Storage Strength and Retrieval Strength: Why Forgetting

‍

‍

What Storage Strength and Retrieval Strength Mean

Bjork and Bjork (1992) proposed that every memory has two separable properties, each of which follows its own rules and responds differently to practice.

Storage strength is a measure of how thoroughly a piece of knowledge is encoded in long-term memory. It accumulates incrementally over time, and once established at a high level, it does not decay. This is an important point: you do not gradually lose well-stored knowledge. Storage strength is relatively permanent. It is also largely invisible, in the sense that you cannot introspect on your own storage strength for a given item of knowledge; you can only infer it from your ability to retrieve that knowledge under different conditions.

Retrieval strength is a measure of how accessible a memory is at a given moment. Unlike storage strength, retrieval strength fluctuates substantially. It is highest immediately after study or practice, drops sharply over hours and days without use, rises again with successful retrieval, and is sensitive to context: retrieval strength is typically higher in familiar environments, with familiar cues, and in low-stress conditions than in novel ones.

The critical insight is that these two properties are independent of each other. You can have high storage strength and low retrieval strength (the "knew it but couldn't recall it" experience in an exam). You can also have low storage strength and high retrieval strength (you can answer the question easily right after the lesson, but the knowledge will be gone within days). These two failure modes look identical from the outside, but they have completely different implications for what a teacher should do next.

‍

The Four Quadrants of Memory Knowledge

A useful way to understand the theory is to arrange the two properties on two axes, producing four combinations. Each quadrant corresponds to a recognisable situation in the classroom.

Knowledge seems forgotten but learners can recall it later (Bjork, 1975). It is temporary inaccessibility, not complete forgetting (Tulving & Pearlstone, 1966). Retrieval practice fixes this common problem (Roediger & Karpicke, 2006).

Quadrant	Storage Strength	Retrieval Strength	What This Looks Like	Teacher Response
1	High	High	Student recalls fluently; knowledge is durable and accessible. This is the target state for key content.	Maintain with widely-spaced retrieval; move on to new material.
2	High	Low	A single retrieval practice event will restore access rapidly. Do not reteach from scratch.
3	Low	High	Student can answer correctly right now, but the knowledge is shallowly encoded. It will be gone within days. This is the cramming quadrant. Also the "seems to understand in the lesson" quadrant.	The student needs spaced retrieval practice, not more exposure to the content. Additional input will not fix shallow encoding.
4	Low	Low	Student neither recalls it nor has it stored durably. Genuine gap: either the content was never taught, was taught inaccessibly, or prerequisite knowledge is missing.	Reteach. Check prerequisites. Address cognitive load barriers before expecting encoding to occur.

‍

The practical value of this table is that it forces a diagnostic question. When a student fails to recall something, the instinctive teacher response is to reteach it (Quadrant 4 response). But if the student is actually in Quadrant 2, reteaching is wasteful. A five-minute retrieval activity would restore access far more efficiently, and the act of retrieval would also increase storage strength further, moving the student securely into Quadrant 1.

‍

The New Theory of Disuse Explained

Bjork and Bjork (1992) called their account the 'New Theory of Disuse' to distinguish it from older models, which held that memories simply decay or fade through lack of use, in the way that a unused path through a field gradually becomes overgrown.

The new theory proposes something quite different. Storage strength, once built, does not decay. What decays is retrieval strength, and this decay is not accidental but adaptive. The human memory system manages an enormous number of stored representations. If every stored item were equally and permanently accessible, the system would become unworkable: every attempt to retrieve one thing would be swamped by interference from thousands of related items. The depression of retrieval strength for infrequently used information is the system's way of managing this interference. Think of it less as forgetting and more as filing: the information moves from the top of the pile to a drawer, but it has not been discarded.

The practical consequence of this model is that forgetting is not the enemy of learning. It is a necessary intermediate state that the learning system passes through on the way to durable encoding. A memory that has never been retrieved under conditions of reduced retrieval strength has never been genuinely tested. Its storage strength remains uncertain. A memory that has been allowed to become partially inaccessible and has then been successfully retrieved has demonstrated genuine storage strength and has had that storage strength reinforced by the retrieval event itself (Bjork, 1994).

