November 24, 2025 | ACM VRST paper by Duke I3T Lab examines what types of visual stimuli distract users in AR
New paper led by I3T Lab's PhD students presented at ACM VRST 2025 by Duke I3T PhD student Sihun Baek: AR-TMT: Investigating the Impact of Distraction Types on Attention and Behavior in AR-based Trail Making Test.
This study was motivated by our experimental "struggles", in our work on developing algorithms for detecting that users of AR devices are distracted -- we had a great deal of difficulty when attempting to distract the users with highly salient stimuli, such as a visual of Duke's mascot, Blue Devil, running around and dancing right in users' field of view. Appearing exceptionally salient to an outside observer, such visuals were all but ignored by the users focused on their AR-supported real-world tasks.
This study investigates why that is the case. Via AR-TMT, an AR adaptation of a popular Trail Making Test, we examined the impact of three different types of distractions on user task performance and gaze patterns in AR. In a user study with 34 participants, we demonstrated that highly salient, "bottom-up", distractors that we were originally designing, engage users' attention originally, but not over the course of the entire test; they do not have a statistically significant impact on user performance on the task. By contrast, the presence of contextually relevant, "top-down", distractors led to a statistically significant reduction in performance on the task. The study also considered the impact of individual differences in attention control abilities on users' ability to resist different types of distractors.
Slides presented at ACM VRST 2025 in Montreal, Canada by Duke I3T PhD student Sihun Baek: