Objects, parts, and categories

Concepts may be organized into taxonomies varying in inclusiveness or abstraction, such as furniture, table, card table or animal, bird, robin. For taxonomies of common objects and organisms, the basic level, the level of table and bird, has been determined to be most informative (Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976). Psychology, linguistics, and anthropology have produced a variety of measures of perception, behavior, and communication that converge on the basic level. Here, we present data showing that the basic level differs qualitatively from other levels in taxonomies of objects and of living things and present an explanation for why so many measures converge at that level. We have found that part terms proliferate in subjects' listings of attributes characterizing category members at the basic level, but are rarely listed at a general level. At a more specific level, fewer parts are listed, though more are judged to be true. Basic level objects are distinguished from one another by parts, but members of subordinate categories share parts and differ from one another on other attributes. Informants agree on the parts of objects, and also on relative "goodness" of the various parts. Perceptual salience and functional significance both appear to contribute to perceived part goodness. Names of parts frequently enjoy a duality not evident in names of other attributes; they refer at once to a particular appearance and to a particular function. We propose that part configuration underlies the various empirical operations of perception, behavior, and communication that converge at the basic level. Part configuration underlies the perceptual measures because it determines the shapes of objects to a large degree. Parts underlie the behavioral tasks because most of our behaviors is indirect toward parts of objects. Labeling appears to follow the natural breaks of perception and behavior; consequently, part configuration also underlies communication measures. Because elements of more abstract taxonomies, such as scenes and events, can also be decomposed into parts, this analysis provides a bridge to organization in other domains of knowledge. Knowledge organization by parts (partonomy) is contrasted to organization by kinds (taxonomy). Taxonomies serve to organize numerous classes of entities and to allow inference from larger sets to sets included in them. Partonomies serve to separate entities into their structural components and to organize knowledge of function by components of structure. The informativeness of the basic level may originate from the availability of inference from structure to function at that level.     

Family resemblances: Studies in the internal structure of categories

Six experiments explored the hypothesis that the members of categories which are considered most prototypical are those with most attributes in common with other members of the category and least attributes in common with other categories. In probabilistic terms, the hypothesis is that prototypicality is a function of the total cue validity of the attributes of items. In Experiments 1 and 3, subjects listed attributes for members of semantic categories which had been previously rated for degree of prototypicality. High positive correlations were obtained between those ratings and the extent of distribution of an item's attributes among the other items of the category. In Experiments 2 and 4, subjects listed superordinates of category members and listed attributes of members of contrasting categories. Negative correlations were obtained between prototypicality and superordinates other than the category in question and between prototypicality and an item's possession of attributes possessed by members of contrasting categories. Experiments 5 and 6 used artificial categories and showed that family resemblance within categories and lack of overlap of elements with contrasting categories were correlated with ease of learning, reaction time in identifying an item after learning, and rating of prototypicality of an item. It is argued that family resemblance offers an alternative to criterial features in defining categories.     

Evidence for a basic level in event taxonomies

The proposition that event taxonomies have a “basic” level was tested. In the first two tasks of the experiment, taxonomically related event categories were elicited. Nine taxonomies were constructed from subjects’ responses on these tasks. In the third task, subjects listed attributes for events categories from the three levels of abstraction of these taxonomies. A single, common level of abstraction (the basic level) was identified. Subjects listed significantly more attributes for basic categories than for superordinate categories, but not significantly more attributes for the lower, subordinate-level categories. Further analyses showed that basic and subordinate categories elicited a greater number of concrete and event-related features than action and personrelated features, and that superordinate categories showed the opposite trend. These findings are discussed in terms of general principles of categorization applied to events, objects, and scenes, and the particular characteristics of events.     

EFFECTS OF CATEGORIZATION ON COLOR PERCEPTION

Abstract— Subjects were shown simple objects and were asked to reproduce the colors of the objects Even though the objects remained on the screen while subjects reproduced the colors and the objects' shapes were irrelevant to the subjects' task, subjects' color perceptions were influenced by the shape category of an object. For example, objects that belonged to categories with redder objects were judged to be more red than identically colored objects belonging to another category. Further experiments showed that the object categories that subjects use, rather than being fixed, depend on the objects to which subjects are exposed.     