This reframing has a direct implication for how teachers think about revision and review. The common approach of reviewing material when it is still fresh (high retrieval strength) produces the comfortable experience of fluent recall, but adds little to long-term retention. Reviewing material after a gap (low retrieval strength) feels harder and produces more errors, but those errors and the effort of retrieval are precisely the conditions that drive storage strength upward.

‍

Why Cramming Creates High Retrieval, Low Storage

Cramming is the most widely observed application of the storage/retrieval distinction, and it illustrates the theory cleanly. A student who revises for an examination by reading through all their notes the night before will enter the examination with very high retrieval strength for the material. They will feel prepared. The information is highly accessible. They may perform adequately in the examination if it occurs within twelve to twenty-four hours.

However, because the student has been reviewing material with already-high retrieval strength, almost nothing has been done to increase storage strength (Bjork & Bjork, 1992). Within days of the examination, retrieval strength decays sharply, and because storage strength is low, the information becomes genuinely inaccessible. This is the mechanism behind the phenomenon that teachers describe as "forgetting everything after the exam." It is not that students have stopped caring about the subject, or that their memory has mysteriously erased itself. The content was encoded shallowly and the cramming strategy did nothing to change that.

Spaced retrieval helps learners more than cramming. Learners remember information better with gaps between practices (Bjork, 1992). They might struggle just before exams, but knowledge sticks long term. This improved encoding lasts (Karpicke & Roediger, 2008).

Classroom example (Year 11 Biology, GCSE): A teacher notices that students who revised enzyme function by rereading notes the night before a test scored well on the test but could not answer questions about enzymes six weeks later during a practice paper. She introduces a low-stakes retrieval quiz on enzyme function every three weeks throughout the year, using questions the students answer from memory without notes. Students initially find these quizzes uncomfortable. By March, they can answer enzyme questions reliably, even for content taught in September. The retrieval strength fluctuates between quizzes, but storage strength has been built across multiple retrieval events.

‍

Forgetting Curves and Storage Strength

Ebbinghaus (1885) showed new information retention drops quickly at first, then slows. Bjork's framework explains the forgetting curve Ebbinghaus found.

Ebbinghaus was measuring retrieval strength: his forgetting curve shows retrieval strength declining from 100% at the time of study to roughly 20% after a month without re-exposure. Bjork's contribution is to distinguish this observable retrieval strength curve from the less visible storage strength curve. Storage strength, in Bjork's model, does not track the forgetting curve. It is built during retrieval events, particularly those that occur when retrieval strength is low.

This distinction matters for practical planning. If you look only at the Ebbinghaus curve, the implication seems to be that you should re-study material as frequently as possible to keep retrieval strength from falling. Bjork's theory reveals that this is exactly wrong. Allowing retrieval strength to fall, then practising retrieval at the trough of the forgetting curve, produces the largest possible increase in storage strength. The goal is not to prevent the forgetting curve from dropping. The goal is to exploit the drop by timing retrieval practice to coincide with moments of reduced retrieval strength (Bjork & Bjork, 1992).

A practical implication is that the optimum spacing of review sessions is not constant. The first review after learning should occur relatively soon (within one to two days), when storage strength is low and retrieval strength has dropped to a level where retrieval requires effort but is still achievable. Subsequent reviews should be spaced further apart, because each successful retrieval event increases storage strength and slows the subsequent decay of retrieval strength. This expanding-interval pattern is the basis of spaced practice systems. For a detailed guide to implementing this in your classroom, see the article on spaced practice.

‍

Why Re-Study Fails at High Retrieval Strength

There is an asymmetry in the storage/retrieval relationship that has direct implications for lesson design. Bjork's research shows that the benefit of a study or practice event is inversely related to the current retrieval strength of the material being studied. When retrieval strength is high, a study event produces a small gain in storage strength. When retrieval strength is low, the same study event produces a much larger gain.

