Jia Jun gives the answer his younger self was once praised for.

That is the problem.

Years earlier, a simple classroom rule helped him make sense of a difficult idea.

It was useful.

Memorable.

Good enough for the questions he was then expected to answer.

Now the curriculum has become more precise.

The old shortcut is no longer merely incomplete.

It gives the wrong answer in the new case.

The tutor has a choice.

Keep both explanations alive and teach Jia Jun when to use each?

Or deliberately retire the old rule because it no longer deserves to govern any part of the target territory?

This is not an ordinary interference problem.

In The Interference Test, two valid rules may compete because the learner has not yet learned the boundary between them.

Here, the deeper question is whether the old rule still deserves a boundary at all.

When new learning arrives, should the old rule remain part of the learner’s usable repertoire—or has the new model genuinely superseded it?

That is the Supersession Test.

Quick Read: The Supersession Test in One Sentence

Keep an old rule when it remains valid in a clearly defined territory; narrow it when its scope has shrunk; retire it only when the newer model explains the same job more accurately without losing useful capability.

This distinction matters because education often uses temporary simplifications.

Young learners need manageable models.

Beginners need scaffolds.

Early procedures reduce cognitive load.

Rules of thumb can help a learner move before full theory is available.

But temporary tools create a later responsibility:

Know when the learner has outgrown the simplification.

1. Old Learning Has Four Possible Futures

When new learning arrives, the old learning can enter one of four states.

Coexist.
Both old and new methods remain valid in different situations.

Narrow.
The old rule remains useful, but only inside a smaller and more explicit domain.

Integrate.
The old representation becomes one special case inside the newer, broader model.

Supersede.
The old rule is no longer an appropriate target response and should stop controlling future decisions.

The tutor’s job is not to prefer newness automatically.

It is to determine which state is intellectually correct and educationally useful.

2. Newer Does Not Automatically Mean Better

A new method can be more sophisticated and still not replace the old one.

Bar models do not become useless because algebra is learned.

Mental arithmetic does not become obsolete because calculators exist.

A diagram does not become childish because an equation can represent the same relationship.

A familiar planning scaffold does not automatically become wrong because a student can write more freely.

Sometimes the old tool remains faster, clearer or pedagogically useful.

The Supersession Test therefore begins with a refusal:

Do not delete useful knowledge merely because something more advanced has arrived.

3. But Familiar Does Not Automatically Mean Worth Preserving

The reverse mistake is sentimental education.

“This method worked when she was younger.”

“This is how he was taught.”

“She feels comfortable with this rule.”

Comfort is not enough.

If the old rule now produces systematic errors, blocks a better model or requires constant exceptions, preserving it can become expensive.

Learning has to be allowed to upgrade.

4. Ask the First Supersession Question: Is the Old Rule Still True Anywhere?

This separates replacement from narrowing.

Consider the beginner heuristic:

Multiplication makes numbers bigger. Division makes numbers smaller.

For whole numbers greater than one, this may appear to work often enough to support early intuition.

Then fractions and decimals arrive.

Multiply 8 by 0.5.

The result becomes smaller.

Divide 8 by 0.5.

The result becomes larger.

The old statement cannot remain a universal rule.

It may be retained only as a tightly scoped observation about particular number ranges—or better, replaced by a more relational understanding of multiplication and division.

Scope decides its future.

5. Ask the Second Question: Does the New Model Contain the Old One as a Special Case?

Sometimes new learning does not destroy old learning.

It explains why the old rule seemed to work.

A narrow pattern becomes one special case inside a broader structure.

This is educationally powerful because the learner does not have to treat the past as nonsense.

The old rule becomes:

something that worked under these conditions for this reason.

That is integration, not erasure.

6. Ask the Third Question: Does the Old Rule Still Add Useful Capability?

Two methods can produce the same answer.

Should both remain?

Perhaps.

An old visual method may reveal structure more clearly.

A newer symbolic method may be faster at scale.

