Previous studies of multitasking have highlighted the importance of cognitive load in interruptibility by showing that forced interruptions are least disruptive when cognitive load is low, and also that users prefer to address interruptions at low-load points when given a choice. We present an empirical study that uses a ringing-phone scenario to examine how users manage deferrable interruptions in the presence of varying time constraints. We found that while cognitive load did influence multitasking as expected, the time constraints placed on the user also had a significant impact. In particular, we observed three distinct strategies for addressing interruption: the expected strategy of switching at low-load points, but also two other strategies of continuing on after a low-load point or giving up at a high-load point. The presence of the latter two strategies strongly suggests that users can adapt their multitasking behavior with respect to the time constraints of the interrupting task.

Recent research has found that forced interruptions at points of higher mental workload are more disruptive than at points of lower workload. This paper investigates a complementary idea: when users experience deferrable interruptions at points of higher workload, they may tend to defer processing of the interruption until times of lower workload. In an experiment, users performed a mail-browser primary task while being occasionally interrupted by a secondary chat task, evenly distributed between points of higher and lower workload. Analysis showed that 94% of the time, users switched to the interrupting task during periods of lower workload, versus only 6% during periods of higher workload. The results suggest that when interruptions can be deferred, users have a strong tendency to ''monotask'' until primary-task mental workload has been minimized.

While the collection of behavioral protocols has been common practice in human-computer interaction research for many years, the analysis of large protocol data sets is often extremely tedious and time-consuming, and automated analysis methods have been slow to develop. This paper proposes an automated method of protocol analysis to find canonical behaviors --- a small subset of protocols that is most representative of the full data set, providing a reasonable "big picture" view of the data with as few protocols as possible. The automated method takes advantage of recent algorithmic developments in computational vision, modifying them to allow for distance measures between behavioral protocols. The paper includes an application of the method to web-browsing protocols, showing how the canonical behaviors found by the method match well to sets of behaviors identified by expert human coders.