Experimental Study toward Estimation of a Learner Mental State from Processes of Solving Multiple Choice Problems Based on Eye Movements

Abstract

Recently, the research area of intelligent educational systems has increasingly approached Educational Data Mining, which addressed a variety of learner aspects such as mental states. Although data of eye movements is promising in realizing mental states of learners, we have to empirically understand features of eye movements in advance due to difficulty in handling them. Toward the final goal of automatic estimation of learner mental states based on eyes, this study experimentally described changes of eye movements by confidence of correct answers. In our experiment, participants were asked to answer multiple-choice problems and respond to questionnaires about the problems, with the results indicating that transitions of eyes in initial processes where all choices were scanned differed depending on the levels of confidence.