Head vs Screen-Mounted Eye Trackers: Comparisons Across Ethnic Diversity

Eye tracking accuracy and reliability may be influenced by physiological diversity and eye tracker design, yet systematic biases remain poorly characterized. To test whether features such as iris color, palpebral aperture, or skin type contribute to variability in performance, we conducted a cross-over experiment using two different eye trackers across a diverse range of participants to inform how future work can address equity and device design.

Forty healthy participants were assessed performing identical tasks using two eye trackers: a head-mounted tracker (HMT, Pupil Labs Core) and a screen-mounted tracker (SMT, Tobii Eye Tracker 4 C). Stimuli were presented across various positions of the screen in a controlled environment. Eye tracker performance was compared across participant features (iris color, palpebral aperture, skin type).

The SMT performed significantly better with non-brown irises and lighter skin types, and the HMT with brown irises and darker skin types. Overall, the HMT had a higher mean percentage of valid data; however, this came at the expense of reduced accuracy and precision.

This study identified variation in eye tracker performance depending on the tracker placement, and participant iris color or skin type. Both trackers exhibited similar gaze errors, but only the SMT showed acceptable precision via test-retest reliability. These data highlight the need for further research to ensure equitable application of eye trackers in diverse populations, and how iris color or skin type might affect gaze error and reliability.