A fragile learner can do it — until the conditions change. The child answers correctly today and forgets tomorrow, performs when the worksheet is familiar and hesitates after a short delay, or succeeds with a tutor’s cues but cannot reproduce the same capability alone. Parents often describe this as “inconsistent”, “careless”, “not retaining”, or “we already did this last week”.
For families in Sengkang asking why a child forgets what was taught, needs repeated revision, gets a topic right in tuition but wrong at school, or seems to relearn the same thing again and again, the useful question is not whether the lesson happened. It is whether the learning became retrievable, repeatable and independent after the lesson ended.
At eduKate Sengkang, Fragile means a capability exists but does not yet survive ordinary variation. The learner is no longer Blocked: a route is available. But the route may depend on recent practice, visible notes, familiar wording, a teacher’s sequence or a prompt. The job now is to make success reappear after time, cue removal and modest surface change.
Quick answer: how do you stabilise fragile learning?
Stop measuring only immediate correctness. Build a sequence of delayed, low-support returns. Retrieve before rereading. Correct errors by reconstructing the reasoning, not copying the fix. Space the same capability across several days. Change the surface while preserving the underlying rule. Reduce prompts one layer at a time. Fragile learning becomes Stable when the capability reappears reliably without the original supports.
- Teach or repair the capability clearly.
- Close the notes and retrieve the key idea soon after.
- Return after a delay long enough that retrieval requires effort.
- Use a second representation or wording.
- Mix the capability with nearby alternatives so the learner must select it.
- Track whether the same error family returns.
- Fade tutor and parent prompts until the learner can reconstruct the route.
This is why practice is not simply repetition. The design of the return matters.
What fragile learning looks like at home
- Monday homework is correct; Thursday revision looks unfamiliar.
- The child says “Oh yes, I remember now” as soon as a parent gives one cue.
- A test with the same format goes well; a reordered or reworded test does not.
- The learner can explain while looking at notes but loses the explanation when the page closes.
- Corrections are neat yet the same error returns in the next paper.
- The child performs well immediately after tuition and poorly several days later.
- The right method appears only when the worksheet title announces the topic.
- Confidence rises and falls sharply with familiarity.
Fragility is common during real learning. New capabilities are often unstable before they become reliable. The danger is misreading supported or immediate performance as proof that the learning has already held.
The difference between Blocked and Fragile
| Question | Blocked | Fragile |
|---|---|---|
| Can the learner enter the task? | Not reliably. | Usually yes, especially when the task is familiar or recently practised. |
| What does help do? | Creates the route. | Reactivates a route that exists but is not reliably available. |
| Main instructional job | Build or restore a workable first move. | Make the route retrievable, repeatable and independent. |
| Most revealing test | Can a bounded cue create movement? | Does success reappear after delay and cue removal? |
A learner can move from Blocked to Fragile in one lesson. That is not a small improvement. It means the problem has changed. Teaching should change with it.
Why “we already taught this” is not enough
Teaching creates an opportunity to learn; it does not guarantee durable access. A student can follow an explanation because the teacher’s words, examples and sequence hold the structure externally. When those supports disappear, the learner must recreate the structure from memory and decide when it applies. That second job is where fragile learning becomes visible.
Research on retrieval practice and spacing consistently shows that effortful recall and distributed return can support longer-term retention better than passive review alone. The practical implication is modest: the learner should repeatedly have to produce the knowledge, method or relationship, not just recognise it.
Four kinds of fragility
1. Retrieval fragility
The idea is known when seen but hard to produce cold. Use short closed-book prompts, then check. Keep the retrieval target precise: a formula condition, a causal link, a vocabulary meaning in context, an algebraic identity, a paragraph plan, a Science relationship.
2. Cue fragility
Success depends on chapter labels, colour coding, worked examples left open or repeated teacher wording. Fade those cues deliberately. A learner preparing for a mixed paper must identify the method from the task itself.
3. Representation fragility
The learner can handle a table but not a graph, a bar model but not an equation, a narrative passage but not an infographic, a familiar apparatus diagram but not a new experimental setup. Translate between forms so the underlying relation becomes the invariant.
4. Sequence fragility
The learner succeeds only when practice arrives in the same order in which it was taught. Change the order, interleave related tasks and ask for method selection. Stable learning should not require the teacher’s sequence to remain attached.
