The Forgetting Curve

What Ebbinghaus actually measured, what the forgetting curve tells us, and why review timing matters for memory retention.

The forgetting curve describes a general pattern in which memory decays rapidly after initial learning and then more slowly over time. It was first observed by Hermann Ebbinghaus in the 1880s through self-experiments with meaningless syllables. The practical lesson is not a precise formula but a reliable principle: without review, retention drops — and a timely retrieval can restore it and slow future decay.

Key Takeaways

Ebbinghaus used a single-subject methodology with nonsense syllables — the curve describes a general pattern, not a universal law that applies equally to all material. - Initial forgetting is fastest; decay slows as the memory stabilizes (assuming no review). - The shape of the curve depends on meaningfulness, prior knowledge, sleep, and interference — not just time. - Spaced repetition uses the curve as a scheduling signal: review early when memory is fragile, push intervals out as memory stabilizes. - Each successful retrieval resets the curve at a higher baseline — the memory becomes more stable than before.

What Ebbinghaus actually did

Hermann Ebbinghaus conducted a series of self-experiments in the 1880s, published in 1885 and translated into English in 1913. He wanted to study memory scientifically, but the methods of his era made it hard to control what people already knew or cared about. His solution was to study material that had no prior associations: nonsense syllables — three-letter combinations like "DAX" or "BUP" that were pronounceable but meaningless in German.

He memorized lists of these syllables to criterion (perfect recitation), then waited varying intervals before attempting to relearn them. His measure was not simple recall but the savings score: how many fewer repetitions were needed to reach criterion again compared to the original learning. A high savings score meant memory was still partially intact; a low score meant most had been lost.

From these experiments, he plotted what became known as the forgetting curve: rapid initial loss of savings score, followed by progressively slower decay.

Research Highlight

Ebbinghaus memory experiments (1885)

The core observation was not a universal formula but a replicable shape: large early loss followed by slower decay, measured through savings during relearning.

The general pattern

Despite its narrow experimental basis, the broad shape Ebbinghaus observed has been replicated across many conditions: people, languages, meaningful material, and realistic contexts. The pattern holds up:

Rapid initial decline: A large portion of what was newly encoded becomes inaccessible within hours to days without review.
Decelerating decay: The rate of loss slows over time. What remains after a week tends to persist longer than what was lost in the first hour.
Asymptote: Very old, well-reviewed memories can become extremely stable and show minimal decay over long periods.

This shape is why massed practice (cramming) feels temporarily effective — you're reviewing material while the initial trace is still strong — but produces little durable retention once the trace decays.

What affects the shape

The forgetting curve is not fixed. Several factors shift how quickly a particular memory decays:

Meaningfulness. Meaningful material is encoded more richly than nonsense syllables. A medical student learning a drug name they've heard before in a different context will forget it more slowly than a meaningless string of letters. Prior knowledge creates more retrieval pathways, making individual facts easier to reconstruct.

Prior knowledge and elaboration. Information that connects to existing knowledge decays more slowly. A fact that slots into a conceptual framework has multiple redundant retrieval routes. An isolated fact has only one.

Sleep. Memory consolidation happens largely during sleep. A memory encoded just before sleep is more likely to be stabilized overnight than one encoded early in the day with many hours of interference before sleep. Sleep-deprived learners show substantially faster forgetting.

Interference. Learning similar material close together creates competition between memory traces — both retroactive interference (new material disrupts old) and proactive interference (old material disrupts new). This is one reason studying similar topics back-to-back is generally less efficient than spacing them.

Emotion and salience. Information tagged with emotional relevance tends to be remembered better. This is partly why mnemonics that create vivid, unusual images work: they add a dimension of memorability that pure rote rehearsal lacks.

Why the curve matters for scheduling

The forgetting curve explains why a single review at any arbitrary interval is insufficient for long-term retention. The question is not whether to review but when.

Interactive Curve

Explore how fast unretrieved memory falls off

The curve below is illustrative rather than person-specific, but it shows the pattern Ebbinghaus observed: rapid early loss, then a slower taper as memory stabilizes.

Retention threshold: 35%

If you review too soon (while the trace is still strong), the retrieval is too easy to produce a strong consolidation signal. If you review too late (after the trace has fully decayed), you are effectively relearning from scratch rather than reinforcing an existing memory.

Spaced repetition uses the decay shape as a scheduling signal. A new card is fragile — schedule it again soon. After a successful review, the memory is more stable — schedule the next review farther out. After each retrieval, the curve resets at a higher starting point: the memory is not just restored but made more durable than it was before.

In Neurako

Neurako's FSRS scheduler models the forgetting curve individually for each card using two parameters: stability (how long the memory is expected to last before dropping to the 90% recall threshold) and retrievability (the current probability of successful recall). When you rate a card Again, the stability drops sharply. When you rate it Good or Easy, stability grows. The schedule adapts to each card's actual behavior rather than applying a fixed decay rate to all material.

How retrieval resets the curve

A key insight that goes beyond Ebbinghaus's original work is that retrieval doesn't just interrupt forgetting — it changes the subsequent forgetting curve. After a successful retrieval event, the memory becomes more stable than it was before the retrieval. The next decay cycle starts from a higher point and proceeds more slowly.

This is why spaced retrieval is more powerful than massed rereading. Each successful retrieval is not just confirmation that the memory still exists; it is an encoding event that makes the memory more robust going forward.

Common misconceptions

"The forgetting curve gives us the precise rate of forgetting." It does not. The 20% retention at 24 hours and 10% at one week figures that appear in popular summaries are from Ebbinghaus's specific nonsense syllable experiments and do not generalize to meaningful learning material.

"Everyone has the same forgetting curve." Individual variation is substantial. Factors like sleep quality, stress, learning history, and the specific material all shift the curve. What FSRS does is estimate the curve for you and each card, rather than assuming a fixed universal shape.

"Once forgotten, it's gone." Relearning almost always takes less effort than the original learning, even when the savings score is very low. This is why review systems that catch cards after lapses are still valuable.

Spaced Repetition Explained

How interval scheduling turns the forgetting curve into a review plan.

FSRS Algorithm

How Neurako models stability and retrievability for each card.

Active Recall

Why retrieval events strengthen memory rather than just measuring it.

Sources

Ebbinghaus, H. (1885/1913). Memory: A contribution to experimental psychology. Teachers College, Columbia University. https://psychclassics.yorku.ca/Ebbinghaus/memory6.htm

Ready to turn this into a study ritual?

Start studying with Neurako