Automatic Assessment of Student Understanding Level Using Virtual Reality

The improvement of the efficiency in teaching requires knowing the understanding level of each student. However, it is difficult due to limited time in a class. We propose a Virtual Reality (VR) space imposing assignments on students, to know their understanding level from their behavior which comes from cognitive loads during their answering. The VR space presents a student an assignment and a working space to answer it. In general, students solve assignments, using elements on their short term memory. When students solve same kind of assignments many times, they build generalized solution methods in their long term memory. When they engage in such assignments, their cognitive load is low enough to make them watch only the working spaces, keeping their hands working. On the other hand, when students have no solution pattern, their short term memory works hard. Their high cognitive load often stop their hands, because of confusion. They also look assignments and the working space many times, to reconsider solutions. Since answering behavior of students exposes their cognitive load, a VR space is ideal to estimate cognitive load. We conducted an experiment to evaluate the ability of the method to estimate the cognitive load. We examined the movement of the hand and the edit distance of student's answer from the correct sentence during their answering. We confirmed a fair correlation of the hands' stagnation with the confidence in students of good scores. We also found a relationship of eye movement with the change of the edit distance. The experiment result implies the possibility to estimate the cognitive load. The estimation would enable teachers to know students' understanding faults, which leads to education according to the understanding level.