This paper shows a method of recognizing a learner's action simply with just a few information on learning attitude measurement system. It is assumed that the head, the right hand and the left hand move characteristically during learning, and we measure the motions of the three body parts of the learner. The system recognizes the behavior type comparing with the motion data related of learning behaviors that has been prepared beforehand. It experimented with the above on the nine typical learning attitudes such as “lateral writing”, “mathematical expression”, “document creation with word-processing software” and so on. As a result, it proved that the method is effective for learning attitude recognition.