A home test for fragility
Choose one capability the learner appeared to master. Do not announce it as a test. After an ordinary delay, present one fresh item without notes. Observe three things: how long it takes to start, whether the learner selects the method, and whether the explanation survives. If a tiny cue restores the whole route, the capability is likely present but fragile. If no cue helps, revisit the Blocked route and the prerequisite chain.
| Observation | Likely interpretation | Next move |
|---|---|---|
| Correct immediately after practice, poor after two days | Retention/retrieval fragility | Space retrieval; reduce rereading. |
| Correct only with worksheet title | Cue dependence | Remove labels; use mixed sets. |
| Correct in one representation | Representation fragility | Translate across graph/table/diagram/prose/symbol. |
| Correct with tutor prompt | Prompt dependence | Delay and fade hints; retest independently. |
| Correct on same examples, poor on changed examples | Early transfer gap | Introduce controlled variability after stability. |
English: when language knowledge is there but unstable
A student may know a vocabulary word on a list and miss it in a passage; understand a grammar rule in a drill and violate it in composition; produce a strong inference with a tutor’s questions and fail to generate the same evidence chain alone. Fragility in English often hides because recognition is easy to create. The learner says the explanation sounds familiar, but cannot yet produce it where the language changes.
Use short retrieval prompts inside real reading and writing. Ask for the meaning of a word in a new sentence, the evidence for an inference, the reason a pronoun refers to a particular noun, or the one sentence that would repair a paragraph’s logic. Route deeper English system work through the English Learning Hub and SETC.
Mathematics: when procedures disappear after the worksheet ends
Fragile mathematics often looks like “sudden forgetting”. A child completes ten ratio questions but cannot identify ratio inside a mixed set. An algebraic manipulation is smooth during the lesson and error-prone after a week. The learner recognises a teacher’s model but cannot decide to draw one independently. These are signs that procedure and selection have not yet fused into stable control.
Retrieve methods without the chapter label. Ask what makes two superficially similar questions require different approaches. Return to old content in small doses while new content is taught. For specialist depth, use BukitTimahTutor; for the learner state, keep the stability test here.
Science: when keywords are remembered but explanations are fragile
A learner may recall a keyword after revision but fail to rebuild the causal chain when the diagram changes. Another remembers a model answer but cannot adapt it to a different variable or context. The fix is not more memorised sentences. Ask what relationship must remain true, what evidence supports it, and what changes when the conditions change.
Use delayed explanation from a new representation. If the underlying Science knowledge is incomplete, move into the Science Hub or the canonical knowledge owner. If the knowledge is present but fails to reappear, stay with retrieval and variability.
A-Math: fragile algebra makes later topics look mysterious
Additional Mathematics is unforgiving of fragile foundations. A student can appear comfortable with a current chapter because the examples cue the method, yet weak factorisation, sign control or equation manipulation reappears as an error somewhere else. The later chapter gets blamed for a foundation that never became stable.
Keep a small rotating foundation set beside current work. Do not reteach the entire syllabus each week. Retrieve only the dependencies that repeatedly fail, then put them back into mixed A-Math conditions.
The 2–2–7 stability loop
A simple rhythm for many school capabilities is: an early return after roughly two learning opportunities, another return after a short delay, then a later return after about a week. The exact interval should follow the learner and material; the principle is that the answer should have to be reconstructed after some forgetting has begun.
- First learning: understand and practise with feedback.
- Early retrieval: close the source and produce the key relation.
- Short-delay return: use a fresh item with less cueing.
- Later return: mix it with other content and change the surface.
- Record whether the learner starts, selects and completes independently.
If the learner repeatedly needs the same prompt, the prompt has become part of the performance. Change the teaching rather than merely repeating the prompt louder.
Corrections: copy less, reconstruct more
Fragile learners often produce beautiful corrections that teach very little because the correct route is visible while they write it. A stronger correction asks the learner to locate the first wrong decision, name the principle, repair it, then close the model and solve a fresh item. The correction should change future behaviour, not merely clean the page.
Feedback is useful when the learner can act on it. A comment such as “check your working” is weak if the learner does not know what checking operation to perform. Make the feedback specific enough to change the next attempt, then remove it once the learner can self-generate the check.
How parents can help without becoming the memory system
- Ask for recall before opening notes.
- Use short returns rather than marathon weekend relearning.
- When giving a hint, give the smallest hint that restarts the route.
