Ebbinghaus forgetting curve¶

Definition¶

The Ebbinghaus forgetting curve is the empirical finding, first measured by Hermann Ebbinghaus (1885) through self-experiments on nonsense syllables, that memory retention decays steeply with time when no attempt is made to retain it — newly learned material roughly halves within days or weeks absent review.

Explanation¶

Ebbinghaus quantified retention as 'savings': the reduction in time needed to relearn material compared with first learning it, and fitted his measurements with a logarithmic decay formula. The result gave experimental psychology one of its first well-designed experimental programmes and its practice of testing with controlled artificial stimuli. A 2015 single-subject replication reproduced results similar to the original data, which is notable durability for a 140-year-old self-experiment. The curve is the empirical foundation for spaced repetition: because decay is steep at first and flattens with each successful recall, reviews scheduled at increasing intervals counteract exactly the loss the curve describes. Ebbinghaus himself measured only the forgetting, not the effect of spacing — that derivative practice came later.

Key Properties¶

Retention loss is steepest immediately after learning and flattens over time
Measured via 'savings' — relearning time saved relative to initial learning
Replicated in 2015 with results similar to the 1885 originals
Foundation for spaced-repetition scheduling, which Ebbinghaus himself did not study

Relationships¶

No relationships recorded yet.
LLM-maintained wiki — supplies the mechanism behind the problem the wiki pattern solves — unreviewed knowledge decays on this curve, so distillation into revisable pages replaces re-reading
Fitted spaced-repetition scheduling — the Ebbinghaus forgetting curve is the empirical decay curve fitted spaced-repetition scheduling's memory model formalizes and personalizes — a fixed population-level shape in the original finding, fitted instead to one learner's own review history.

Applications¶

Scheduling spaced-repetition review so recall happens before predicted decay. Setting expectations for how quickly untouched knowledge (or an unreviewed corpus) goes stale.

Sources¶

https://en.wikipedia.org/wiki/Forgetting_curve