Category Exemplars Research Articles

Perceived distributions of the characteristics of in-group and out-group members: empirical evidence and a computer simulation.

This research studied 2 properties of perceived distributions of the characteristics of social category members: the probability of differentiating (making distinctions) among category members and the perceived variability (variance) of category members. The results of 4 experiments supported the hypothesis that greater familiarity with a social group leads to greater perceived differentiation and variability regarding that group. In-group members formed more differentiated and variable distributions for groups defined by age and more differentiated distributions for groups defined by nationality. For gender (where students were roughly equally familiar with people of both genders), no in-group--out-group differences occurred. Also, students perceived greater differentiation and variability among classmates over the course of a semester. To explain these results, we developed PDIST, a multiple exemplar model that assumes that people form perceived distributions by activating a set of category exemplars and then judging the relative likelihoods of different feature values on the basis of the relative activation strengths of these feature values. The results of a computer simulation experiment indicated that PDIST is sufficient to explain the results of our 4 experiments. According to the perceived distributions formed by PDIST, increasing familiarity leads to greater differentiation and variability, has a concave impact, and has greater impact on differentiation than on variability.

Exemplar-based accounts of relations between classification, recognition, and typicality.

Previously published sets of classification and old-new recognition memory data are reanalyzed within the framework of an exemplar-based generalization model. The key assumption in the model is that, whereas classification decisions are based on the similarity of a probe to exemplars of a target category relative to exemplars of contrast categories, recognition decisions are based on overall summed similarity of a probe to all exemplars. The summed-similarity decision rule is shown to be consistent with a wide variety of recognition memory data obtained in classification learning situations and may provide a unified approach to understanding relations between categorization and recognition. Recently, there has been an upsurge of interest among categorization researchers in exploring relations between classification learning and old-new recognition memory. This interest has been fueled by the exemplar view of category representation, which holds that people base classification decisions on similarity comparisons with stored exemplars (Hintzman, 1986b; Medin & Schaffer, 1978; Nosofsky, 1986). Recognition data provide a source of converging evidence bearing on the nature of people's category representations. Presumably, if individual exemplars are being stored in memory, the fact ought to be revealed by postacquisition recognition tests. Indeed, a number of researchers have taken exemplar models to task on grounds of certain dissociations between classification learning and recognition memory, or patterns of recognition data deemed to be inconsistent with the predictions of exemplar-only memory models. In virtually all cases, however, there has been a failure to specify and test an explicit

Inferring category attributes from exemplar attributes: Geometric shapes and social categories.

Three experiments tested the hypothesis that judgments about the attributes of categories are disproportionately based on the characteristics of exemplars that best fit the category. In the first 2 experiments, subjects were presented with good and bad exemplars of categories with defining attributes (rectangles, triangles, pentagons, and ellipses) in which different colors were arbitrarily paired with the good and poor examples. In both experiments, subjects erroneously judged the colors paired with the good exemplars as more frequent than colors paired with the poor exemplars. A third experiment, using social categories, examined whether attributes associated with a single category member were more likely to generalize to the category as a whole for prototypical than for nonprototypical category members. Subjects were presented with information about individual fraternity members who varied in prototypicality, and the tendency to infer a target behavior (liberal vs. conservative voting behavior) from the individual fraternity member to the fraternity as a whole increased with the prototypicality of the category member. Implications for the contact hypothesis, category-exemplar relations, and belief stability are discussed.

Perceptual units in the acquisition of visual categories.

A series of experiments provided evidence that the representational structure of categories comprising dot patterns is based on pattern parts and pattern configuration rather than on pattern elements. We found that similarity judgments and postacquisition classification data could not be explained in terms of element-level perceptual units, even for categories of dot patterns with seven of their eight dots in the exact same relative location. The importance of higher order perceptual units was indicated by evidence that the long-term retention of information specific to previously learned category exemplars, which is typical of natural objects, can also be obtained for artificial dot patterns, providing their structure reflects the perceptual characteristics identified in Tversky and Hemenway's (1984) study of natural objects: Members of the same category had to be perceptually distinctive at the level of pattern configuration and perceptually similar at the level of pattern parts. The level of within-category similarity for a set of categories (relative to between-categories similarity) did not predict whether item-specific information would be retained; long-term retention appears to require both within-category similarity and dissimilarity, but at different levels of perceptual structure.

On exemplar-based exemplar representations: Reply to Ennis (1988).

