Categorizing objects in isolation and in scenes: what a superordinate is good for

Many studies have shown that subjects are faster at categorizing objects into "basic" concepts than into more general superordinate concepts. However, all of these studies have used a categorization task in which single, isolated objects are identified. There is good reason to believe that superordinate concepts are typically used to refer to collections of objects rather than to individual objects. For example, people more often use the term furniture to refer to a number of pieces of furniture rather than to name a single piece. This suggest that superordinate concepts include information about multiple objects and their common relations, particularly the typical scenes in which such objects appear. Four experiments examined this possibility by investigating whether the basic concept advantage will decrease or reverse itself when subjects are asked to categorize an object as part of a scene. The results showed that the basic-superordinate difference did decrease when subjects categorized objects in scenes. Furthermore, when an object was placed in an inappropriate scene, there was more interference for superordinate identifications. The results suggest qualitative differences in the representations of superordinate and basic concepts.     

Ways of probing situated concepts

Two ways of eliciting conceptual content have been to instruct participants to list the intrinsic properties that concept exemplars possess or to report any thoughts that come to mind about the concept. It has been argued that the open, unconstrained probe is better able to elicit the situational information that concepts contain. We evaluated this proposal in two experiments comparing the two probes with regard to the content that they yield for object concepts at the superordinate and basic levels. The results showed that the open probe was better able to elicit situated conceptual knowledge and point out differences in the representations of superordinate and basic concepts.     

Inductive Reasoning in Alzheimer''s Disease

We evaluated knowledge of basic level and superordinate semantic relations and the role of cognitive resources during inductive reasoning in probable Alzheimer's disease (AD). Nineteen mildly demented AD patients and 17 healthy control subjects judged the truthfulness of arguments with a premise and a conclusion that contain familiar concepts coupled with "blank" predicates, such as "Spiders contain phosphatidylcholine; therefore all insects contain phosphatidylcholine." Like healthy control subjects, AD patients were relatively insensitive to the typicality of the premise category when judging the strength of arguments with a conclusion containing a basic-level concept, but were relatively sensitive to typicality during judgments of arguments containing a superordinate in the conclusion. Moreover, AD patients resembled control subjects in judging arguments with an immediate superordinate in the conclusion compared to arguments with a distant superordinate. AD patients differed from control subjects because they could not take advantage of two premises in an argument containing basic-level concepts. We conclude that semantic knowledge is sufficiently preserved in AD to support inductive reasoning, but that limited cognitive resources may interfere with AD patients' ability to consider the entire spectrum of information available during semantic challenges.     

Distortions in judged spatial relations

In three experiments, we explore distortions in subjects' judgments of relative geographical relations. People make large systematic errors in judging the geographical relations between two locations that are in different geographical or political units. There is a strong tendency to distort the judged relation to conform with the relation of the superordinate political unit. To account for this result, we present a model in which spatial information is stored hierarchically. Spatial relations between any two locations are stored explicitly only if those locations are within the same superordinate unit. Spatial relations not stored are inferred by combining the relations from between and within superordinate units.     

Naming practices,typicality, and underextension in child language

Preschool children (3 to 5 years of age) in an initial study underextended superordinate labels. The labels (e.g., food) were not applied to category instances that were rated as atypical by adults (e.g., ketchup). In a second experiment, mothers seldom used a superordinate term in referring to pairs of the same atypical instances, but they did label pairs of typical instances with a superordinate. A third experiment demonstrated the effects of several word-referent modeling conditions on young children's comprehension of a superordinate term. The results were viewed as suggesting that the child undergeneralizes superordinate labels because adult naming practices are such that category instances having certain functional or perceptual attributes may be labeled with a superordinate, while other, dissimilar instances are almost always labeled with subordinate names instead of a superordinate.     

