Abstract

When an automating application needs a user's input or has feedback or other information for that user, it typically engages the user immediately, interrupting the user's current task. To empirically validate why unnecessarily interrupting a user's task should be avoided, we designed an experiment measuring the effects of an interruption on a user's task performance, annoyance, and anxiety. Fifty subjects participated in the experiment. The results demonstrate that an interruption has a disruptive effect on both a user's task performance and emotional state, and that the degree of disruption depends on the user's mental load at the point of interruption. We discuss the implications of these results in terms of building a system to better coordinate interactions between the user and applications competing for that user's attention.

Citations