- Ask “Why does this method fit?” after a correct answer.
- Bring back one old skill beside the current homework.
- Celebrate reliable return, not only first-time correctness.
- Let the learner experience a little retrieval difficulty before rescuing.
Difficulty is not always evidence that learning is failing. Some effort during retrieval is part of making access less dependent on the original lesson.
How a tutor should fade support
| Round | Tutor support | Learner job |
|---|---|---|
| 1 | Model and explain the structure. | Follow and explain back. |
| 2 | Partial example or targeted cue. | Complete the missing reasoning. |
| 3 | Question only. | Retrieve and select. |
| 4 | No cue; fresh item. | Begin and complete independently. |
| 5 | Changed surface / mixed set. | Recognise the invariant and transfer. |
The tutor should not be impressed by support-dependent fluency. The important measurement is how much of the route remains when support is removed.
When fragile becomes stable
Stable does not mean perfect. It means the important capability reappears across ordinary variation: another day, another question, another order, another reasonable wording. Small mistakes may remain. What changes is that the route itself no longer disappears.
- The learner retrieves the core idea without opening notes first.
- Equivalent items are completed after delay.
- A change of numbers or context does not erase the method.
- The learner can detect and repair a familiar error family.
- Prompts are occasional rather than structural.
- Performance survives a mixed set of nearby alternatives.
Then the teaching job changes. Move to the Stable route and begin increasing variability. See How Practice Variability Works.
When not to keep drilling
If repeated retrieval keeps failing in the same way, do not simply increase frequency. Reopen the diagnosis. The learner may still be Blocked at an earlier prerequisite, the representation may be unsuitable, the distinction between two methods may never have been taught, or the current task may ask for more than the repaired capability can support.
Four worked cases: what fragile learning looks like across subjects
Primary 4 English — the grammar rule that vanishes in composition
A learner scores well on subject–verb agreement exercises immediately after teaching. In composition, the same errors return. The rule exists, but it is not available while attention is divided among ideas, spelling, sentence construction and story flow. The repair is to retrieve the rule briefly before writing, edit a small paragraph without a rule label, then write and self-check a fresh paragraph. The target is not another grammar worksheet; it is stable access inside real writing.
Primary 6 Mathematics — ratio is correct in tuition, wrong in a mixed paper
During ratio practice the learner is fluent. Three days later, inside a mixed PSLE set, she chooses percentage instead. The procedural steps were stable enough under a label, but selection was fragile. The tutor returns to contrast practice: ratio versus percentage versus fraction of a whole, with the chapter names removed. The learner must state the relationship before calculation. Later retrieval becomes both memory and discrimination practice.
Primary Science — the keyword survives, the causal chain does not
The child remembers “conduction” but writes an incomplete explanation when the material and diagram change. The fragile unit is not the word; it is the link from particle/material property to heat transfer and observed outcome. The learner practises reconstructing the chain from different diagrams after short delays. Model wording is used only after the learner produces the mechanism.
Secondary 3 A-Math — factorisation works until it becomes a hidden dependency
A student can factorise in a chapter exercise but fails partial fractions because factorisation is now embedded inside another task. The skill was recently available but not yet stable under integration. The tutor keeps a rotating two-question retrieval check at the start of lessons, then uses it inside current work. Old foundations are maintained without restarting the whole syllabus.
Retrieval is not the same as testing for marks
Retrieval practice is often misunderstood as “give more tests”. The useful mechanism is simpler: the learner tries to bring knowledge or a method to mind before seeing the answer. The attempt can be low stakes, brief and diagnostic. A five-minute closed-book recall can be retrieval. Explaining a concept from memory, drawing a diagram without labels, reconstructing a formula condition or outlining a composition before checking notes can all serve the same purpose.
For fragile learning, the attempt should be followed by accurate checking and correction. Retrieval without feedback can rehearse an error. Feedback without a later independent return can create temporary understanding that disappears again. The loop needs both.