Semantic categorization in the human brain: spatiotemporal dynamics revealed by magnetoencephalography

We examined the cortical representation of semantic categorization using magnetic source imaging in a task that revealed both dissociations among superordinate categories and associations among different base-level concepts within these categories. Around 200 ms after stimulus onset, the spatiotemporal correlation of brain activity elicited by base-level concepts was greater within than across superordinate categories in the right temporal lobe. Unsupervised clustering of data showed similar categorization between 210 and 450 ms mainly in the left hemisphere. This pattern suggests that well-defined semantic categories are represented in spatially distinct, macroscopically separable neural networks, independent of physical stimulus properties. In contrast, a broader, task-required categorization (natural/man-made) was not evident in our data. The perceptual dynamics of the categorization process is initially evident in the extrastriate areas of the right hemisphere; this activation is followed by higher-level activity along the ventral processing stream, implicating primarily the left temporal lobe.     

Object identification: The mental representation of physical and conceptual attributes

In two experiments, subjects indicated whether two pictures of familiar objects were equivalent. The picture pairs were identical, showed the same object in various perspectives and states, or showed different objects with varying degrees of conceptual relatedness. In Experiment 1, the equivalence criterion for judging the picture pairs was varied between subjects (identity of pictures; conceptual equivalence of objects at subordinate, basic, or superordinate level). The reaction times of the four subject groups suggest that a pictorial stimulus is not always mentally represented in the same way, and that the instructions given determine which attributes of the stimulus are represented. In Experiment 2, the visual similarity of the picture pairs was varied. The results indicate, at least with a basic-level equivalence instruction, that not only perceptual, but also non-perceptual, functional attributes are represented—namely, those that, according to Rosch et al. (1976), are common to the members of a superordinate category.     

Basic level and superordinate level categorization by autistic, mentally retarded, and normal children

Three experiments designed to test autistic children's nonverbal and verbal categorization abilities are reported in this paper. In the first two experiments, 14 autistic children were compared to 14 retarded and 14 normal children matched on verbal mental age. Their ability to categorize pictures from basic level categories and from biological and artifactual superordinate level categories was assessed using a matching-to-sample procedure. The three groups of subjects were similar in their performance. Basic level categorization was easier than more abstract categorization, and for all three groups, prototypicality played a role in categorizing superordinate level concepts; that is, children in all three groups made more errors categorizing peripheral examples. In the third experiment, a subgroup of 7 autistic children showed evidence that their lexicons were well organized and that they appreciated the meaning relationships among words at the superordinate level. These findings suggest that autistic children do not suffer a specific cognitive deficit in the ability to categorize and form abstract concepts, as has been previously suggested in the literature.     

Conceptual Hierarchies in a Flat Attractor Network: Dynamics of Learning and Computations

The structure of people's conceptual knowledge of concrete nouns has traditionally been viewed as hierarchical ( Collins & Quillian, 1969 ). For example, superordinate concepts ( vegetable ) are assumed to reside at a higher level than basic-level concepts ( carrot ). A feature-based attractor network with a single layer of semantic features developed representations of both basic-level and superordinate concepts. No hierarchical structure was built into the network. In Experiment and Simulation 1, the graded structure of categories (typicality ratings) is accounted for by the flat attractor network. Experiment and Simulation 2 show that, as with basic-level concepts, such a network predicts feature verification latencies for superordinate concepts ( vegetable ). In Experiment and Simulation 3, counterintuitive results regarding the temporal dynamics of similarity in semantic priming are explained by the model. By treating both types of concepts the same in terms of representation, learning, and computations, the model provides new insights into semantic memory.     

Intraverbal acquisition of semantic concepts by preschoolers

Intraverbal learning is a process through which semantic knowledge is acquired from purely linguistic information. The concern of this study is to investigate the role of intraverbal learning in the preschool-aged child's acquisition of semantic concepts. Three- and 4-year-old children were either shown pictures illustrating novel concepts or they were told verbal definitions of novel concepts, the definitions containing information about key physical and functional properties. An additional manipulation consisted of including or omitting information about the superordinate category membership of the concept. Concept acquisition was assessed via a battery of comprehension tasks; these tasks assessed both referential and formal knowledge of the concepts. Results indicate a correspondence between the type of information received and the type of knowledge acquired, with intraverbal information resulting in broader knowledge than pictorial information. Purely pictorial information was associated with referential knowledge only, while intraverbal information enhanced both referential and formal knowledge.     

