Ebbinghaus Forgetting Curve: Why Memory Fades

A familiar page can feel easy to read while its ideas remain hard to recall. For revision, the useful question is what a memory test measures. Ebbinghaus timed relearning, rather than counting remembered facts. This guide follows his experiments, the savings score and later critiques before turning to study plans and classroom use.

On this page

• Ebbinghaus forgetting curve theory: what was measured?
• Ebbinghaus used nonsense syllables to limit meaning
• Forgetting curve formula and savings score
• How much do we forget in 24 hours?
• Is the forgetting curve real? Evidence and critiques
• What are the four types of forgetting?
• Learning quality, interference and individual factors shape retention
• Sleep and emotion shape what learners retain
• What are the 7 common causes of forgetfulness?
• Meaningful learning still needs review
• Spaced practice supports longer recall than cramming
• Forgetting curve study schedule: spaced repetition
• Recall before checking makes study gaps visible
• Teachers can combine spaced review, retrieval and memory cues
• Forgetting curve resources: PDFs, slides, podcast and quiz
• Frequently Asked Questions
• Further reading and research references

‍

Ebbinghaus forgetting curve theory: what was measured?

These tests ask different questions. A savings score tells us how much faster relearning was, rather than how many facts someone could recall.

The original task measured relearning after a delay. It differs from the short-term demands described in our guide to working memory in the classroom.

What does each study establish? Ebbinghaus measured the effort saved on relearning syllable lists. Murre and Dros (2015) repeated that task with one participant. Cepeda et al. (2006) reviewed 839 assessments from 317 experiments in 184 articles. Their findings support spaced practice.

Later work on retrieval practice examines remembering through tests. Broader metacognitive strategies involve planning and monitoring learning as well as review.

The observed loss was fastest early in the delay. Later points fell more slowly. Ebbinghaus fitted equations to his results. A good fit describes those data; it leaves the causes of forgetting open to further study.

The steep early part and the slower tail are features of the fitted results. They describe the rate of change, rather than two separate kinds of memory. Later research tests how that shape changes with the task and the material.

‍

Key Insights:

• For a related account of demands during learning, see cognitive load theory.
• The Ebbinghaus Forgetting Curve illustrates the rapid loss of information if it isn't revisited, emphasising the need for consistent review.
• Early savings fell faster than later savings. The graph supports a pattern of change, rather than a universal two-hour deadline.

‍

‍

Ebbinghaus used nonsense syllables to limit meaning

‍

Who is the father of memory research?

Hermann Ebbinghaus is the name usually meant by 'father of memory research'. He brought controlled experiments to learning and forgetting, using himself as the participant and nonsense syllables as test material.

Ebbinghaus (1850-1909) was a German psychologist. He reported experiments in 1880 and published Über das Gedächtnis in 1885. Its English translation, Memory: A Contribution to Experimental Psychology, appeared in 1913. Murre and Dros (2015) explain the method and its history.

Ebbinghaus used nonsense syllables, short consonant-vowel-consonant items such as BOK or DAX. They reduced links to familiar words and existing knowledge. He studied lists and then tested himself until he could recite them correctly. The controlled material made it easier to compare learning and relearning.

He then learned the same lists again after a delay. Savings captured how much less effort the second attempt required. Faster relearning could reveal a remaining memory trace even when free recall was weak. The starting level of learning and the standard used for relearning both matter.

The design used one participant: Ebbinghaus himself. Repeated trials helped him compare delays within that person. They also raised a question for later researchers: would the pattern hold with other people and other material? Replication is one way to test that question.

The material also shaped the task. Familiar facts come with meanings and links that syllable lists were designed to reduce. Ebbinghaus's method gave researchers a controlled starting point. Later experiments could vary meaning, list position and test type to explore how each affected forgetting.

