You see an unfamiliar bird.
Small.
Brown.
Quick movements.
Short beak.
You do not begin by consulting a perfect abstract definition of every bird species you have ever encountered.
Sometimes another process happens.
The new bird reminds you of one you have seen before.
Then another.
Specific remembered cases become reference points.
Those remembered instances are exemplars.
Quick Read
An exemplar is a specific remembered instance that can be used as a reference when judging, classifying, interpreting or responding to a new case.
In exemplar theories of categorisation, people are modelled as retaining information about individual category members and comparing a new item with stored instances.
This differs from:
- prototype: a central or summary-like category representation,
- rule: an explicit condition–classification structure,
- schema: a richer structured body of prior knowledge,
- example: any case shown or mentioned, whether remembered as a cognitive reference or not.
One-sentence answer: An exemplar is a specific remembered case that helps the mind judge a new case by similarity, contrast or analogy to something previously encountered.
An Exemplar Is a Specific Case
You remember one teacher.
One difficult examination question.
One restaurant.
One bird.
One failed project.
Each can become a reference case.
The power of exemplar reasoning comes from preserving details that abstractions often remove.
Exemplar Is Not Prototype
eduKateSG already has Prototype Theory in English Vocabulary.
That page owns prototype theory, typicality and category centrality.
The distinction is foundational.
Prototype account:
Store or construct a summary of what members of the category are typically like.
Exemplar account:
Retain information about specific members and compare the new item with those remembered instances.
The APA Dictionary defines exemplar theory in precisely this contrast: categorisation based on specific remembered instances rather than an abstract prototype or feature-based defining rule.
Exemplar Is Not Rule
Rule:
If an integer is divisible by 2, classify it as even.
Exemplar:
2, 8, 14 and 26 are remembered members of the category.
A rule can classify a new case without comparing it with remembered examples.
An exemplar strategy relies on stored instances.
Human cognition can use both.
A classic study, Rules and Exemplars in Category Learning, found evidence that people can combine rule induction with exemplar encoding rather than operating through one mechanism exclusively.
Exemplar Is Not Schema
Schema:
what restaurants generally contain and how restaurant events usually unfold.
Exemplar:
the tiny noodle shop you visited last Saturday where customers ordered at the counter.
Schema compresses across episodes.
Exemplar preserves one episode or instance as a reference.
Both can contribute to judgement.
Exemplar Is Not Simply “Example”
A teacher can invent an example on the board.
It may never become a durable memory reference.
An exemplar in cognitive theory is more specific:
a stored or remembered instance participating in later judgement.
Every exemplar is an instance.
Not every example becomes an exemplar.
Why Keep Specific Cases?
Because summaries lose detail.
The category “bird” may be represented abstractly.
But remembering a penguin preserves a crucial unusual case.
The category “fraud” may have general indicators.
But one memorable fraud case preserves a configuration that no short checklist captures.
Exemplars retain local structure.
Similarity Is the Bridge
A new item arrives.
How similar is it to stored members of Category A?
How similar to Category B?
Exemplar models often formalise classification through summed similarity to remembered instances.
That does not mean human minds consciously calculate a numerical similarity score.
It means similarity to stored cases is a plausible computational basis for some categorisation behaviour.
Similarity Itself Depends on Attention
Two objects can be similar in colour and different in shape.
Which dimension matters?
Category learning includes learning where to place attention.
A new mushroom resembles an edible species in colour but a poisonous species in gill structure.
Similarity is useless unless the relevant features are weighted correctly.
This is why exemplar reasoning is not merely “looks like.”
It depends on learned feature relevance.
The Exemplar–Prototype Debate
Cognitive psychology has debated whether categories are represented mainly through prototypes, exemplars, rules or combinations of systems.
The debate has produced strong models on several sides.
A review titled Is There an Exemplar Theory of Concepts? argues that exemplars are clearly important in some categorisation judgements and category-learning experiments but cautions against treating exemplar theory as a complete theory of all human concepts.
Likewise, research comparing prototype and exemplar models has produced mixed results depending on task and modelling assumptions.
Cognitive Art therefore does not declare one winner.
Specific remembered instances are one important resource for human judgement; they are not the whole architecture of concepts.
Multiple Systems May Cooperate
The classic Annual Review of Human Category Learning distinguishes several kinds of category-learning tasks and argues that human category learning involves qualitatively different systems.
More recent reviews continue to debate exactly how many systems are needed and how they interact.
The practical consequence is simple.
Do not expect one teaching method to build every kind of category knowledge.