Keeping both creates representational flexibility.

But if the old rule contributes no unique understanding, no useful speed, no checking function and no special-case value—while continuing to create errors—its maintenance cost may exceed its educational value.

Then supersession becomes more plausible.

7. Ask the Fourth Question: Is the Old Rule Creating Interference?

This connects directly to The Interference Test.

If old Rule A and new Rule B are both valid, teach discrimination.

If old Rule A is no longer a valid response anywhere the learner currently needs it, continued activation of A is pure interference.

Then the teaching objective changes.

Do not maintain A beside B.

Teach the learner to inhibit A and reconstruct the decision around B.

That is supersession.

8. Ask the Fifth Question: Is the Old Rule a Scaffold Rather Than Knowledge?

Some things that look like knowledge are actually temporary support structures.

A writing frame.

A colour-coding system.

A fixed checklist.

A bar model drawn in a prescribed format.

A mnemonic.

The scaffold may have done its job when the learner can perform the underlying function without it.

In that case, the scaffold should not be maintained as if it were the learning itself.

The function remains.

The visible support can disappear.

9. Supersession Is Not the Same as Scaffolding Fade

Scaffolding fade asks how external help reduces as competence rises.

Supersession asks whether an internal rule or representation should remain part of the learner’s future decision system.

A checklist may fade because the learner internalised the process.

A false heuristic may be superseded because the learner now possesses a more accurate model.

The two can occur together.

They are not the same educational job.

10. Conceptual Change Is Often Revision, Not Simple Deletion

Research on conceptual change is useful here because it warns against a naive replacement metaphor.

Older misconceptions do not always vanish when correct scientific knowledge is learned.

Research reviews describe cases in which old and new conceptions continue to coexist, with learners needing to select or inhibit the older response when the context calls for the scientifically accepted one.

That means educational supersession is not necessarily neurological deletion.

The practical goal is stronger:

The obsolete rule should stop governing the learner’s decisions even if traces of the old idea remain retrievable.

11. The Supersession Test Begins With Territory

Draw two circles mentally.

Territory A:

where the old rule gives an appropriate answer.

Territory B:

where the new rule gives an appropriate answer.

Now ask:

• Are the circles separate?
• Do they overlap?
• Does B fully contain A?
• Does B replace A completely?
• Does A still offer a useful shortcut inside part of B?

This prevents a vague “old versus new” argument.

Supersession is fundamentally a scope decision.

12. The Coexistence Case

Maya learns bar models.

Later she learns algebra.

Should algebra supersede bar models?

No automatic reason exists.

Bar models can remain useful for visualising relationships, checking reasonableness and communicating structure.

Algebra can be more general and efficient for complex symbolic manipulation.

The learning job is selection.

Which representation exposes this problem best?

That is coexistence, not supersession.

13. The Narrowing Case

A beginner rule works only inside a restricted set of conditions.

The learner once treated it as universal.

New learning reveals the boundary.

The correct move is not necessarily to erase the rule.

Rewrite it.

Old:

Multiplication makes bigger.

Narrowed:

Multiplying a positive quantity by a number greater than 1 increases its magnitude; multiplying by a positive number between 0 and 1 decreases it.

The learner now owns the conditions.

14. The Integration Case

Old knowledge becomes a special case of a broader relationship.

This is often the most elegant outcome.

The learner does not maintain two disconnected rules.

They maintain one larger model that explains why the old case worked.

Integration reduces memory clutter.

It also strengthens transfer because the learner can see relationships rather than isolated procedures.

15. The True Supersession Case

True supersession is rarer than people think.

It requires the old rule to have no remaining useful decision territory for the target work—or to create enough systematic harm that preserving it is not worth the benefit.

A false misconception corrected by a scientific model can be a supersession target.

A temporary writing scaffold that has become a rigid dependency can be retired as a visible rule while its underlying rhetorical function remains.

An obsolete shortcut that repeatedly fails under the learner’s current number domain can be replaced by a relational rule.