Structure of the Indonesian Emotion Lexicon

Based on a prototype approach to emotion concepts, two studies were conducted: (1) to identify the mental state words that Indonesian speakers are most certain name emotions ( perasaan hati ) and (2) to map the hierarchical and family-resemblance structure of the top 124 emotion concepts. As in an earlier study of emotion terms in American English (Shaver, Schwartz, Kirson, & O'Connor, 1987), cluster analysis of sorting data collected in Indonesia revealed five basic-level emotion categories: cinta (love), senang (happiness), marah (anger), kawatir/takut (anxiety/fear), and sedih (sadness). Also in line with the American results, the five basic-level categories formed two large categories at the superordinate level: positive emotions and negative emotions. Each of the five basic-level categories contained several subordinate-level categories, totaling 31 in all. The results suggest that the emotion lexicons, and corresponding conceptualizations of the emotion domain, in Indonesia and the U.S.A. are similar at the superordinate and basic levels but somewhat variable at the subordinate level. This outcome – like other kinds of psychological research on emotions and emotion concepts – suggests that the gross structure of representations of the emotion domain are similar worldwide, perhaps for biological reasons, but that different cultures make different fine-grained distinctions and emphasize different subordinate-level emotion concepts.     

The effect of hierarchical levels of categories on children's deductive inferences about inclusion

Although the hierarchical levels of categories have been recognized as a major factor of variation in categorical reasoning, few studies have examined its effect on the understanding of inclusion. This issue was approached by varying the levels (subordinate, basic, and superordinate) of categories involved in inference tasks assessing 5‐, 7‐, and 9‐year‐old children's understanding of transitivity and asymmetry of inclusive relations in the dog hierarchy. Children were administered both a qualitative inference task and a quantitative class‐inclusion task, each presenting different hierarchical levels. Results showed that the ability to make qualitative inferences assessing transitivity varied with age. Although children of all ages demonstrated a high rate of success at these inference questions, 7‐ and 9‐year‐olds had better performance than 5‐year‐olds, suggesting that the capacity to understand the transitivity of inclusive relations still develops until at least 7 years. However, the hierarchical levels of categories had no effect on children's performance either in qualitative inferences requiring transitivity understanding or in class‐inclusion problems. In contrast, for qualitative inferences assessing asymmetry, children's performance varied with the hierarchical level of the categories involved. Inferring from a superordinate to a basic level category, inferring from a superordinate to a subordinate level category, and inferring from a basic to a subordinate level category appeared as three levels of increasing difficulty. Our analyses also revealed that 7‐year‐olds were better at grasping the asymmetry in the superordinate‐to‐basic relation than in any other relation between categories of different hierarchical levels, and that their 9‐year‐old peers mastered the asymmetric nature of inclusion in both superordinate‐to‐basic and superordinate‐to‐subordinate relations. This might indicate that the different levels of difficulty observed are developmentally grounded. Though exploratory, these findings help to clarify the steps through which the child comes to grasp the difficult concepts of inclusion and asymmetry and give some indications on the possible constraints that may affect their acquisition.     