This asymmetry means that massed re-study is self-defeating. A student who reads through a chapter, then immediately re-reads it, gains little from the second reading because retrieval strength is still maximal from the first. The same student who reads the chapter, waits a day, then attempts to recall the main points from memory (with retrieval strength now reduced) will gain substantially more from that retrieval event than from any amount of immediate re-reading.

The practical implication is that the structure of practice matters more than the quantity. Fifteen minutes of spaced retrieval distributed over a week produces more durable learning than an hour of massed re-reading on a single evening. This finding has been replicated across subjects, age groups, and types of knowledge, from vocabulary learning in language classes to procedural skills in mathematics and science (Soderstrom & Bjork, 2015).

Classroom example (Year 9 French Vocabulary): A teacher sets homework using a vocabulary learning application that shows students words they already know at high frequency alongside new words. Students find this enjoyable: they are mostly getting correct answers because retrieval strength is high for known words. The teacher replaces this with a distributed retrieval task: on day one, students learn twelve new words; on day two, they retrieve all twelve from memory before seeing them; on day five, they retrieve again; on day fourteen, they retrieve again. The second approach produces slower apparent progress in weeks one and two but substantially better retention at six weeks.

‍

Why Students and Teachers Misread Performance

One of the most practically significant implications of Bjork's framework is that it reveals the gap between performance and learning. Performance is observable: it is what a student can do right now, in this lesson, under current conditions. Learning is the durable change in knowledge or skill that persists over time and transfers to new contexts. Performance during a lesson correlates poorly with long-term learning (Soderstrom & Bjork, 2015).

This means that the signals teachers typically use to assess whether students have learned something are unreliable. A student who can answer questions fluently during a lesson has high retrieval strength for the material right now. That high retrieval strength may reflect genuine high storage strength (Quadrant 1: excellent), or it may reflect the recent exposure to the material (Quadrant 3: fragile). The teacher cannot tell from the in-class performance which quadrant the student is in.

The converse is equally important. A student who struggles to answer a retrieval question three weeks after the lesson is not necessarily in Quadrant 4 (genuine gap requiring reteaching). They may be in Quadrant 2 (high storage, low retrieval), and a brief retrieval practice event will restore access quickly. Teachers who treat every retrieval failure as evidence of insufficient learning will over-reteach, which is time-inefficient and misses the opportunity to use the retrieval failure itself as a productive learning event.

Formative assessment connects to a wider framework. Good assessment finds temporary and real learning gaps. Asking learners to recall knowledge tests them better than testing straight after teaching (Bjork & Bjork, 2011).

‍

Working Memory and Retrieval Reconstruction

Retrieval practice boosts memory by engaging working memory (Bjork, 1992). Learners try to recall information, activating related memory cues (Tulving, 1983). Working memory reconstructs knowledge from partial information (Anderson, 1983; Reder, 1982). This strengthens long-term storage (Baddeley, 2000).

Constructing memories is hard work, say Karpicke and Blunt (2011). Re-reading allows learners to recognise information, not actively recall it. Recognition and recall differ (Anderson, 1983; Bjork, 1975). Recognition uses presented cues, while recall builds memories from within.

The effort of recall under reduced retrieval strength is what drives storage strength upward. This is also why simply making material more vivid, colourful, or interesting does not reliably increase long-term retention. Visual presentation affects encoding quality (which is the domain of cognitive load theory), but it does not address the retrieval dynamics that determine whether encoded information becomes durably stored. Storage strength is built through retrieval, not through re-exposure, regardless of how well-designed that re-exposure is.

Classroom example (Year 7 History): A teacher creates a beautiful and well-organised revision display on the causes of the Norman Conquest. Students enjoy consulting it and can answer questions about it confidently while it is on the wall. When the display is removed and students are asked to recall the causes from memory two weeks later, performance drops sharply. A parallel class instead uses five-minute brain dumps at the start of three subsequent lessons, writing everything they remember from memory. Their recall two weeks later is significantly better than the class with the display, despite (or because of) the greater apparent difficulty of the task.