Supersession should be earned, not assumed.

16. The Counterexample Test

Ask the learner to apply the old rule to a case where the newer model predicts failure.

If the old rule survives only because the learner has never seen its counterexample, the curriculum has reached a useful turning point.

The tutor should not use the counterexample merely to embarrass the old idea.

Use it to ask:

What assumption did the old rule depend on that this example violates?

This converts contradiction into model revision.

17. The Coverage Test

Give the learner several cases.

Some where the old rule works.

Some where it fails.

Some where both old and new methods produce correct results.

Ask the learner to map the territory.

If the old rule has a clean, useful region, keep it with a boundary.

If every useful case is already handled more accurately by the new model and the old rule adds no unique value, supersession becomes stronger.

18. The Exception Burden Test

Old rules can become expensive because they require too many exceptions.

“Use this rule—except here.”

“And here.”

“And when this number appears.”

“And not under these conditions.”

At some point the exception list is evidence that the learner should move to the more general model.

A compact general rule can replace a fragile patchwork of remembered exceptions.

19. The Error-Cost Test

Some old rules are harmless simplifications.

Others generate recurring marks lost under current curriculum conditions.

If the learner repeatedly applies the old heuristic where it is now false, the cost of coexistence rises.

This can move the issue up The Priority Queue.

Supersession is not philosophical housekeeping.

It is warranted when the old response is actively damaging performance.

20. The Retrieval Test

After teaching the newer model, return later.

Present a case that used to trigger the old heuristic.

Which model comes first?

If the obsolete response still wins under delay, the conceptual update is not yet behaviourally dominant.

Do not assume one good lesson erased the old pathway.

Conceptual-change research gives a useful caution here: prior misconceptions can continue to coexist and interfere even after correct knowledge has been learned.

The supersession job therefore includes later selection.

21. The Explanation Test

A learner can choose the new answer because the tutor recently said so.

That is not yet a robust conceptual replacement.

Ask:

Why did the old rule appear to work before, and why does it fail here?

A strong answer contains history and boundary.

The learner understands the old model rather than merely being told to suppress it.

22. The Generation Test

Ask the learner to create:

• a case where the old rule works;
• a case where it fails;
• a case where the new model explains both.

If the learner can generate the boundary, the conceptual architecture is becoming theirs.

This is stronger than repeating the newer sentence.

23. The Compression Test

Good advanced learning often reduces the number of disconnected rules.

Three memorised exceptions become one general relationship.

Five separate procedures become one structural model plus a selection rule.

If the newer model explains more cases with fewer arbitrary exceptions, it may deserve to become the learner’s primary representation.

Supersession can therefore be a form of knowledge compression.

24. Primary 1: Preserve Useful Simplicity

Young learners need simple rules.

The goal is not to replace every beginner explanation with adult technical language immediately.

A Primary 1 rule can be developmentally appropriate even if later learning will refine it.

The tutor’s responsibility is to avoid presenting a temporary simplification as an eternal law when a small qualification would protect future learning.

Good beginner teaching leaves room for upgrade.

25. Primary 2: Retire Visible Scaffolds Before They Become Identity

Ethan uses a sentence checklist.

Capital letter.

Complete idea.

Punctuation.

Useful.

Months later, he checks those things automatically.

Does the paper checklist still need to remain on every task?

No.

The functions remain.

The visible scaffold can be superseded by an internal routine.

This is successful teaching, not loss of support.

26. Primary 3 and Primary 4: Science Needs Model Upgrades

Science learning repeatedly asks children to revise explanations.

A familiar everyday explanation may conflict with a scientific model.

The tutor should not merely say:

“That is wrong. Memorise this.”

Instead:

• surface the old prediction;
• test it against evidence;
• introduce the better model;
• show which observations the newer model explains;
• return later to see which model governs prediction.

Science supersession should be evidence-driven.

27. Primary 5: Watch for Old Heuristics Becoming Too Broad

Primary 5 is full of moments where a rule that worked on simpler material meets a larger domain.

