We Taught It. They Forgot It. We Moved On.

There is a finding in cognitive psychology so robust, so consistently replicated, and so directly relevant to what happens in every classroom in Britain every day, that its continued absence from how we design schools, timetables, and curricula should count as a professional embarrassment. It has been available to us since 1885. Most of the system ignores it.

The finding is this: without any form of reinforcement, human beings forget approximately 70% of new information within 24 hours of encountering it. Within a week, that figure climbs closer to 90%. The curve of forgetting is steep, rapid, and merciless - and it applies to every lesson taught in every school, to every student sitting in every classroom, right now.

But before we reach for the easy outrage - before this becomes another piece about teachers failing to read the research - it is worth being precise about what this story is actually about. It is not a story about ignorant professionals. Most teachers are neither unaware of memory science nor indifferent to it. It is a story about a system so structurally committed to coverage that retention has been rendered structurally impossible. The architecture of schooling - its timetables, its specifications, its accountability frameworks - does not just fail to support long-term memory. In several important respects, it actively works against it.

That is the argument here. Not that teachers should do better, but that the system within which they work was never designed with memory (or anything long-term for that matter) in mind - and that this should trouble anyone who believes education is supposed to produce something durable.

Ebbinghaus, and the limits of his own discovery

Hermann Ebbinghaus was a German psychologist working in Berlin in the 1880s who did something both admirable and slightly peculiar: he spent years using himself as his sole experimental subject, memorising lists of meaningless nonsense syllables - DAX, BUP, ZOL - and carefully tracking how quickly he forgot them. The result was the forgetting curve: a mathematical description of the rate at which information decays from memory without reinforcement, and one of the most replicated findings in the history of psychology.

His work also revealed the concept of the spacing effect: that distributing practice across time produces dramatically stronger retention than concentrating it in a single session. Study something today, revisit it in three days, again in a week, and again in a month - and you will remember far more of it, far more reliably, than if you spent the same total time on it in one sitting. This finding, too, has been replicated extensively across more than a century of subsequent research.

There is, however, a legitimate criticism of Ebbinghaus that deserves to be named rather than glossed over. His experiments used one person - himself - memorising material that was deliberately stripped of meaning, context, and relevance. The cognitive psychologist Ulric Neisser argued forcefully that laboratory memory research of this kind tells us relatively little about how meaningful knowledge is actually acquired and retained in real-world contexts. Remembering nonsense syllables is not the same thing as remembering why the First World War started, how to construct a conditional clause in French, or what happens when you add acid to an alkali.

The criticism is fair, and it matters - because arguments built on Ebbinghaus alone are arguments built on somewhat shaky ground. But the forgetting curve does not depend on Ebbinghaus alone. Harry Bahrick, a psychologist at Ohio Wesleyan University, spent decades studying long-term retention of genuinely meaningful content. In one landmark study, he tracked people’s retention of high school Spanish over periods of up to fifty years - real content, learned in real classrooms, by real students. The findings were sobering. Without any rehearsal or use of the language, participants forgot the overwhelming majority of what they had learned within three to six years of leaving school. The curve was less steep than Ebbinghaus suggested for meaningless material, but the direction was the same. Time, without reinforcement, dissolves learning.

I know it anecdotally myself. In school, not that long ago (!), I studied Latin & Classics (I know, I am cultured after all). I got A* at GCSE and loved it. I read the classics like Homer and Pliny, studied Greek and Roman comedy & tragedy, and could conjugate verbs until the cows came home! I remember learning 1500 Latin words for that exam. Now, I am limited to amo amas amat amamus amatis amant - the different expressions of the verb to love. (Ok, the odd Latin etymology in Pompeii was familiar but nothing of real significance.)

Bahrick called the residual knowledge that does persist across decades permastore - a relatively small core of deeply rehearsed, highly connected knowledge that becomes genuinely durable. The implication is not trivial: most of what is taught in schools never reaches permastore. It passes through students like water through a sieve, appearing briefly on an examination paper before dissolving back into forgetting.

Ebbinghaus opened the door. A century of subsequent research confirmed what was behind it. The question is why so little of it has changed the building.