The multivariate theory of similarity discussed by Ennis (1988) entails the assumption that individual category exemplars are themselves represented psychologically as distributions of individual exemplars. The potential utility of this exemplar-distribution approach for theories relating selective attention, similarity, and categorization is briefly discussed. Nosofsky (1986) proposed and tested a unified quantitative approach to modeling performance in tasks of multidimensional stimulus identification and categorization. The approach was based on the assumption that both identification and categorization decisions are governed by similarity comparisons with individually stored exemplars. By modeling patterns of identification confusions, a multidimensional scaling solution for the exemplars was derived. This solution was then used in conjunction with the exemplar-similarity model to predict performance in four separate categorization conditions with the same set of stimuli. A key component assumption was that selective attention processes may systematically modify distances among exemplars in the multidimensional psychological space.

Relationship of Preference Judgments to Typicality, Novelty, and Mere Exposure

Two theories of aesthetic preference are reviewed. According to Berlyne's psychobiological theory, the main determinants of aesthetic preference are collative variables such as complexity and novelty, and preference is generally related to its determinants in an inverted-U manner. Recent findings contrary to these predictions are briefly reviewed. According to Martindale's cognitive theory, the main determinant of aesthetic preference is prototypicality or meaningfulness, and preference is usually related to its determinants in a monotonic or U-shaped fashion. Three experiments concerning preference for semantic category exemplars are reported. Typicality accounted for eight or nine times more of explained variance in preference than did long-term novelty, short-term novelty, or mere exposure. Preference was related to both typicality and collative variables by monotonic or U-shaped functions.

Memorization and “feature selection” in the acquisition of natural concepts in pigeons

Pigeons were trained on a successive discrimination task using complex visual stimuli. In Experiment 1, each photographic slide that contained a person had a corresponding “matched background” slide, one that showed the same scene with the person removed. Birds trained on a human positive discrimination acquired the matched pairs problem, but birds trained on a human negative discrimination performed poorly. This suggests a feature-positive effect for complex stimulus categories. Memorization control groups that were trained on a human-irrelevant discrimination also performed poorly with matched slides. However, subsequent experiments demonstrated that these effects depended on the use of matched pairs of slides. The human-as-feature effect was not obtained when human positive and human negative groups were subsequently trained with non-matched slides (Experiment 2), and memorization control groups acquired a human-irrelevant discrimination when trained with nonmatched slides (Experiment 3). Additional tests conducted in Experiments 2 and 3 found that performance was not disrupted when either the reinforced or nonreinforced slides were replaced. This effect was obtained when the category was relevant to the discrimination (Experiment 2) and when the category was irrelevant to the discrimination (Experiment 3). Finally, Experiment 4 demonstrated that memorization of a set of slides is possible when slides are sufficiently dissimilar, (i.e., nonmatched) but performance is not as good when the category exemplars are irrelevant to the discrimination.

A ROWS is a ROSE: spelling, sound, and reading.

Skilled readers generally are assumed to make little or no use of words’ phonological features in visual word identification. Contrary to this assumption, college students’ performance in the present reading experiments showed large effects of stimulus word phonology. In Experiments 1 and 2, these subjects produced larger false positive error rates in a semantic categorization task when they responded to stimulus foils that were homophonic to category exemplars (e.g., ROWS for the category A FLOWER) than when they responded to spelling control foils. Additionally, in Experiment 2, this homophony effect was found under brief-exposure pattern-masking conditions, a result consistent with the possibility that phonology is an early source of constraint in word identification. Subjects did, however, correctly reject most homophone foils in Experiments 1 and 2. Experiment 3 investigated the source of this ability. The results of Experiment 3 suggest that subjects detected homophone impostors, such as ROWS, by verifying target foil spellings against their knowledge of the correct spellings of category exemplars, such as ROSE.

Alternative Models of Categorization: Toward a Contingent Processing Framework

Widely different accounts of how people categorize new instances have been advanced in recent years. This article reviews these alternative formulations with a particular focus on the use of concrete category exemplars (from prior experience) as an alternative to category-defining rules and prototypes. It advances a contingent processing formulation that emphasizes the flexibility of the information processing system in its response to important contextual factors, and describes empirical procedures useful in identifying categorization processes.

Effect of Prime Type on Lexical Decision Time

The present investigation concerns the issues of the control condition and type of related prime-target relationship operationalization in the lexical-decision paradigm. It is shown that the use of a row of asterisks produces strong inhibitory effects on reaction time to the target relative to a control condition formed with the word “neutro” (“neutral”). The operationalization of prime-target relatedness by means of association of category norms seems equally adequate, although category exemplars do not prime category exemplar targets. Both sets of data are discussed in relation to current research trends using lexical-decision time.