Numbers widen.

Fractions and percentages interact.

Science explanations become more conditional.

Writing becomes less template-driven.

This is a good year to ask:

Which Primary-school shortcuts are still useful, and which now need a boundary or upgrade before PSLE practice hardens them?

28. Primary 6: Do Not Introduce Unnecessary Conceptual Revolution Near PSLE

Supersession has timing.

Near a high-stakes examination, replacing a functioning method simply because another method is more elegant may be irresponsible.

If the old method is valid, stable and fast enough, preserve it.

Supersede only when the current method is producing consequential errors or blocking required curriculum understanding.

Educational elegance does not outrank examination stability.

29. Secondary 1: Transition Creates Legitimate Supersession Opportunities

Secondary 1 introduces more symbolic and abstract representations.

Some Primary procedures should coexist.

Some should become checking tools rather than primary methods.

Some beginner heuristics need explicit narrowing because the number system or problem space has expanded.

The transition is not:

Primary wrong, Secondary right.

It is:

Which earlier representations remain useful inside a more general system?

30. Secondary 2: Procedures Should Begin Becoming Principles

A learner may have memorised several algebraic transformations separately.

Expand.

Factorise.

Rearrange.

Collect terms.

As understanding grows, the learner should increasingly see these as transformations governed by equivalence and structure rather than unrelated commands.

The older procedural list is not necessarily deleted.

It is integrated into a more general model.

That reduces interference and supports transfer.

31. Secondary 3: Additional Mathematics Exposes Obsolete Shortcuts

Additional Mathematics punishes rules that were memorised without scope.

Maya may have an old algebra shortcut that worked on simple expressions.

Inside functions, trigonometry or calculus, the shortcut begins producing structural errors.

The tutor should ask whether the shortcut can be narrowed safely.

If not, replace it with a more general structural rule before the old response becomes entrenched inside A-Math.

32. Secondary 4: Stability Can Beat Novelty

An examination year changes the supersession threshold.

Suppose Ethan uses a valid but slightly longer method.

A newer shortcut is faster.

Should the tutor replace Ethan’s method six weeks before O-Level?

Only if the expected benefit clearly exceeds the destabilisation risk.

Supersession is not a fashion contest.

A mature tutor sometimes leaves a working system alone.

33. JC: Advanced Learning Often Integrates Rather Than Replaces

At JC, many earlier rules become special cases of broader mathematics.

The ideal learner no longer stores them as unrelated school-year memories.

They become connected representations inside a larger structure.

This is why higher-level teaching should repeatedly ask:

What does this new model explain that the old one could not?

Where does the old result appear inside the new model?

What conditions changed?

Integration produces a coherent mathematical world.

34. English: Writing Frames Should Eventually Lose Their Authority

Hana learns a paragraph frame.

Claim.

Evidence.

Explanation.

This can be extremely useful.

But if every future paragraph must visibly reproduce the same sequence and sentence pattern, the scaffold has begun controlling style rather than supporting reasoning.

What should be superseded?

The rigid surface frame.

What should remain?

The responsibility to make a claim, support it and explain the relationship.

Good supersession preserves function while retiring unnecessary form.

35. English: “Never” Rules Often Need Scope

Beginner language teaching sometimes uses absolute-sounding rules because they are easy to remember.

As proficiency grows, students need to distinguish grammar, convention, register, style and rhetorical choice.

A rule that says “never” may eventually become:

usually avoid this in this formal context unless you have a clear rhetorical reason.

The older absolute was useful for control.

The newer model adds register and purpose.

That is narrowing and integration rather than reckless rule-breaking.

36. Vocabulary: Definitions Should Become Concept Boundaries

A young learner memorises:

reluctant = hesitant.

Useful beginning.

Later, Hana learns that the words are not perfectly interchangeable.

Reluctance carries unwillingness.

Hesitation can arise from uncertainty or delay.