Coverage culture and how it got built

I reckon that I could walk into almost any secondary school in England and ask a head of department how they plan their schemes of work. The answer will almost always be some version of the same thing: we work backwards from the exam. The specification determines what must be covered. The timetable determines how much time is available. The scheme of work divides the content across the available lessons. And the pressure - from line managers, from Ofsted, from performance management frameworks, from the entirely understandable anxiety of being held accountable for results - is to get through it all.

This is coverage culture. It is not a conspiracy or a collective professional failure. It is the entirely rational response of intelligent people to the incentive structure they have been placed inside. When the metric is whether students can demonstrate knowledge of specific content on a specific day, the system will optimise for producing that demonstration. Whether the knowledge persists beyond that day is, structurally speaking, somebody else’s problem.

The American cognitive psychologist Robert Bjork has spent much of his career studying what he calls desirable difficulties - the somewhat counterintuitive finding that the conditions which produce the fastest apparent learning are often the worst conditions for long-term retention. Massed practice - studying the same material intensively in a single block - produces rapid performance gains that feel like learning. But those gains are shallow and fragile. Spaced practice - distributing study across time, returning to material after gaps - produces slower apparent progress but dramatically stronger long-term retention.

“The most important message is that learners should break away from the misconception that the most effective ways of learning are those that make learning easy. The experience of having to expend effort, generate errors, or work hard to achieve understanding should not be interpreted as evidence of one's inadequacy as a learner, but, instead, as important steps towards actual long-term learning and comprehension." Robert Bjork

The implication for schools is uncomfortable. Bjork’s research suggests that a lesson which feels productive - in which students appear to be absorbing new material quickly, in which the teacher judges the class to be performing well - may be producing exactly the kind of shallow, rapidly-forgotten learning that Ebbinghaus and Bahrick documented. Meanwhile, the approaches that would actually produce durable memory - spaced repetition, retrieval practice across gaps, interleaving of topics - involve short-term performance dips that, in a system where teachers are observed and judged on in-lesson progress, look uncomfortably like failure.

There is incredible structural absurdity in this position. A teacher who introduces spaced repetition into their practice - who deliberately returns to content from three weeks ago rather than pressing forward with new material - may find themselves explaining to an observer why their students are struggling with things they were apparently taught already. The teacher who presses forward, who maintains pace, who covers the specification on schedule, looks professional. The irony is that the former is building memory; the latter is almost certainly not.

This is not a teacher problem. It is an architecture problem. The timetable, the specification, the observation framework, the performance management system - all of them reward coverage. None of them reward retention.

What the evidence actually supports - and where it stops

The science of memory offers three interventions with consistently strong evidence behind them. They are not new ideas. They are not experimental. They are among the most replicated findings in educational psychology. And they are still treated, in most schools, as optional enrichments rather than foundational practice.

Retrieval practice is the act of recalling information from memory rather than re-reading or reviewing it. Testing, in other words - not as an assessment tool but as a learning tool. The testing effect shows that the act of retrieving information from memory strengthens the memory trace far more effectively than passively re-encountering the same material. Low-stakes quizzing, flashcards, written recall - all of these, used regularly, produce measurably better long-term retention than conventional review.

Spaced repetition is the practice of distributing encounters with material across time rather than concentrating them. Return to content after one day, then three days, then a week, then a month, and the material becomes progressively more resistant to forgetting. Sebastian Leitner’s spaced repetition system, developed in the 1970s, formalised this into a practical method. Pimsleur embedded it in language learning programmes in 1967. The evidence has been available for half a century. Its adoption in school curriculum design remains sporadic although there are some good modern examples.

Interleaving is the practice of mixing different topics or problem types within a single study session rather than blocking them - spending twenty minutes on algebra, then twenty on geometry, then twenty on statistics, rather than an hour on algebra alone. It feels harder. Students typically rate it as more frustrating and less effective than blocked practice. The research consistently shows it produces better long-term retention and better transfer of learning to new problems.