Exemplars Preserve Exceptions Better Than Summaries
Prototype:
birds usually fly.
Exemplar memory:
penguin.
The unusual instance protects against turning typicality into definition.
Exemplars can therefore act as anti-overgeneralisation anchors.
The Nearest Counterexample Is Often an Exemplar
Rule:
all members look like this.
Remembered case:
except that one.
The exemplar does not automatically disprove a probabilistic tendency.
But it reminds cognition that the distribution has edges.
Frequency Matters
You have encountered one strange case once.
You have encountered another case fifty times.
The memory landscape is not equal.
Repeated exemplars can gain retrieval strength or influence category judgement more strongly.
Erickson and Kruschke’s work on rules and exemplars found that instance frequency affected generalisation and could be captured by models combining rule and exemplar representations.
Recency Can Matter Too
One recent failure feels vivid.
Ten older successes become quiet.
The recent exemplar may dominate judgment even when it is statistically unrepresentative.
This is where exemplar reasoning can collide with availability and salience.
Memorable is not the same as representative.
The Famous Case Problem
One dramatic accident becomes famous.
People estimate the entire technology through that case.
One exceptional student becomes the model for what “works for everyone.”
One business success becomes a universal entrepreneurship lesson.
The exemplar is real.
The inference from exemplar to population may be wrong.
Exemplar Is Evidence of One Case
A specific case tells you:
this happened at least here, under these conditions.
It does not automatically tell you:
this is typical, causal or general.
This is a crucial epistemic boundary.
Case-based reasoning is powerful when the new case genuinely shares relevant structure.
It becomes anecdotal overreach when one story substitutes for population evidence.
Exemplars and Similarity
Similarity is multidimensional.
Two students both scored 45%.
Superficially similar.
Student A:
- strong concepts,
- slow execution.
Student B:
- fast execution,
- weak concepts.
Same score.
Different causal structure.
Exemplar matching needs relevant similarity, not surface resemblance.
The Similarity Audit
When a new case “reminds you of” an old one, ask:
- Which features are shared?
- Which are merely vivid?
- Which mechanism is shared?
- Which constraint differs?
- Is the operating regime the same?
This turns anecdotal resemblance into disciplined comparison.
Exemplar and Comparison
The Cognitive Art article on comparison owns alignment between cases.
Exemplar supplies one of the cases that can be aligned.
Remembered instance:
last year’s similar problem.
Current instance:
today’s problem.
Comparison asks which relations match.
Exemplar and Abstraction
Several exemplars can be compared.
Repeated structure emerges.
That structure can support abstraction.
But exemplar theories demonstrate why abstraction is not always necessary for classification.
A new case can sometimes be judged by its relation to stored instances directly.
Exemplar and Prototype
Suppose you know five dogs.
Prototype-like strategy:
compare the new animal with an abstract central tendency of those dogs.
Exemplar-like strategy:
compare the new animal with the remembered dogs individually or collectively.
Behaviour can often be fitted by more than one model.
This is why theoretical humility matters.
Exemplar and Rule
Rules are excellent when category boundaries can be stated cleanly.
Exemplars are valuable when local detail matters or boundaries are irregular.
A robust learner can carry both:
the rule + the memorable cases that test its edges.
Exemplar and Counterexample
A counterexample has a logical job:
one valid case violates a universal claim.
An exemplar has a memory-and-judgement job:
one remembered instance becomes a reference for a later case.
A case can be both.
But the concepts are not synonyms.
Exemplar and Schema
Schemas compress across many episodes.
Exemplars preserve local detail.
Strong cognition can move between them.
Use schema for broad expectation.
Use exemplar when the current case resembles a specific known configuration strongly enough that detail matters.
Exemplars in Mathematics
A student meets an unfamiliar algebra question.
It resembles a worked problem from last week.
The remembered solution becomes an exemplar.
This can help:
- identify structure,
- select a method,
- anticipate intermediate steps.
It can also hurt if the learner copies the surface method into a problem where one important condition changed.
The Worked-Example Trap
New question looks familiar.
Student remembers:
last time we differentiated first.
They differentiate immediately.
But the new question belongs to a different structural family.
Exemplar retrieval outran rule selection.
Good teaching asks:
What makes this case like the old one—and what makes it not like the old one?
Exemplars in English
A learner wants to write a strong introduction.
They remember one excellent introduction.
That exemplar can reveal:
- tone,
- pacing,
- claim placement,
- reader orientation.
But copying the exact sentences produces imitation rather than transferable writing.
Use exemplar to extract structure.
Do not turn exemplar into script.
