ExemplarTheorie

ExemplarTheorie is a cognitive theory of categorization and recognition that holds that knowledge of categories consists of memories for individual encountered examples, or exemplars, rather than a single abstract prototype. When judging a new item, the observer compares it to stored exemplars and assigns it to the category whose exemplars it most closely resembles. The strength of the match can depend on feature similarity, typicality, and the weight given to different features.

Origin and development: ExemplarTheorie emerged in the 1970s and 1980s as an alternative to prototype theory in categorization research. Influential work by researchers such as David Nosofsky introduced exemplar-based models, including the Generalized Context Model, and later refinements allowed attention weights and memory decay over time.

Mechanisms and predictions: The model stores many exemplars, each labeled with its category. Classification proceeds by computing similarity between the new item and each exemplar, then summing similarities within each category to choose the category with the highest total similarity. This framework naturally produces typicality effects, where some exemplars are more influential than others, and it can account for context effects where current experience alters judgments.

Evidence and criticisms: Proponents report good fits to data in perceptual categorization and memory tasks, particularly for abstract or nuanced categories. Critics point to potential memory demands and computational complexity, and some data are better explained by prototype models or hybrid approaches. The theory has influenced areas such as case-based reasoning in artificial intelligence, where new cases are classified by similarity to stored cases.

Related concepts include prototype theory, case-based reasoning, and kernel-based learning.