‍

Forgetting curve formula and savings score

Ebbinghaus fitted a logarithmic equation to his savings data: b = 100k / ((log t)^c + k). Here, b is savings as a percentage and t is the delay in minutes. The constants c and k are fitted to the data. The equation describes his results over the measured delays.

Murre and Dros (2015) compare this fit with power and exponential functions. These are different ways to describe a curve. Choosing a good mathematical fit and explaining the memory process are separate tasks.

The savings score is Q(t) = (L - L_t) / L. L is the effort needed to learn the material first. L_t is the effort needed to relearn it after delay t. Multiply Q(t) by 100 to express savings as a percentage.

A simple exponential model is often written as R = e^-t/S. R represents a model's retention value, t is elapsed time and S controls its rate of decline. Larger S gives a slower drop. This simplified model differs from Ebbinghaus's original logarithmic fit.

Compare curves only when their measures are clear. A timed relearning task, a recognition test and a free-recall test can produce different scores for the same material. A useful graph names the measure on its vertical axis and the delay on its horizontal axis.

‍

How much do we forget in 24 hours?

After one day, Ebbinghaus's relearning savings was about 34%. Relearning saved roughly a third of the original effort. This measure differs from the percentage of facts recalled from memory.

The earlier savings values were about 58% after 20 minutes and 44% after one hour (Murre and Dros, 2015, reviewing the original data). The drop was steeper early on. Those figures describe his syllable-list task.

After six days, savings was about 25%. A quarter of the first learning effort was still saved when the list was relearned. Slower change at later delays means the curve has a long tail. The result describes relearning, rather than a fixed amount of knowledge left.

A review schedule is a separate decision. It depends on the material, the test date and how well recall is holding up. The schedule later in this guide gives a starting pattern. Use later spacing evidence when choosing the gaps between study sessions.

‍

Is the forgetting curve real? Evidence and critiques

Yes, Murre and Dros (2015) replicated the broad savings pattern with a similar syllable-list task. This does not establish a universal percentage of forgetting for every person, subject or memory test.

The original study and the 2015 replication each used one participant. That allows close tracking of one person's results. It does not show how much people differ, or establish the same pattern across ages, languages and settings. The five issues below explain what a reader should check.

The curve circulates on X both as a number and as a shape. These are different claims.

‍

Ebbinghaus forgetting curve vs linear forgetting

Fisher and Radvansky (2019) found that a straight line fitted forgetting better in three experiments using well-learned material. The shape of forgetting can depend on what is learned and tested.

The difference concerns the rate of loss. A curved pattern falls steeply at first and then slows. A straight line describes a more steady decline over the observed period. Each is a description of results within a task.

Radvansky et al. (2022) propose the Memory Phases Framework. It separates working, early, transitional and long-lasting memory. These phases help explain why one continuous function may fit some periods better than others.

‍

Read Ebbinghaus's 'Memory' (1885) for background. Brown, Roediger, and McDaniel's 'Make It Stick' applies this to learning. Journals share spaced repetition research. The Learning Scientists offer free classroom resources. Cognitive science features in teacher training.

The following reference list is retained for further reading. Its topics range from human relearning to models of production and artificial synapses. These studies address different tasks. Read their methods before treating them as evidence for the same form of forgetting.

1. Murre, J., & Dros, J. (2015). Replication and Analysis of Ebbinghaus’ Forgetting Curve. PLoS ONE, 10.
2. Roe, D. G., Kim, S., Choi, Y. Y., Woo, H., Kang, M., Song, Y., Ahn, J., Lee, Y., & Cho, J. (2021). Biologically Plausible Artificial Synaptic Array: Replicating Ebbinghaus’ Memory Curve with Selective Attention. Advanced Materials, 33.
3. Jaber, M., & Bonney, M. (1996). Production breaks and the learning curve: The forgetting phenomenon. Applied Mathematical Modelling, 20, 162-169. 
4. Miller, R. R. (2021). Failures of memory and the fate of forgotten memories. Neurobiology of Learning and Memory, 181.
5. Hewitt, D., Sprague, K., Yearout, R., Lisnerski, D., & Sparks, C. (1992). The effects of unequal relearning rates on estimating forgetting parameters associated with performance curves. International Journal of Industrial Ergonomics, 10, 217-224.

