Concept learning is the process by which we acquire categories — the mental groupings that allow us to treat different objects, events, or ideas as equivalent for some purpose. Concepts are fundamental to cognition: they allow us to go beyond individual experiences to make inferences, communicate efficiently, and organize knowledge. A toddler learning "dog" must somehow extract what is common across diverse instances (from Chihuahuas to Great Danes) while excluding similar but different categories (cats, wolves).
Classical View: Defining Features
The classical view held that concepts are defined by necessary and sufficient features: a "bachelor" is defined as "unmarried adult male." Every bachelor has all three features, and every entity with all three features is a bachelor. While this works for some formal concepts, it fails for most natural categories. What are the defining features of "game," "furniture," or "bird"? Wittgenstein noted that the members of many categories share only "family resemblances" rather than common defining features.
Prototype Theory
Eleanor Rosch's prototype theory proposed that categories are organized around prototypes — the most typical, most representative members. Category membership is graded: a robin is a better example of "bird" than a penguin. Typicality effects are pervasive — typical members are verified faster, listed more frequently, learned first by children, and more likely to serve as cognitive reference points. The prototype is the central tendency of the category, and new items are categorized by their similarity to it.
Exemplar models (Medin and Schaffer, 1978; Nosofsky, 1986) propose that categories are represented not by an abstract summary (prototype) but by stored memories of individual encountered examples. A new item is categorized by comparing it to all stored exemplars and assigning it to the category whose exemplars it most resembles. Exemplar models can account for sensitivity to within-category variability and to the specific distribution of encountered instances — effects that prototype models struggle with.
Rule-Based and Theory-Based Accounts
Some concepts appear to be learned through explicit rules (e.g., "prime numbers are divisible only by 1 and themselves"). Theory-based accounts propose that concepts are embedded in naive theories about how the world works: we categorize whales as mammals not because they look like other mammals but because we understand their biological properties. These approaches highlight that categorization involves more than surface similarity — it involves understanding the causal and explanatory structure of categories.