Abstract
Most of eye-tracking devices provide information not only about gaze location but also pupil size, which is not commonly used. Pupil dilation is known as indicative factor for cognitive load, memory operations, emotions, involvement, anticipation of the rivalry switch, pain. The reflection of mental processes in pupil size make the latter a good psychophysical metric, which can be measured distantly. However, pupil changes are also affected by amount of light, perceived by the eyes, also called light reflex. The balance between mental factors and light level and it's influence on pupil size is still a matter of discuss. In this paper, we used our eye-tracking dataset to check correlation between pupil size and luminosity level. Dataset represents the data of users performing everyday activities, such as web browsing, video watching, reading, social networking. It contains 19 hours of simultaneous recording of webcam, screen and eye-tracking data of 32 users. For each screen video frame was calculated average screen brightness (luminance). Then, smoothing and filtering methods we used to reduce eye-tracker noise. To estimate the effect of luminance changes on pupil size we apply time series analysis. The results show moderate correlation between PD and screen luminance and no correlation between PD and webcam frame luminance. However, the analysis of PD dynamics shows strong delayed response to changes in both lighting and screen luminance, which reaches its maximum in about 1 second after the change. The observed delay in pupil response should be taken into account in experiments where pupil size serves as an indicator of mental processes. Despite weak or zero correlation between current environment luminance and PD, changes in luminance have prolongated impact on dynamics of pupillary response.
Meeting abstract presented at VSS 2018