Taken together, these approaches form the backbone of what Tom Sherrington and others have popularised as Rosenshine’s Principles of Instruction - a set of evidence-based teaching strategies that have gained significant traction in English schools since the mid-2010s, supported in part by the Education Endowment Foundation’s engagement with cognitive science. The Knowledge Organiser movement represents a related attempt to give these ideas practical form: structured reference documents designed to support retrieval and spaced practice. The intention is sound. The implementation, however, is frequently disconnected from the spacing architecture that makes retrieval meaningful. Giving students a knowledge organiser without building in structured opportunities to retrieve across time is a bit like installing a gym in a building with no changing rooms. The infrastructure is there. The conditions for its use are not.

But intellectual honesty requires acknowledging the limits of this research territory. Retrieval practice works most reliably for declarative, factual knowledge - the kind of content that can be stated as a proposition and tested with a question. For complex conceptual understanding, for the development of critical thinking, for creative and analytical capability, the evidence is thinner and the picture more contested. A curriculum designed entirely around low-stakes quizzing and flashcard retrieval might produce students who can recall a great deal of content while remaining unable to think flexibly with it. Oh, and AI in the mix make the latter less important than many people first thought.  

There is also an emerging body of concern about anxiety. Post-pandemic, in schools where student anxiety is already elevated, cultures of relentless low-stakes testing can exacerbate distress - particularly for students who already struggle with performance pressure. The evidence base for retrieval practice does not override the obligation to know your students. These are tools with genuine power and genuine limitations, and the honest position is to hold both.

The case against - and why it matters

Before arriving at what any of this means in practice, it is worth taking seriously the two most substantial objections to the argument being made here - because they are not trivial, and dismissing them would be intellectually dishonest.

The first comes from the sociologist of education Michael Young, whose work on powerful knowledge has been enormously influential in English curriculum debates over the past decade. Young’s argument is that broad, ambitious curriculum coverage has its own profound value - not despite the fact that students may forget much of it, but partly because of the exposure itself. To understand what is worth knowing deeply, you first need wide acquaintance with the landscape of a discipline. You cannot meaningfully prioritise what to retain until you have encountered enough to know what matters. On this view, a curriculum that slows down in pursuit of retention risks producing students with deep knowledge of a narrow range of content, and no map of the wider territory.

It is a serious argument. The response is not to deny it but to observe that it does not actually justify the current situation. The problem is not that schools teach broadly. The problem is that schools teach broadly and then do nothing architecturally to consolidate what matters most from that broad exposure. Wide coverage and structured retention are not mutually exclusive. They require different curriculum design - but they can coexist. The current system does not choose coverage over retention as a considered philosophical position. It defaults to coverage because retention is harder to measure and slower to produce.

The second objection is more philosophically fundamental, and in some ways more interesting. An education system is not simply a memory system. Schools are not just in the business of installing information into children’s heads for future retrieval. They are involved in socialisation, identity formation, relationship-building, cultural transmission, the development of character. Many of the things that matter most about a good education - a love of books fostered by a single brilliant English teacher, a moment of genuine intellectual excitement in a history lesson, the slow development of confidence through sport or music or drama - cannot be measured by retention rates and should not be reduced to them.

John Dewey, whose philosophy of education has shaped progressive pedagogy (and my own) for over a century, argued that genuine learning is fundamentally reconstructive - that knowledge is not passively received but actively rebuilt by the learner through experience and reflection. A rigid focus on what is retained risks reducing education to a transmission model that Dewey spent his career arguing against. 

“I believe finally, that education must be conceived as a continuing reconstruction of experience; that the process and the goal of education are one and the same thing.” John Dewey

Alfred North Whitehead had a related concern: what he called inert knowledge - facts that can be recalled in an examination but are never transferred, never applied, never genuinely integrated into a person’s thinking. The danger of a purely retention-focused education is that it produces exactly this: students who remember more, but think no more deeply.

These are not objections to be swept aside. But they do not ultimately undermine the core argument. They reframe it. The question is not should schools optimise for retention? The question is: whatever we believe education is for - memory, character, socialisation, identity, critical thinking, cultural transmission - is a system in which students forget the overwhelming majority of what they are taught within days actually serving any of those purposes? You do not need to be a retrieval-practice evangelist to find that question troubling. You only need to believe that education should produce something that lasts.

