A Support Vector Machine Application for the Detection of Pupillary Markers of Cognitive Workload
Kouadio Niamba, Frank Schieber · Proceedings of the Human Factors and Ergonomics Society Annual Meeting · 2022
The visual system is a core processing system and an assessment of the cognitive workload generated by vision-related tasks is made possible through pupillometry. Pupillometry, the study of pupil dimensions and their reactivity to mental effort, proposes that the diameter of the pupil expands concomitantly with the amount of mental resources mobilized (Beatty, 1982). However, considering the pupil’s responsiveness to the brightness of the surrounding, an application of pupillometry in the absence of controlled illumination represents a challenge. Overcoming this challenge is the objective of the present study. Using a head-mounted eye-tracking device, the present investigation induces transient levels of brightness in the continuous performance of cognitive tasks. Most importantly, the novel implementation of Support Vector Machine algorithms suited for recognition across intermittent time series enables the discrimination of frequency-specific pupillary markers of cognition from frequency-specific pupillary markers of brightness.