Exemplars in Reading
A reader encounters an unreliable narrator.
A remembered earlier novel becomes a comparison case.
The exemplar can help detect:
- contradictions,
- self-deception,
- selective reporting.
But literary similarity is not identity.
The new text must remain primary evidence.
Exemplars in Science
Case reports can reveal phenomena that broad models overlooked.
One unusual organism.
One rare failure.
One unexpected experimental result.
A specific case can generate a new hypothesis.
But scientific generalisation requires more than one compelling exemplar.
Case can open inquiry.
Population evidence establishes reach.
Exemplars in Medicine
Experienced clinicians often recognise patterns partly through memories of prior cases.
“This reminds me of…” can be diagnostically valuable.
It can also create premature closure if one memorable case dominates objective evidence.
Clinical case comparison should therefore remain subordinate to current measurements, differential diagnosis and professional standards.
Exemplars in Law
Legal systems often reason from prior cases through precedent and analogy.
This is not identical to psychological exemplar theory.
But the structural resemblance is useful.
Specific prior cases guide judgement when legally relevant similarities hold.
The discipline lies in identifying which similarities are legally material rather than merely vivid.
Exemplars in Organisations
“Last time we launched this way, the handoff failed.”
Useful exemplar.
But ask:
- Was the scale the same?
- Was the team the same?
- Was the technology the same?
- Was the market regime the same?
Organisational memory improves when cases carry context instead of becoming slogans.
Case Libraries
Experts often accumulate internal or external case libraries.
Successful cases.
Failed cases.
Borderline cases.
Rare exceptions.
Good case libraries preserve enough context to make later comparison meaningful.
The Context-Stripped Exemplar
“Apple succeeded by…”
“Singapore succeeded by…”
“This student improved because…”
The case gets compressed into one lesson.
Then exported to another context.
But the omitted conditions may have caused the original result.
Exemplar reasoning fails when the case loses its regime, constraints and causal structure.
Exemplar Detail Has a Resolution Problem
Remember too little:
the case becomes a slogan.
Remember too much:
comparison becomes overwhelmed by irrelevant detail.
Useful exemplars preserve the features most likely to matter for later classification or action.
Exemplar Selection Matters
Teach only central examples.
Learner acquires a narrow category.
Add boundary cases.
Category becomes more robust.
Add contrasting non-members.
Decision boundary becomes clearer.
Curriculum design is partly exemplar design.
The Typical-Only Curriculum
Every triangle is upright.
Every bird flies.
Every algebra question uses familiar notation.
The learner performs beautifully in the training distribution.
Then the first rotated triangle or flightless bird feels like a different category.
Exemplar diversity is a robustness tool.
Nearest Exemplars Are Often Powerful Teachers
To teach square versus rectangle, compare:
one square + one rectangle that differs by only one critical relation.
Far-apart examples teach category centres.
Near neighbours teach boundaries.
Exemplars and Transfer
One remembered example can seed transfer.
But transfer fails if the learner copies surface features rather than relational structure.
Strong transfer asks:
- What is shared?
- What changed?
- Which invariant matters?
- Which old action no longer applies?
Exemplars and Expertise
Experts have seen more cases.
But raw case count is not enough.
Expert advantage comes from:
- better feature weighting,
- better indexing,
- better comparison,
- better knowledge of exceptions,
- better integration with rules and schemas.
A novice remembers stories.
An expert remembers discriminating structure inside stories.
Expertise Can Also Overweight Familiar Cases
“I have seen this before.”
Maybe.
But one new variable changed.
Regime changed.
Causal structure changed.
The familiar exemplar creates premature closure.
Expert case libraries need active mismatch detection.
Exemplar and Availability
Some exemplars are easy to retrieve because they are:
- recent,
- dramatic,
- emotionally intense,
- frequently discussed.
This can give them disproportionate influence.
A case may dominate because it is retrievable, not because it is representative.
The fix is not to ignore exemplars.
It is to ask how the remembered case sits inside the larger distribution.
Exemplar and Evidence
A specific case can be:
- direct evidence that something can occur,
- a source of mechanism clues,
- a hypothesis generator,
- a stress test for a universal rule.
It is weaker evidence for:
- frequency,
- average effect,
- population prevalence,
- general causal strength.
Evidence weight depends on the claim.
Exemplar and Uncertainty
The new case partly resembles three old cases.
None is a perfect match.
Good exemplar reasoning should preserve uncertainty.
“Closest case” is not “same case.”
Similarity creates evidence.
Difference creates uncertainty.
The Exemplar Audit
- Which specific remembered case is influencing this judgement?
