The impact of cognitive hints on mathematical problem solving: A webcam-based eye-tracking study

Source abstract

This study investigates the impact of cognitive hints on mathematical problem-solving performance and the role of individual learner characteristics in engaging with hints, while simultaneously evaluating the technical feasibility of webcam-based eye-tracking with a single calibration session for large-scale learning analytics. Utilizing a webcam-based eye-tracking approach, we recorded the gaze patterns of 5th-grade students (n = 134) as they solved mathematical tasks across four thematic sets. The study employed a within-subject design with rotated thematic sets and on-demand access to step-by-step hints. Data were analyzed using Linear Mixed Models (LMM) to account for the nested structure of the observations. A key methodological objective was to determine whether lower-quality webcam data, which can be easily gathered at scale without compromising participant privacy, can still reliably support the analysis. Findings indicate that while hints have a negligible effect on high-performing students, they significantly reduce the knowledge gap for students with lower initial ability levels. Furthermore, a significant learning effect was observed for subsequent tasks within the same thematic set. Despite the inherent noise in webcam-based data, the eye-tracking metrics explained a substantial part of the variance between experimental conditions. We found that students with higher levels of math anxiety and intrinsic motivation exhibited significantly longer total fixation time on hints once accessed and a lower overall fixation count in this area. Additionally, female students demonstrated a general tendency for higher visual engagement with hint areas, regardless of whether they chose to open them.