成り立つ心 : 赤ちゃんの心の発達 (特集 : 「心の科学」)

Abstract

In the present article, I reviewed recent stream of developmental sciences, especially of development in infancy. In the first section, the importance of the perspectives of evolutionary developmental psychology in infant study is emphasized. In the same section, connection of infant study and brain science were described. From now onward, non-invasive techniques, such as functional Magnetic Resonance Imaging (fMRI), or Near Infra-Red light spectroscopic Topography : NIRT) are providing a large contribution in the infant studies. In the next section, the development of the ability to find other's goal, intention, and mind, so-called mentalizing system was reviewed. According to Butterworth and his colleague, by 18-month-old, infants developed joint visual attention in three mechanisms, "ecological mechanism", "geographical mechanism", and "representational mechanism." Gergely et al. found that infants aged between 9 and 12 months expect agents to approach a goal in the most economic way. We also found that children aged between 24 and 36 months imitated humanoid robot's action, and they completed the action even it is failed attempt when the robot gazed at the human partner before and after it began action. We tried to investigate whether young children infer a robot's mental state in a standard False Belief task. We employed autonomous robot, named Robovie. All of the stimuli were presented in the video monitor. There were two versions of video stimuli. One of the scene of the video was as follows : Robovie puts the doll away in a particular location (Box A), then leaves the room. During Robovie's absence the man removes the doll from Box A, and places the doll, not back in Box A, but in Box B. The other of the scene of the video was the same as robot version except that human was appeared instead of the robot. Each subject was shown these two types of scenes, and given five questions just after watching each video scene individually. The order of presenting the stimulus was counterbalanced. Four questions are as follows : 1) "Where will it/he look for a doll?" (we called this "question of prediction"), 2) "Which box does it/he think the doll is in?" (We called this "question of representation"), 3) "Which box has a doll?" (We called this "question of reality"), 4) "Which box had a doll at first?" (We called this "question of memory"). There are two main findings in this study. First, older children answered correctly whether the agent was human or robot. Second, the children tended to answer more correctly in the human condition when they were asked by using psychological verb, such that "What does it/he think?." This seems to be very interesting results. The older children can predict both human and robot behavior correctly in FB task, however, they showed different responses when psychological verb was used in the question.