Category-structure effects in picture comparisons

Subjects indicated for a series of trials whether or not two pictures of common objects were from the same category. Reaction time (RT) was recorded. The pool of stimuli used in the series of trials consisted of two pictures representing each of several categories. Half were S categories, members of which were similar; the other half were D categories, which had dissimilar members (this variable is termed category structure). In Experiments 1 and 3, the similarity manipulation defining category structure was based on the concept of basic level categories (Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976). In Experiment 2, category labels were homonyms, and the two members of S or D categories represented one or two senses of the labels, respectively. The principal finding was an effect of category structure (faster RT on trials using stimuli from S categories rather than D categories). These results are interpreted in terms of a model that postulates the use of semantic codes in the task, even on trials where the two stimuli are physically identical.     

Parts and the basic level in natural categories and artificial stimuli: Comments on Murphy (1991)

Natural taxonomies consist of categories that vary in level of abstraction. Categories at the basic level, such as chair and apple, are preferred in a broad range of situations (Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976). Several studies have revealed qualitative differences between the basic level and other levels. For example, Tversky and Hemenway (1984) presented evidence that parts proliferative at the basic level; they proposed that parts link the appearance of category members with their functions. Although not taking issue with these findings, Murphy (1991) investigated whether parts are necessary or sufficient for a basic level. In an attempt to demonstrate that parts are not necessary, Murphy used artificial stimuli that did not capture the essential features of natural taxonomies. These discrepancies preclude any conclusions based on his studies. Murphy's data also do not support his claim that parts are not sufficient for a basic level. Finally, it is unlikely that pursuing questions of necessity or sufficiency will produce insights into human categorization.     

Acquisition of basic object categories

Two implications of best-example theory for category acquisition were considered. The first is that categories which people acquire based on initial exposure to good exemplars should be learned more easily and (at first) more accurately than categories based on initial exposure to poor exemplars. The second is that people should generally learn that the best exemplars are category members, before learning that the poor exemplars are category members. These implications are based on the premise that people generalize based on similarity, and that the best example has maximal within-category similarity and minimal extra-category similarity, while the poor examples have minimal within-category similarity and relatively high extra-category similarity. Both implications were strongly supported by the present research. It was also found that when pressure to communicate was removed, comprehension and production of category names were virtually identical. The predictions of best-example theory concerning the conceptual structures underlying the words used by children who are just beginning to talk were discussed briefly. This research also allowed the replication of several important categorization results which had previously been found with real-world categories, with a set of artificial concrete object categories.     

Cognitive reference points

Two methods were used to test the hypothesis that natural categories (such as colors, line orientations, and numbers) have reference point stimuli (such as focal colors, vertical and horizontal lines, and numbers that are multiples of 10) in relation to which other stimuli of the category are judged. In Experiment I, subjects placed pairs of stimuli into sentence frames consisting of linguistic “hedges” such as “— is essentially—.” Results were that the supposed reference stimuli were most often placed in the second (reference) slot. In Experiment II, the subject placed a stimulus in physical space to represent his feeling of the psychological distance of that stimulus from another spatially fixed stimulus. Results showed that, when supposed reference stimuli were fixed, other stimuli were placed closer to them than vice versa. The results have substantive implications for the understanding of internal structure of categories and methodological implications for the mapping of reference points, quantification of linguistic intuitions, and the assumption of symmetry in psychological distance judgments.     

How causal knowledge affects classification: A generative theory of categorization

Several theories have been proposed regarding how causal relations among features of objects affect how those objects are classified. The assumptions of these theories were tested in 3 experiments that manipulated the causal knowledge associated with novel categories. There were 3 results. The 1st was a multiple cause effect in which a feature's importance increases with its number of causes. The 2nd was a coherence effect in which good category members are those whose features jointly corroborate the category's causal knowledge. These 2 effects can be accounted for by assuming that good category members are those likely to be generated by a category's causal laws. The 3rd result was a primary cause effect, in which primary causes are more important to category membership. This effect can also be explained by a generative account with an additional assumption: that categories often are perceived to have hidden generative causes.     

