Exploring eye-tracking data for detection of mind-wandering on web tasks

Abstract

Mind-wandering (MW) is a phenomenon that affects most of us; it affects our interactions with information systems. Yet the literature on its effects on human-computer interaction is only scant. This research aims to contribute to establishing eye-tracking measures that could be used to detect periods of MW while a user is engaged in interaction with online information. We conducted a lab study (N = 30) and present an exploratory analysis of eye-tracking data with a focus on finding differences between periods of MW and not-MW. We found 12 eye tracking measures that were significantly different between periods of MW and not-MW. We also show promising classification results of the same variables. Our results indicate plausibility of using eye-tracking data to infer periods of MW.