Where are my students looking at? Using Gaze Synchronicity to Facilitate Online Learning

Giving effective lectures requires managing all students' attention. In lecture rooms, such crowd attention is typically intuitively assessed by judging students' gaze cues. However, in online teaching, such gaze cues are often missing. In a series of studies, we aimed at facilitating online teaching by visualizing eye movement synchronicity and predicting learning success: (1) We recorded students‘ gazes (fixation positions) on PowerPoint slides and visualized the aggregated gazes to the lecturers as a heat map, an ellipse map, two moving bars, and one vertical bar superimposed on instructional videos. We found that the lecturers preferred the ellipse for live teaching and the heat map for retrospective analysis because they valued the spatial proximity of the visualization to the content. (2) In addition to passively watching visualizations of students‘ attention, signaling (with a digital laser pointer) is often used to actively direct and synchronize learners‘ attention. We investigated to which extent learners follow signals on PowerPoint slides and whether this synchronization predicts post-quiz performance. We found that the average distance between the learner’s gaze and the pointer position predicts the student’s post-quiz performance. (3) So, looking at the same point as the lecturer seems to be beneficial for learning success. But does synchronicity between individuals (how synchronous one is to the average of all learners) predict performance as well? In principle, yes. However, we were not able to reproduce the findings with unreliably-sampled webcam data, highlighting the challenges lying ahead of webcam-based eye tracking in practice. Overall, we found that measuring and visualizing gaze synchronicity can be a useful tool for investigating the students‘ attention in applied settings. It is also beneficial for online education by providing direct feedback to the lecturer and provide quizless assessment of the learner. These insights can help in creating automated immediate-feedback systems for educational videos.