Contrast effects in typicality judgements: A hierarchical Bayesian approach

We examine the influence of contrast categories on the internal graded membership structure of everyday concepts using computational models proposed in the artificial category learning tradition. In particular, the generalized context model (Nosofsky, 1986), which assumes that only members of a given category contribute to the typicality of a category member, is contrasted to the similarity–dissimilarity generalized context model (SD-GCM; Stewart & Brown, 2005), which assumes that members of other categories are also influential in determining typicality. The models are compared in a hierarchical Bayesian framework in their account of the typicality gradient of five animal categories and six artefact categories. For each target category, we consider all possible relevant contrast categories. Three separate issue are examined: (a) whether contrast effects can be found, (b) which categories are responsible for these effects, and (c) whether more than one category influences the typicality. Results indicate that the internal category structure is codetermined by dissimilarity towards potential contrast categories. In most cases, only a single contrast category contributed to the typicality. The present findings suggest that contrast effects might be more widespread than has previously been assumed. Further, they stress the importance of characteristics particular of everyday concepts, which require careful consideration when applying computational models of representation of the artificial category learning tradition to everyday concepts.     

Reasoning with uncertain categories

Five experiments investigated how people use categories to make inductions about objects whose categorisation is uncertain. Normatively, they should consider all the categories the object might be in and use a weighted combination of information from all the categories: bet-hedging. The experiments presented people with simple, artificial categories and asked them to make an induction about a new object that was most likely in one category but possibly in another. The results showed that the majority of people focused on the most likely category in making inductions, although there was a group of consistently normative responders who used information from both categories (about 25% of our college population). Across experiments the overall pattern of results suggests that performance in the task is improved not by understanding the underlying principles of bet-hedging but by increasing the likelihood that multiple categories are in working memory at the time of the induction. We discuss implications for improving everyday inductions.     

归类优势与基本水平效应的再探讨

类别特异性分数是预测归类优势的有效指标。实验1考察该分数能否预测基本水平归类优势, 结果：分数较高的表现归类优势。实验2检验有归类优势的下位类别的分数, 结果：它们都有较高分数, 同时发现下位类别的特异性可以影响分数。实验3探讨相同的基本水平与特异性不同的下位组合的归类表现, 结果：下位特异性低就表现基本水平归类优势, 高则相反。据此提出“经验说”, 分数高低与人们在经验中对各层次类别形成的表征特异性程度相关。     

The development of category learning strategies: What makes the difference?

Category learning can be achieved by identifying common features among category members, distinctive features among non-members, or both. These processes are psychologically and computationally distinct, and may have implications for the acquisition of categories at different hierarchical levels. The present study examines an account of children’s difficulty in acquiring categories at the subordinate level grounded on these distinct comparison processes. Adults and children performed category learning tasks in which they were exposed either to pairs of objects from the same novel category or pairs of objects from different categories. The objects were designed so that for each category learning task, two features determined category membership whereas two other features were task irrelevant. In the learning stage participants compared pairs of objects noted to be either from the same category or from different categories. Object pairs were chosen so that the objective amount of information provided to the participants was identical in the two learning conditions. We found that when presented only with object pairs noted to be from the same category, young children (6 ? YO ? 9.5) learned the novel categories just as well as older children (10 ? YO ? 14) and adults. However, when presented only with object pairs known to be from different categories, unlike older children and adults, young children failed to learn the novel categories. We discuss cognitive and computational factors that may give rise to this comparison bias, as well as its expected outcomes.     

