Proceedings of the Annual Meeting of the Cognitive Science Society

Sustained attention is a dynamic process with rich temporalstructure. Eye-tracking provides a tool for capturing rich tem-poral data relevant to sustained attention, but extracting rele-vant insights from this rich data is nontrivial. This paper stud-ies eye-tracking data collected from children, aged 3-5, per-forming the TrackIt task, a visual object tracking paradigm de-signed for studying sustained attention development in youngchildren. Building on a hidden Markov model paradigm re-cently proposed for analyzing eye-tracking data with TrackIt,we explore characterizations of participant behavior, such ascontinuously maintaining attention on an object and transition-ing attention between objects, that provide richer insights thantask performance alone. In particular, our results suggest thatimprovement in TrackIt performance that accompanies devel-opment in this age range may stem more from improved abilityto return to task after distractions, rather than from improve-ments in ability to continuously maintain attention on the task.