‍

What are the four types of forgetting?

Four useful explanations of forgetting are trace decay, interference, retrieval failure and weak consolidation. These concern changes to the memory trace, competition between memories, missing cues and limited stabilisation after learning.

These are ways to organise the topic. Their roles overlap, and researchers use other groupings too. The curve describes changes in performance over time. Wixted (2022) examines the competing explanations of that change.

A cue can make a memory easier to access. Multiple-choice tests supply possible answers, while free recall asks for an answer without those options. Recognition and recall therefore reveal different aspects of performance (Tulving, 1974; Brown, 1976).

The curve describes a general trend to forget new information when there is no attempt to retain it. The shape of the curve varies between individuals. 

According to research, learners remember key information better. Taking a new perspective can bring back details that were not recalled before (Anderson & Pichert, 1978). This helps learners retain facts for longer periods of time (Bransford et al., 2000).

‍

‍

Learning quality, interference and individual factors shape retention

The total delay matters, but so do the material and the test. Ebbinghaus's savings fell most quickly early on. Other tasks may show a different pace of loss. Comparisons are clearest when the study method and test remain the same.

Bjork and Bjork (1992) proposed that how well something is learned (storage strength) and how easily it can be retrieved (retrieval strength) are separate, and that retrieval strength falls with disuse. Craik and Lockhart (1972) proposed that deeper processing leads to more durable memories.

Learners recall practical facts well when the material connects to their existing knowledge. Motivation and personal relevance support stronger encoding, in line with the levels-of-processing framework (Craik & Lockhart, 1972). Information that learners process for personal meaning is easier to retrieve later. When new information is similar or related to prior learning, it can impact the decline in memory in both directions.

New information sticks when learners link it to prior knowledge (Bartlett, 1932). Existing knowledge offers helpful cues for recalling new facts (Anderson & Pichert, 1978). This helps embed learning in long-term memory (Schank, 1982).

Learning new things can interfere with older memories. Ringbom (2007) examined how similarity between languages shapes foreign language learning.

There will be individual differences in memory strength, even for nonsense syllables (three-letter ‘words’ that have no meaning). Some of the reasons for this variation include:

‍

Use these figures to distinguish savings from correct answers on a recall test. Source: Murre and Dros (2015), reviewing Ebbinghaus’s data.

‍

Sleep and emotion shape what learners retain

Sleep, emotion, and memory retention are closely connected factors in how new learning is stored, strengthened and remembered over time. What happens after learning truly matters. Studies by Walker and Stickgold show that sleep strengthens new memories. The brain organises and secures new facts while we sleep. If learners meet a key idea once and never revisit it, that memory will fade.

Emotion also shapes what is remembered. McGaugh's work on emotional memory shows that information linked to meaning, surprise or relevance is often retained more strongly than material presented in a flat way. In lessons, this does not mean adding drama for its own sake. Use a vivid example, a puzzling question, or a real-life application. These give learners a reason to attend carefully and connect new learning to something that matters.

However, strong negative feelings can harm memory. When learners are anxious, tired, or overloaded, their focus shrinks. This makes early learning weaker and forgetting more likely. Calm routines, clear teaching, and short recall tasks help.

These are very useful before tests. They build focus without turning practice into a threat. Memory improves when teaching makes learners feel alert, safe, and ready to think.

‍

What are the 7 common causes of forgetfulness?

For study tasks, check weak attention, shallow initial learning, a long delay, interference, missing retrieval cues, poor sleep and high stress. Each of these seven factors can affect remembering.

This is a learning checklist, rather than a clinical list of causes. Several factors can overlap. A forgetting graph alone leaves the cause unclear; Wixted (2022) reviews theories that seek to explain the change.