The one-word equivalence should be superseded by a richer semantic map.

The original synonym did its job.

Now precision needs more.

37. Mathematics: A Shortcut Should Never Outrank the Relationship

Shortcuts are useful only while they preserve the mathematics.

If a remembered shortcut works under one surface but fails under a slightly changed one, the learner needs the relationship underneath.

The Supersession Test asks:

Can the shortcut be derived from the general model?

Does it remain safe?

Can the learner identify its conditions?

If not, retire it.

Speed that produces systematic error is not efficiency.

38. Mathematics: Do Not Supersede Multiple Valid Methods Needlessly

Factorisation.

Completing the square.

Quadratic formula.

These are not a historical ladder where each newer method deletes the previous one.

They remain useful under different conditions.

The correct learning job is method selection.

This is precisely why the Supersession Test must come after the Interference Test, not replace it.

39. Science: Replace the Prediction, Not the Learner’s Dignity

Misconceptions often feel reasonable because they fit everyday experience.

When evidence contradicts them, the tutor should not frame the learner as foolish.

Ask:

What did the old idea predict?

What happened instead?

What does the new model explain?

Where would the two models make different predictions?

This creates conceptual change through evidence rather than authority alone.

40. Science: Coexistence Can Be Necessary

Not every apparently competing scientific representation should be deleted.

Different models can be useful at different scales or for different questions.

A school-level model may remain useful even when later science introduces a more detailed one, provided the learner understands the scope and limitations.

This is an important scientific habit:

Models are tools with domains, not eternal pictures of reality.

Supersession should therefore be based on explanatory need, not on the belief that sophistication always destroys simpler models.

41. Additional Mathematics: New Structure Should Supersede Unsafe Algebra Habits

Ethan has an old habit:

expand first.

It worked often enough in earlier algebra.

In A-Math, expanding everything can destroy useful structure and create unnecessary work.

Should “expand first” coexist as a general strategy?

No.

It should be superseded by:

Inspect structure first; transform only when the transformation serves the next mathematical goal.

Expansion remains a valid operation.

The obsolete part is the unconditional trigger.

42. Supersede the Trigger Before Superseding the Procedure

This is a subtle but powerful distinction.

Sometimes the procedure remains mathematically valid.

What has become obsolete is the rule that says when to deploy it.

“Always expand.”

Supersede.

“Expand when it exposes or enables the structure needed next.”

Keep.

“Always use this paragraph frame.”

Supersede.

“Make sure the paragraph contains enough reasoning to connect evidence to claim.”

Keep.

Often the old action survives while the old trigger is upgraded.

43. Supersession Can Reduce Interference Load

A learner maintains seven separate rules.

Each has exceptions.

Each competes for retrieval.

Then a more general model arrives that explains five of them.

Integrate.

The learner now maintains one model plus a small number of special cases.

This reduces selection burden.

It also makes the Maintenance Window cheaper because fewer disconnected objects need deliberate return.

44. Supersession Can Reduce Reactivation Cost

Disconnected shortcuts are easy to forget.

A general model may be easier to reconstruct.

When old learning is integrated into a broader relationship, the learner can often derive the special case rather than memorise it separately.

This can lower The Reactivation Cost.

The knowledge no longer needs to remain permanently hot.

It can be rebuilt from structure.

45. Supersession Needs an Evidence Threshold

Do not replace a stable method because one teacher prefers another.

Do not retire a useful scaffold because the student succeeded twice without it.

Do not declare a misconception corrected because the learner repeated the new explanation immediately after instruction.

The Evidence Threshold still applies.

For a small reversible trial, little evidence may be needed.

For permanent removal of a useful method, the threshold should be higher.

46. Supersession Needs a Transfer Gate

The learner uses the new model in the teaching example.

Good.

Now:

• change the surface;
• remove the label;
• mix old-trigger and new-trigger cases;
• delay;
• return to school.

If the old response returns under every changed condition, supersession is incomplete.

