An Implicit Dialogue Injection System for Interruption Management

This paper presents our efforts in redesigning the conventional on/off interruption management tactic (a.k.a. "Do Not Disturb Mode") for situations where interruptions are inevitable. We introduce an implicit dialogue injection system, in which the computer implicitly observes the user's state of busyness from passive measurement of the prefrontal cortex to determine how to interrupt the user. We use functional Near-Infrared Spectroscopy (fNIRS), a noninvasive brain-sensing technique. In this paper, we describe our system architecture and report results of our proof-of-concept study, in which we compared two contrasting interruption strategies; the computer either forcibly interrupts the user with a secondary task or requests the user's participation before presenting it. The latter yielded improved user experience (e.g. lower reported annoyance), in addition to showing a potential improvement in task performance (i.e. retaining context information) when the user was busier. We conclude that tailoring the presentation of interruptions based on real-time user state provides a step toward making computers more considerate of their users.