- Why was that case retrieved?
- Which features are genuinely shared with the current case?
- Which features are merely vivid?
- Is the old case typical, rare or unknown in frequency?
- Is the operating regime comparable?
- What rule or schema connects the cases?
- What nearest contrasting exemplar should also be considered?
- Am I using one case to make a population claim?
- What evidence would show that this analogy is misleading?
A Practical Exercise: Build a Three-Case Set
For one concept, select:
- one typical exemplar,
- one boundary exemplar,
- one non-member that looks deceptively similar.
Ask what feature actually controls the classification.
This trains both category centre and boundary.
A Practical Exercise: Case Without Context
Take one famous success story.
List the contextual variables often omitted when people retell it.
- time,
- market,
- resources,
- selection,
- competition,
- luck.
Now ask which of those variables must match before the exemplar should guide a new decision.
A Practical Exercise: Retrieve the Opposite Case
When one exemplar strongly supports your conclusion, deliberately retrieve one case with the opposite outcome.
Compare the pair.
What feature discriminates them?
This prevents one memorable instance from monopolising the model.
A Practical Exercise: From Exemplar to Rule
Take three successful examples.
Extract one proposed invariant.
Write a candidate rule.
Then search for a counterexample.
You have moved from case memory toward abstraction while preserving a path back to evidence.
A Primary-to-Adult Progression in Exemplar Thinking
Primary: learn through varied concrete cases
Children build categories from examples and benefit when typical cases are paired with unusual members and clear non-members.
Lower secondary: compare examples by relevant structure
Students learn to ask why a new problem resembles a worked example instead of copying procedures from visual familiarity.
Upper secondary: separate case evidence from population evidence
Learners distinguish anecdote, counterexample, representative case and statistical pattern while building richer case libraries.
Adulthood: use cases as probes, not dictators
Professional expertise draws on accumulated cases while checking causal structure, regime, base rates and current evidence before transferring conclusions.
Five Exemplar Failures
1. Famous-Case Capture
A dramatic remembered case receives more weight than its representativeness warrants.
2. Surface-Similarity Transfer
A new case looks like the old one but differs in the causal or structural feature that matters.
3. Typical-Only Library
Only central examples are stored or taught, leaving category boundaries fragile.
4. Case-to-Population Leap
One valid instance becomes a claim about what usually happens.
5. Context-Stripped Analogy
The memorable lesson survives while the conditions that made the original case work disappear.
Frequently Asked Questions
What is an exemplar in psychology?
It is a specific remembered instance that can serve as a reference when judging or categorising a new item.
What is exemplar theory?
It is a family of categorisation models in which category judgements depend on similarity between a new item and stored category instances rather than solely on an abstract prototype or explicit defining rule.
Is exemplar theory proven to explain all concepts?
No. Exemplars clearly contribute to some categorisation and learning tasks, but researchers debate their scope and how exemplar processes interact with prototype, rule and other representations.
What is the difference between exemplar and prototype?
An exemplar is a specific stored category member. A prototype is a central or summary-like representation capturing what is typical of a category.
Why are exemplars useful in teaching?
Concrete cases make abstract structure visible. A varied set containing typical, boundary and contrasting cases helps learners discover which features actually determine classification or method selection.
Why can exemplars mislead?
Memorable cases may be unusual, surface similarity can hide structural differences, and one instance cannot establish population frequency or general causal strength.
Research Notes and Further Reading
For a concise definition of exemplar theory, see the APA Dictionary of Psychology.
For the limits of treating exemplar theory as a complete theory of concepts, see Murphy, Is There an Exemplar Theory of Concepts?.
For evidence that rule and exemplar representations can interact in category learning, see Erickson and Kruschke, Rules and Exemplars in Category Learning. For a broad multi-system framework, see Ashby and Maddox, Human Category Learning.
Comparisons between exemplar and prototype accounts remain theoretically contested. Cognitive Art therefore treats exemplar as one important reader-facing cognitive resource—specific remembered cases guiding new judgement—without claiming that all human concepts are stored as instance lists.
Final Thought: Sometimes Intelligence Remembers the Rule; Sometimes It Remembers the Case
The new bird lands.
You could ask for a definition.
You could compare it with a prototype.
Or a remembered bird quietly returns.
Same beak.
Similar movement.
Different tail.
That remembered instance becomes a bridge from the unfamiliar to the known.
This is the gift of exemplars.
They preserve the grain that abstraction removes.
But one case must never become the whole world.
Use the remembered case to ask a better question of the present case—not to answer it before the present evidence has spoken.