The Transfer Gate tests whether the new model truly owns the territory beyond the lesson.

47. Supersession Needs a Counterfactual Check

Performance improves after the old shortcut is retired.

Did supersession cause the improvement?

Perhaps.

The Counterfactual Check asks what else changed and whether the improvement matches the mechanism.

If the exact errors generated by the old heuristic disappear while unrelated performance remains similar, the supersession story becomes more plausible.

Again:

use evidence to assign credit proportionately.

48. The Supersession Ladder

A practical sequence is:

Level 1 — Surface the old rule.
What does the learner currently believe or habitually do?

Level 2 — Map its territory.
Where does it still work, and where does it fail?

Level 3 — Introduce the newer model.
Show what additional cases it explains.

Level 4 — Decide coexist, narrow, integrate or supersede.
Do not use one response for every old/new relationship.

Level 5 — Teach the boundary or replacement explicitly.
The learner should know what changed.

Level 6 — Test mixed selection.
Does the learner choose the correct model without labels?

Level 7 — Delay.
Does the newer model remain dominant after immediate instruction has cooled?

Level 8 — School return.
Does the upgraded model govern authentic performance?

This is an explanatory sequence, not a validated measurement scale.

49. The Learner Should Be Told What Is Happening

“But you told me this last year.”

Students notice contradictions.

Do not respond defensively.

Say:

“Yes. That rule was useful for the simpler cases you were learning then. Now the domain is larger, so we need a rule that explains both the old cases and these new ones.”

This teaches something important about knowledge itself.

Models can be improved.

Understanding can become more precise.

Learning is allowed to revise itself.

50. Parents Need the Same Explanation

A parent sees the child abandon a familiar method.

“Why change it? She was getting correct answers.”

Sometimes that is the right challenge.

The tutor should be able to answer:

• what the old method still does well;
• where it now fails;
• what the new model adds;
• whether both will remain available;
• why the timing of the change is justified.

If the tutor cannot explain the supersession, perhaps the change is merely preference.

51. The Parent Should Not Force the Old Method Back Into Service

“That is not how I learned it.”

This sentence can create unnecessary interference.

A parent’s method may still be valid.

Or the curriculum may now frame the idea differently.

Before reintroducing the old route at home, ask whether the school and tutor are deliberately building a newer representation.

Family Life Education sometimes means resisting the urge to add another correct-but-competing explanation at the kitchen table.

52. The Tutor Should Not Supersede for Branding

A tuition centre invents a branded method.

Now every school method is treated as inferior.

That is not evidence.

A new method should supersede an old one because it improves accuracy, generality, efficiency, understanding or transfer enough to justify the change.

Not because the tutor owns the newer label.

The learner’s cognition is not a marketing surface.

53. The Tutor Should Not Maintain Obsolete Rules for Nostalgia

Tuition has the opposite temptation too.

“I have taught this shortcut for twenty years.”

Longevity is not evidence of current usefulness.

If the curriculum, notation, problem types or learner level now make the shortcut unsafe, upgrade.

Experience should make a tutor better at retiring obsolete tools, not more attached to them.

54. The Supersession Test Protects Against Interference

If both old and new rules remain in active maintenance when only one deserves future use, the learner pays a selection tax forever.

Every relevant question activates both.

The wrong one must be inhibited.

If the old rule contributes no remaining value, that tax is unnecessary.

Supersession reduces the number of competitors the learner must manage.

55. The Supersession Test Protects Against Premature Forgetting

Deleting old knowledge too aggressively also has a cost.

A visual method may remain valuable for checking.

A slower method may remain robust under pressure.

An old model may remain the clearest way to explain a special case.

A learner may need multiple representations to become flexible.

The Supersession Test therefore protects both directions:

do not keep what has expired.

Do not discard what still works.

56. The Supersession Test Protects Maintenance Time

Every rule kept alive has a maintenance cost.

Retrieval.

Practice.

Selection.

Interference management.

