A unified quantitative approach to modeling subjects' identification and categorization of multidimensional perceptual stimuli is proposed and tested. Two subjects identified and categorized the same set of perceptually confusable stimuli varying on separable dimensions. The identification data were modeled using Sbepard's (1957) multidimensional scaling-choice framework. This framework was then extended to model the subjects' categorization performance. The categorization model, which generalizes the context theory of classification developed by Medin and Schaffer (1978), assumes that subjects store category exemplars in memory. Classification decisions are based on the similarity of stimuli to the stored exemplars. It is assumed that the same multidimensional perceptual representation underlies performance in both the identification and Categorization paradigms. However, because of the influence of selective attention, similarity relationships change systematically across the two paradigrns. Some support was gained for the hypothesis that subjects distribute attention among component dimensions so as to optimize categorization performance. Evidence was also obtained that subjects may have augmented their category representations with inferred exemplars. Implications of the results for theories of multidimensional scaling and categorization are discussed. Bruner, Goodnow, and Austin marveled at the capacity of people to discriminate stimuli and to identify them as unique items. At the same time they stressed the importance of categorization, the process by which discriminably different things are classified into groups and are thereby rendered equivalent. In one sense the processes of identification and categorization seem diametrically opposed, the former dealing with the particular and the latter with the general. Yet similar principles may underlie subjects' identification and categorization of multidimensional stimuli, and performance in these tasks may be highly related. Indeed, the present research renews the issue explored previously by Shepard, Hovland, and Jenkins (1961) and Shepard and Chang (1963)namely, Do the principles of stimulus generalization underlying identification performance also underlie categorization performance? Furthermore, given knowledge of performance in an identification paradigm, can one predict performance in a categorization paradigm using the same set of stimuli? In their 1956 classic, A Study of Thinking,This article is based on a PhD dissertation submitted to Harvard University.

The authors propose and test an exemplar-based random walk model for predicting response times in tasks of speeded, multidimensional perceptual classification. The model combines elements of R.M. Nosofsky's (1986) generalized context model of categorization and G. D. Logan's (1988) instance-based model of automaticity. In the model, exemplars race among one another to be retrieved from memory, with rates determined by their similarity to test items. The retrieved exemplars provide incremental information that enters into a random walk process for making classification decisions. The model predicts correctly effects of within-and between-categories similarity, individual-object familiarity, and extended practice on classification response times. It also builds bridges between the domains of categorization and automaticity.Models of multidimensional perceptual classification have grown increasingly powerful and sophisticated in recent years, providing detailed quantitative accounts of patterns of classification learning, transfer, and generalization (e.g., Anderson, 1991;Ashby, 1992;Estes, 1986Estes, , 1994Nosofsky, 1992b;Shanks & Gluck, 1994). However, a fundamental limitation of all the major competing models in the field today is that they offer no processing account of the time course of classification. Because response times provide a window into understanding the nature of cognitive representations and decision processes, it is vital to move in the direction of models that account for this form of data. In this article we propose and test a process-oriented model for predicting response times in tasks of speeded perceptual classification.Our proposed model follows in the spirit of some leading extant models of categorization by assuming that people represent categories in terms of stored exemplars (Hintzman, 1986;Medin & Schaffer, 1978;Nosofsky, 1986). Classification decisions are made by retrieving these stored exemplars from memory. In the newly proposed model, retrieved exemplars are used to drive a random walk process (e.g., Luce, 1986;Townsend & Ashby, 1983) in which evidence accrues to alternative categories over time. Random-walk models have been successful at accounting for performance in tasks of memory, decision making, sensory discrimination, and unidimensional absolute judgment (e.g., Busemeyer, 1985; Karpiuk, Lacouture, & Marley, in press; This work was supported by Grant PHS RO1 MH48494-05 from the National Institute of Mental Health.Jerome Busemeyer, A. A. J. Marley, Richard Shiffrin, and James Townsend provided extensive commentary and discussion. We also wish to thank John Anderson, Rob Goldstone, John Kruschke, Roger Ratcliff, Roger Shepard, and Trisha van Zandt for their comments, discussion, and advice.Correspondence concerning this article should be addressed to Robert M. Nosofsky, Department of Psychology, Indiana University, Bloomington, Indiana 47405; or to Thomas J. Palmeri, Department of Psychology, Vanderbilt University, Nashville, Tennessee 37240. Electronic mail may be sent vi...

Medin and Schaffer's (1978) context theory of classification learning is interpreted in terms of Luce's (1963) choice theory and in terms of theoretical results obtained in multidimensional scaling theory. En route to this interpretation, quantitative relationships that may exist between identification and classification performance are investigated. It is suggested that the same basic choice processes may operate in the two paradigms but that the similarity parameters that determine performance change systematically according to the structure of the choice paradigm. In particular, when subjects are able to attend selectively to the component dimensions that compose the stimuli, the similarity parameters may tend toward what is optimal for maximizing performance.

No abstract

The relationship between subjects' identification and categorization learning of integral-dimension stimuli was studied within the framework of an exemplar-based generalization model. The model was used to predict subjects' learning in six different categorization conditions on the basis of data obtained in a single identification learning condition. A crucial assumption in the model is that because of selective attention to component dimensions, similarity relations may change in systematic ways across different experimental contexts. The theoretical analysis provided evidence that, at least under unspeeded conditions, selective attention may play a critical role in determining the identification-categorization relationship for integral stimuli. Evidence was also provided that similarity among exemplars decreased as a function of identification learning. Various alternative classification models, including prototype, multiple-prototype, average distance, and "value-on-dimensions" models, were unable to account for the results.This article seeks to characterize performance relations between the two fundamental classification paradigms of identification and categorization. Whereas in an identification paradigm people identify stimuli as unique items (a one-to-one stimulus-response mapping), in a categorization paradigm people classify items into groups (a many-to-one stimulus-response mapping). The present study of the identification-categorization relationship is motivated by the recent "exemplar view" of categorization proposed by investigators such as Brooks (1978), Hintzman and Ludlam (1980), and Medin and Schaffer (1978). According to the exemplar view, people represent categories by storing individual category exemplars in memory. Classification decisions are based on the similarity of stimuli to the stored exemplars. This view contrasts with some other approaches that assume that people form category "summary" representations such as a prototype or a rule.A suggestion that follows from the exemplar view is that there may be highly regular and systematic relations between identification and categorization performance. Presumably, when subjects learn to identify stimuli, a unique representation of each stimulus is stored in memory. Furthermore, the extent to which individual stimuli are confused during identification is This article is b~tsed on portions of a PhD dissertation submitted to Harvard University and on subsequent work conducted at Indiana University.The work was supported by Grants BNS 80-26656 from the National Science Foundation and MH 37208 from the National Institute of Mental Health to Harvard University, and by Grant BNS 85-19573 from the National Science Foundation to Indiana University.I would like to express my thanks to William Estes for his guidance and support, and to Jerome Busemeyer, John Flowers, Douglas Medin, and Linda Smith for their criticisms and suggestions regarding earlier versions of this article.Correspondence concerning this article should be addressed to Robert M. Nosof...