Ebbinghaus's forgetting curve: the original 1885 data (not the meme)

What the 1885 experiments actually measured

In 1885, Ebbinghaus 1885Memory: A Contribution to Experimental PsychologyTeachers College, Columbia University (1913 translation) · 1885Measured his own retention over time without rehearsal and found forgetting is steep in the first hours and days, then levels off — the original forgetting curve. memorized lists of meaningless syllables (chosen specifically to avoid any prior association that would make some easier to remember than others) and then tested his own retention at increasing intervals afterward. The resulting curve drops fast in the first hour, keeps dropping through the first day, and then flattens into a long, shallow decline. The headline finding isn't “you forget things” — it's that almost all of the forgetting happens early, and what survives the first day or two tends to be comparatively durable.

That shape matters practically because it tells you when intervention is highest-leverage: not weeks later, but in the window right after first exposure, before the steepest part of the drop has happened. A single review during that early window does more to flatten the curve than the same review done much later, once most of the forgetting has already occurred.

What the meme gets wrong

Popular summaries often invent a tidy percentage — “you forget X% in Y hours” — and present it as a universal law. Ebbinghaus's own work was more careful and more limited: one learner, nonsense material, specific intervals, self-testing. The shape is robust enough to teach from; the meme numbers floating around the internet usually are not. Treat the curve as a family of steep-then-shallow declines under no-review conditions, not as a single branded percentage.

Another common distortion is treating the curve as destiny: once something fades, it is gone. That is not what the original work implies, and it is not what later spacing research implies. The curve describes what happens with no further successful retrieval. Add retrieval — especially early, while the drop is still steep — and the next decline can flatten. The practical question is not “how doomed am I?” It is “when do I force the next production attempt?”

What later research added for readers

The original curve was for nonsense syllables with no contextual structure, which is the most pessimistic case — real material with logical structure, like a book's argument, tends to be more durable than Ebbinghaus's own results because it has more to attach to. Cepeda et al. 2008Spacing effects in learning: A temporal ridgeline of optimal retentionPsychological Science · 2008A meta-analysis of spacing studies showing retention depends on the ratio between study gap and test delay, not a fixed interval that works for everyone.'s meta-analysis extended the picture considerably: rather than one universal curve, retention depends on the relationship between the interval between study sessions and the interval before the information is needed again, which is why “one good review a week later” isn't the same fix for material you need to retain for a month versus a decade.

Later work also clarifies what counts as a review. Passive re-exposure can feel like progress because familiarity returns, but Koriat & Bjork 2005Illusions of competence in monitoring one's knowledge during studyJournal of Experimental Psychology: Learning, Memory, and Cognition · 2005Readers systematically overrate how well they've learned something they've merely re-seen — recognition gets mistaken for recall. found that exposure-driven familiarity is a poor predictor of actual retrievability. What changes the curve's shape is retrieval — trying to recall the material without looking — not another smooth pass over the same page. A reminder that only shows you the text again mostly resets recognition; a reminder that forces production first is the intervention the curve is asking for.

A free experiment for one chapter: after you finish it, wait until recall feels slightly effortful — often later the same day or the next — then write the central claim and one support without looking. Check, note what you missed, and attempt those misses again after another gap. You will feel the early steep drop personally, and you will also feel how a single retrieval attempt flattens the next decline, without needing a branded schedule of fixed review days.

To see the measured 1885 points against a modeled spaced-review curve you can toggle, use the interactive forgetting-curve tool — that interactive page is the primary forgetting-curve asset on this site. the full research page collects the fuller citation list, and the Field Guide annotates what each shape implies for a reader specifically. If you want to try this on Thus Spoke Zarathustra — or browse a book you're already reading. Start there.