Webcam-based eye tracking to detect mind wandering and comprehension errors.

Area of Science:

• Computer Vision
• Educational Technology
• Cognitive Science

Background:

• Webcam-based eye tracking offers scalable solutions for monitoring user engagement.
• Online educational technologies require adaptive responses to student experiences.
• Detecting covert cognitive states is crucial for understanding learning processes.

Purpose of the Study:

• To evaluate the efficacy of webcam-based eye tracking (WebGazer) for detecting cognitive states during online reading.
• To assess the accuracy and precision of WebGazer in predicting task-unrelated thought and reading comprehension.
• To examine the generalizability and performance of WebGazer across diverse populations and conditions.

Main Methods:

• Utilized WebGazer, a webcam-based eye-tracking system, for data collection.
• Employed an online reading-comprehension task to elicit cognitive states.
• Conducted two studies with distinct participant demographics (university students and Prolific users).
• Performed slicing analyses to assess performance under varying conditions (e.g., lighting, glasses).

Main Results:

• WebGazer provided accurate and precise gaze measurements for predicting task-unrelated thought and reading comprehension.
• Initial evidence of predictive validity was found, with a positive correlation between predicted task-unrelated thought and comprehension scores.
• The system demonstrated generalizability across different datasets and robustness under various environmental conditions.

Conclusions:

• Webcam-based eye tracking is a viable tool for monitoring cognitive states in online learning environments.
• This technology supports the development of adaptive educational tools and facilitates remote research.
• Further research can explore the application of webcam eye tracking in diverse remote learning and research contexts.