‍

‍

How Desirable Difficulties Exploit This Distinction

Bjork (1994) called helpful challenges 'desirable difficulties'. These slow immediate gains, but improve lasting recall. Each difficulty uses storage and retrieval processes uniquely.

Spaced practice works by allowing retrieval strength to drop between practice sessions. The gap is not wasted time; it is the condition that makes the subsequent retrieval event maximally productive for storage strength.

This active recall strengthens memory (Kang, 2016). Interleaving forces learners to choose the right method (Rohrer, 2012). They identify the problem before solving it (Bjork, 1994). This knowledge retrieval improves learning, not just following patterns (Brown, Roediger & McDaniel, 2014).

The testing effect is the finding that testing produces better retention than re-studying, even when the test produces errors. It works directly on storage strength: each retrieval attempt, successful or not, reconstructs and reinforces the memory trace. Even a failed recall attempt primes subsequent learning of the correct answer.

Bjork (1994) found "desirable difficulties" create short-term retrieval problems. These problems actually boost long-term learning for the learner. Teachers can use this knowledge with flexibility (Bjork & Bjork, 2011). Knowing the logic allows application over rigid plans (Bjork, 1994).

Interleaving deserves closer examination because it is the desirable difficulty teachers find most counterintuitive and students find most uncomfortable. In blocked practice, a student completes ten problems of the same type, then ten of a different type. In interleaved practice, problem types are mixed. Blocked practice produces better immediate performance; interleaved practice produces better retention and transfer (Kornell & Bjork, 2008).

The storage/retrieval framework explains why. During blocked practice, the student solves the first problem of a type, establishing the retrieval pathway. Problems two through ten travel that same pathway with high (and rising) retrieval strength, so each adds minimal storage strength. During interleaved practice, retrieval strength for any given approach drops between instances because other problem types intervene. Reinstating the retrieval pathway is a genuine storage-strength-building event.

The practical difficulty is that students experiencing interleaved practice will often tell teachers they are confused. Their current performance is visibly worse, and this is uncomfortable for both parties. The research evidence is clear that this discomfort is not only acceptable but necessary. Kornell and Bjork (2008) found that students consistently preferred blocked practice and rated it as more effective, even after their own test results demonstrated the opposite. This is the metacognitive illusion created by the performance/learning distinction.

A Year 10 maths teacher mixed simultaneous equations, quadratics, and inequalities. Learners found these weekly problem sets tough. After six weeks, they beat another class in topics and combined problems. This retrieval difficulty strengthened their understanding, as found by (Bjork & Bjork, 1992).

Teachers using interleaving need to explain the rationale to students explicitly. Without this explanation, students interpret the difficulty as evidence that they are failing, rather than as evidence that the practice design is working. Sharing the storage/retrieval distinction is one way to do this, and is addressed in the section below.

‍

How to Explain This to Students

Learners often have wrong ideas about memory and learning. This leads to poor revision, like rereading notes (Bjork & Bjork, 2011). Explain storage and retrieval to learners. This metacognitive instruction changes how they study.

The explanation does not need to be complex. A straightforward classroom version runs as follows:

"Your memory has two separate dials. One is how well something is stored, deep down. The other is how easy it is to get it out right now. The problem is that studying something when you can already get it out easily does almost nothing to store it more deeply. That's why rereading your notes before a test feels helpful but often isn't. Your brain already knows the information is there, so it doesn't bother making a stronger copy. What actually stores things deeply is retrieving them when it's hard. When you have to struggle to get something out of your memory, that's when your brain makes the strongest copy."

Explain to learners that interleaved practice builds stronger memories. This reframes difficulty, showing learners their brains are working well. Soderstrom & Bjork (2015) found this reduces resistance to helpful challenging activities.

For younger students, a physical analogy can help. Ask them to imagine that remembering something is like lifting a weight. Lifting a weight that is already in your hands (high retrieval strength) is easy but does not build strength. Picking up a weight that has been put down and feels heavy (low retrieval strength) is harder but builds strength faster. This metaphor captures the key asymmetry without requiring students to understand the theoretical framework.