If three obsolete heuristics can be replaced by one general model, the Maintenance Window becomes cheaper.

Knowledge architecture matters because attention is finite.

57. The Supersession Test Protects Family Time

A student appears confused between three old approaches and the school’s new approach.

The family response is often:

practise all four.

That may deepen the competition.

First decide the architecture.

Which methods coexist?

Which old rule needs narrowing?

Which scaffold can disappear?

Which misconception should stop being rehearsed?

Then practise the correct architecture.

Less volume.

More clarity.

58. The Tutor’s Supersession Test

• What exactly is the old rule, method, model or scaffold?
• What job did it originally solve?
• Is it still valid anywhere in the learner’s current curriculum?
• Does the new model contain the old case as a special case?
• Does the old method still add speed, clarity, checking value or representation value?
• How many exceptions does the old rule now require?
• Is the old rule causing recurrent interference?
• Can the learner explain where the old rule fails?
• Should the outcome be coexistence, narrowing, integration or supersession?
• Has the newer model passed a Transfer Gate?
• Does it remain dominant after delay?
• Would replacing the old method destabilise performance near an examination?
• What should stop being maintained if supersession succeeds?

These questions prevent “new method” from becoming an automatic virtue.

59. The Student’s Supersession Test

An older learner can ask:

• What did I used to believe or do?
• Where did that work?
• Where does it now fail?
• What does the newer model explain?
• Do I still need the old method anywhere?
• Can the new model generate the old result as a special case?
• Am I mixing the two because both are valid—or because one is obsolete?
• Can I recognise a case that should trigger the newer model without being told?

This is metacognition about knowledge architecture itself.

60. The Parent’s Supersession Test

A parent can ask the tutor:

• Is the old method actually wrong now, or merely less efficient?
• Will the child still need it for checking or explanation?
• Are both methods valid under different conditions?
• What evidence shows that changing now is worth the disruption?
• How will you test whether the newer model is truly independent?

These are reasonable questions.

Good tuition should be able to answer them without defensiveness.

61. The Long Return: Experts Regularly Supersede Rules

Adult expertise is full of supersession.

A programmer stops using a deprecated approach when the new system offers a safer replacement.

An engineer updates a design rule when better evidence changes the standard.

A scientist replaces an explanation when a stronger model accounts for the evidence.

A professional retains an older method when it remains useful under a special condition.

Expertise is not the accumulation of every rule ever learned.

It is the continual maintenance of a coherent, current and appropriately scoped repertoire.

62. Back to Jia Jun

The tutor places the old explanation beside the newer one.

“Why did this older rule seem to work before?”

Jia Jun studies the examples.

“Because all the questions had the same condition.”

“And now?”

“Now that condition isn’t always true.”

“So should we keep both rules?”

Jia Jun thinks.

“Maybe not as two rules. The new one explains the old case too.”

There it is.

Not deletion.

Not contradiction.

Compression.

The old classroom rule has not become a shameful mistake.

It has become a limited case inside a better model.

The tutor does not ask Jia Jun to forget his learning history.

He asks him to understand why the history had to advance.

That is the Supersession Test doing its real job.

---

The Supersession Test in One Page

1. Newer does not automatically mean better.
Preserve useful old methods when they still own valid territory.

2. Familiar does not automatically mean worth preserving.
Old shortcuts can become costly when the domain expands.

3. Choose among four futures.
Coexist, narrow, integrate or supersede.

4. Map territory.
Where is the old rule still valid?

5. Ask whether the new model contains the old case.
Integration is often better than memorising two disconnected rules.

6. Measure unique value.
Keep the old representation if it still adds speed, clarity, checking or explanatory power.

7. Count the exception burden.
A rule requiring constant patches may deserve replacement by a general model.

8. Inspect interference.
If the old response is obsolete yet still intrudes, teach inhibition and replacement.

9. Distinguish scaffold from knowledge.
Retire visible support when the underlying function has become internal.

10. Teach why the old rule once worked.
Revision is stronger than arbitrary contradiction.

11. Test with counterexamples.
Find cases where the old and new models make different predictions.

12. Test after delay.
The old pathway may return after immediate instruction cools.

13. Protect high-stakes stability.
Do not replace a valid working method near an examination merely for elegance.

14. Remove obsolete items from maintenance.
Do not spend future retrieval time keeping expired heuristics alive.

15. Let knowledge upgrade without erasing its history.
A mature learner understands not only the current model, but why it became necessary.

