Four studies examined adults' and children's understanding of polymorphs -- organisms that undergo a radical transformation (e.g., caterpillar-butterfly) -- which have been used as critical tests of essentialism in concepts. In Study 1, adults judged the similarity of juvenile organisms to their adult versions and perceptual controls; in Study 2, children made forced-choice similarity judgments. Both age groups were influenced by both biological and perceptual relatedness. Age differences were largely explainable by children's knowledge of the particular metamorphosis rather than changes in overall strategy. In Study 3, adults made inductions of different kinds of properties from juveniles to other animals; biological relatedness was most important for “deep$rdquo; properties and perceptual resemblance for “superficial$rdquo; properties. Study 4 tested children's inductions on biological properties and found results paralleling their similarity judgments. These studies revealed that children and adults use both the biological relation across life stages and perceptual⁄morphological similarity in their thinking about these organisms. Development was largely in the direction of being better able to coordinate these two sources of information. The findings speak against accounts that emphasize perception or underlying knowledge alone.
It has been well documented that nonhuman animals, particularly pigeons, can classify photographs that are not defined only by a few simple features but contain a particular type of natural object. Categorization is of great relevance for humans and nonhuman animals to cope with the variability and complexity of their environments, allowing them to respond appropriately to new instances after having learned only a few instances from a given category. The advantage of animal studies is to provide insights about nonverbal, pre-linguistic cognitive mechanisms and to uncover the essential roles of human language in the development of cognitive behaviors. This is a review of recent developments of the categorization studies in animals. The topics included are discrimination of artificial categories created by mimicking the supposed structure of natural categories, theoretical issues surrounding the prototype effects shown by animals with artificial polymorphous, prototype, and family-resemblance categories, and learning of functional associations among highly varied members of a category.
What is ontology? There is a theoretical disagreement in the proper status of ontology. One group regards ontology as a tool for characterizing human knowledge; the other considers ontology to be a systematic method for representing reality. This article addresses basic questions surrounding ontology engineering and human concepts from a vantage point of cognitive science. Specifically, it is argued that upper-level ontology -- basic divisions of being -- is reflected in human cognition and affects various aspects of our everyday judgments. The authors think that this is because human cognition is organized in the way that it can help approximate the invariant structure of the environment. Possible research directions that help integrate ontology engineering, cognitive science and philosophy are also discussed.
Thus far, cognitive studies on concept have not focused on its creative features. The process of concept generation is still an open issue. In order to clarify the concept generation process, we discuss design from the perspective of creativity. First, we consider previous studies on design and segregate design into three categories: drawing, problem solving, and pursuit of the ideal. Next, we discuss each category from three perspectives: time direction, driving force, and creativity as a rational novelty. Following this, we define the generation process of a design image as a model of concept generation related to metaphors, abductions, and operations of abstract concepts. Finally, we redefine design -- that is the process of composing a desirable figure toward the future -- and address that creativity in design is a criterion for the desirable figure on the basis of the knowledge gained from our studies.
Several studies on categorization have revealed that the strategy of categorization varies according to the category structures of the stimuli and the procedures used in the experiment. This study examined how categorization could vary by the distribution of learning exemplars, which has not been examined in the previous studies. The stimuli were star-shaped geometric figures, and they were divided into two categories: fat and thin. Two conditions were set for the learning exemplars. In the dense condition, the exemplars were located slightly far from the boundary of the two categories. In the prominent condition, one exemplar was located far from the boundary (the prominent exemplar) and the others, near the boundary. Participants were told to rate the typicality for each exemplar in the test phase. In the dense condition, the responses of many participants revealed that the typicality ratings tended to be the highest for the learning exemplars located near the center. On the other hand, in the prominent condition, the rating values of many participants increased according to the increase in the distance from the boundary. These results showed that the difference in the distribution of learning exemplars caused the change of categorization. Additionally, the prominent exemplar facilitated the comparison between the two categories and contributed largely to summarize information in the prominent condition.
