The Forgetting Curve & Memory

Your brain is a leaky bucket — without reps, fresh memories drain away fast. The forgetting curve maps how quickly knowledge decays, and spaced repetition is the spotter that lets you re-rack the weight just before it falls.

What it is

The forgetting curve is a model of how memory fades over time when you don't reinforce it. First mapped by German psychologist Hermann Ebbinghaus in his 1885 work Über das Gedächtnis (Memory: A Contribution to Experimental Psychology), it describes a steep initial drop in retention followed by a slow flattening. You learn something today; by tomorrow a big chunk is already gone; a week later most of it has evaporated — unless you do something about it.

That "something" is spacing. Each time you successfully recall a fading memory, the curve resets and decays more slowly. Review at the right moments and the same item that vanished in days can stick for months, then years. This is the engine behind every flashcard app and the reason cramming feels productive but disappears by Friday.

In gym terms: a single workout doesn't build muscle. Muscle is built by repeated, well-timed sessions with progressive load. Memory works identically. One exposure to a word is one rep — it does almost nothing alone. Spaced, repeated retrieval over weeks is the training program that turns a wobbly first lift into something your brain holds without strain. The forgetting curve isn't your enemy; it's the feedback signal telling you exactly when to do your next set.

The crucial mindset shift: forgetting is normal and even useful. A memory that's a little hard to retrieve, then successfully retrieved, comes back stronger than one you reviewed while it was still fresh. Difficulty is the load. No load, no gains.

The evidence

Ebbinghaus (1885) ran his experiments on himself, memorizing lists of nonsense syllables (like "WID" or "ZOF") to strip out the effect of prior meaning, then testing how much he retained after delays ranging from 20 minutes to a month. His data produced the classic exponential decay shape and two findings that still hold up: retention drops fastest right after learning, and spaced reviews dramatically slow the decay — what he called the advantage of distributed over massed practice. He also documented the savings effect: relearning something you'd "forgotten" is much faster than learning it the first time, meaning the trace was never fully gone.

A century-plus of research has confirmed and refined this. The spacing effect — that learning spread over time beats the same study packed together — is one of the most robust findings in cognitive psychology, replicated across thousands of studies. The related testing effect (also called retrieval practice) shows that the act of recalling information, not just re-reading it, is what cements it; pulling a word out of your head is far stronger than seeing it again. Researchers like Roediger and Karpicke have demonstrated this repeatedly: students who self-test outperform students who simply restudy, often by large margins on delayed tests.

For vocabulary specifically, Paul Nation's work on how words are learned shows that a word typically needs many meaningful encounters — often cited in the range of a dozen or more — before it's reliably known, and that those encounters work best spread across contexts and time. This is the bridge from dry memory science to language learning: you don't "memorize" a word once; you meet it again and again, ideally inside comprehensible input, until your brain stops treating it as foreign.

Two honest caveats. First, Ebbinghaus used meaningless syllables; real language is full of meaning, context, and emotion, all of which make memories stickier than his nonsense lists. So your real-world curve is gentler than the textbook one — especially when you learn words inside stories and conversations rather than on isolated lists. Second, no spacing trick replaces understanding. Spacing optimizes how you keep what you've genuinely encountered; it can't manufacture comprehension you never had. That's why this works as a support tool for input, not a substitute for it.

How to actually use it

You don't need to memorize the math. You need a system that schedules your reps for you and a steady stream of input that feeds it. Here's the program:

1. Get input first, cards second. Don't sit down to "memorize 50 words." Read, watch, and listen to stuff you mostly understand, and let words you actually meet become your review material. A word ripped from a list with no context is a dead lift with bad form. A word you met in a sentence you understood already has a memory hook. See finding comprehensible input.

1. Let an SRS schedule the reps. A spaced repetition system like Anki does the only thing Ebbinghaus proved you must do: show each item again right as it's about to fade, then stretch the interval each time you recall it. You rate how hard the recall was; the algorithm sets the next date. Today, tomorrow, in four days, in ten, in a month — the gaps widen as the memory hardens.

1. Retrieve, don't re-read. When a card comes up, force yourself to produce the answer before flipping. The struggle is the workout. Flipping immediately because "I kinda know it" is skipping the rep. If you peek, you cheated the set.

1. Keep cards lean and personal. One word or one short sentence per card, ideally pulled from your own reading via sentence mining. Add audio. Bloated cards with paragraphs of notes don't get reviewed — they get suspended in shame.

1. Show up daily, keep the load light. Ten to twenty minutes of reviews a day beats a two-hour binge on Sunday — that's the spacing effect applied to your routine, not just your cards. A short daily session is the difference between a training habit and an injury. Build it into your daily routine.

1. Let cards die. If a word keeps failing, the problem is usually that you haven't met it enough in real input yet. Delete or suspend it, and trust that you'll re-add it naturally when it shows up again in your reading. Forcing a word your brain isn't ready for is ego lifting.

1. Don't confuse reps with reading. SRS is the accessory work — the spotter that catches the high-value words. The main lift is always immersion. If your flashcard streak is long but your input is thin, you're polishing dumbbells you never actually press.

The honest expectation: this is a long game. Spacing makes your memory efficient, not instant. You will still forget things; the system just makes sure you re-meet them at the cheapest possible moment, again and again, until forgetting stops winning. No "fluent in 30 days." Just reps that compound.

Resources

• Anki↗ — the free, open-source spaced-repetition flagship. Brutal interface, unbeatable algorithm. (AnkiMobile on iOS is paid; everything else is free.) See our full Anki guide.
• Hermann Ebbinghaus, Memory: A Contribution to Experimental Psychology (1885) — the original. Old but readable; search the title for free English translations in the public domain.
• Paul Nation, Learning Vocabulary in Another Language — the authority on how many encounters a word really needs and how to engineer them. Search his name plus "vocabulary" for free papers and resources on his university page.
• "Make It Stick" by Brown, Roediger & McDaniel — the most accessible book on the testing effect and spacing for general learners. Search the title.
• Language Reactor↗ / LingQ — tools that turn the input you actually consume into reviewable items, so your SRS feeds on real context, not lists. See Language Reactor and LingQ.

Related

• Spaced Repetition (SRS)
• Anki: The Complete Guide
• Retrieval Practice & Interleaving
• Sentence Mining
• Vocabulary Acquisition
• Comprehensible Input (Krashen)
• The Goldlist Method
• Building Your Daily Routine