How to choose spacing intervals without turning home into a timetable factory
There is no universal perfect interval. The interval should be long enough that retrieval is not automatic, but short enough that the learner can still recover the route with reasonable effort. Early in learning, that may mean a return later the same day or the next day. As stability improves, returns can spread across several days and weeks.
| If the learner… | Try… | Then… |
|---|---|---|
| Cannot retrieve anything after one day | Shorten the next return and repair understanding | Lengthen only after successful recovery |
| Retrieves instantly every time | Increase delay or reduce cues | Add a changed surface |
| Needs one small cue | Keep the delay and fade the cue | Retest independently |
| Retrieves but selects the wrong method in mixed work | Keep retrieval but add discrimination | Contrast confusable methods |
| Maintains the skill over a week | Move toward Stable practice | Increase variability rather than raw frequency |
Spacing is a tool for observing and strengthening memory, not a rigid calendar. The learner’s evidence should drive the interval.
The fragile-learning error log: track mechanisms, not just questions
An error log is useful only if it helps the next decision. Writing “Question 7 wrong” has little value. Record the earliest mechanism: misread condition, retrieval failure, wrong method selected, sign error, incomplete causal link, evidence not cited, unit not converted, prompt dependence, or answer changed without reason.
| Date | Task | First error mechanism | Repair | Delayed retest |
|---|---|---|---|---|
| Example | P6 ratio word problem | Selected percentage because ‘out of 100’ cue dominated | Compare ratio/percentage contrast pair | Fresh mixed item in 3 days |
| Example | P5 comprehension inference | Claim plausible but no textual evidence | Claim → clue → inference routine | New passage in 2 days |
| Example | Sec 3 A-Math | Forgot factorisation identity | Closed-book retrieval + two examples | Integrated question next lesson |
When the same mechanism stops recurring, retire it from the active log. The log should shrink as control grows.
Fragile confidence: why recent success can mislead both child and parent
A learner can leave tuition feeling highly confident because the material is fresh and the tutor’s cues are still active. That feeling is not false, but it may describe the supported state rather than later independent access. Parents sometimes interpret the next poor school result as a mysterious collapse. The two performances occurred under different conditions.
Calibrate confidence by asking for a prediction before a delayed retrieval: “How sure are you that you can do this without notes?” Then compare prediction with result. The goal is not to make the learner pessimistic. It is to help them distinguish familiarity from durable readiness.
The prompt-fading plan
- Full prompt: “This is a ratio problem; draw a model.”
- Reduced prompt: “What relationship do you see?”
- General prompt: “What is your first useful representation?”
- Metacognitive prompt: “What makes you confident that method fits?”
- No prompt: learner reads and begins independently.
The same logic works beyond Mathematics. “Look for the evidence in paragraph 3” can fade to “What evidence supports your answer?”, then to no cue. “Use heat transfer” can fade to “What mechanism explains the observation?”, then to independent explanation.
Mixed practice for Fragile learners: not too early, not too late
A learner who has only just formed a procedure may need a short block of similar practice to avoid unnecessary search. But if blocked practice continues too long, the worksheet itself becomes a cue. Introduce a small amount of mixing once the learner can execute the new method: perhaps two target items among three familiar alternatives. Increase the mix as selection stabilises.
The purpose is not to make every session feel difficult. It is to prevent the learner from confusing “I can execute this when told what it is” with “I can recognise and use it independently”.
How fragile learning behaves around school holidays
Long breaks are natural stress tests for stability. A learner who appears secure at the end of term may return with large gaps if the capability was held mostly by recent exposure. The answer is not to fill the holiday with daily worksheets. Use compact retrieval of high-dependency knowledge: multiplication facts where needed, fraction relationships, grammar principles, core vocabulary in context, Science mechanisms, algebraic identities and other foundations that current learning depends on.
A small rotating set can preserve access while leaving the holiday largely intact. If a foundation is truly stable, it should not require constant maintenance at high volume.
Fragile learning during the PSLE and Secondary examination years
Examination years expose fragility because revision mixes old and new material. A student who looked strong chapter by chapter may suddenly appear inconsistent when the paper stops announcing the topic. This is not necessarily decline. It may be the first honest measurement of what is independently retrievable and selectable.
Use topical repair only for the mechanisms that fail, then return quickly to mixed conditions. A fragile capability should not live forever in a protected worksheet. As stability increases, the learner should spend more time in mixed and eventually paper-level work under Examination Craft.
