松香 敏彦

Particle Swarm Optimization (PSO) is a type of meta-heuristic optimization method built on the basis of the principle of collective behaviors exhibited by simple organisms. Although PSO is a model of social behaviors, the present research attempts to model learning behaviors of an individual human with PSO in order to evaluate our hypothesis that the dynamics of knowledge that are being acquired and updated in our mind resemble the dynamics of social interactions exhibited by swarms. A simulation study showed that a cognitive model with PSO was able to replicate not only manifested cognitive behaviors but also latent cognitive behaviors, resulting in the acquisition of at least two dissimilar yet functional solutions for a given task.

There is an ever increasing number of electronic documents available today and the task of organizing and categorizing this ever growing corpus of electronic documents has become too large to perform by analog means. In this paper, we have proposed an augmented version of the bisecting k-means clustering algorithm for automated text categorization tasks. In our augmented version, we have added (1) a bootstrap aggregating procedure, (2) a bisecting criteria that relies on dispersions of data within clusters, and (3) a method to automatically terminate the algorithm when an optimal number of clusters have been produced. We have performed text categorization experiments in order to compare our algorithm against the standard bisecting k-means and k-means algorithms. The results showed that our augmented version improved approximately 15% and 20% in classification accuracies compared to the standard bisecting k-means and k-means, respectively.

Recent cognitive modeling studies suggest the effectiveness of metaheuristic optimization in describing human cognitive behaviors. Such models are built on the basis of population-based algorithm (e.g., genetic algorithm) and thus hold multiple solutions or notions. There are, however, important yet unaddressed issues in cognitive mechanisms associated with possession of multiple notions. The issues we address in the present research is about how multiple notions are organized in our mind. In particular, we paid close attention to how each notion interact with other notions while learning a new concept. In so doing, we incorporated Particle Swarm Optimization in a cognitive model of concept learning. Three PSO-based concept learning models were developed and compared in the present exploratory cognitive modeling study.

The majority of previous computational models of high-order human cognition incorporate gradient descent algorithms for their learning mechanisms and strict error minimization as the sole objective of learning. Recently, however, the validity of gradient descent as a descriptive model of real human cognitive processes has been criticized. In the present paper, we introduce a new framework for descriptive models of human learning that offers qualitatively plausible interpretations of cognitive behaviors. Specifically, we apply a simple multi-objective evolutionary algorithm as a learning method for modeling human category learning, where the definition of the learning objective is not based solely on the accuracy of knowledge, but also on the subjectively and contextually determined utility of knowledge being acquired. In addition, unlike gradient descent, our model assumes that humans entertain multiple hypotheses and learn not only by modifying a single existing hypothesis but also by combining a set of hypotheses. This learning-by-combination has been empirically supported, but largely overlooked in computational modeling research. Simulation studies show that our new modeling framework successfully replicated important observed psychological phenomena. (C) 2008 Elsevier B.V. All rights reserved.

The majority of previous computational models of high-order human cognition incorporate gradient descent algorithms for their learning mechanisms and strict error minimization as the sole objective of learning. Recently, however, the validity of gradient descent as a descriptive model of real human cognitive processes has been criticized. In the present paper, we introduce a new framework for descriptive models of human learning that offers qualitatively plausible interpretations of cognitive behaviors. Specifically, we apply a simple multi-objective evolutionary algorithm as a learning method for modeling human category learning, where the definition of the learning objective is not based solely on the accuracy of knowledge, but also on the subjectively and contextually determined utility of knowledge being acquired. In addition, unlike gradient descent, our model assumes that humans entertain multiple hypotheses and learn not only by modifying a single existing hypothesis but also by combining a set of hypotheses. This learning-by-combination has been empirically supported, but largely overlooked in computational modeling research. Simulation studies show that our new modeling framework successfully replicated important observed psychological phenomena. (C) 2008 Elsevier B.V. All rights reserved.

In two empirical studies of attention allocation during category learning, we investigate the idea that category learners learn to allocate attention optimally across stimulus dimensions. We argue that "optimal" patterns of attention allocation are model or process specific, that human learners do not always optimize attention, and that one reason they fall to do so is that under certain conditions the cost of information retrieval or use may affect the attentional strategy adopted by learners. We empirically investigate these issues using a computer interface incorporating an "information-board" display that collects detailed information on participants' patterns of attention allocation and information search during learning trials. Experiment 1 investigated the effects on attention allocation of distributing perfectly diagnostic features across stimulus dimensions versus within one dimension. The overall pattern of viewing times supported the optimal attention allocation hypothesis, but a more detailed analysis produced evidence of instance- or category-specific attention allocation, a phenomenon not predicted by prominent computational models of category learning. Experiment 2 investigated the strategies adopted by category learners encountering redundant perfectly predictive cues. Here, the majority of participants learned to distribute attention optimally in a cost-benefit sense, allocating attention primarily to only one of the two perfectly predictive dimensions. These results suggest that learners may take situational costs and benefits into account, and they present challenges for computational models of learning that allocatc attention by weighting stimulus dimensions.