Sleep helps learners remember things. Physical health matters too.

Encoding means forming a memory from an experience. Attention, meaning and links to existing knowledge shape that process (Craik and Lockhart, 1972). Stronger first learning gives later remembering a better starting point. Sleep and emotion are discussed separately below.

‍

‍

Meaningful learning still needs review

A common myth is that the same forgetting percentages apply to everyone. Ebbinghaus controlled his learning material with nonsense syllables. Everyday study adds meaning, prior knowledge and different goals. Check the task and measure before applying a chart's numbers to a course.

Another misconception is that meaningful learning is somehow exempt from forgetting. Ideas linked to prior knowledge, stories or clear examples tend to last longer. This fits with schema theory and research on depth of processing. However, understanding does not remove the need for review.

If you teach photosynthesis through a memorable practical, learners may grasp it well on the day. They still benefit from short retrieval questions in the next lesson and again a week later.

Low stakes quizzes, mini whiteboard checks and last lesson, last week, last month starters all turn forgetting into a useful teaching tool. The goal is not perfect recall after one explanation, but stronger recall after repeated retrieval.

A final myth is that revision only matters before an assessment. Spaced review works best when it is built into normal classroom routines, not saved for a revision week. Teachers can revisit key vocabulary with brief cumulative quizzes, ask learners to explain a concept in their own words, or interleave older material into homework and do-now tasks. These small moves respect the original insight behind the curve while adapting it to real learning, where meaning, attention and practice all matter.

‍

Spaced practice supports longer recall than cramming

Cramming is a form of massed practice that boosts short-term fluency but weakens recall compared with spaced repetition. The problem is that this fluency is often mistaken for real learning. Ebbinghaus' work on forgetting, and later research on the spacing effect, shows that massed practice may boost short-term performance. But it leads to much weaker recall after a few days. For teachers, this matters because a lesson that seems secure on Friday may be largely forgotten by Monday.

Spaced repetition works differently because it asks learners to return to ideas after a gap, when some forgetting has already begun. That small struggle to remember is useful, it strengthens the memory and makes future recall easier. Cepeda and colleagues' research on distributed practice suggests that learning spread over time leads to better long-term retention. The comparison is with the same amount of study packed into one sitting. In classroom terms, learning needs planned return points, not just initial exposure.

A simple example is vocabulary teaching. Instead of spending twenty minutes drilling ten new words once, introduce them briefly. Revisit them the next day with a low-stakes quiz. Bring them back a week later in a sentence task.

The same principle works in maths, where a method taught on Tuesday can reappear in a mixed starter the following week. These short return visits are often more effective than one large revision block.

‍

Forgetting curve study schedule: spaced repetition

‍

What is the 2-7-30 rule for memory?

The 2-7-30 rule is a study convention that schedules reviews 2, 7 and 30 days after first learning. It offers a starting timetable that you can adjust as recall improves or weakens.

Day 0 below means the first learning session. All review dates are counted from that session. The pattern is a planning aid. The studies cited here support spacing and adjusting gaps, rather than these three exact dates.

Cepeda et al. (2006) found that the best study gap depends on the retention interval, the time until the final test. Mawson and Kang (2025) reviewed 22 classroom reports and 31 effect sizes. Spaced practice showed a moderate average advantage over massed practice (d = 0.54).

The effect size expresses a difference in standard-deviation units. It is not a percentage gain in marks. For a test next week, fit several short sessions before it. For a later exam, spread reviews across the longer period and keep checking recall.

Spaced repetition means returning to learned material after gaps. Our guide to spaced practice explains how to plan it. It draws on the spacing effect: practice spread over time often leads to better delayed recall than the same practice in one sitting. The useful gap depends partly on the final test date.

One useful strategy is the cumulative starter task. Learners answer two questions from yesterday. They answer two from last week, and one from last month. Another method uses flashcards or retrieval grids.