Where This Fits in eduKatePunggol

• How Tuition Works at eduKatePunggol — the broad tuition relationship.
• How Tuition Works | The Weak Link — identify the first limiting mechanism.
• The Flight Recorder — preserve the history showing what the learner previously used successfully.
• The Learning Dispatcher — route the upgraded skill to the appropriate environment.
• The Exit Ramp — remove external supports when the underlying function becomes internal.
• The Priority Queue — promote obsolete rules when they create recurrent high-cost errors.
• The Translation Layer — describe precisely what old response is appearing and under which conditions.
• The Transfer Gate — test whether the newer model governs fresh and authentic contexts.
• The Diagnostic Probe — distinguish obsolete knowledge from retrieval failure, representation failure or ordinary interference.
• The Evidence Threshold — determine when the evidence is strong enough to retire or integrate an old rule.
• The Counterfactual Check — assess whether replacing the old rule plausibly improved performance.
• The Maintenance Window — stop deliberately maintaining rules that have been legitimately superseded.
• The Reactivation Cost — understand whether a useful older method remains cheap to recover when it is kept as a special-case tool.
• The Interference Test — distinguish two valid competing rules from a genuinely obsolete rule that should no longer control decisions.

This completes Batch 04:

• The Maintenance Window — when should stable learning return?
• The Reactivation Cost — how expensive is restoration after fading?
• The Interference Test — are neighbouring rules competing?
• The Supersession Test — should the old rule still exist as an active option at all?

The maintenance architecture now has a full lifecycle:

Maintain what remains useful → reactivate what cools → discriminate what competes → supersede what has genuinely expired.

---

Research Foundations

The Supersession Test is an explanatory tuition model, not a validated clinical or psychometric instrument. It draws on conceptual-change research showing that learning often involves revising prior conceptions and that older ideas may continue to coexist with newer knowledge rather than disappearing cleanly.

• Vosniadou — Conceptual Change, Oxford Research Encyclopedia of Education (2020). The review describes conceptual change as revision of prior conceptions and notes evidence that misconceptions may remain available and compete with scientifically accepted conceptions even after learning.
• Blown & Bryce — Ausubel’s Meaningful Learning Re-Visited, Current Psychology (2023). The review discusses conceptual change as involving coexistence and revision rather than assuming that earlier knowledge is simply deleted.
• Pacaci — Effectiveness of Conceptual Change Strategies in Science Education: A Meta-Analysis, Journal of Research in Science Teaching (2024). The meta-analysis reviews several approaches to helping learners revise misconceptions, including cognitive conflict, cognitive bridging and ontological-category change.
• Butterfuss & Kendeou — Reducing Interference From Misconceptions: The Role of Inhibition in Knowledge Revision, Journal of Educational Psychology (2020). The experiments examine how prior misconceptions can interfere with processing corrected information and how inhibition relates to knowledge revision.
• American Psychological Association — How Do I Get My Students Over Their Alternative Conceptions?. The teaching resource summarises why entrenched prior conceptions can interfere with new learning and why conceptual change requires more than simply adding another fact.

The practical conclusion is simple.

Do not keep every rule forever.

Do not delete every old rule because something newer appears.

Ask what territory the old rule still owns.

Ask what the newer model explains.

Ask whether both create useful flexibility or needless competition.

Then choose deliberately:

coexist.

narrow.

integrate.

or supersede.

Learning should remember enough of its past to understand why it changed—and be free enough to stop carrying what no longer helps.