A 30-day fragile-to-stable plan
| Week | Focus | What to collect |
|---|---|---|
| 1 | Identify 3–5 high-value fragile capabilities and establish closed-book baselines. | What is retrieved, what needs cues, which errors recur. |
| 2 | Use spaced retrieval and precise correction. | Number of prompts falls; delayed accuracy improves. |
| 3 | Add mild representation and wording changes; begin small mixed sets. | Learner recognises the capability without labels. |
| 4 | Run fresh delayed samples across two or three conditions. | Capability reappears independently enough to classify as Stable. |
Do not carry twenty fragile targets at once. Choose dependencies that unlock downstream work. Stability of a few foundations can improve several visible symptoms.
A Sengkang weekday rhythm: 15 minutes of maintenance without creating another school day
- Five minutes: retrieve two older high-dependency items with notes closed.
- Five minutes: one current-school item or correction.
- Three minutes: one changed or mixed item.
- Two minutes: learner states what to return to and when.
This is a maintenance rhythm, not a universal prescription. On heavy school days, less may be more. The goal is to prevent fragile learning from requiring full relearning every weekend.
How to coordinate fragile learning across tutor, school and parent
The learner benefits when adults use compatible evidence, not identical teaching. A tutor can report which cue was needed and what delayed retest is planned. A parent can notice whether homework begins independently. School assessment provides a less-supported sample. Together these observations reveal whether the capability is stabilising across contexts.
Avoid sending the learner the message that every adult is watching every error. Share only what changes instruction. The purpose of evidence is better routing, not pressure.
What progress looks like before the marks rise
- The learner needs fewer reminders to retrieve before looking.
- The same correction is not repeated as often.
- Old skills take less time to reactivate.
- Method selection improves inside mixed work.
- The learner can explain why an answer is wrong before seeing the correction.
- Confidence predictions become more accurate.
- A week away from a topic produces hesitation rather than total loss.
These changes are valuable because they reduce the cost of future learning. A student who can maintain foundations has more attention available for new work.
When to reclassify: Fragile, Stable, or back to Blocked?
Reclassify to Stable when the capability repeatedly survives delay, cue removal and ordinary variation. Move back to Blocked if the learner cannot reconstruct a workable route even with small cues and the prerequisite chain appears missing. Stay Fragile if the capability is present but intermittent. Accurate reclassification is more important than protecting a narrative of constant upward progress.
The anatomy of fragility: where stability can leak
Once a route exists, fragility can appear at several points. A learner may retrieve the fact but not the condition, execute the method but forget how to choose it, remember the explanation but lose it under a different representation, or perform independently only when practice is recent. Treating all of these as “poor memory” is too broad.
| Fragility point | What survives | What fails | Repair emphasis |
|---|---|---|---|
| Fact access | Recognition | Cold retrieval | Short retrieval with checking |
| Condition access | Rule | When the rule applies | Examples/non-examples and contrast |
| Method selection | Procedure | Choosing it in mixed work | Interleaving and discrimination |
| Representation | Concept in one form | Same concept in another form | Translation and invariant mapping |
| Independence | Supported route | Route without prompts | Prompt fading and delayed return |
| Integration | Component skills | Using them together | Gradual combination |
The first four weeks should target the actual leak rather than assigning the same revision routine to every fragile learner.
What should happen in the first lesson with a Fragile learner?
Do not reteach immediately. First sample the learner without recent support. Use one familiar item, one delayed retrieval prompt and one small variation. The aim is to distinguish “forgotten”, “recognised but not retrievable”, “method known but not selected” and “support-dependent”.
- Ask for a cold attempt before notes open.
- Ask the learner to rate confidence before checking.
- Inspect the earliest failure, not only the final mark.
- Give one precise feedback move.
- Use a second item after the feedback.
- Schedule a later fresh item so the next lesson starts with a stability check.
A fragile first lesson should therefore create a maintenance plan, not a new pile of content.
The first four weeks: make the route reappear
| Week | Focus | Successful evidence |
|---|---|---|
| 1 | Identify high-dependency fragile skills; establish retrieval baselines. | We know what disappears and under what cue conditions. |
| 2 | Space retrieval; reconstruct corrections; fade obvious prompts. | Learner recovers more with less support. |
| 3 | Mix with confusable alternatives; change representation modestly. | Selection survives without chapter labels. |
| 4 | Run delayed fresh samples and one larger mixed task. | Capability recurs often enough to call Stable. |
If the learner needs full reteaching on every return, the state may actually be Blocked or the original explanation may never have produced a usable model.
Is a small-group class a good fit for a Fragile learner?
