Towards a Physiological Model of User Interruptability

User interruptability has become an important topic of study in Human-Computer Interaction (HCI). However, automatically determining the availability of users is still problematic. In this paper, we present a preliminary study of the use of physiological measurements for predicting user interruptability status. We measured Heart Rate Variability (HRV) and Electromyogram (EMG) signals whilst users performed a variety of tasks; including reading, solving word puzzles, mental arithmetic, typing, and playing a racing game. Results show high correlations for both HRV (r = 0.96) and EMG (r = 0.85) with user self-reports of interruptability. We combined these two measures into a single linear model, which predicted user interruptability with a combined r of 0.95, or 95% of the variance. Please note that our model, at this stage, describes interruptability across users rather than per individual. We describe an application of our findings in the Physiological Weblog, or ’Plog, a system that uses our model for automating online messaging status.