When to Interrupt: A Comparative Analysis of Interruption Timings Within Collaborative Communication Tasks (original) (raw)

This study seeks to determine if it is necessary for a software agent to monitor the communication channel between a human operator and human collaborators to effectively detect appropriate times to convey information or "interrupt" the operator in a collaborative communication task. The study explores the outcome of overall task performance and task time of completion (TOC) at various delivery times of periphery task interruptions. A collaborative, goaloriented task is simulated via a dual-task where an operator participates in the primary collaborative communication task and a secondary keeping track task. User performance at various interruption timings: random, fixed, and humandetermined (HD) are evaluated to determine whether an intelligent form of interrupting users is less disruptive and benefits users' overall interaction. There is a significant difference in task performance when HD interruptions are delivered in comparison with random and fixed timed interruption. There is a 54% overall accuracy for task performance using HD interruptions compared to 33% for fixed interruptions and 38% for random interruptions. These results are promising and provide some indication that monitoring a communication channel or adding intelligence to the interaction can be useful for the exchange.