Often yes. Fragile learners benefit from short independent retrieval attempts followed by feedback, which fits well inside a three-student rotation. While the tutor works with another student, the fragile learner gets exactly the condition we need to observe: can the route remain active without continuous prompting?
The class is less suitable if the learner requires constant cueing just to remain on task or if important access needs cannot be met. Again, fit is about instructional conditions, not status.
Questions parents should ask a tutor about Fragile learning
- Which capabilities are fragile enough to deserve maintenance?
- How often are they being retrieved without notes?
- What prompts are being faded?
- Are the same corrections recurring or shrinking?
- When do you mix old and new content?
- How do you distinguish forgetting from misunderstanding?
- What delayed evidence will justify calling the skill Stable?
- What should we do at home that will not interfere with the measurement?
A strong answer should mention delayed evidence. Immediate post-lesson correctness alone is not a stability measure.
The Fragile learner’s progress receipt
- Capability: the exact knowledge or method under maintenance.
- Delay: how long since the last supported practice.
- Cue level: none, general, targeted or structural.
- Result: correct/incorrect plus the first failure mechanism.
- Recovery: whether feedback enabled independent correction.
- Next return: when and under what changed condition.
Over time the receipt should show longer delays, lower cue levels and better recovery. That pattern matters more than one perfect session.
Why rereading feels good and can still leave learning fragile
Rereading restores familiarity quickly. That feeling can be reassuring, especially before examinations. But familiarity answers a different question from independent retrieval. A learner may recognise every sentence in notes and still be unable to produce the idea when the paper asks for it.
Use rereading when understanding needs repair. Then close the source and retrieve. The sequence matters: understand → retrieve → check → return later. Rereading alone should not be asked to do the work of memory testing.
How to keep revision from expanding forever
Fragile learners can accumulate huge revision lists because every old topic seems to need another look. Prioritise dependencies. A shaky multiplication fact system can affect fractions, ratio and algebraic substitution; weak sentence parsing can affect comprehension and writing; unstable causal language can affect many Science explanations.
Repair and maintain high-leverage foundations first. When they stabilise, several downstream problems may improve at once. This is the same first-weak-link logic used elsewhere in the Atlas.
The subject-route handoff
State pages should not become replacement textbooks. When a fragile learner lacks actual content knowledge, route to the relevant subject owner, teach the knowledge properly, then return here to test whether it survives. Sengkang owns the state and next action; the subject hubs own the detailed curriculum routes.
This boundary protects the ecosystem from duplication. It also makes the learner’s journey clearer: “learn the idea here, then prove it holds under these conditions.”
Fragile learning and the first school test after tuition
The first school test is useful evidence precisely because it removes some tuition supports. Do not overreact to one result. Compare which capabilities survived, which collapsed, and whether the errors match the fragility profile observed in tuition. If the same mechanism appears, the route was not stable enough. If a new mechanism appears, update the model.
The test is not merely a verdict on tuition. It is another observation point in the runtime.
What if the learner is improving but the total mark is flat?
A total mark can stay flat while the error distribution improves. The learner may stop repeating one old misconception but lose marks on newly taught material; finish more of the paper but still make local slips; or improve comprehension while composition remains unchanged. Inspect the components before declaring no progress.
Eventually the mark should matter, especially when examinations approach. But early repair is better understood through mechanisms plus marks than through marks alone.
Myths about Fragile learners
| Myth | Better interpretation |
|---|---|
| “They have a bad memory.” | Identify which knowledge, cue and delay conditions fail before generalising. |
| “They need daily drilling.” | Some need shorter intervals; others need better retrieval and discrimination, not more volume. |
| “If they got it wrong after a week, the lesson failed.” | The delayed failure may be the first useful evidence of what still needs stabilising. |
| “Notes make learning secure.” | Notes support understanding and storage; closed-source retrieval tests access. |
| “Mixed practice is always better.” | Early learning may need some blocking; mixing becomes useful once basic execution exists. |
The state model replaces absolutes with timing and conditions.
Decision tree: right today, wrong tomorrow
- Can the learner explain the idea while looking at notes? If no, repair understanding.
- Can the learner retrieve it with notes closed now? If no, build retrieval immediately.
- Can the learner retrieve it after a delay? If no, shorten spacing and repeat.
- Can the learner select it without a chapter label? If no, add contrast/mixed practice.
- Can the learner use it in another representation? If no, add translation.
