Questions: Prototypes and Exemplars in Category Learning

Question 1

A novice medical student describes a diagnosis as 'this case looks like what lung cancer usually looks like.' An expert radiologist says 'this reminds me of patient 47 from 2018.' This difference most likely reflects:

• A. The expert using prototype theory more efficiently because experience refines the prototype
• B. The novice using exemplar theory because they have fewer stored instances to compare
• C. A shift from prototype-based to exemplar-based classification as expertise develops in domains requiring sensitivity to specific prior cases
• D. Both using the same cognitive mechanism, with the expert simply having a more accurate prototype

Answer: (C)

Expert classification in domains like radiology is well-documented to rely heavily on specific remembered cases rather than abstract central tendencies. The expert's retrieval of a specific patient is exemplar-based classification. Novices, lacking stored exemplars, rely more on prototypes (general feature summaries). Options A and D fail to capture this distinction; option B reverses the pattern — novices default to prototypes precisely because they lack enough exemplars.

Question 2

Which empirical finding most challenges a pure prototype account of categorization but is straightforwardly explained by exemplar theory?

• A. People verify 'a robin is a bird' faster than 'a penguin is a bird'
• B. People rarely list ostriches when asked to name a bird
• C. People correctly classify a very atypical category member they have personally encountered before
• D. Natural categories tend to have family resemblance structure rather than necessary and sufficient features

Answer: (C)

Exemplar theory's key advantage is explaining sensitivity to atypical members that someone has actually seen. If you've encountered a pet penguin, you have a stored exemplar and can classify it correctly even though it doesn't match the bird prototype. Prototype theory predicts classification by similarity to the central tendency — an atypical penguin should be misclassified or slow. Options A, B, and D are actually *predicted* by prototype theory; they are the typicality effects that Rosch's work documented.

Question 3

Exemplar theory predicts that category classification accuracy should continue to improve with more training examples, even for rare atypical members.

• T. True
• F. False

Answer: True

This is a distinctive prediction of exemplar theory: since classification draws on stored memories of specific instances, more stored exemplars improve accuracy — especially for atypical members, whose unusual features are preserved in exemplar storage rather than averaged away in a prototype. Prototype theory predicts that once the central tendency is learned, additional typical instances yield diminishing returns and atypical members remain hard to classify.

Question 4

Prototype theory holds that nearly every member of a category should share the defining features of the prototype.

• T. True
• F. False

Answer: False

This is a fundamental misunderstanding of prototype theory. Rosch explicitly rejected the classical 'necessary and sufficient features' view. In prototype theory, category membership is graded — determined by *degree of similarity* to the prototype — and no single feature is necessary for membership. That's why penguins count as birds despite lacking wings adapted for flight: they share enough other features. Prototype theory describes categories as having fuzzy boundaries and family resemblance structure, not strict definitional boundaries.

Question 5

Why do experts in fields like medicine or law often show better sensitivity to unusual or atypical cases than novices, even though novices sometimes have more recently studied formal definitions and rules?

Model answer: Experts have accumulated large libraries of stored exemplars — memories of specific cases — that allow them to recognize atypical presentations by similarity to a specific prior case, even when the presentation doesn't match a prototype or rule. Exemplar-based classification preserves the co-occurrence of unusual features and the variability within categories, which prototype abstraction smooths over. Novices relying on prototypes or rules will systematically underperform on atypical cases that fall outside the central tendency.

The key insight is that exemplar storage preserves information about variability that prototype abstraction discards. A radiologist remembers not just 'typical lung cancer' but specific unusual presentations — and can match a new unusual case to a remembered one. This is why deliberate practice with varied cases, not just formal instruction, builds expert classification skill.