Categorization
Sorting the endless particulars of experience into a manageable set of kinds.
What it means
Categorization is the cognitive process of grouping distinct entities as instances of the same kind, enabling inference, communication, and economical use of memory by letting one treat novel members like known ones. Classical theory took categories to be defined by necessary and sufficient features, but evidence of graded membership, unclear boundaries, and typicality effects showed most natural concepts are not so crisp. Modern accounts model categories around prototypes, stored exemplars, theory-based knowledge, or some blend, each capturing different aspects of how people learn and use concepts. Categorization is what allows induction — predicting unseen properties of a new member — but the same gist-extraction underlies stereotyping and the loss of individuating detail. It is the substrate on which schemas, language, and much of higher cognition are built.
The classical theory and why it fell
For most of the twentieth century a category was assumed to be a mental definition: a list of features every member had and non-members lacked. The trouble is that natural concepts rarely submit to one. No single property is shared by all games, all chairs, or all birds; ask where fruit ends and vegetable begins and agreement dissolves. Rosch and Mervis showed instead that members share overlapping family resemblances, with the most typical items being those whose features recur across the category and seldom outside it. Typicality, not definition, predicts how quickly people verify membership and which examples they volunteer first.
Prototypes versus exemplars
Two accounts dominate the modern picture. Prototype theory says we abstract a summary representation, an average or ideal member, and judge new items by their resemblance to it. Exemplar theory, formalized in Nosofsky's generalized context model, says we store individual instances and classify a new item by its summed similarity to all remembered examples, with attention weighting the dimensions that matter. The two often predict the same behavior, but they diverge on telling cases: exemplar models capture sensitivity to specific old items and to the variability within a category, information a single prototype discards. Most researchers now treat both as partly right, sometimes running together.
When similarity is not enough
Similarity alone cannot fix what belongs together, because any two things share indefinitely many features and something must decide which ones count. Murphy and Medin argued that background knowledge and naive theories do that work: watching a man leap fully clothed into a pool, we file him as drunk on the strength of an explanation rather than any perceptual match, because a causal story binds the category. The same reasoning explains psychological essentialism, the intuition that category members share a hidden inner nature, the birdness that keeps a plucked, painted robin a bird. On this view coherence comes from explanation, not from counting shared attributes.
One system, or several?
Do all categories form the same way? Ashby and Maddox's influential account holds that a verbal, rule-testing system and a slower procedural system compete, and that the two are dissociable in the brain and in patients. But the evidence is contested. Newell, Dunn and Kalish argued that much of the best evidence for two systems can be reproduced by a single flexible mechanism, and that the reported dissociations are weaker than claimed. A 2024 review by Minda and colleagues reaches a similarly cautious verdict. The honest summary is that categorization is plural in behavior, while the count of underlying processes remains genuinely open.
The cost of extracting a gist
The efficiency has a price. Because categorizing keeps the kind and discards the token, it can manufacture memories that were never encoded: after studying bed, rest and tired, people confidently recall the unpresented word sleep. The same abstraction feeds out-group homogeneity, the sense that they are all alike, because members of unfamiliar groups are stored as category-level gist rather than as individuals. None of this is a malfunction; it is the ordinary operation of a system built to generalize. Knowing exactly where the individuating detail drops out is what lets a careful reader, recruiter, or clinician deliberately put it back.
Examples
On seeing an unfamiliar animal with feathers and a beak, a person instantly categorizes it as a bird and infers it probably lays eggs and can likely fly.
You meet an unfamiliar fruit in a market, file it as citrus, and instantly expect sour segments that would work in a drink — none of which anyone actually told you.
A recruiter skims a CV and files the candidate as 'agency background.' The gist arrives fast and free; the individuating details that would complicate it quietly drop out.
A triage nurse hears 'crushing chest pain spreading to the left arm,' files it as a likely cardiac event, and orders an ECG before any test confirms it.
A sommelier tastes a blind pour, files it as cool-climate Pinot Noir, and predicts its acidity, light body, and rough price bracket, none of which the glass announced.
First described in Rosch (1970s); Bruner, Goodnow & Austin (1956).
Key references
- Minda, J. P., Roark, C. L., Kalra, P. B., & Cruz, A. (2024). Single and multiple systems in categorization and category learning. Nature Reviews Psychology, 3, 536-551. doi.org/10.1038/s44159-024-00336-7
- Newell, B. R., Dunn, J. C., & Kalish, M. (2011). Systems of category learning: Fact or fantasy? Psychology of Learning and Motivation, 54, 167-215. doi.org/10.1016/B978-0-12-385527-5.00006-1
- Ashby, F. G., & Maddox, W. T. (2005). Human category learning. Annual Review of Psychology, 56, 149-178. doi.org/10.1146/annurev.psych.56.091103.070217
- Murphy, G. L., & Medin, D. L. (1985). The role of theories in conceptual coherence. Psychological Review, 92(3), 289-316. doi.org/10.1037/0033-295X.92.3.289
- Nosofsky, R. M. (1986). Attention, similarity, and the identification-categorization relationship. Journal of Experimental Psychology: General, 115(1), 39-57. doi.org/10.1037/0096-3445.115.1.39
- Rosch, E., & Mervis, C. B. (1975). Family resemblances: Studies in the internal structure of categories. Cognitive Psychology, 7(4), 573-605. doi.org/10.1016/0010-0285(75)90024-9