What knowing this actually requires

The gap between knowing something and acting on it is, of course, its own kind of forgetting curve. Educational research has a long and inglorious history of findings that are widely known, widely cited, and widely not acted upon. The question is what would actually have to change - at classroom level and at systemic level - for memory science to become something other than a CPD slide deck that gets forgotten by the following Monday.

At classroom level, some things are genuinely within a teacher’s control without requiring anyone’s permission. The ‘do now’ activity at the start of a lesson that retrieves content from two weeks ago rather than yesterday. Homework (if it has to exist 🙄) designed as spaced retrieval rather than extension tasks. Unit planning that revisits prior topics rather than treating each unit as a sealed container. None of this requires a structural revolution. It requires a shift in the logic of lesson planning - from what are we covering today to what do they need to have encountered, retrieved, and encountered again.

At departmental level, the conversation is about curriculum design. A scheme of work that builds in retrieval checkpoints - not as assessment events but as memory consolidation moments - looks different from one that simply maps content to lessons. Interleaving, where feasible within a specification, distributes topics rather than sealing them off. End-of-unit tests, used well, are not just summative tools - they are the most powerful retention mechanism available to a teacher, if the results are used to inform what gets revisited rather than simply recorded in a markbook.

But the honest acknowledgment is that classroom and departmental-level changes can only go so far. The specification that demands coverage of a volume of content incompatible with deep consolidation creates a structural ceiling. The accountability framework that measures in-year progress and terminal examination performance, with nothing in between that captures durable learning, incentivises exactly the coverage culture that produces forgetting. The assessment system that prioritises high stakes terminal exams at the expense of any other ways of checking understanding, even if the overwhelming evidence shows that exams are engineered for certain types of learners and have little impact on whether knowledge has gone into long-term memory, motivates rote memorisation and cramming.  

Leaders who take this seriously need to be asking uncomfortable questions about their own institutional architecture. What does our CPD programme actually teach about how memory works - and does it do so in a way that is itself spaced and retrieved, rather than delivered once and never revisited? What does our curriculum design process prioritise - the completeness of coverage or the durability of learning? What would our timetable look like if it were designed with memory consolidation as a genuine constraint rather than an afterthought?

These are not questions with easy answers, in part because they sit inside a national accountability framework that most individual schools cannot unilaterally redesign. But they are questions worth asking anyway - because the alternative is to continue planning lessons, building schemes of work, and designing curricula in ways that are structurally incompatible with the thing we say we are doing, which is education.

Key Takeaways

1. The forgetting curve is real, but Ebbinghaus alone is not sufficient authority - Bahrick’s research on meaningful content confirms that most school learning is forgotten within years without reinforcement, regardless of how relevant or well-taught it was.

2. Coverage culture is not a failure of teacher professionalism - it is the rational response to an accountability system that rewards demonstrable content delivery and has no meaningful metric for long-term retention.

3. Retrieval practice, spaced repetition, and interleaving have the strongest evidence base of any classroom-level interventions for durable learning but they work best for declarative knowledge, carry caveats around student anxiety, and are not a complete theory of education.

4. The strongest counter-argument is not that the science is wrong but that education is more than memory - socialisation, identity, and cultural transmission matter too; the honest response is that none of those purposes are served by a system in which students forget almost everything they are taught.

5. Classroom-level changes are within a teacher’s control - retrieval starters, spaced homework, units that revisit rather than seal off prior content - and do not require structural permission, but they cannot solve what the timetable and the specification are actively working against.

6. The question leaders need to ask is not whether their staff know about memory science - many do - but whether their institutional architecture makes acting on it possible; timetable design, curriculum planning, CPD structure, and accountability frameworks are all part of the answer.

There is something quietly revealing about the fact that a finding this robust, this replicated, and this practically significant has spent 140 years sitting largely unincorporated into the design of the institutions most responsible for learning. It is not that the evidence is obscure. It is not that the interventions are complex or expensive or beyond the reach of ordinary schools. It is that the system was built around a different priority - coverage, compliance, demonstrable delivery - and it has proved remarkably resistant to having its foundations questioned. If you are serious about what education is for, the forgetting curve is not an interesting neuroscience footnote. It is a description of what your institution is currently producing. The question is whether that is acceptable - and if not, what you intend to do about it.