A rational model of human categorization behavior is presented that assumes that categorization reflects the derivation of optimal estimates of the probability of unseen features of objects. A Bayesian analysis is performed of what optimal estimations would be if categories formed a disjoint partitioning of the object space and if features were independently displayed within a category. This Bayesian analysis is placed within an incremental categorization algorithm. The resulting rational model accounts for effects of central tendency of categories, effects of specific instances, learning of linearly nonseparable categories, effects of category labels, extraction of basic level categories, base-rate effects, probability matching in categorization, and trial-by-trial learning functions. Al-though the rational model considers just I level of categorization, it is shown how predictions can be enhanced by considering higher and lower levels. Considering prediction at the lower, individual level allows integration of this rational analysis of categorization with the earlier rational analysis of memory (Anderson & Milson, 1989). Anderson (1990) presented a rational analysis ot 6 human cog-nition. The term rational derives from similar "rational-man" analyses in economics. Rational analyses in other fields are sometimes called adaptationist analyses. Basically, they are ef-forts to explain the behavior in some domain on the assump-tion that the behavior is optimized with respect to some criteria of adaptive importance. This article begins with a general char-acterization ofhow one develops a rational theory of a particu-lar cognitive phenomenon. Then I present the basic theory of categorization developed in Anderson (1990) and review the applications from that book. Since the writing of the book, the theory has been greatly extended and applied to many new phenomena. Most of this article describes these new develop-ments and applications. A Rational Analysis Several theorists have promoted the idea that psychologists might understand human behavior by assuming it is adapted to the environment (e.g., Brunswik, 1956; Campbell, 1974; Gib-

SUSTAIN (Supervised and Unsupervised STratified Adaptive Incremental Network) is a model of how humans learn categories from examples. SUS-TAIN initially assumes a simple category structure. If simple solutions prove inadequate and SUSTAIN is confronted with a surprising event (e.g., it is told that a bat is a mammal instead of a bird), SUSTAIN recruits an additional cluster to represent the surprising event. Newly recruited clusters are available to explain future events and can themselves evolve into

We previously introduced an exemplar model, named GCM-ISW, that exploits a highly flexible weighting scheme. Our simulations showed that it records faster learning rates and higher asymptotic accuracies on several artificial categorization tasks than models with more limited abilities to warp input spaces. This paper extends our previous work; it describes experimental results that suggest human subjects also invoke such highly flexible schemes. In particular, our model provides significantly better fits than models with less flexibility, and we hypothesize that humans selectively weight attributes depending on an item's location in the input space. We need more flexible models of concept learning Many theories of human concept learning posit that concepts are represented by prototypes (Reed, 1972) or exemplars (Medin & Schaffer, 1978). Prototype models represent concepts by the "best example" or "central tendency" of the concept. 1 A new item belongs in a category C if it is relat...

Past researchers studied prototype learning by asking subjects to categorize exemplars constructed from different prototypes. This procedure is less than ideal because learning must be inferred from the percentage of correct categorizations pooled across many trials or subjects or both. An alternative procedure is proposed in which subjects are asked to reproduce their estimate of the prototype on each trial, thereby providing trial-by-trial information about changes in the estimated prototype. This procedure provides straightforward tests of three basic properties implied by several prototype learning models: additivity across exemplars, noninterference among features, and time invariance of serial position effects. An experiment is reported and the results provide reasonably good support for the properties of additivity and noninterference, but clear violations of time invariance were observed. The implications of the results for distributedmemory models and multiple-trace models of prototype learning are discussed. It seems quite easy to produce an image of an ideal circle despite the fact that our experience is based on thousands of and reproduce a single image from a myriad of examples is

List of tables and figures............................................................................................. 4 Executive summary..................................................................................................... 5 1.0 Introduction..................................................................................................... 8 2.0 Why worry about simulation training effectiveness?........................................ 10 2.1 The frequency of use of force decisions...................................................... 10 2.2 Explaining use of force decisions............................................................... 12 2.3 Defending use of force training programs................................................... 13

Results from the classic dot pattern distortion paradigm have sometimes yielded prototype enhancement effects that could not be accounted for by exemplar models of categorization. However, in these experiments the status of the prototype was confounded with certain stimulus-specific properties as well as with the frequency of presentation of the prototype during testing. In two mock-subliminal experiments, participants made categorization judgments to patterns that were generated as prototypes, low-level distortions, or high-level distortions. The participants rated the prototypes as being more likely to be members of a category, although no patterns were presented during training, and there was no objective category structure. In two other experiments, greater prototype enhancement effects were observed when the prototype and low-level distortions were presented with greater frequency during transfer. These results suggest that classic prototype enhancement effects may not be due to the abstraction of a prototype at time of original learning, but rather to other factors not formalized in extant models.