Prototype and Exemplar-Based Information in Natural Language Categories

Abstract

Two experiments are reported in which four dependent variables; typicality ratings, response times, category-naming frequencies, and exemplar-generation frequencies of natural language concepts, were predicted by two sorts of prototype predictors and by an exemplar predictor related to Heit and Barsalou's (1996) instantiation principle. In the first experiment, the exemplar predictor was compared to a prototype predictor calculated as in Hampton (1979). The four dependent variables were either predicted better by the exemplar measure than by the prototype predictor or the predictive value was about equal. In the second experiment, a new prototype predictor was calculated based on Rosch and Mervis ' (1975) classic family resemblance measure. The results showed that the exemplar predictor accounted better for the dependent variables than Hampton's and Rosch and Mervis' prototype measures. The differences between the prototype measures were not significant.

Uncovering mental representations with Markov chain Monte Carlo

“…Cognitive psychologists have extensively studied the ways in which people group objects into categories, investigating issues such as whether people represent categories in terms of exemplars or prototypes (J. D. Smith & Minda, 1998;Nosofsky & Zaki, 2002;Minda & Smith, 2002;Storms et al, 2000;Nosofsky, 1988;Nosofsky & Johansen, 2000;Zaki et al, 2003;E. E. Smith et al, 1998;Heit & Barsalou, 1996).…”

“…In categorization studies, this goal is generally accomplished by asking participants for typicality ratings of stimuli (e.g., Storms et al, 2000;Bourne, 1982). This procedure was originally used to justify the existence of a category prototype (Rosch & Mervis, 1975;Rosch, Simpson, & Miller, 1976), but typicality ratings have also been used to support exemplar models (Nosofsky, 1988).…”

“…The ratings can be typicality judgments of single objects (e.g., Storms et al, 2000) or a pairwise judgment of which object is more typical (e.g., Bourne, 1982). The pairwise judgment is exactly the task used in the MCMC method.…”

“…Categorization research has benefited from intensive investigation using a large number of mathematical models. These models have been developed to fit the data from a variety of tasks including binary classifications (e.g., Nosofsky et al, 1994;Ashby & Gott, 1988;Nosofsky, 1987), typicality judgments (e.g., Storms et al, 2000;Nosofsky, 1988), or stimulus reconstructions (e.g., Huttenlocher et al, 2000;Vevea, 2006). In order to produce predictions, the models need to be adapted for each task in which the researcher is interested.…”

“…Most of these training experiments use simple stimuli with only a few dimensions, such as binary dimensional shapes , lengths of a line (Huttenlocher, Hedges, & Vevea, 2000), or Munsell colors (Nosofsky, 1987;Roberson, Davies, & Davidoff, 2000), though more complicated stimuli have also been used (e.g., Posner & Keele, 1968). Research on natural perceptual categories has focused on exploring the relationships of category labels (Storms, Boeck, & Ruts, 2000;Rosch & Mervis, 1975;Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976), with a few studies examining members of a perceptual category using simple parameterized stimuli (e.g., Labov, 1973).…”