These cycle back to earlier topics. This helps in subjects like science, history, and languages. Precise knowledge matters in these subjects. Homework can also follow a spaced pattern.

Short review tasks can revisit earlier material. This is better than only practising the most recent lesson.

Organisation makes this manageable. Pick the knowledge you need, set a review date and record what you recalled. A flashcard score can help choose the next gap. Keep time for checking errors and explaining the idea, as well as answering from memory.

‍

Forgetting curve in workplace training

For workplace training, schedule short checks of key procedures after the first session, then review errors before the next practice task. Use different scenarios to test whether the knowledge transfers.

Evidence on spacing includes verbal recall and classroom learning. Check the effect in the workplace task itself, rather than assuming a laboratory curve predicts job performance.

‍

‍

Recall before checking makes study gaps visible

A useful study habit tests what you can bring to mind after a delay. Try to answer first, then check against your notes. This makes gaps visible. You can then correct errors and choose what to revisit in the next session.

Roediger and Karpicke (2006) found that students who took tests remembered more a week later than students who only restudied. It works without needing special techniques. Agarwal and Bain (2019) state that regular recall boosts long-term memory.

Active recall asks you to retrieve information before checking the answer. A question is itself a cue, so recall can be prompted. The key is whether you work out an answer from memory instead of copying it from material in view.

Retrieval practice helps learners remember facts (Karpicke & Blunt, 2011). Active recall means answering questions, defining words, or taking tests (Brown, Roediger & McDaniel, 2014).

Some study tasks keep the source in view and offer little chance to check recall. Multiple-choice questions can still involve active thinking when you choose an answer and use feedback. The process matters more than the task label.

• Re-reading notes
• Highlighting text
• Looking at answers before trying a question

Summarising a page of text or using flashcards can involve either active or passive recall. It depends on how each task is approached.

Mnemonic techniques are one way to help learners remember lists of words. Learners can use dual coding with vivid images to memorise key words or definitions (Paivio, 1971). This makes the material more memorable for learners. For further reading on this topic, explore our guide to Declarative Memory.

‍

‍

Teachers can combine spaced review, retrieval and memory cues

Teachers can plan spaced review, ask learners to retrieve answers and make links to prior learning clear. Use a recall check to see which ideas need another explanation. The Ebbinghaus graph starts the discussion; later studies provide more direct evidence about classroom practice. For related tasks, see our active learning techniques.

Use one row to plan a recall task. Check the answer before choosing the next review.

‍

Year 8 science: checking particle explanations after a gap

For Year 5 French, use the vocabulary routine described above: recall words after a gap, then use them in sentences. If an older word replaces the target word, compare the pair before another recall check.

‍

Repeated Retrieval Practice

Retrieval practice combats forgetting. Learners should recall new information soon after learning, then again at gaps suited to when they need it (Cepeda et al., 2008).

Keep revisiting after recall succeeds, because repeated retrieval strengthens long-term retention (Roediger & Butler, 2011).

‍

Techniques to Improve Memory

Mnemonic devices focus on encoding new information in a way that will make it easier to retrieve it in the future. Using acronyms can be very successful when it is necessary to learn the order of a list of words:

‍

Memory Prompts During Learning

Link new facts to ideas learners already know. In a schema, related knowledge forms an organised pattern. A learner can use those links to explain a new idea or recall a related fact (Bartlett, 1932).

Physical cues or the setting can help learners remember. The place where learning happened can act as a cue (Godden & Baddeley, 1975; Smith, 1979). Learners can also write key terms in colour or capitals and keep them in a set place on the page.

Smith (1979) examined how the setting can cue recall. Familiar surroundings may help bring learned material to mind. For classroom review, vary the prompts and check whether an answer is still available in a fresh context. This tests how much recall depends on the original cues.

‍

Memory science in classroom practice

Plan review across lessons, rather than waiting for one revision week. A quick check tells you which facts learners can retrieve. A later check tests whether recall lasts. Feed those results into the next lesson and the next return to the topic.