- Can the learner do all of the above across several returns? If yes, move to Stable.
Return to Atlas V2.0 for the whole map or Parents’ Guide when the question is whether tuition is producing durable change.
Subject mini-labs: test durability without turning everything into an exam
English mini-lab: retrieve the rule inside real language
Choose one grammar, vocabulary or comprehension capability that looked secure in recent work. Present it inside a fresh sentence or short passage without naming the rule. Ask the learner to identify what matters and explain the answer. If the learner can perform only after the rule name is supplied, the knowledge is still cue-fragile. If the learner retrieves and applies it independently, increase the delay next time.
Mathematics mini-lab: one old item inside a new set
Place one previously learned method among two or three current-school items. Do not label the old topic. Ask the learner to identify the method before calculating. This takes only a few minutes and provides better stability evidence than reassigning a whole old worksheet. If selection is correct but execution contains one local error, the learning may already be becoming Stable.
Science mini-lab: explain after the picture changes
Return to a known mechanism using a new diagram or everyday context. Ask the learner to explain cause and effect before checking notes. Fragile memorisation often breaks when the picture changes; durable understanding can reconstruct the explanation. Record whether the learner needed a keyword prompt, a mechanism prompt or no prompt.
A-Math mini-lab: prerequisite pulse check
Select two high-dependency foundations from earlier chapters and place them at the start of the lesson without announcement. Keep the check short. The purpose is to see whether algebraic access remains available, not to consume the lesson with old content. Fragile foundations should become easier to retrieve over successive weeks.
How to talk to a learner who is inconsistent
Inconsistency can feel personal to a child: “I could do it yesterday, so why am I stupid today?” Replace the self-judgement with a memory model: learning can be available under one condition and unavailable under another; our job is to make access more reliable. This gives the learner something to do rather than something to be ashamed of.
| Instead of | Try | What it teaches |
|---|---|---|
| “You forgot again.” | “What cue did you need to reactivate it?” | Access can be measured and improved. |
| “We just did this.” | “What survives before you look back?” | Retrieval before review. |
| “Your memory is bad.” | “Which part is stable and which part drops?” | Specificity over identity. |
| “Revise more.” | “Let’s schedule the next return and change one condition.” | Revision becomes designed practice. |
| “Don’t make the same mistake.” | “What will you check next time before the error happens?” | Correction becomes prevention. |
A Fragile learner needs evidence that successful retrieval is becoming more independent. Naming that improvement helps motivation far more than repeating “remember better”.
Fragile learning close to examinations
The examination calendar can tempt families into massed revision: long sessions, repeated reading, multiple full papers. Some of that work may be necessary, but fragile capabilities benefit from deliberate recurrence. An old method should reappear across days and inside mixed sets, not only during one intensive weekend.
Use full papers to sample fragility, not as the only repair tool. After a paper, classify errors. Pull out one or two fragile mechanisms, repair them in focused practice, then reinsert them into another mixed or timed context. This preserves the connection between diagnosis and performance.
As the examination approaches, the goal is to shrink the list of fragile high-dependency skills. Some low-frequency details may remain imperfect. Prioritise what repeatedly causes downstream mark loss.
What success looks like six months later
A learner who was Fragile may still forget occasionally six months later. The difference is that forgotten knowledge is easier to reconstruct, old errors recur less often, mixed work no longer feels like a complete reset, and the learner knows how to retrieve before reaching for notes. Support has become lighter and more strategic.
This is a major change in the cost of learning. New topics no longer sit on foundations that have to be rebuilt every week. The learner can devote more attention to interpretation, transfer and examination craft. That reduced maintenance burden is one of the clearest signs that Fragile has become Stable.
Subject mini-labs: test durability without turning everything into an exam
English mini-lab: retrieve the rule inside real language
Choose one grammar, vocabulary or comprehension capability that looked secure in recent work. Present it inside a fresh sentence or short passage without naming the rule. Ask the learner to identify what matters and explain the answer. If the learner can perform only after the rule name is supplied, the knowledge is still cue-fragile. If the learner retrieves and applies it independently, increase the delay next time.
Mathematics mini-lab: one old item inside a new set
Place one previously learned method among two or three current-school items. Do not label the old topic. Ask the learner to identify the method before calculating. This takes only a few minutes and provides better stability evidence than reassigning a whole old worksheet. If selection is correct but execution contains one local error, the learning may already be becoming Stable.