It is widely acknowledged that categorically organized abstract knowledge plays a significant role in high-order human cognition. Yet, there are many Unknown issues about the nature of how categories are internally represented in our mind. Traditionally, it has been considered that there is a single innate internal representation system for categorical knowledge, such as Exemplars, Prototypes, or Rides. However, results Of recent empirical and computational Studies collectively suggest that the human internal representation system is apparently capable of exhibiting behaviors consistent with various types Of internal representation schemes. We, then, hypothesized that humans' representational system as a dynamic mechanism, capable of selecting a representation scheme that meets situational characteristics, including complexities of category structure. The present paper introduces a framework for a cognitive model that integrates robust and flexible internal representation machinery. Three simulation studies were conducted. The results showed that SUPERSET, Our new model, successfully exhibited cognitive behaviors that are consistent with three main theories of the human internal representation system. Furthermore, a simulation study oil social cognitive behaviors showed that the model was capable of acquiring knowledge with high commonality, even for a category structure with numerous valid conceptualizations. @ 2007 Elsevier Ltd. All rights reserved.

It is widely acknowledged that categorically organized abstract knowledge plays a significant role in high-order human cognition. Yet, there are many Unknown issues about the nature of how categories are internally represented in our mind. Traditionally, it has been considered that there is a single innate internal representation system for categorical knowledge, such as Exemplars, Prototypes, or Rides. However, results Of recent empirical and computational Studies collectively suggest that the human internal representation system is apparently capable of exhibiting behaviors consistent with various types Of internal representation schemes. We, then, hypothesized that humans' representational system as a dynamic mechanism, capable of selecting a representation scheme that meets situational characteristics, including complexities of category structure. The present paper introduces a framework for a cognitive model that integrates robust and flexible internal representation machinery. Three simulation studies were conducted. The results showed that SUPERSET, Our new model, successfully exhibited cognitive behaviors that are consistent with three main theories of the human internal representation system. Furthermore, a simulation study oil social cognitive behaviors showed that the model was capable of acquiring knowledge with high commonality, even for a category structure with numerous valid conceptualizations. @ 2007 Elsevier Ltd. All rights reserved.

In the human mind, high-order knowledge is categorically organized, yet the nature of its internal representation system is not well understood. While it has been traditionally considered that there is a single innate representation system in our mind, recent studies suggest that the representational system is a dynamic, capable of adjusting a representation scheme to meet situational characteristics. In the present paper, we introduce a new cognitive modeling framework accounting for the flexibility in representing high-order category knowledge. Our modeling framework flexibly learns to adjust its internal knowledge representation scheme using a meta-heuristic optimization method. It also accounts for the multi-objective and the multi-notion natures of human learning, both of which are indicated as very important but often overlooked characteristics of human cognition.

It is well known that our prior knowledge and experiences affect how we learn new concepts. Although several formal modeling attempts have been made to quantitatively describe the mechanisms about how prior knowledge influences concept learning behaviors, the underlying cognitive mechanisms that dive rise to the prior knowledge effects remains unclear. In this paper, we introduce a computational cognitive modeling framework that is intended to describe how prior knowledge and experiences influence learning behaviors. In particular, we assume that it is not simply the prior knowledge stored in our memory trace influencing our behaviors, but it is also the learning strategies acquired through previous learning experiences that affect our learning behaviors. Two simulation studies were conducted and the results showed promising outcomes.

We present a computational model of human category learning that learns the essential structures of the categories by forgetting information that is not useful for the given task. The model shifts attention to salient information and learns associations between items and categories. Attention and association strengths are adjusted according to the degree of prediction errors the model makes. The attention and association weights are interpreted as memory strengths in the model and decay over time, allowing the model to focus on the salient structures. Using memory decay mechanisms, our model simultaneously explained human recognition and classification performances that previous models could not.