Ebbinghaus' research shows how learners forget. Teachers can adapt practice to improve memory retention. Consider these approaches, based on Ebbinghaus (1885). Use strategies from researchers like Baddeley (1986) and Brown et al (2014). Review practices suggested by Bjork & Bjork (1992) too.

‍

Supporting learners with special educational needs and disabilities (SEND)

Supporting SEND learners with memory review involves maintaining curriculum ambition while reducing memory load through carefully structured support. This means keeping high expectations for the core curriculum. At the same time, we must reduce their memory load. We do this through clear and well-paced support.

Some learners struggle with working memory or attention control. Others have specific learning difficulties. Their memory loss speeds up if teaching is rushed. It also happens when lessons use too much language.

This shows how SEND adaptations fit into quality-first teaching. Keep the core curriculum ambitious for all learners. However, we must change how we present and rehearse information.

A practical response is micro-spacing: revisiting key knowledge in short, frequent bursts rather than waiting for a weekly recap. A teacher might say, “Tell your partner the two causes of evaporation in ten seconds,” as a short prompt. Repeat it later in the lesson with a visual cue and again the next morning. Learners might first produce single words, then a full sentence, then a labelled diagram. This shows retrieval building over time rather than being judged in one high-pressure moment.

This matters a lot. Retrieval practice only helps SEND learners if the task is accessible. Low-stakes checks must reduce barriers, not add them. You can offer sentence starters and dual coding.

You can also use partly finished examples. Try oral practice before independent writing. The Education Endowment Foundation (EEF) has guidance on Special Educational Needs in Mainstream Schools.

It says targeted scaffolds must help learners access the same learning. They should not divert learners onto a separate track (EEF, 2020).

‍

How adaptive software schedules revision

Review software can use previous answers to estimate when another check may help. It gives different questions or intervals to different learners. These estimates describe performance in the tool. Teachers still choose the content and check the errors that need explanation.

In practice, the software handles the timing maths instead of the teacher. A Year 8 history teacher might end a lesson by setting four retrieval questions. However, each learner gets a different set of questions. One learner might confuse alliance and entente.

The software will test them again tomorrow. Another learner might explain militarism well. The software will test them next week.

Both write short answers in their books. The teacher then receives the automated results.

It found a 31 per cent reduction in lesson-planning time (EEF, 2024). For busy staff, the test is simple. If dynamic quizzing saves planning time and surfaces misconceptions earlier, it is worth a look.

Teachers choose which knowledge to revisit and whether a question tests understanding. A class may need another explanation before more recall practice. Automated timing is one part of that planning. It works best alongside checks of what learners actually say or write.

‍

Educational technology (EdTech) tools for review automation

Review platforms can schedule questions and show which answers were weak. They use quiz results to plan another check. That can help a teacher track review across lessons. The dashboard needs to be read alongside the questions and the learners' explanations.

In a Year 8 science lesson, the teacher says, "Open today's retrieval starter. Yours will not all look the same." One learner gets three questions on particle movement because she missed them last week. Another gets acids and alkalis from last month. Both see instant feedback before improving their answers.

That kind of real-time memory tracking is not a perfect map of memory. It is a practical way of building personalised forgetting curves from accuracy, timing and prior errors.

A dashboard can show which questions need another explanation. Use those results to choose the next starter or plan a class review. Check that the answers reflect the knowledge you wanted to teach. The software supports a teaching decision; the recall task supplies the evidence.

‍

‍

Forgetting curve resources: PDFs, slides, podcast and quiz

The existing Memory Retention Teaching Pack offers posters, desk cards and continuing professional development (CPD) material. The page also includes a slide deck, podcast and revision quiz. Use the quiz to check the theory, then revisit the studies behind any missed answer.

For how this idea fits among the other major models, see our overview of learning theories. Readers comparing accounts of memory and thinking often set this alongside information processing theory.