Beyond visual comfort: comparing pupillometry and subjective measures of cognitive strain in workplace lighting

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Introduction This study evaluated the discriminative sensitivity of pupillometry, relative to subjective self-report measures, as an objective indicator of cognitive load during screen-based tasks performed under two workplace illuminance levels (300 and 500 lux). Inadequate workplace illuminance is associated with visual fatigue, stress, and impaired attention, yet its effect on cognitive strain during screen-based tasks remains underexplored through objective physiological indicators, particularly given the confounding influence of the pupillary light reflex (PLR) on pupillometric measures. Methods Forty-eight working-age adults performed writing and working memory (N-back) tasks under both conditions in a controlled laboratory replicating office environments; pupil diameter was recorded continuously using Pupil Labs Core eye-tracking glasses, while the NASA-TLX and Visual Analog Scales (VAS) assessed subjective workload, stress, fatigue, and eye strain. Results A direct comparison between illuminance conditions (Holm-corrected) revealed no significant effects of illuminance on pupillary parameters (|d| ≤ 0.19), but subjective workload increased with illuminance during the most demanding task ( d = 0.81). Both pupillometry and subjective measures detected significant pairwise differences between task demand levels consistently under both illuminance conditions (Kendall's W ranging from 0.29–0.35 for pupil diameter variance to 0.34–0.58 for NASA-TLX and VAS Stress). No significant differences were found for the writing task or the 2-back task across any measure. Discussion The results indicate that pupillometry and subjective self-report measures are comparably sensitive to task demand, while only subjective workload was sensitive to illuminance, underscoring the value of combining objective and subjective indicators in occupational lighting research.