The uncanny valley is a great issue in the implementation of facial expression to a human-like agent. Recent studies are giving an explanation for this negatively emotional response to the agent from the perspective of \prediction error," the contradictory to their prediction that an agent should be a human. We hypothesized that that prediction error caused the negative affect of the facial expression of the agent, such as its eerie smile, as well. The qualitative computational model was proposed, providing an explanation for how human predicts the other's facial movement of the facial expression, based on cerebellar function. As a result, the rapid emotional processing was required and the thalamus was expected to play an essential role in it.

In the current study, we investigated the relationship between cognitive loads and cognitive processes during learning through problem solving. We proposed a cognitive model and tested this model with Reversi game as an experimental task. To control three types of cognitive load, we developed a learning environment in which participants were given the best move as help information and discs on the board could be indistinct. In this environment, we controlled intrinsic load with presenting the help information, extraneous load with making the board indistinct, and germane load with instruction. Experimental results revealed that the cognitive model we proposed was effective and implied that model-based research on cognitive load theory might be promising.

In this paper, learning support system for physical dynamics (especially impulse) using multiple tablet PCs and portable haptic devices (called as SPIDAR-tablet) is proposed. In this system, the virtual environment constructed and displayed on tablets PCs. Two users using this system can access this virtual environment from each tablet PCs. And by using haptic devices, users can feel various forces (touch force, impulse and so on) in virtual environment. In addition, tablet PCs can share a virtual environment therefore it is expected that proposed system is used for collaborative learning. In the verification, it was confirmed learner's discrimination threshold on virtual environment. From these result, high limit of power ratio (coefficient of restitution) used in virtual environment is determined as 0.8. And learning experiments using proposed system is performed.

This paper reports on the design, implementation and results of the Young Researcher Workshop on Educational Systems in 2015. We held the workshop under the theme:``Bonds of Young Researchers on Visualizing the Educational System Research'' in Boso Hanto, Chiba. The workshop design based on reflection and consisted of individual work, group discussion and reflection session. Forty-four faculty members and students from nineteen universities/institutes contributed to the presentation and discussion on their own research topics as well as new research conjunction possibilities. The design of the workshop was favorably evaluated by the participants.

Management of the learning activity based on the learning conditions permits smooth learning progress to meet learning object. This study intends to visualize learning conditions and to provide the information to teachers for realizing the management of learning activity. The group activity result is formed based on the opinion and understanding of the constituent members. Therefore, group activity result can be predicted from individual activity result. This paper reports the prediction approach and verification experiment for the approach.

This study aimed to develop a Thought Experiment Externalizer (TE-ext). TE-ext enables students to visualize a problem situation. Users of TE-ext can implement changes in the situation and see the result as an animation. Additionally, they can make reference to physical laws if they need. We expected that the students would have realized the benefits of thought experiment by solving the problem successfully using TE-ext. However, no one was able to reach the correct answer in our experiment. The students had a problem with carrying out an operation that made the situation different from the initial one.

Concept map is utilized as a technique for classroom assessment widely. Educational researchers try to apply concept map to various learning situations and propose several methods for evaluating concept map. In this research, we categorize the evaluation method of concept map to three types. First is the structural scoring method that takes an interest to the structure of concept map. Second is the semantic interpretation method that gives precedence to the quality of proposition. The last method is the semantic comparison method that compares the propositions in the concept maps of learners with the ones of their teacher. From these type of methods, we choose the Novak and Gowin's method, McClure and Bell's method and Kit-Build concept map in our preliminary experiment for comparing these types of methods. The preliminary experiments are designed to find suitable situation for using Kit-Build concept map and confirm our hypothesis that Kit-Build is good for knowledge-sharing situation. We assigned two situations that are reading and teaching situation for the experiment. Reading situation is that the learners read material and build their own understanding while teaching situation is that they try to understand instructor's interpretation. From the experiment results, although Kit-Build concept map has limitations that it needs the teaching and the ideal map (goal map), Kit-Build concept map is more effective in teaching than reading situation because the provided nodes and links can help learners to recognize exist knowledge. In addition, it can help instructors to check their learners' understanding by using the group-goal difference map that is an automatic diagnose. While the semantic interpretation method is good for representing learners' own understanding but it cannot help learners to understand teacher's interpretation and it takes a long time to score learner maps. And the structural scoring method can be used in both situation. But it returns the results that cannot explain the learners' understanding clearly because this method ignores meaning.

In our research, we have developed Error-based Simulation for simulating errors in force, acceleration and velocity in behaviors. In this EBS, a learner is required to input not only forces by also acceleration and velocity. The EBS generates simulation for the three types of inputs. Then, errors are visualized not only by the strange behavior but also by the differences between the generated behaviors. The experimental use of this EBS by students of a national institute of technology (equal to senior high school students) is also reported.

