Appearance
3.2 — Forgetting, and the Spacing Effect
Between 1879 and 1885 a German psychologist named Hermann Ebbinghaus performed one of the most tedious experiments ever conducted, on a single subject: himself.
He invented lists of meaningless three-letter syllables — WUX, ZOF, KEL — specifically so that no existing knowledge could help. He memorised a list to the point of perfect recitation, recorded how long it took, then waited a set interval and measured how long it took to relearn it. The difference was his measure of what remained.
He did this for six years, thousands of lists, keeping his conditions fanatically constant.
The curve he produced is the oldest quantitative result in psychology and it has held up.
Most of what you lose, you lose fast. For meaningless material, a large fraction is gone within a day. Real material with meaning and connections decays more slowly, and the shape is the same.
The second, less famous half of Ebbinghaus's work is the part that matters. He found that the same total study time produces far more retention when it is spread out than when it is done in one sitting. That is the spacing effect, and it is one of the most robust findings in the whole of psychology — reproduced for over a century, across ages, subjects, and species.
Why spacing works
Three mechanisms, and they are not mutually exclusive.
Retrieval difficulty is the point. When you review immediately, the material is still sitting in working memory and retrieval is trivial, so no pathway is strengthened. When you review after a gap, you have to actually reconstruct it, and the reconstruction is what strengthens the trace. The effort you resent is the mechanism.
Each review happens in a different context — a different room, mood, time of day, and a slightly different you. Each one attaches new retrieval cues, so the memory becomes reachable from more directions rather than only from the context you first learned it in.
Consolidation needs time and sleep. A gap that spans a night allows the transfer described in 2.8 to run before the next round.
How long the gaps should be
The most useful practical result: the optimal gap scales with how long you need to remember the material.
A rough guide, from studies that varied the gap systematically and tested at different delays:
| Need to remember for | First gap should be around |
|---|---|
| A week | 1 day |
| A month | 2–4 days |
| A year | 2–4 weeks |
| Indefinitely | Expanding: 1 day, 3 days, 1 week, 3 weeks, 2 months |
The pattern is roughly ten to thirty per cent of the target retention interval. Do not treat these as precise — the effect is broad, and any spacing beats none by a wide margin. The difference between one gap and no gap is much larger than the difference between a good gap and a slightly wrong one. Do not spend energy optimising the schedule; spend it on having one at all.
Expanding intervals — each gap longer than the last — are the usual recommendation and are what every flashcard program implements. The evidence that expanding beats a fixed sensible interval is weaker than people assume; both work well. Expanding is more efficient because it spends fewer reviews on things you already know.
Spaced repetition as a system
An algorithm can do the scheduling. You see an item, you rate how hard it was, and the program decides when to show it again — sooner if you struggled, much later if it was easy. Over time, well-known items appear every few months and shaky ones every few days.
What it is genuinely excellent for: vocabulary in a new language, medical terminology, drug names and doses, anatomy, legal definitions, formulae, historical dates, anything where a large body of discrete facts must be available instantly and permanently. Medical students and language learners use it heavily for good reason.
What it is bad for, and this is where most people waste months: understanding, procedures, and judgement. You cannot flashcard your way to knowing how to design a system, structure an argument, or handle a difficult conversation. Turning a concept into a card usually shreds it into fragments that can be recited without being understood.
The honest division: facts go on cards. Understanding is built by explaining, solving and doing (3.3, 3.4).
The three rules that decide whether a card system works or becomes a burden:
Make your own cards. Someone else's deck contains their compressions of their understanding, and reviewing it teaches you the words without the meaning. Writing the card is a large fraction of the learning.
One fact per card. A card asking for five things fails if you get one wrong, and you will then re-review the four you knew. Split them.
Only add what you actually want to know for years. The single reliable way to abandon a system is to add three hundred cards in a week. Reviews compound; a card added today costs you time for a decade. Ten good cards a day is sustainable and adds up to thousands a year.
Cramming, fairly assessed
Cramming works. That is why people do it, and pretending otherwise makes the advice easy to dismiss.
Massed study the night before produces a real, measurable performance boost on a test the next morning. What it does not do is last. In direct comparisons, crammed material is largely gone within weeks while spaced material persists for months to years, from the same total hours.
So the honest position: if you need to pass something tomorrow and have nothing prepared, cram — it is the correct move in that situation. If you need to know something for a career, a language you will speak, or a subject you are building on, cramming is throwing away the time you spent.
And there is a specific trap worth naming. Cramming produces exactly the fluency illusion from 3.1. The material feels extremely well known at 2 am because it is still active in working memory. That feeling has almost no relationship to what will be available in three weeks.
Interleaving, which belongs here
A closely related finding, and one people find even harder to accept.
Study three types of problem — call them A, B and C — either in blocks (AAAA BBBB CCCC) or interleaved (ABCACBBAC). Blocked practice produces better performance during the session and worse performance on a later test. Interleaving produces the opposite, often dramatically.
In one well-known study of learning to identify painters' styles, students who saw the artists interleaved were substantially better at classifying new paintings than those who studied one artist at a time — and, when asked afterwards, the majority of students believed the blocked method had worked better for them. They were wrong about their own learning while holding the evidence.
The reason: blocked practice lets you apply the same approach repeatedly without deciding which approach to use. Real situations do not announce which type they are. Interleaving trains the discrimination, which is the part you actually need, and it also introduces spacing automatically since each type is revisited after a gap.
3.4 covers how to apply it. The short version: mix problem types, mix topics, and resist the strong pull to finish one thing completely before starting the next.
What to do with this
Build one habit and you have most of the value: review anything you want to keep on day one, day three, and around day ten. Three short reviews. Not re-reading — recalling with the material closed (3.3). Fifteen minutes total per topic, and it moves material from gone-in-a-fortnight to available-in-a-year.
Put the reviews in the calendar when you first learn the thing. Not on a list of intentions. A dated entry, because the whole system fails at the point where you have to remember to remember.
Never study the same thing twice in a row in one session. If you have three hours, do topic A, then B, then C, then back to A. The same hours, rearranged, produce measurably more.
Use a spaced-repetition program for discrete facts only, make your own cards, one fact each, and add slowly.
And expect it to feel worse. Spaced, interleaved practice produces more errors during study, slower sessions, and a persistent sense that you are not getting anywhere. That is what it feels like from inside when it is working, and the confident, smooth feeling of blocked re-reading is what it feels like when it is not. The discomfort is not a sign to change method. It is the method.
Next: 3.3 covers the single most powerful thing you can do with material once it is in — and the reason that closing the book is worth more than opening it again.