Science mini-lab: explain after the picture changes
Return to a known mechanism using a new diagram or everyday context. Ask the learner to explain cause and effect before checking notes. Fragile memorisation often breaks when the picture changes; durable understanding can reconstruct the explanation. Record whether the learner needed a keyword prompt, a mechanism prompt or no prompt.
A-Math mini-lab: prerequisite pulse check
Select two high-dependency foundations from earlier chapters and place them at the start of the lesson without announcement. Keep the check short. The purpose is to see whether algebraic access remains available, not to consume the lesson with old content. Fragile foundations should become easier to retrieve over successive weeks.
How to talk to a learner who is inconsistent
Inconsistency can feel personal to a learner: “I could do it yesterday, so why can’t I do it today?” Replace the self-judgement with a learning model: a capability can be available under one condition and unavailable under another; our job is to make access more reliable. This gives the learner something to do rather than something to be ashamed of.
| Instead of | Try | What it teaches |
|---|---|---|
| “You forgot again.” | “What cue did you need to reactivate it?” | Access can be measured and improved. |
| “We just did this.” | “What survives before you look back?” | Retrieval before review. |
| “Your memory is bad.” | “Which part is stable and which part drops?” | Specificity over identity. |
| “Revise more.” | “Let’s schedule the next return and change one condition.” | Revision becomes designed practice. |
| “Don’t make the same mistake.” | “What will you check next time before the error happens?” | Correction becomes prevention. |
A Fragile learner needs evidence that successful retrieval is becoming more independent. Naming that improvement helps motivation far more than repeating “remember better”.
Fragile learning close to examinations
The examination calendar can tempt families into massed revision: long sessions, repeated reading, multiple full papers. Some of that work may be necessary, but fragile capabilities benefit from deliberate recurrence. An old method should reappear across days and inside mixed sets, not only during one intensive weekend.
Use full papers to sample fragility, not as the only repair tool. After a paper, classify errors. Pull out one or two fragile mechanisms, repair them in focused practice, then reinsert them into another mixed or timed context. This preserves the connection between diagnosis and performance.
As the examination approaches, the goal is to shrink the list of fragile high-dependency skills. Some low-frequency details may remain imperfect. Prioritise what repeatedly causes downstream mark loss.
What success looks like six months later
A learner who was Fragile may still forget occasionally six months later. The difference is that forgotten knowledge is easier to reconstruct, old errors recur less often, mixed work no longer feels like a complete reset, and the learner knows how to retrieve before reaching for notes. Support has become lighter and more strategic.
This is a major change in the cost of learning. New topics no longer sit on foundations that have to be rebuilt every week. The learner can devote more attention to interpretation, transfer and examination craft. That reduced maintenance burden is one of the clearest signs that Fragile has become Stable.
Frequently asked questions
Is forgetting normal?
Yes. Some forgetting is normal. The educational question is whether planned returns become easier and more independent over time. A capability that never becomes easier to recover needs a closer look.
Should my child revise every day?
Not every capability needs daily review. Short distributed returns are often more useful than constant massed review. The schedule should protect current learning while keeping important older dependencies alive.
Why does tuition look better than school tests?
Tuition often provides recent practice, topic cues and immediate feedback. School tests remove some of those supports and mix content. If performance changes sharply, the learning may be fragile or not yet transferable.
Should I make more notes?
Notes can support understanding, but notes left open can also hide retrieval weakness. Use notes to build the model, then close them and test what the learner can reproduce.
Where this page sits in Atlas V2.0
This is the second learner-state owner under Start Here / When Learning Slips. If the learner still cannot reliably begin, use the Blocked route. If the capability now survives delay and ordinary cue removal, continue to Stable. Use How We Know Learning Has Really Held for the evidence standard and Learning Atlas V2.0 for the wider map.
Continue the learner-state route
If the learner still cannot reliably begin, return to The Blocked Learner. When learning now survives delay and ordinary cue removal, continue to The Stable Learner. Return to Learning Atlas V2.0 for the whole map.
Evidence and further reading
- Education Endowment Foundation: Feedback
- Nature Reviews Psychology: spacing and retrieval practice
- Educational Psychology Review: variability, retrieval and transfer
Last updated: 26 September 2026. Learner-state language describes instructional evidence, not a fixed trait or clinical condition.