Prototype theory of categorization and category learning assumes that a category is simply represented by its central tendency. The theory accounts for many psychological phenomena associated with categories, yet it is shown to be incapable of accounting for some important aspects of categories and concepts. For example, Prototype theory, due to its simplistic representation, cannot describe how people make inference about variabilities and correlations among feature dimensions within categories. In addition, it cannot learn categories that are not linearly separable. The present research extends Prototype theory of category learning in order to improve its explanatory capability while maintaining the simplistic representation mechanism. Our theory assumes that a category is not only represented by its central tendency but also by an abstracted within-category structure. In order to evaluate its descriptive validity, we developed a computational model built on the basis of the theory. In our model, called STRAP for STRucture Abstracing Prototype, a central tendency is represented by a mean vector (i.e., centroids) and an abstracted within-category structure by a covariance matrix. Three simulation studies were conducted and the results showed that STRAP successfully accounted for empirical phenomena that have not been replicated by existing prototype models: it acquired knowledge that is necessary for making inferences about variabilities and correlations among feature dimension within categories; it learned to categorize linearly non-separable categories; it reproduced A2 advantage, which is a tendency that people categorize a less “prototypical” stimulus A2 more accurately than more “prototypical” stimulus, invalidating some criticisms against Prototype theory. More important, STRAP and thus our theory accounts for these psychological phenomena with distinctive cognitive information processes, as compared with those of other successful models, providing new insights into how categories are represented in our mind.
A semantic space model provides a framework of semantic representation. In this model, each word is represented by a high-dimensional vector and the degree of semantic similarity between any two words can be easily computed as the cosine of the angle formed by their vectors. Recently, a number of methods have been proposed for constructing semantic spaces, but little has been known about the properties of different semantic spaces, in particular what kinds of semantic relations can be represented by what kinds of semantic spaces. In this study, we constructed fourteen different semantic spaces using three corpora (i.e., Japanese newspaper articles, Japanese novels, and Japanese dictionary), two construction methods (i.e., term frequency (TF) and term cooccurrence (CO)) and three context-window sizes (i.e., article, paragraph, and sentence). We then examined the properties of these spaces by comparing the ability to represent three semantic relations (i.e., coordination⁄synonymy, superordination, and collocation) and their eight subrelations. As a result, we demonstrated that, regardless of construction method and window size, the coordination⁄synonymy relation was better represented by the dictionary-based semantic spaces, but the collocation relation was better represented by the newspaper- and TF-based spaces. We also found that the superordination relation was better represented by the TF-based spaces with paragraphs as a window size, and corpus difference between dictionary and newspaper did not affect the representational ability of superordination. In addition, we investigated the effects of dimensionality reduction by singular value decomposition. The overall result was that the performance in predicting word association was degraded, but the performance of typicality judgment for the coordination⁄synonymy relation was improved by dimentionality reduction.
The purpose of this study to clarify the cognitive process associated with comprehension of metaphorical expression “B is like A” using model simulations and psychological experiments. The process of understanding the metaphor in the form of “B is like A” is regarded as the process that transforms the meaning of the target “B” into the meaning of the metaphor. In our model, nouns are represented probabilistically by meaning vectors based on the concept structure estimated using statistical analysis of Japanese corpus. This model assumes that metaphor understanding consists of two processes. The first is the categorization process; according to Class Inclusion Model (Glucksberg \& Kayser, 1990), a target is assigned to an ad hoc category of which the vehicle is a prototypical member. The second is the dynamic interaction process; the target assigned to the ad hoc category is influenced by dynamic interaction among features. Feature emergence occurs through this dynamic interaction. In order to verify the psychological validity of our model, we first conducted psychological experiments followed simulation studies based on the experiments. The results showed that our model that incorporates both categorization and dynamic interaction processes performed better than a model with only categorization process in prediction human cognitive behaviors, providing evidence for its psychological validity.
