Ebbinghaus on forgetting
📚SOURCE
💎PERMANENT NOTES
🧵SYNTHESIS
In the early 1880s 🏛️ Hermann Ebbinghaus decided that memory could be measured, and measured his own. He was the only subject. To keep prior knowledge out of it he invented a stock of some 2,300 meaningless syllables, consonant, vowel, consonant, drew them at random into lists, and read each list aloud at a fixed pace until he could recite it once without error.
His difficulty was how to measure what was left later, when a list could no longer be recited at all. The answer is the savings method. Relearn the same list and count how much less time it takes than the first learning. If it took twenty minutes originally and thirteen now, a third of the work has been saved, and that third is a measure of what remained even though none of it could be recalled.
With this he timed the loss. It is steep at first and then slow: well under half the saving was left after an hour, about a third after a day, a quarter after a week, and still roughly a fifth after a month. He also found that longer lists cost disproportionately more to learn, that meaningful verse needed about a tenth of the effort of syllables, and that repetitions spread over days beat the same number crowded into one.
✨KEY IDEAS
Most of what will be lost is lost within the first hour, and the decline then flattens.
Something can be impossible to recall and still be quicker to relearn.
A stanza of poetry took roughly a tenth of the repetitions that the same amount of nonsense did.
Thirty-eight repetitions spread across three days did the work of sixty-eight in one sitting.
Seven syllables could be learned in a single reading. Twelve needed many more.
📝READING NOTES
Paraphrase, not quotation. First published in German in 1885. I read the English translation of 1913, in a scan, on 25 and 26 April. It is dry and very short, and he is scrupulous about what he does not know.
Why a book from 1885
The midterm on 1 April came back as a 68. Three weeks earlier I had known the material. I wanted to see the original evidence for the 🧩 Forgetting curve before building an experiment on a diagram from a slide.
What the method teaches, apart from the result
- He controlled what he could: same hour of day, same pace, same criterion.
- He reports his averages with their variation, and worries in print about whether one man's memory generalises. Hale would approve.
- He picked a measure that still works at zero. Recall hits the floor quickly. Savings keeps registering.
Cautions
- One subject. Himself, in his thirties, highly practised by the end.
- Nonsense material. The famous curve is for content with no meaning, learned to one bare recitation. My lecture notes are not that, and by his own data meaningful material behaves far better. So the popular curve overstates how fast I forget something I understood.
- The percentages are often quoted as how much is remembered. They are savings in relearning time. Those are different quantities.
- No formula of his gets used today in the form he gave it. The fitted curve in the formula note is a later convenience.
So I take the shape and not the numbers: fast, then slow.
What I built on it
🧪 Spaced review of lecture notes, six weeks before the spring final. The design follows his spacing result: short reviews at widening gaps, the first within a day of the lecture.
And two permanent notes:
- 💎 Forgetting is fastest right after learning, a seed of two sentences.
- 💎 Notes should be written for the person who has forgotten everything. This is the one that changed how I write here. The reader of this note is me in a year, holding perhaps a fifth of it, which is his one-month figure and probably generous. So every note has to carry its own context. 📖 Matuschak on evergreen notes reaches the same rule from the design side.
The final was a 74. Six marks is not proof: the exams differed and so did the weeks before them. I am recording it as consistent with the experiment and no more than that.
Reading record
- Edition: English translation of 1913, read as a scan
🗣️IN SHORT
- In : Memory: A Contribution to Experimental Psychology, 1885, English translation 1913
- Ebbinghaus argues that : retention can be measured exactly, and that it falls quickly just after learning and slowly thereafter
- by stating that : relearning a list of meaningless syllables takes less time than first learning it, that the saving shrinks steeply in the first hour and gently over the following month, and that spaced repetition and meaning both slow the loss