Adult age differences in the perception and learning of artistic style categories.

Adult age differences in conceptual behavior were studied using informationally complex stimuli from real-world categories: paintings by two impressionist artists. In Experiment 1 we examined perceptions of category structure by having subjects sort paintings according to style similarity. Young adults were observed to depend more on abstract information in making style judgments, whereas older adults relied more on similarity in content. This resulted in different category structures between age groups, but similarity judgments in both groups appeared to correspond to actual style differences between the two artists. In Experiment 2, learning efficiency was shown to increase with a painting's category centrality, but older adults had particular trouble learning noncentral items. At transfer, both age groups were able to use abstracted central tendency information to categorize new paintings, although young adults appeared to have better access to information about specific category exemplars from acquisition. The results are generally consistent with those from studies using simpler artificial stimuli.

Effects of speaking rate on the perceived internal structure of phonetic categories

It is well known that when listeners identify consonants as voiced or voiceless on the basis of voice onset time (VOT), they do so in a rate‐dependent manner: As rate of speech decreases such that overall syllable duration increases, the category boundary moves toward a longer VOT value. We have been examining whether this alteration in boundary location is accompanied by a change in the specific stimuli judged to be the best category exemplars. For our experiment we created two /bi/‐/pi/ series with overall syllable durations of 125 and 325 ms, respectively. VOT varied from 10–120 ms for the 125‐ms series and from 10–320 ms for the 325‐ms series. We asked listeners to judge each syllable for its goodness as a member of the /p/ category, using a scale from 1–10; the higher the number, the better the exemplar. For each series, as VOT increased there was an initial increase in rating response, followed by a subsequent decline. Thus for both the short and long syllables, the /p/ category was perceived as having internal structure, with a limited range of stimuli serving as the best exemplars. However, the entire function for the longer syllables was shifted toward longer VOT values. That is, as syllable duration increased, so too did the VOT values of the stimuli judged to be the best category members. Moreover, in a subsequent control experiment we established that the shift in rating funtion was not due to range or frequency effects, but to the alteration in syllable duration, per se. These findings strongly suggest that the adjustment for speaking rate, whatever the underlying mechanism, involves a comprehensive modification in the mapping between acoustical signal and phonetic structure.

Array models for category learning

A family of models for category learning is developed, all members being based on a common memory array but differing in memory access and decision processes. Within this framework, fully controlled comparisons of exemplar-similarity, feature-frequency, and prototype models reveal isomorphism between models of different types under some conditions but empirically testable differences under others. It is shown that current exemplar-memory models, in which categorization judgments are based on similarities of perceived and remembered category exemplars, can be interpreted as generalized likelihood models but can be modified in a simple way to yield pure similarity models. Distance-based exemplar models are formulated that provide means of investigating issues concerning deterministic versus probabilistic decision rules and links between categorization and properties of perceptual dimensions. Other theoretical issues discussed include aspects of similarity, the role of memory storage versus computation in category judgments, and the limits of applicability of array models.

Classification of empirically derived prototypes as a function of category experience.

In three experiments, subjects learned to classify dot patterns into three categories represented by either three, six, or nine exemplars. Following learning, subjects were tested on an additional set of patterns, which included patterns from the learning phase, the category objective prototypes, and new distortions of the objective prototypes. Also included in the test set were empirical prototypes and distortions of empirical prototypes. Empirical prototypes were derived by averaging feature values of category exemplars in the learning phase. The overall results revealed that empirical prototypes were classified more accurately than were objective prototypes. In addition, a pattern of convergence in error rates was observed for distortions of the objective and empirical prototypes as category set size increased, but this same pattern was not observed for the objective and empirical prototypes themselves. This lack of convergence for the prototypes is inconsistent with explanations of the category set size effect that rely on the central limit theorem.

The child's conception of food: Differentiation of categories of rejected substances in the 16 months to 5 year age range

Children (N = 54) ranging in age from one year four months to five years were offered over 30 items to eat. The items included normal adult foods and exemplars of different adult rejection categories: disgust (e.g. grasshopper, hair), danger (liquid dish soap), inappropriate (e.g. paper, leaf) and unacceptable combinations (e.g. ketchup and cookie). We report a high to moderate level of acceptance (item put into mouth) of substances from all of these categories in the youngest children. Acceptance of disgusting and dangerous substances decreases with increasing age, while acceptance of inappropriate substances remains at moderate levels across the age range studied. Although the youngest children accepted more disgust items, the majority rejected most of the disgust choices. Almost all children at all ages tested accept combinations of foods which, although individually accepted by adults, are rejected in combination. No significant differences were observed between 'normal' children and those with a history of toxin ingestion, although there was a tendency of ingesters to accept more inedible items. In general, the results suggest that a major feature of the development of food selection is learning what not to eat.

