거울 과제에 대한 자기조절 과정의 fNIRS 두뇌 활성 규명과 딥러닝 기반 예측 모델 탐색 (Identification of fNIRS Brain Activity and Exploration of Deep Learning-Based Predictive Model in Self-Regulation Process Taking Mirror Task)

서지정보

· 발행기관 : 한국교원대학교 뇌·AI기반교육연구소
· 수록지 정보 : Brain, Digital, & Learning / 10권 / 4호 / 365 ~ 376페이지
· 저자명 : 권승혁, 박상희, 박진선, 황나래, 권용주

초록

Self-regulation is one of important cognitive processes in learning to change the learner's schema by itself. However, its definition and process have yet been clearly clarified and no strategies that can be systematically measured have been proposed. In this study, we tried to find out the neural characteristics of self-regulation process consisting of assimilation, conflict and accommodation by analyzing subject’s brain activity using fNIRS. In addition to the brain activity, a deep learning model was also developed for the process prediction of self-regulation. Forty-six high-school students were administered to take a mirror task inducing self-regulation, and their brain activities were measured and analyzed using fNIRS. As a result, the activities of both OFC and left DLPFC were found in the entire process of self-regulation. It can be seen that the process of self-regulation is mainly based on monitoring and evaluation of one's performance. In addition, the activities of FP and VLPFC were additionally shown in the accommodation phase. These results are presumably proposed that the goal-oriented multitasking and clarification using language might be two of major players in this process. In addition, a predictive model was developed through a deep learning process with an accuracy of 72.8% and a loss value of 0.309. These results are possibly suggested to know the characteristics of the brain level in self-regulation process and provided a basic information for the development and improvement of self-regulation. In addition, it can be used as a basis for developing a system capable of fast and accurate diagnosis and treatment for self-regulation.

영어초록

Self-regulation is one of important cognitive processes in learning to change the learner's schema by itself. However, its definition and process have yet been clearly clarified and no strategies that can be systematically measured have been proposed. In this study, we tried to find out the neural characteristics of self-regulation process consisting of assimilation, conflict and accommodation by analyzing subject’s brain activity using fNIRS. In addition to the brain activity, a deep learning model was also developed for the process prediction of self-regulation. Forty-six high-school students were administered to take a mirror task inducing self-regulation, and their brain activities were measured and analyzed using fNIRS. As a result, the activities of both OFC and left DLPFC were found in the entire process of self-regulation. It can be seen that the process of self-regulation is mainly based on monitoring and evaluation of one's performance. In addition, the activities of FP and VLPFC were additionally shown in the accommodation phase. These results are presumably proposed that the goal-oriented multitasking and clarification using language might be two of major players in this process. In addition, a predictive model was developed through a deep learning process with an accuracy of 72.8% and a loss value of 0.309. These results are possibly suggested to know the characteristics of the brain level in self-regulation process and provided a basic information for the development and improvement of self-regulation. In addition, it can be used as a basis for developing a system capable of fast and accurate diagnosis and treatment for self-regulation.

참고자료

· 없음