This study provides an explanation about a general concept structure underlying inductive reasoning that enables context-dependent knowledge selection. The similarity that determines inductive strength in an argument is considered to be based on features of entities in the propositions. It is thought that, instead of exploiting all possible features, contexts in a proposition select a certain set of those features. However, no general theory about concept structure has been proposed that enables such a context-dependent feature selection. In the present study, we focus on the syntactic dependency structure of propositions in an argument that determines the relationship of entities to the predicate in those propositions. We also focus on the issue that the statistical analysis of such a relationship reveals the concept structure underlying inductive reasoning, in which nouns are arranged by the semantic roles in verbs that could describe these nouns (a verb-centric concept structure). Such a concept structure enables the specification of the appropriate knowledge to be induced, and the similarity based on that selected knowledge determines the inductive strength. We explained the mechanism by a computational model with data from a statistical analysis of syntactic dependency. The model simulation predicts some phenomena that are examined and validated by empirical studies. Finally, we discuss the issue that the domain-inclusive verb-centric concept structure that enables the context-dependent knowledge selection underlies inductive reasoning.
In 2004, Ninio suggested that the process of comprehending noun phrases such as “black shoes” included two steps: comprehending the noun “shoes” and the addition of the attribution “black.” Based on this assumption, Ninio predicted that the developmental process of comprehending noun phrases must progress through a phase in which children understood only the noun. She demonstrated the validity of her assumption through experiments conducted in Hebrew. However, her results were confounded by the word order of the noun phrase in the Hebrew language, in which the noun is the first word and the headword. In this study, we used Japanese phrases to eliminate the artifacts of the above study, because in the Japanese language, the noun is neither the first word in a noun phrase, nor the headword in an adjective phrase. Results indicated that also in Japanese, there is a developmental phase in which children comprehended only the noun in the noun phrases, which confirmed Ninio's assumption. However, there was also a phase in which children comprehended only the noun in the adjective phrases. These results cast doubt on Ninio's suggestion that the process of comprehending the noun phrase includes two steps, or that this two-step process results in a phase in which children comprehend phrases based only on the noun. We think current results relate the idea of noun dominance in children's learning of new words as described by Gentner in 1982, and this dominance influences children's attention by directing it to the noun when comprehending phrases.
In this paper, we propose a new model of human hypothesis testing process. The model predicts a sequence of testing when people continuously receive positive results that confirm the hypothesis. First we constructed a Bayesian model which addresses 5 situations to be tested based on information theory. In a simulation, we identified 5 qualitative characteristics of the model's behavior, some of which were corroborated in a subsequent experiment. The results of the experiment, however, also suggested that the participants did not correctly estimate the size of the hypothesis. By reconstructing the model, we simplified the model to explain participants' data. As a consequence, we found that a much simpler model can sufficiently explain the data. In conclusion, it was suggested that people are roughly Bayesian but they use heuristic strategies in hypothesis testing.
The purpose of this study is to investigate whether copying works of art facilitates creative drawing. Thirty undergraduates, not majoring in art, individually participated in a three-day experiment. Each one was assigned to one of three experimental groups: the Copy and Create (CC) group, the Copy and Reproduce (CR) group, or the control (No copy but Create, NC) group. The participants who copied an artist's drawings (CC group) produced more creative drawings in the post-test than those who did not (NC group). Furthermore, the designs of the drawings by the CC group were different from those by the CR group. Analyses of questionnaires and verbal protocols indicate that copying an unfamiliar style of art (in this case, the abstract style) facilitated a change in the participants' mental representation of artistic creation through the following two cognitive processes: First, it relaxed the constraint of their existing idea about drawing, i.e., ‘a realistic drawing style is important’, and second it helped them to construct a new idea about drawing, i.e., ‘Subjective impression or evaluation is essential’, so that they were able to have a new perspective in drawing their original pictures. They consciously tried to express such impressions or evaluations in the post-test drawing, and the result was that they produced more creative works. Although it has been claimed that copying often plays a negative role in creation, our findings suggest that active interaction with the ideas of others through copying their works of art has great potential to facilitate student's artistic creation.