The purpose of our study is to investigate human teaching behavior and robot learning behavior when a human teaches a robot. Agents for learning support need to build a pedagogical relationship, in which a teacher agent and a student agent change their behaviors as they recognize the other's characteristic behaviors. In order to investigate how a robot that behaves as a student should respond to humans' teaching behaviors in a pedagogical relationship between human and robot, we conducted a case study using a game played on a tablet with a robot. In the case study, we analyzed how humans changed their teaching behaviors when the humanoid robot failed to understand what they taught. From the results of this case study, we observed that some subjects carefully taught the robot in each trial in order to allow the robot to understand the subjects. Moreover, we also observed that subjects' teaching behavior changed when the subject received feedback from the robot about the teaching.