Quick Read
Transfer asks whether a learner can use an idea in a new situation.
Generalisation asks a harder question:
What part of the idea should remain true across many situations—and how far can we extend it before the conditions change?
This is where learning begins to move from examples toward principles.
The One-Sentence Answer
Generalisation is reliable when the learner can identify the invariant relationship beneath several examples, extend it to genuinely similar cases, and preserve the conditions that stop the rule from becoming too broad.
Where the Generalisation Test Sits in Curie
The Transfer Test changes the surface and asks whether the learner can still use the capability.
The Boundary Test asks where the rule must stop.
The Generalisation Test joins them: what is stable enough to extend, and what boundary must travel with the extension?
The Diagnostic Problem: Students Can Copy Variation Without Seeing the General Rule
A learner may solve ten related examples because each one resembles the previous one.
Then ask, “What is true in all ten?”
If the learner cannot answer, the practice may have produced local familiarity without a general model.
The opposite problem also occurs: the learner extracts a rule too quickly and applies it everywhere.
Good generalisation therefore needs both extension and restraint.
Find What Stayed the Same While the Examples Changed
| Change across examples | Possible invariant | Question to ask |
|---|---|---|
| Different story contexts | Same quantitative relationship | What relationship survives the story change? |
| Different texts | Same evidence-to-inference logic | What makes an inference justified in all of them? |
| Different experiments | Same causal mechanism | Which mechanism explains the shared pattern? |
| Different algebraic forms | Same structural property | What remains true despite the form changing? |
The invariant is what makes the generalisation useful.
A General Rule Should Explain More Than It Memorises
A slogan can sound general without containing usable structure.
“Always show evidence.”
Useful—but what counts as evidence, how much is required, and what relationship must the evidence support?
A stronger generalisation tells the learner what remains stable and under what conditions.
English: Generalise the Relationship, Not the Sentence Template
A student may learn one successful comprehension phrase or essay structure and then overextend it.
The stronger generalisation is not “always use this sentence.” It is “claims need support proportionate to the question, evidence and purpose.”
For developmental English progression, continue through the English Tutor route. Curie keeps the generalisation principle visible across subjects.
Mathematics: From Repeated Cases to Structure
Mathematics generalisation often begins when the learner sees what remains invariant across changing values, diagrams or examples.
Patterns become rules. Particular equations become relationships. Concrete ratios become proportional reasoning.
The existing Mathematics estate provides the subject-specific derivations and proof. For developmental Mathematics progression, continue through the Mathematics Tutor route.
Science: Extend a Mechanism Only as Far as the Evidence Allows
Science generalisation moves from repeated observations toward patterns and mechanisms, but it must retain conditions and uncertainty.
A learner should be able to say not only “this happened in the experiment,” but “this mechanism should also apply in these related conditions—and here is why.”
The Boundary Test then asks where that extension should stop.
Additional Mathematics: General Forms Compress Many Cases
A-Math increasingly uses symbolic forms that represent whole families of cases.
The learner must understand what features of the specific example are incidental and what features belong to the general structure.
This is one reason algebraic abstraction becomes powerful: it carries the invariant relationship beyond one set of numbers.
What Counts as Evidence of Generalisation?
- The learner identifies what stayed constant across several cases.
- They can state the general relationship in their own words or symbols.
- They can predict a new case not previously practised.
- They can explain why the new case belongs.
- They can reject a near-miss case that violates the conditions.
- They can revise the generalisation if new evidence exposes a limit.
Overgeneralisation Is Not the Opposite of Learning
Sometimes overgeneralisation is evidence that the learner has begun to see a pattern but has not yet located its boundary.
That is useful diagnostic information.
Instead of simply saying “wrong,” ask what feature made the learner believe the new case belonged. Then use a contrast or boundary case to refine the rule.
Repair: Build the Generalisation From Cases, Then Test Its Limits
- Present several varied examples with one shared structure.
- Ask what changed.
- Ask what stayed the same.
- Have the learner state the general relationship.
- Ask for a prediction in a new valid case.
- Introduce a near-miss outside the rule.
- Refine the condition.
- Return later with a new context.
Prediction Is a Strong Test of a General Rule
Prediction Before Feedback asks what the learner expects before the answer arrives.
A generalisation becomes more credible when it correctly predicts new cases that were not part of the original pattern-finding exercise.
Tutorial Should Vary Surface While Protecting Structure
Inside a Tutorial, the tutor can design a family of examples where the surface changes deliberately but the deep relationship remains.
Then one final example crosses the boundary.
The learner has to say which cases belong, why they belong, and why the final case does not.
What Tutor Should Say
The Tutor dashboard might say:
“You can apply the pattern to familiar examples, but your rule is still too broad and includes cases that violate the original condition.”
Or:
“You identified the invariant relationship, predicted two new cases correctly and rejected the near-miss because the required condition was absent.”
What Parents Can Ask
- What stayed the same across these examples?
- What changed but did not matter?
- Can you make a new example that should work?
- Can you make one that almost works but fails?
- What condition separates the two?
Voyage: Education Moves From Cases Toward Principles
Young learners need many concrete cases. Across the Voyage, education increasingly asks them to compress experience into principles, models and transferable rules without losing contact with evidence.
University and adult learning depend heavily on this move: identify what is general, recognise what is local, and know when conditions have changed enough to require a new model.
Darwin: General Rules Must Remain Correctable
The Darwin Series protects generalisation from becoming dogma. A useful general rule should survive many cases, but when repeated evidence crosses its boundary, the model must be allowed to evolve.
Boundaries
- Do not demand abstraction before learners have enough cases to abstract from.
- Do not reward broad rules merely because they sound elegant.
- Do not confuse one successful transfer with a robust generalisation.
- Always carry assumptions and conditions with the general rule.
- Allow new evidence to refine or narrow the learner’s claim.
The Core Rule of the Generalisation Test
Help the learner extract the invariant relationship from varied cases, extend it to genuinely new situations, carry its conditions with it, and stop the extension when evidence crosses the rule’s boundary.
Frequently Asked Questions
How is generalisation different from transfer?
Transfer asks whether a capability works in another case. Generalisation asks what principle explains the family of cases and how far that principle can extend.
Is overgeneralisation always bad?
No. It can show that the learner has detected a pattern but has not yet located its boundary. That makes it useful evidence for refinement.
What is the strongest simple test?
Ask the learner for one new case that should work and one near-miss case that should not, then require an explanation of the condition separating them.
Marie Curie Series · Transfer Test · Boundary Test · Prediction Before Feedback · Contrast Test · Tutor · Voyage · Darwin
eduKate Sengkang | Small groups of up to 3 | 1.5-hour tutorials | Properly Taught Kids Shine a Bright Light Into the Future.
