In this article, we report the differences in eye movement between experts and novices in “Mind Map,” a graphical note-taking method. Using a total of six different mind maps, we examine how subjects' eye movements followed the basic structure of each Mind Map stimulus. Each subject was asked to comprehend the content of the mind maps on a display. By using an eye-tracker built into the display, we were able to obtain both subjects’ gaze information and eye-movements without restraining their natural observing behavior. Our findings indicate that while novices observe the mind map on a branch-to-branch basis, as if they were following the note-taking process itself, experts first browse key information located around the central image of the note, then move onto more de tailed content. That is, while novice Mind Mappers thoroughly scan entire branches attached to a trunk, and then shift their attention to the next trunk, experienced Mind Mappers first browse the trunk, possibly to obtain a general idea of what the mind map is about, then shift their attention to detailed branches. This appears as though they were spontaneously building their own table-of-contents.
Many types of humanoid robots have been developed recently, and they are mainly designed for social interaction with human beings. The most communicative partners for human beings are other humans. Therefore, to develop successful communicative robots, it is important to know how closely they resemble a human. In the present study, we attempted to evaluate the human likeness of a humanoid robot (Robovie) by using near-infrared spectroscopy (NIRS). Since activity of the human mirror neuron system (MNS) is thought to reflect the perceived human likeness of observed agents, we compared MNS activity during observations of an action performed by a human and the robot. Seven male and ten female participants were included in the study, and eventually, fourteen of them were analyzed. NIRS probes were placed at the bilateral premotor and primary motor areas, which are components of the MNS. Under obser vation conditions, stimuli were presented live or on a video monitor; there were four observation conditions, namely, live-human, live-robot, video-human, and video-robot. After the observation conditions, the participants executed the same action performed by the human agent in the observation conditions by themselves (execution condition). We identified the NIRS channels in which significant activation was induced under both the observation and execution conditions, and used this information to determine the possible regions reflecting MNS activity. We found no significant effect of the agent (human/robot) on MNS activity, and this indicated that MNS response in the motor related area is relatively analogous irrespective of the agent (human/robot). However, the effect of the mode of presentation (live/video) was found in a few channels. Two channels corresponding to the left ventral premotor cortex were activated more strongly in the live condition than in the video condition, particularly when the agent was the human. In contrast, one channel corresponding to the right primary motor cortex was activated more strongly in the video condition than in the live condition only when the agent was the robot. These findings suggest that live presentation of action is necessary to reveal true brain activity in actual situations.