Establishing visual category boundaries between objects: a PET study

Individuals with Alzheimer's disease (AD) often have problems in recognizing common objects. This visual agnosia may stem from difficulties in establishing appropriate visual boundaries between visually similar objects. In support of this hypothesis, showed that AD subjects have difficulties in establishing visual category boundaries between continuously graded shapes. In an attempt to investigate the neural basis of these impairments, the current study required a group of neurologically healthy elderly participants to categorically classify a series of ellipses varying in width while regional blood flow changes were measured using positron emission tomography (PET). Two categorization conditions were compared in order to isolate changes in cortical activity that dissociated the categorization of shapes situated either near or far from a category boundary that divided the width continuum. The participants produced a discontinuity in the probability and speed of categorizing the shapes at some point along the continuum, suggesting that the objects were classified into distinct categories. Moreover, a comparison of the PET scans obtained while the subjects were categorizing the shapes situated near vs. far form the category boundary revealed significant differences in cortical activity in the parietal and frontal brain areas. These findings suggest that both visuo-spatial and decision making mechanisms may be involved in establishing categorical distinctions between continuously graded stimuli. It is proposed that the functional role of the parietal and frontal cortical regions in establishing visual boundaries between categories of objects may be relevant for understanding object recognition impairments in AD.     

The cause of infant categorization?

We asked whether infants are sensitive to causal relations between objects and outcomes and whether this sensitivity supports categorization. Fourteen- and 18-month-old infants were familiarized with objects from a novel category. For some, the objects caused an electronic toy to activate. For others, the objects were present during activation of the toy, but did not cause the event. For the remaining infants, the events were never activated. Infants were asked to select another category member from a pair of previously unseen objects (one from the familiar, and one from a novel, category). Infants were more likely to select the category match in the causal than the non-causal and no outcome conditions, suggesting that they capitalize on causal information in forming object categories.     

Failure to find evidence of stimulus generalization within pictorial categories in pigeons

Pigeons' key pecks were reinforced in the presence of pictures from one of two categories, cats or cars. A single picture associated with reinforcement was used in Experiment 1, and 20 pictures from the same category were associated with reinforcement in Experiment 2. Pigeons then were presented with novel test pictures from the training category and from the other, previously unseen, category. During Session 1 of testing, pigeons pecked no more often at pictures from the reinforced category than at pictures from the previously unseen category. When pigeons were trained with pictures associated with reinforcement or its absence from different categories in Experiment 3, differential responding to novel pictures from different categories appeared during Session 1. These findings argue against a process of automatic stimulus generalization within natural categories and in favor of the position that category distinctions are not made until members of at least two categories are compared with one another.     

实物形状的知觉相似性对视觉隐喻加工的影响

视觉隐喻是众多研究者比较关注的领域, 当前多以物体简笔画、漫画及广告等方面的研究为主, 但对与生活密切相关的实物的视觉隐喻研究甚少。因此研究采用掩蔽启动范式、李克特9点量表及语音生成等研究方法, 在物体概念类型(概念相同、概念相异)及形状类型(形状相似、形状相异)条件下, 针对实物图片的形状知觉相似性是否对视觉隐喻加工产生影响进行探讨。结果发现无论物体概念是否相同, 形状相似性对个体的物体识别具有“促进”作用, 并有利于视觉隐喻加工; 在物体识别中, 当两个物体概念不同时, 个体会试图建立一个特设类别来对形状相似的物体进行归类, 从而实现了视觉隐喻的加工。因此研究进一步证明了实物形状的知觉相似性有利于视觉隐喻的加工。     

Folk psychology of mental activities

A central aspect of people's beliefs about the mind is that mental activities--for example, thinking, reasoning, and problem solving-- are interrelated, with some activities being kinds or parts of others. In common-sense psychology, reasoning is a kind of thinking and reasoning is part of problem solving. People's conceptions of these mental kinds and parts can furnish clues to the ordinary meaning of these terms and to the differences between folk and scientific psychology. In this article, we use a new technique for deriving partial orders to analyze subjects' decisions about whether one mental activity is a kind or part of another. The resulting taxonomies and partonomies differ from those of common object categories in exhibiting a converse relation in this domain: One mental activity is a part of another if the second is a kind of the first. The derived taxonomies and partonomies also allow us to predict results from further experiments that examine subjects' memory for these activities, their ratings of the activities' importance, and their judgements about whether there could be "possible minds" that possess some of the activities but not others.