Work

CaliGaze: A Validated Webcam Eye-Tracking Framework for Attentional Drift-Diffusion Modeling

Year: 2026
Type: preprint
Abstract: Eye-tracking has become central to process-tracing research in judgment and decision-making, yet access to this paradigm remains constrained by the cost and logistical demands of research-grade infrared systems. We present CaliGaze, an open-source webcam eye-tracking framework that uses geometric iris landmark estimation, individually calibrated piecewise spatial mapping, and a recorded single-eye eyelid-opening validity flag to support ROI-level gaze classification in binary choice experiments. In a within-subject validation study (N = 51), CaliGaze was evaluated against a Tobii Pro Spark infrared eye tracker while participants completed a risky binary choice task with simultaneous dual-system recording. CaliGaze showed strong cross-device agreement in ROI-level dwell proportions and reproduced the core fixation-choice signatures of the task. Both gaze streams consistently supported gaze-dependent over gaze-independent evidence accumulation, and attentional-weighting estimates showed strong cross-device correspondence in between-participant rank ordering. These results suggest that CaliGaze can support ROI-level process tracing and gaze-dependent computational modeling in binary choice tasks, offering a scalable and accessible alternative for decision researchers, with particular potential for large-scale, remote, or resource-limited settings.
Language: English
Open Access status: green
OpenAlex ID: W7226025927
Cites:
Cited by:
Related to: 10