Concept maps are widely used in education area that are applied to represent and assess knowledge of learners. Several researchers propose various approaches for concept map evaluation. Scoring concept maps refer to a lot of criteria and concept map rubric. The scorings are also used for knowledge assessment that is to measure the level of success at the end of instruction unit. Automatic assessment of concept map is an alternative technique learner evaluation. In this paper, we compare automatic concept map assessment methods with Kit-build frameworks, and discuss in a different aspect of assessment process and effective feedback. We propose to construct Kit-build workflow for support formative assessment in classroom. Formative assessment is continually evaluated to make sure that students understand as expected during the learning process. The advantages of Kit-build over other automatic concept map assessment systems are the automatic diagnose and the immediate feedback are given from individual- or group- goal difference map. All of learner maps are overlaid to be group map. The deference map is generated from diagnosis process that is enhanced by filtering function to control intensity of excessive, lacking and leaving link. These links provide informative information for thorough analysis. The framework can reduce time for assessment and analysis. It also supports instructor to recognize misunderstanding or confusion of learners in the topic commonly and give immediate feedback in good time that help to improve learner in learning process.

Concept map is utilized as a technique for classroom assessment widely. Educational researchers try to apply concept map to various learning situations and propose several methods for evaluating concept map. In this research, we categorize the evaluation method of concept map to three types. First is the structural scoring method that takes an interest to the structure of concept map. Second is the semantic interpretation method that gives precedence to the quality of proposition. The last method is the semantic comparison method that compares the propositions in the concept maps of learners with the ones of their teacher. From these type of methods, we choose the Novak and Gowin's method, McClure and Bell's method and Kit-Build concept map in our preliminary experiment for comparing these types of methods. The preliminary experiments are designed to find suitable situation for using Kit-Build concept map and confirm our hypothesis that Kit-Build is good for knowledge-sharing situation. We assigned two situations that are reading and teaching situation for the experiment. Reading situation is that the learners read material and build their own understanding while teaching situation is that they try to understand instructor's interpretation. From the experiment results, although Kit-Build concept map has limitations that it needs the teaching and the ideal map (goal map), Kit-Build concept map is more effective in teaching than reading situation because the provided nodes and links can help learners to recognize exist knowledge. In addition, it can help instructors to check their learners' understanding by using the group-goal difference map that is an automatic diagnose. While the semantic interpretation method is good for representing learners' own understanding but it cannot help learners to understand teacher's interpretation and it takes a long time to score learner maps. And the structural scoring method can be used in both situation. But it returns the results that cannot explain the learners' understanding clearly because this method ignores meaning.

In this study, we conducted experiments to examine the factors that facilitate shifts in explanations using short story in which participants were required an explanation reconstruction. In the experiment, we controlled the time of presentation of a key fact that contradicts an initial explanation and has a central role in its reconstruction, reflective thinking, and the two together. The results are summarized as follows. First, when the prior explanation was rejected, attention to the key fact was inhibited although a new explanation was required. Second, the successful group increased their attention on the key fact just before the explanatory shift. Third, protection of the preceding explanation with unobserved facts was inhibited by guiding the participants' attention toward the key fact. Finally, although the initial explanation was not completely shifted, it has been suggested that a explanatory shift was prepared by activating reflective thinking with attention to the key fact.

This document describes the evaluation of tablet terminals, real projectors and mobile real projectors, from the practical examples to acquire tacit knowledge for learning at schools and museums.

We have built an instructional design support system called ``FIMA-Light'' which reasons about teacher's intentions from his/her lesson plan and automatically produces I_L event decomposition trees. The decomposition tree expresses the ways of achieving a learner's state change that should be realized in a whole lesson in the form of a tree structure. In this paper, we discuss an effective use of I_L event decomposition trees produced by FIMA-Light in teacher education program. And, we report about a practical use of FIMA-Light in teacher training education in the university in order to investigate its effect.

In this study, we experimentally investigated the influence of a three-dimensional (3D) graphic image and a 3D-printed object on a spatial reasoning task in a situation where liver resection surgery was presupposed. The results of the study indicated that using a 3D-printed object produced more accurate and faster mental model construction of a liver structure than a 3D image. Using a 3D-printed object was assumed to reduce cognitive load and information accessing cost more than using a 3D image.

This research describes a system supporting research activity of students through providing appropriate know-how information by using two models: know-how model and user model. In this paper, we discuss approaches to present know-how information based on relationships between tagged know-how in the know-how model. Adequacy of user models which represents features and trends of know-how information retrieved and evaluated by students is also evaluated after constructing the models based on the students' usage history of prototype system composed of the know-how model.

People perform tasks using their knowledge such as driving a car, cooking, and processing arithmetic tasks. Such knowledge for performing a variety of cognitive activities is referred to as procedural knowledge. In learning, it is important to understand what procedural knowledge is used for performing tasks. However, it is also difficult to monitor procedural knowledge automatically performed after repetitive experiences. In this study, we promote learners' understandings of procedural knowledge by having them engage in the externalization of procedural knowledge. We developed a rule-based cognitive modeling system for training of the externalization of procedural knowledge. An important feature of this system is the function to refer other members' models in group activities. With this feature, it is expected that more learners are guided to create cognitive models. We conducted two experiments for the evaluation of this system. In Experiment 1, it was confirmed that, in the solo learning setting, two thirds of learners create models using the system. In Experiment 2, it was also confirmed that, in the group learning setting, more learners successfully create plausible models with the function for group activities. Pre and post tests were performed in Experiment 2, indicating that externalizing procedural knowledge promotes the understandings of the knowledge.