Xiaorui Xue¹, Ido Davidesco¹, Na’ama Av-Shalom¹, Marina Vasilyeva¹, Nanyu Zhang¹, Rachael Sabelli¹, Jason Geller¹; ¹Boston College

Attention is highly dynamic, and students frequently shift their focus from the learning task to unrelated thoughts—a phenomenon known as mind wandering (MW). Although MW can sometimes be beneficial, it is generally viewed as detrimental to learning. Prior estimates suggest that students spend roughly 30% of class time MW, but these estimates rely on self-reports, which could be inaccurate. Laboratory studies suggest that there is a tight link between attention shifts and eye movements, but it remains unclear whether eye-movement patterns can reliably index MW in authentic learning environments. To address this gap, we conducted a classroom-based study using wearable eye-tracking glasses (Pupil Labs Neon). Data were collected from 15 college students over an 8 week period. Each student completed two regular class sessions while wearing eye tracking glasses. During each session, students received 6–8 probes asking whether they had been mind wandering and to report their cognitive load immediately beforehand. At the end of each session, students completed a short quiz to assess their learning. Preliminary analyses show that students reported MW on approximately 28% of the probes. Furthermore, the frequency of mind wandering reports was negatively correlated with quiz scores (r = -.27, p = .035), suggesting that greater mind wandering was linked to poorer learning outcomes. Reports ofMW were associated with longer fixations and reduced blink frequency in the 60 second period before each probe.These preliminary results support the use of eye-tracking metrics as indicators of attention fluctuations in real-world environments. This pilot study demonstrates the potential of using eye-tracking measures to capture the dynamics of attention in real-world classrooms and highlights opportunities to advance methodological approaches for detecting MW.

Acknowledgements: This material is based upon work supported by the National Science Foundation under Grant No. 2526250