Can EEG signal predict learners’ perceived difficulty?

Abstract

This study presents an approach to predict learner’s perceived difficulty using features extracted from electroencephalography (EEG) data. We demonstrate how EEG signals can be used effectively to estimate learner’s perceived difficulty of learning content. Student self-reports of perceived difficulty and EEG data were gathered from 9 participants who watched a video lecture. A machine learning model with random forest classifier achieved a maximum accuracy of 75.24% in estimating perceived difficulty. Furthermore, the model predicted the difficulty level of the entire video lecture for individuals fairly well. Our results have implications for intelligent tutoring systems which aim at providing the learner with an adaptive and personalized learning environment.