Predicting Intrinsic and Extraneous Cognitive Load with Oculo- and Bio-metric Indicators

This study focused on the prediction of cognitive load types using eye-tracking metrics, heart rate variability, and galvanic skin response using a machine learning model. Intrinsic cognitive load is associated with the inherent complexity of the mental task, whereas extraneous cognitive load is related to the distracting and unrelated elements in the task. Thirty-four participants (aged 21.18 ± 3.42) performed different levels of mental calculations to induce intrinsic cognitive load in the first task and a visual search task to manipulate extraneous cognitive load in the second task. During both tasks, participants’ eye movements, heart rate, and galvanic skin response were continuously recorded. The subjective cognitive load was also assessed following each experimental task. Participants’ working memory was controlled. The discriminant model, which consisted of ocular- and bio-metric indicators, demonstrated that these indicators have a high level of discriminant power to separate low and high cognitive load in mental calculations and visual search. In particular, the average fixation duration, average saccade amplitude, and K coefficient have a greater impact on the model. In addition, the trial difficulty of the tasks may be separated by low and high index of pupillary activity and heart rate variability.