High-order human cognition involves processing of abstract and categorically represented knowledge. Traditionally, it has been considered that there is a single innate internal representation system for categorical knowledge. However, on the basis of the previous empirical and simulation studies, we view the representational system as a dynamic mechanism, capable of selecting a representation scheme that meets situational characteristics, including complexities of category structure. The present paper introduces a framework for a cognitive model that integrates robust and flexible internal representation machinery. A set of three simulation studies were conducted. The results showed that SUPERSET, our new model, successfully exhibited cognitive behaviors that are consistent with rule- (Simulation 1A), prototype- (Simulation 1B), and exemplar-like (Simulation 1C) internal representation schemes.

Is the visual analysis of human actions modulated by the emotional content of those actions? This question is motivated by a consideration of the neuroanatomical connections between visual and emotional areas. Specifically, the superior temporal sulcus (STS), known to play a critical role in the visual detection of action, is extensively interconnected with the amygdala, a center for emotion processing. To the extent that amygdala activity influences STS activity, one would expect to find systematic differences in the visual detection of emotional actions. A series of psychophysical studies tested this prediction. Experiment 1 identified point-light walker movies that convincingly depicted five different emotional states: happiness, sadness, neutral, anger, and fear. In Experiment 2, participants performed a walker detection task with these movies. Detection performance was systematically modulated by the emotional content of the gaits. Participants demonstrated the greatest visual sensitivity to angry walkers. The results of Experiment 3 suggest that local velocity cues to anger may account for high false alarm rates to the presence of angry gaits. These results support the hypothesis that the visual analysis of human action depends upon emotion processes.

Cognitive models has been a main tool for quantitatively testing theories on human cognition. The results of previous cognitive modeling research collectively suggest the comparative advantage of exemplar over prototype accounts in human cognition. However, we hypothesized that unsuccessful outcomes by traditional prototype models may be the unforeseen consequences of the algorithmic constraints imposed on the models, but not of the implausibility of the theory itself. To test this hypothesis, a new cognitive model based on prototype theory with a more complex and realistic attention system is introduced and evaluated in the present study. A simulation study shows that a new model termed CASPRE resulted in a substantial improvement as compared with the traditional prototype model in replicating empirical findings and that it performed marginally better than an exemplar model, thus confirming our hypothesis.

This paper introduces a new cognitive model of human learning, specifically applied for category learning. Our new model, called SCODI, assumes that human learning is driven by heuristically controlled optimization processes of subjectively and contextually defined utility of knowledge being acquired, and offers hypothesis-testing-like interpretations with emphasis on stochastic processes. SCODI is built on an algorithm that (a) allows the utilization of past experience to retrospectively evaluating the current hypotheses set in order to revise knowledge and concepts, (b) is capable of generating and testing more than one set of hypotheses for a given corrective feedback datum, and (c) adapts to dynamically fluctuating contextual factors in learning. SCODIs effectiveness in replicating observed human data was established by two simulation studies.

Human category learning has been modeled using exemplar, prototype, and rule-based theories. Rule-based models are the least discussed. This paper presents a rule-based model based on evolutionary computation techniques. Such techniques allow for the combination of concepts, an important aspect of human cognition that has been largely overlooked in previous cognitive modeling research. We also include other human-like characteristic in the model, namely a simplicity bias and instance-based learning. The results suggest that such an algorithm can replicate well-known results in human category learning. We discuss the broader issue of which of the three models of categorization make sense in particular situations.

The primary focus in Computational modeling research in high order human cognition is to compare how realistically embedded algorithm describes human cognitive processes. However, several Current models incorporated learning algorithms that apparently have questionable descriptive validity or qualitative plausibleness. The present research attempts to bridge this gap by identifying five critical issues overlooked by previous modeling research and then introducing a modeling framework that addresses the issues and offers better qualitative plausibleness. A simulation study utilizing the present framework with two distinctive implementation approaches shows their descriptive validity.

The majority of previous computational models of high-order human cognition incorporate gradient descent algorithms for their learning mechanisms and strict error minimization as the sole objective of learning. Recently, however, the validity of gradient descent as a descriptive model of real human cognitive processes has been criticized. In the present paper, we introduce a new framework for descriptive models of human learning that offers qualitatively plausible interpretations of cognitive behaviors. Specifically, we apply a simple multi-objective evolutionary algorithm as a learning method for modeling human category learning, where the definition of the learning objective is not based solely on the accuracy of knowledge, but also on the subjectively and contextually determined utility of knowledge being acquired. In addition, unlike gradient descent, our model assumes that humans entertain multiple hypotheses and learn not only by modifying a single existing hypothesis but also by combining a set of hypotheses. This learning-by-combination has been empirically supported, but largely overlooked in computational modeling research. Simulation studies show that our new modeling framework successfully replicated observed phenomena.