Visual and Affective Multimodal Models of Word Meaning in Language and Mind

One of the main limitations of natural language-based approaches to meaning is that they do not incorporate multimodal representations the way humans do. In this study, we evaluate how well different kinds of models account for people's representations of both concrete and abstract concepts. The models we compare include unimodal distributional linguistic models as well as multimodal models which combine linguistic with perceptual or affective information. There are two types of linguistic models: those based on text corpora and those derived from word association data. We present two new studies and a reanalysis of a series of previous studies. The studies demonstrate that both visual and affective multimodal models better capture behavior that reflects human representations than unimodal linguistic models. The size of the multimodal advantage depends on the nature of semantic representations involved, and it is especially pronounced for basic-level concepts that belong to the same superordinate category. Additional visual and affective features improve the accuracy of linguistic models based on text corpora more than those based on word associations; this suggests systematic qualitative differences between what information is encoded in natural language versus what information is reflected in word associations. Altogether, our work presents new evidence that multimodal information is important for capturing both abstract and concrete words and that fully representing word meaning requires more than purely linguistic information. Implications for both embodied and distributional views of semantic representation are discussed.     

A large rectangle delays the perception of a separate small rectangle

Deciding whether two objects, rather than one, are present takes longer for large-small and small-large pairs of rectangles than for large-large and small-small pairs of rectangles. This large-small slowdown was eliminated when the large rectangle was slightly modified, when the large and small rectangles were contiguous, or when the task was to identify the large rectangle. However, it did occur when the task was to identify the small rectangle. These results suggest that the large rectangle delayed the perception of the small rectangle. Codes for features did not cause this delay. Instead, the evidence indicates that each rectangle evoked its own superordinate code, and that the large-rectangle superordinate code produced the gestalt of a large rectangle and also inhibited the small-rectangle superordinate code, thereby delaying the perception of the gestalt of a small rectangle. Thus, superordinate codes may also be the direct cause of other gestalts and their associated perceptual outcomes.     

Structure of the Basque emotion lexicon

Using a prototype approach to emotion concepts, two studies were conducted in the Basque Country, where an ancient non-Indo-European language is still spoken, to identify the mental state words that Basque speakers are most certain name emotions (emozioak) and to map the hierarchical and family resemblance structure of the most prototypical 124 emotion concepts. Cluster analysis of sorting data collected in the Basque Country revealed five basic level emotion categories similar to those found in American English and Indonesian (love, happiness, anger, sadness, and fear) as well as five other small positive emotion categories. All major categories found at the basic level contained several terms that are not traceable to Romance languages. Also in line with the American and Indonesian results, the basic level categories in Basque fell within two large superordinate categories: positive and negative emotions. Each of the five large basic level categories contained several subordinate level categories. The results suggest that the emotion lexicons, and corresponding conceptualisations of the emotion domain, in the Basque Country, Indonesia, and the US are similar, although there are some important differences.     

Hierarchical categorization by bilingual Latino children: does a basic-level bias exist?

Research has shown that children classify most easily at the basic level where objects in the same category look similar enough to each other to be grouped together but are distinct enough from objects in other categories to be discriminated (e.g., animal/bird/duck). In this article, the authors report on 2 experiments they conducted to determine whether children maintain this basic category bias when the perceptual similarity of stimuli at different hierarchical levels is equalized. Pictures within and across 3 hierarchical levels were made perceptually equivalent and shown to 71 Latino children who were bilingual in Spanish and English. In Experiment 1, the pictures used as exemplars could be categorized on any of the 3 hierarchical levels. In Experiment 2, example pictures unambiguously defined the level of categorization that would be accurate, and linguistic cues were given that might assist in the selection of the correct category. In both experiments, the children sorted pictures from all 3 levels equally well, but they found it harder to justify their sorting of superordinate pictures. English competence predicted sorting on the more ambiguous sorting task in Experiment 1; and English competence predicted verbal justifications in both experiments, even though the experiments were conducted in Spanish. Competence in Spanish or English was an equally good predictor of sorting in the better defined sorting task in Experiment 2. These findings indicate that a superordinate level deficiency remains after perceptual differences are eliminated and that the deficiency is cognitive in nature. Differences in the performances of children who differed in bilingualism support the hypothesis that a threshold of proficiency in both languages is an important determinant of the effect of bilingualism on categorization.     

