The Forgetting Curve: Why You Forget Almost Everything You Study (and What Actually Slows It Down)
Published: September 10, 2026 | By: Roman Vasylyshyn, founder of Lambdio | Reading time: 8 min
Here's an uncomfortable fact about studying: if you don't review something again, you will forget most of it within a month — often within days. This isn't a personal failing or a sign you studied "wrong." It's the default behavior of human memory, and it was measured with startling precision over 130 years ago.
Understanding why forgetting happens on a predictable curve — and what specifically bends that curve — is more useful than almost any study tip, because it explains why some study habits (rereading, highlighting, cramming) feel productive but barely work, while others (that feel harder) work dramatically better.
Table of Contents
- The Experiment That Started It All
- What the Curve Actually Looks Like
- Why Memory Decays This Way
- The One Thing That Reliably Bends the Curve
- Modeling Forgetting: From Ebbinghaus to FSRS
- What This Means for How You Should Actually Study
- How Lambdio Applies This
The Experiment That Started It All
In the 1880s, a German psychologist named Hermann Ebbinghaus wanted to study memory in its purest form — without the interference of meaning, association, or prior knowledge. So he invented nonsense: thousands of meaningless three-letter syllables like "WID" or "ZUC," memorized lists of them himself, and then tested his own recall at increasing delays — 20 minutes later, an hour later, a day later, a week later, a month later.
He was, essentially, running a spaced-repetition experiment on himself before the term existed, using nothing but a notebook and enormous patience. The pattern he found was so consistent that it's still called Ebbinghaus's forgetting curve, and it's held up remarkably well against more than a century of follow-up research using real, meaningful material rather than nonsense syllables.
What the Curve Actually Looks Like
The shape of forgetting isn't linear — it isn't as if you lose a fixed percentage of a memory every day. It's steep at first and flattens out over time: you lose a large chunk of new information within the first day, then the rate of loss slows considerably.
Here's the general shape, based on Ebbinghaus's original retention data and modern replications, for information studied once and never reviewed again:
Roughly 40% of newly learned material is gone within an hour if it's never revisited. By a month out, most people retain only around a fifth of it. This isn't a reason to despair — it's actually the key insight that makes spaced repetition possible at all, as you'll see below.
Why Memory Decays This Way
The forgetting curve isn't a flaw in how the brain is built — it's closer to a feature of an efficient filing system. A brain that retained every sensory detail of every day with equal strength forever would be paralyzed by irrelevant information. Instead, memory appears to work on something like a use-it-or-lose-it principle: information that's never retrieved gets weaker, while information that's successfully retrieved gets reinforced and becomes more resistant to future decay.
That second part is the important one. It means the forgetting curve isn't fixed — it's a description of what happens to a memory if you do nothing. The moment you retrieve that memory again, successfully, the entire curve changes shape.
The One Thing That Reliably Bends the Curve
Ebbinghaus also tested what happened when he re-studied the same lists after a delay, rather than only measuring decay from a single session. The re-learning was faster each time, and — critically — each successful review made the next forgetting curve flatter. A memory reviewed right as it's about to be forgotten decays more slowly the second time around than it did the first.
This is the empirical basis for spaced repetition as a technique: if you review information right before you'd otherwise forget it, each review buys you a longer interval before the next one is needed. Reviewed enough times, at the right intervals, a memory that would have vanished in a week can last years.
Compare the single-session curve above with what happens when the same material gets three well-timed reviews:
The sawtooth pattern in the second line is the actual mechanism of spaced repetition: retention drops, a review pulls it back up close to 100%, and each subsequent drop is shallower and slower than the last, because the underlying memory has been strengthened. This is why spaced repetition schedules stretch out over time — daily, then every few days, then weeks, then months — rather than staying fixed.
Modeling Forgetting: From Ebbinghaus to FSRS
Ebbinghaus described the shape of forgetting; he didn't have a way to predict it precisely for a specific fact, for a specific person, at a specific moment. For most of the 20th century, spaced repetition software approximated this with simple heuristics — Piotr Wozniak's SM-2 algorithm (1987), used by classic Anki, is the best-known example: track one number per fact and multiply the interval by it after each review.
More recent algorithms attempt to model the curve itself, explicitly, per fact. FSRS (Free Spaced Repetition Scheduler) — now the modern standard, and the algorithm Lambdio runs on — represents each piece of material with three values: Difficulty, Stability, and Retrievability, where retrievability is a direct, continuously decaying estimate of exactly where you currently sit on your personal forgetting curve for that fact, right now. We covered how that model works in detail in our FSRS vs. SM-2 breakdown — the short version is that instead of eyeballing the shape of forgetting the way Ebbinghaus's contemporaries had to, modern systems calculate it.
What This Means for How You Should Actually Study
A few direct, practical conclusions follow from the shape of this curve:
- Cramming produces an illusion of mastery that evaporates fast. A study session that ends with you feeling confident is often measuring short-term familiarity, not the kind of stable memory that survives a week — the initial part of the curve is the steepest.
- Rereading notes doesn't create a review event in the sense that matters. Passive re-exposure barely dents the curve compared to active retrieval — actually trying to recall the answer, and struggling a bit, is what reinforces the memory. This is why testing yourself beats rereading, even when rereading feels more thorough.
- The right time to review is uncomfortable, not comfortable. If recall still feels easy, you're reviewing too early and wasting time. The ideal review happens close to the point where you'd otherwise have forgotten — which is exactly what makes spaced repetition scheduling algorithms useful instead of guessing.
- One good review schedule beats studying longer in a single sitting. Because each successful retrieval flattens the subsequent curve, four well-timed 10-minute reviews spread over a month will typically outperform a single 40-minute cram session for long-term retention, even though the cram session feels like "more studying."
How Lambdio Applies This
Every chapter in Lambdio has its own individual forgetting curve, tracked per student. When you finish a learning or review session — in Standard, Socratic, Chat Review, Feynman, or Quiz mode — your Again/Hard/Good/Easy rating updates that chapter's estimated difficulty and stability, and the FSRS engine recalculates exactly when retrievability is predicted to drop to the target threshold (90% by default). That's the review date you see next.
In practice, this means the platform is constantly doing what Ebbinghaus did by hand with nonsense syllables and a notebook — just per student, per chapter, recalculated after every single review, instead of once, on one person, in a Leipzig laboratory in the 1880s.
If you want to see your own forgetting curve in action rather than read about someone else's, courses are free to start at lambdio.com.
Sources: H. Ebbinghaus, "Über das Gedächtnis" (1885); Cepeda et al. (2006), meta-analysis of spaced vs. massed practice; FSRS4Anki Wiki.