The gradient descent optimization method has been a de facto standard learning algorithm in computational models of category learning. However, it can be considered as a normative (vs. descriptive) model of human learning processes. In particular, there are three concerns associated with the learning algorithm-namely, complexity, regularity, and context independency. In response to these limitations, the present study introduces an alternative, hypothesis-testing-like learning algorithm on the basis of a stochastic optimization method. The new learning model, termed SCODEL, provides qualitatively simple interpretations for its implied category-learning processes. Moreover, SCODEL is the first modeling attempt to depict individually unique and context-dependent learning processes. Four simulation studies were conducted and showed that the present model has the competence to operate as several different types of learners in various plausibly real-life situations.

Humans have the ability to flexibly adjust their information processing strategy according to situational characteristics. However, such ability has been largely overlooked in computational modeling research in high-order human cognition, particularly in learning. The present work introduces frameworks of cognitive models of human learning that take contextual factors into account. The framework assumes that human learning processes are not strictly error minimization, but optimization of knowledge. A simulation study was conducted and showed that the present framework successfully replicated observed psychological phenomena.

Haxby et al. [Science 293 (2001) 2425] recently argued that category-related responses in the ventral temporal (VT) lobe during visual object identification were overlapping and distributed in topography. This observation contrasts with prevailing views that object codes are focal and localized to specific areas such as the fusiform and parahippocampal gyri. We provide a critical test of Haxby's hypothesis using a neural network (NN) classifier that can detect more general topographic representations and achieves 83% correct generalization performance on patterns of voxel responses in out-of-sample tests. Using voxel-wise sensitivity analysis we show that substantially the same VT lobe voxels contribute to the classification of all object categories, suggesting the code is combinatorial. Moreover, we found no evidence for local single category representations. The neural network representations of the voxel codes were sensitive to both category and superordinate level features that were only available implicitly in the object categories. (C) 2004 Elsevier Inc. All rights reserved.

In the present study, GECLE (Matsuka, 2003) was used as a general modeling framework to systematically compare the plausibility of two prominent assumptions about internal representations of neural network (NN) models of human category learning. In particular, exemplar-model friendly Medin and Schaffer's 5/4 stimulus set (1978) was used for comparing prototype-and exemplar-based NN models. The results indicate that some prototype-based models performed as good as or better than an exemplar-based model in replicating the empirical classification profile. In addition, a phenomenon called A2 advantage (i.e., people tend to categorize the less "prototypical" stimulus A2 more accurately than more "prototypical" stimulus A1) reported in empirical studies (e. g., Medin & Schaffer 1978) was also successfully reproduced by these prototype-based NN models.

Items that violate a category rule axe remembered better than items that follow the rule. This finding cannot be predicted by exemplar models when all exemplars share the same attention along a dimension. With dimension-wide attention, violating and rule-following items are treated equally. When each exemplar selects which dimensions to attend to, exemplar models can predict the memory advantage for violating items. With exemplar-specific attention, attention is distributed uniformly for exemplars encoding violating items but is allocated to the rule dimension of exemplars encoding rule-following items. This differential attention makes violating items distinctive in memory. In addition to exemplar-specific attention, exemplar models need the ability to distinguish important errors from negligible ones to predict better memory for items that violate a stronger than a weaker rule.

Many neural network (NN) models of categorization (e.g., ALCOVE) use a gradient algorithm for learning. These methods have been successful in reproducing group learning curves, but tend to underpredict variability in individual-level data, particularly for attention allocation measures (Matsuka, 2002). In addition, many recent models of categorization have been criticized for not being able to replicate rapid changes in categorization accuracies and attention processes observed in the empirical studies (Macho 1997; Rehder & Hoffman, 2003). In this paper we introduce stochastic learning algorithms for NN models of human category learning and show that use of the algorithms can result in (a) rapid changes in accuracies and attention allocation, and (b) different learning trajectories and more realistic variability in individual-level.

One problem in evaluating recent computational models of human category learning is that there is no standardized method for systematically comparing the models' assumptions or hypotheses. In the present study, a flexible general model (called GECLE) is introduced that can be used as a framework to systematically manipulate and compare the effects of a limited number of assumptions at a time. Two simulation studies are presented to show how the GECLE framework can be useful in the field of human high-order cognition research.