‍

What This Theory Does Not Tell Us

Bjork's framework is powerful, but it is important for teachers to understand its limits. The theory describes the relationship between storage strength, retrieval strength, and practice, but it does not specify what constitutes sufficient prior knowledge for retrieval practice to work. If a student has genuinely never encountered a concept, or lacks prerequisite schema structures, retrieval practice will produce confusion rather than learning. The theory presupposes that there is something stored to retrieve, even weakly.

Spacing and testing work well for facts and procedures. This is shown in many studies (Soderstrom & Bjork, 2015). Evidence for complex knowledge is less clear. However, spacing and testing are not harmful to learners.

Desirable difficulties may not help all learners, (Bjork, 1994). Anxious or disengaged learners might struggle with retrieval practice's cognitive benefits. Teachers must tackle motivation and anxiety first for retrieval to work well. Self-regulated learning models support this (Winne & Hadwin, 1998).

‍

Connecting to EEF Evidence and Classroom Impact

Metacognition and self-regulation show good impact, says the EEF. Learners gain seven months' progress (Teaching and Learning Toolkit). Spacing, retrieval, and elaboration help learning, say researchers. These strategies exploit storage/retrieval processes.

The EEF evidence on formative assessment (four months' average impact) also connects to this framework. Formative assessment designed around retrieval rather than recognition gives teachers more accurate information about storage strength, not just current retrieval strength. A teacher who uses a low-stakes retrieval quiz two weeks after teaching a topic is measuring something closer to genuine learning than a teacher who asks questions during the lesson when retrieval strength is still high.

This matters for planning. If you want to know whether students have learned something, ask them to retrieve it from memory under conditions of reduced retrieval strength: at least a day after teaching, without access to notes, in a format that requires recall rather than recognition. The results will be less impressive and more accurate than a lesson-end check. They will also be more useful for deciding what to do next, because they distinguish Quadrant 2 students (who need a retrieval prompt) from Quadrant 4 students (who need reteaching).

‍

Common Study Strategies Compared

Retrieval practice improves recall, but teachers often overlook storage strength. (Bjork & Bjork, 1992). Teachers can assess advice using this framework. (Kang, 2016; Karpicke, 2016). This helps learners remember information better. (Smith et al., 2023).

Strategy	Effect on Retrieval Strength	Effect on Storage Strength	Verdict
Rereading notes	Raises temporarily	Minimal gain	Poor long-term return
Highlighting	Raises slightly during activity	Minimal to no gain	No evidence of benefit over rereading
Summarising	Raises during activity	Small to moderate gain if done from memory	Better if summary is written without notes
Mind mapping from notes	Raises during activity	Small gain	Better if done from memory (then becomes retrieval practice)
Retrieval practice (flashcards, free recall, quizzing)	Rises after each attempt	Large gain, especially when retrieval is effortful	Most effective single strategy
Spaced retrieval practice	Fluctuates between sessions; rises after each	Very large cumulative gain	Most effective overall approach
Practice tests under exam conditions	Variable (stress can suppress retrieval)	Large gain, particularly after feedback	Effective, especially with corrective feedback

‍

Retrieval practice boosts memory, research shows (Karpicke & Blunt, 2011). Learners gain more by recalling information than passively reviewing it. Teachers should encourage recall-based revision like flashcards instead of just reading (Dunlosky et al., 2013).

‍

Why Students Forget Everything After an Exam

The experience of students forgetting all examination content within days of the exam is a diagnostic signal, not a mystery. Using Bjork's framework, the mechanism is completely clear.

Learners often review content before exams for correct answers. This works, as retrieval peaks around exam time. But rereading familiar material helps less, according to Bjork and Bjork (1992). Each review gives small storage gains when retrieval is high. Learners maintain retrieval but don't build storage, state Bjork and Bjork (1992).

Within one to two weeks of the examination, retrieval strength decays (as it always does without use), and because storage strength is modest, the material crosses below the threshold of accessible recall. This is not a failure of memory, intelligence, or effort. It is the predictable outcome of using the wrong practice strategy.

