The Forgetting Curve, Explained

In the 1880s, Ebbinghaus memorised long lists of nonsense syllables — meaningless three-letter combinations chosen so no prior knowledge could help. He then tested his own recall at increasing delays and plotted the results.

The curve dropped steeply in the first hour, more gently over the first day, and settled into a long slow decline over weeks. About half of what he learned was gone within 20 minutes; only a small fraction remained after a month without review.

New memories are physically fragile. The neural connections that encode them are still weak, and without reinforcement they are progressively pruned or overwritten. Sleep helps consolidate some of what you learned during the day, but consolidation is a partial rescue, not a permanent save.

The more meaningful the material — the more it connects to what you already know — the slower it decays. This is why memorising unrelated facts is harder than memorising facts that fit into a story you already understand.

Each successful review pushes the curve out to the right. A memory that would have been gone in a day becomes stable for a week. Reviewed again, it becomes stable for a month, then a season, then a year. This is the mechanism spaced repetition exploits.

The timing matters. Review too soon and you waste effort on material that is still fresh. Review too late and the memory has already collapsed and has to be rebuilt from scratch. The sweet spot is right as the memory is starting to fade — which is exactly what an adaptive spaced-repetition schedule targets.

The forgetting curve was measured on nonsense syllables. Real, meaningful material decays more slowly, especially when it is deeply understood or emotionally significant. The curve is a lower bound on how bad forgetting can be, not a strict prediction for every subject.

It also assumes no review. Even a single retrieval attempt within the first 24 hours changes the shape of the curve substantially.

The practical rules are simple. Review new material within 24 hours. Review again 2 to 3 days later. Review again a week later, then two weeks, then a month. Adjust based on how well each item is holding — struggle means a shorter interval, easy means a longer one.

You can do this by hand with paper flashcards and a calendar. Software just automates the bookkeeping so you can focus on the retrieval itself.