is the same phenomenon spaced repetition organises — one is the description, the other is the schedule.
Common mistakes
- Using the curve as an excuse. The fall is steepest when reviews are absent, and that is the one variable you control.
- Scheduling by the calendar instead of by the item. Every card in a deck sits on its own curve.
- Treating a precise-looking percentage from a blog post as a fact about your own memory.
Things you can try today
- Pick something you studied days ago and write down what is left, without notes. One point on your own curve beats the quoted ones.
- Put a first review inside a day of learning something new, before the steep part finishes.
- Lengthen a gap when review feels easy and shorten it when it fails. The strain is the readout.