Revising differently helps more than revising more. Learners using spaced retrieval (Weeks & months) retain info (Roediger & Butler, 2011). They recall material for years after (Karpicke & Blunt, 2011). Exam performance improves; retrieval remains strong even when stressed (Bjork, 1992).

‍

A Practical Planning Guide for Teachers

Researchers (Bjork & Bjork, 1992; Karpicke, 2016) suggest spacing learning. Teachers should plan for learners to revisit material later. Frequent retrieval practice, as shown by Roediger & Butler (2011), strengthens memory. This approach applies to any subject area.

First, build retrieval events into your scheme of work rather than treating review as an end-of-unit activity. A minimum effective dose is three spaced retrieval practice events for each major piece of content, timed to coincide with periods of reduced retrieval strength. For content taught in Week 1, a retrieval event in Week 2, another in Week 5, and a third in Week 10 will produce substantially better retention than three revision lessons in Weeks 9, 10, and 11.

Second, use the four quadrants as a diagnostic lens. Before deciding whether to reteach, ask whether a retrieval event might restore access first. If students struggle with a retrieval starter, wait for their responses before judging their knowledge. Students in Quadrant 2 will often recall more than they initially think if given a little time and a few retrieval cues.

Brown et al. (2014) suggest teaching learners about storage and retrieval. Explain it clearly for their age. Learners understanding productive struggle try harder with revision. This approach links to self-regulated learning. You need no special training (Bjork et al., 2015).

Fourth, align your formative assessment practices to measure storage strength rather than retrieval strength. Ask students to retrieve content from memory with at least a day's gap from the last teaching of that content. Use low-stakes conditions to reduce the anxiety that can suppress retrieval strength independently of storage strength.

In your next lesson, identify one piece of content that you taught at least one week ago and design a five-minute blank-page retrieval activity around it. Ask students to write everything they remember, without notes or prompts. Observe their responses using the four-quadrant lens: who recalls it fluently (Quadrant 1), who struggles but gets there with effort (Quadrant 2), and who has genuinely no access (Quadrant 4). Let the data determine whether your next move is to maintain, prompt, or reteach.

---

‍

‍

‍

Frequently Asked Questions

‍

Further Reading: Key Papers on This Topic

Bjork, R.A. & Bjork, E.L. (1992) "A new theory of disuse and an old theory of stimulus fluctuation" View study ↗

Bjork and Bjork (1992) presented the storage/retrieval distinction. They argued retrieval strength weakens without use, while storage strength stays stable. This difference explains why spaced and retrieval practice work better than cramming. Teachers should consider this theory for understanding learning difficulties.

Soderstrom, N.C. & Bjork, R.A. (2015) "Learning versus performance: An integrative review" View study ↗

This review paper synthesises decades of research distinguishing performance during learning (current retrieval strength) from actual learning (storage strength). Soderstrom and Bjork examine why the two dissociate and what conditions create the largest gap between them. Directly relevant to teachers who want to understand why in-class performance is a poor predictor of long-term retention.

Kornell, N., Bjork, R.A. & Garcia, M.A. (2011) "Why tests appear to prevent forgetting" View study ↗

Kornell, Bjork and Garcia (2011) say tests curb competing memories, not just boost recall. Their research suggests testing improves storage strength, not retrieval. This affects how teachers should create low stakes quizzes.

Storm, B.C., Bjork, R.A. & Storm, J.C. (2010) "Optimizing retrieval as a learning event" View study ↗

Storm, Bjork, and Storm (2010) studied successful learning retrieval conditions. They found that harder retrieval leads to better long-term retention. This only works if the learner successfully remembers, however (Storm et al., 2010).

Bjork, E.L. & Bjork, R.A. (2011) "Making things hard on yourself, but in a good way" View study ↗

Bjork and Bjork (date not provided) reviewed desirable difficulties (spacing, interleaving, testing, variation). They used storage/retrieval. They explored metacognitive illusions, explaining why learners pick weaker study methods. This is good for teachers new to Bjork's (date not provided) research.