HPC › Learning › Glossary
Forgetting Curve is the exponential decay function first described by Hermann Ebbinghaus in 1885, quantifying how recently acquired information degrades over time without deliberate review. The rate of loss is steepest immediately after encoding and decelerates progressively. Spaced repetition exploits this pattern by scheduling retrieval at intervals that rebuild memory strength before decay becomes irreversible.
The exact mathematical form (exponential, power function, or composite) remains contested; the curve is better understood as a family of decay models than a single equation.
How it works
Ebbinghaus established the forgetting curve in 1885 by memorising and relearning lists of nonsense syllables, measuring the time saved on relearning compared to the original study session. 1 He found that retention loss was sharpest in the first 20 to 60 minutes after encoding, then decelerated as time passed. A 2015 replication by Murre and Dros applied the same savings method and closely matched his original data: retention fell to approximately 56% after one hour, 44% after nine hours, and 26% after 31 days for nonsense-syllable material. 2
The mathematical form of the curve remains contested. Ebbinghaus proposed an exponential decay formula, but subsequent researchers have found that power functions or composite exponential models may better capture individual-level forgetting, particularly over extended retention intervals. 2 The rate of forgetting is also modulated by contextual factors: Radvansky et al. found that event boundaries (shifts in physical or mental context between encoding and retrieval) steepen the curve, indicating that the underlying mechanism is not uniform decay but is sensitive to how tightly encoding and retrieval conditions are linked. 4
Example
A team completes a half-day compliance or technical training session. Within 24 hours, most participants have lost the majority of the specific procedural details covered. When tested a week later without any intervening review, retention of precise rules and sequences is low enough to create operational risk. The same material, distributed across three shorter sessions over a week with brief retrieval practice between each, yields substantially higher retention at the point of actual need.
The forgetting curve does not penalise learning; it penalises the assumption that a single exposure is sufficient for durable retention.
Why it matters
The forgetting curve matters because most educational and training systems are designed around massed study: intensive blocks of instruction followed by an extended gap before any further encounter with the material. Under this model, the steep initial loss documented by Ebbinghaus and confirmed by Murre and Dros 2 means that the gap between instruction and application is precisely where retention collapses. The curve is also not uniform across material types: arbitrary or decontextualised content decays faster than material that is emotionally salient, richly contextualised, or connected to prior knowledge. 4
The practical counter-strategy is distributed practice. Cepeda et al.'s synthesis of 317 experiments demonstrated that separating study into multiple sessions with gaps between them consistently produces superior long-term retention compared to equivalent time in a single block, with optimal gap length scaling to the target retention interval. 3 Spaced repetition systems formalise this principle by scheduling each item for review just before its predicted decay point, allowing large knowledge bases to be maintained with a fraction of the review time that unstructured restudying would require.
How quickly does the forgetting curve drop after learning?+
The drop is steepest in the first hour after learning. Murre and Dros's 2015 replication recorded only 56% retention after 60 minutes and 44% after nine hours. Without any review, the majority of specific verbal material is typically gone within a day.
Can the forgetting curve be flattened through spaced repetition?+
Yes. Spacing study across multiple sessions with gaps between them reliably produces superior long-term retention compared to massed practice. Cepeda et al.'s analysis of over 300 experiments confirmed that even modest spacing dramatically reduces the effective decay rate, and that the optimal gap grows with the length of the desired retention interval.
Does the forgetting curve apply equally to all types of material?+
Not equally. The curve is steepest for isolated, decontextualised material and shallower for content that is emotionally significant or connected to an existing knowledge framework. Depth of initial encoding is the key moderating variable: the richer the encoding conditions, the slower the subsequent rate of decay.
What is the savings method Ebbinghaus used to measure forgetting?+
The savings method measures forgetting indirectly by comparing the time required to relearn material to a set criterion against the original study time; a shorter relearning time indicates that some memory trace persists. Ebbinghaus used this approach because it is more sensitive to residual memory than simple recall tests.
