Predicting the Optimal Time for Interruption using Pupillary Data and Classification

AbstractIn the current study we present an air traffic control (ATC) task in which we measured pupil dilation to determine high and low workload periods. We manipulated working memory (WM) requirements across three conditions: a no WM condition, a passive WM condition in which information was accumulated, and an active WM condition in which information had to be added to and removed from WM. Results showed that no WM resulted in the least dilation, but that passive WM and active WM did not differ. Next, we used the pupil data to train a range of classifiers to differentiate between high and low workload periods to create an online task-independent interruption management system (IMS). The best predicting features were the median and a polynomial fit, going 12 second back in time. Our classifier was able to predict workload at high accuracy (77%).

Return to previous page