语言和情境对具体概念感知运动仿真的影响

具体概念加工中的感知运动仿真是概念表征的核心加工过程。本研究系统探讨了语言因素(语言类型：第一语言和第二语言)和情境因素(知觉情境：空间知觉情境和语义知觉情境)对具体概念加工中感知运动仿真的影响。实验1采用语义相关判断范式探讨了第二语言加工过程中是否存在感知运动仿真, 以及第二语言中的仿真和第一语言中的仿真是否存在差异。研究发现, 第二语言加工中依然存在感知运动仿真, 但是第一语言的感知运动仿真具有一定的优势。实验2采用语义相关判断范式及其变式, 通过2个分实验分别探讨了个体加工概念时空间知觉信息和语义知觉信息对于概念表征过程中感知运动仿真的影响。研究发现, 无论是在较弱的空间信息知觉条件下还是在较浅的语义信息知觉条件下, 具体概念加工过程中都产生了感知运动仿真。本研究的发现弥补了知觉符号理论未对第二语言中的感知运动仿真提出针对性预测的不足, 表明感知运动仿真具有一定的跨语言稳定性。同时, 具体概念加工中的感知运动仿真不受空间信息和语义信息的调节, 这表明感知运动仿真能在一定程度上自动化产生。     

Object manipulability affects children’s and adults’ conceptual processing

Research on kinds of concepts indicates that children use perceptual and functional information differently to form natural and artifact concepts. Beyond object domain, object manipulability appears to be a decisive factor in adult conceptual processing. Thus, the effect of object manipulability on conceptual processing was tested in 5- and 7-year-olds and adults using a picture matching task. Reaction times for identifying conceptual relations on the basis of perceptual similarity (e.g., jacket-coat) and contextual/functional information (e.g., jacket-hanger) were analyzed according to object manipulability and domain. Both children and adults were faster to identify contextual/functional relations for manipulable than for nonmanipulable objects. Conversely, they were faster to identify perceptual similarity relations for nonmanipulable than for manipulable objects, particularly for natural concepts. Results reveal an early distinction between concepts of manipulable and nonmanipulable objects. Implications for further research on concept formation and for embodied views of concepts are discussed.     

A connectivity model for semantic processing

Summary A connectivity model for semantic processing, based on two major assumptions, is proposed and tested. The representation assumption states that semantically related features are represented by an interconnected structure of features. A network is termed interconnected if each node is linked to at least two other nodes. According to the processing assumption, indirect activation becomes increasingly effective as the number of interconnected nodes increases. The connectivity model thus predicts that more complex concepts will be processed faster than less complex concepts. Experiment 1 was performed to determine the number of features for a set of 72 words and to show that subordinate concepts share common features with their natural superordinate concept. In Experiments 2 and 3 subjects had to judge whether or not a word belonged to a previously defined natural superordinate concept. The results confirm the connectivity model and show that concepts with many features can be classified faster than concepts with only a few features.This research was supported by a Grant of the Max Kade Foundation     

Art educators report grades at which they teach 11 drawing concepts

The purpose of this study was to solicit information about the applicability of 11 basic perceptual concepts for instruction in drawing to students in Kindergarten through Grade 12. The study is based on responses to a national survey of 750 art educators, 250 in elementary, 250 in middle, and 250 in high school teaching assignments. Among these, 87% were female, 8% were 20 to 29 years old, 15% were 30 to 39 years, 48% were 40 to 49 years, 27% were 50 to 59 years, and 2% were 60 years or older. As few as 35.5% to as many as 48.7% (per concept) of the art educators reported that they teach each of the 11 concepts in elementary grades. Also, from 33.3% to 38.4% (per concept) of these teachers indicated that they stop teaching the 11 target concepts to students in high school. Empirical evidence from psychologists suggested that children gain comprehension of the concepts at differing points in their development. Were art educators employing the information identified by psychological research, their reports might have presented clear distinctions between grades at which each of the 11 concepts are taught. The results are interpreted in light of five possible explanations.