Recently, remote education methods have been needed due to the globalization of production activity in companies. Moreover, it is becoming difficult to perform conventional on-the-job training (OJT) due to a reduction in the number of skilled domestic engineers. Therefore, we are developing a New OJT system as a remote e-learning skill education support system. The prototype of the system has now been developed. However, it is only able to provide education in low-level skills (education of skill level I is described later). The purpose of this research is to investigate education in high-level skills using the New OJT system, and to raise the level of the learner's skills. According to the fording of brain science, the discovery of "novelty" attracts attention. In addition, it is assumed that the selection of an action becomes adequate by correcting the incorrect answer action. Therefore, the following three methods are proposed for improving the learner's skill level. (1) Mistakes and problems are experienced as incorrect answer actions in the skill education. (2) A mental model to deal with the mistakes and problems that occur by accident. (3) A cause conception training method that is concerned with the causes and measures of the mistakes and problems. In an experiment, a learner studied the operation of the Numerical Control (NC) machine using the above-mentioned three methods. As skill level is classified into four classes, the learner's skill was evaluated on each class. Moreover, improvement in the learner's skill level was observed. Thus, the effectiveness of the proposed skill level improvement method was shown.