Distinguishing Intrinsic and Extraneous CognitiveLoad Using Eye-Tracking and Wavelet-Based Features

This study investigates the differentiation of cognitive load types—intrinsic and extraneous—using only eye-tracking data enhanced by our proposed wavelet-based features (WF) and selected eye-tracking metrics. Intrinsic load reflects task complexity and primarily taxes working memory, while extraneous load arises from inefficient information presentation, engaging visual attention. In a controlled experiment, 21 participants (aged 24.76 ± 3.29 years) performed standard tasks designed to induce both types of load. Eye-tracking data were collected using a Tobii Pro Nano Eye-Tracker, and WF were introduced to capture time-sensitive eye movement patterns. Alongside traditional metrics, we employed a participant-specific modeling approach to account for individual variability, training models on each individual’s data to optimize classification performance. Our results show that WF features, fixation metrics, and saccade peak speed are particularly effective in distinguishing between intrinsic and extraneous load. These findings highlight the potential of using eye-tracking data alone to model cognitive load types, offering a non-intrusive means of real-time cognitive state assessment. This work has practical implications for adaptive systems that dynamically respond to users’ cognitive demands, enhancing usability, efficiency, and overall user experience in human-computer interaction contexts