The role of knowledge base in the memory performance of good and poor readers

Good and poor junior high school readers, matched for IQ, were seen in four experiments. In the first experiment, subjects rated items from 12 categories in terms of typicality. These ratings were used to select self-defined typical and atypical items, individually for each child, for an incidental cued-recall task (Experiment 2). In Experiment 3, subjects were given a list comprised of typical and atypical items for free recall, with item typicality determined based on each subject's typicality norms from Experiment 1 (self-generated lists). Experiment 4 was identical to Experiment 3, with the exception that typicality of the items was based on adult norms (adult-generated lists). In Experiment 1, poor readers selected significantly fewer items as appropriate category exemplars than did good readers, and correlations of ratings with a group of adults were greater for the good than for the poor readers for each of the 12 categories. Recall was greater for the typical than for the atypical items in each of the three memory experiments for both the good and poor readers. Recall performance of the poor readers was comparable to that of the good readers for both the incidental cuedrecall and self-generated free recall tasks. Good readers recalled more than poor readers only with the adult-generated lists in Experiment 4. A series of analyses indicated no apparent differences in strategy use between the good and poor readers in either free recall experiment. The results were interpreted as reflecting the fact that poor readers have a different knowledge base for familiar categories than do good readers, and that cognitive differences between good and poor readers are related to differences in their semantic memories and not to differences in their ability or inclination to use mnemonic strategies. Possible ways in which differences in knowledge base may affect memory performance were also discussed.

Memory storage and retrieval processes in category learning.

The detailed course of learning is studied for categorization tasks defined by independent or contingent probability distributions over the features of category exemplars. College-age subjects viewed sequences of bar charts that simulated symptom patterns and responded to each chart with a recognition and a categorization judgment. Fuzzy, probabilistically defined categories were learned relatively rapidly when individual features were correlated with category assignment, more slowly when only patterns carried category information. Limits of performance were suboptimal, evidently because of capacity limitations on judgmental processes as well as limitations on memory. Categorization proved systematically related to feature and exemplar probabilities, under different circumstances, and to similarity among exemplars of categories. Unique retrieval cues for exemplar patterns facilitated recognition but entered into categorization only at retention intervals within the range of short-term memory. The findings are interpreted within the framework of a general array model that yields both exemplar-similarity and feature-frequency models as special cases and provides quantitative accounts of the course of learning in each of the categorization tasks studied.

Memory storage and retrieval processes in category learning.

The detailed course of learning is studied for categorization tasks defined by independent or contingent probability distributions over the features of category exemplars. College-age subjects viewed sequences of bar charts that simulated symptom patterns and responded to each chart with a recognition and a categorization judgment. Fuzzy, probabilistically defined categories were learned relatively rapidly when individual features were correlated with category assignment, more slowly when only patterns carried category information. Limits of performance were suboptimal, evidently because of capacity limitations on judgmental processes as well as limitations on memory. Categorization proved systematically related to feature and exemplar probabilities, under different circumstances, and to similarity among exemplars of categories. Unique retrieval cues for exemplar patterns facilitated recognition but entered into categorization only at retention intervals within the range of short-term memory. The findings are interpreted within the framework of a general array model that yields both exemplar-similarity and feature-frequency models as special cases and provides quantitative accounts of the course of learning in each of the categorization tasks studied.

Attention, similarity, and the identification-categorization relationship.

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 Shepard'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 paradigms. 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.

Contrasting the representation of scripts and categories

Recent findings have suggested that script actions are not sequentially organized in memory. Instead investigators have argued that (1) the positions of actions within scripts are represented by temporal properties stored with actions, this being analogous to how properties represent the size, color, and other characteristics of exemplars in categories, and (2) actions are organized by centrality in scripts, this being analogous to how exemplars are organized by typicality in categories. To compare the representation of scripts and categories, this study observed timed production of script actions and category exemplars. Subjects generated actions from scripts either in whatever order they came to mind, from most to least central, from first to last, or from last to first. Subjects generated exemplars from categories either in whatever order they came to mind, from most to least typical, from smallest to largest, or from largest to smallest. The results clearly demonstrated that script actions are sequentially organized in memory and that properties are not the means by which script actions are normally ordered during production. The results are also not consistent with the view that ease of production simply depends on the efficiency of retrieval strategies. It is concluded that scripts and categories share invariant properties of abstract representations and that their differences reflect constraints associated with their respective domains.