Abstract:
The majority of previous research on interrupted task performance has primarily focused
on the few actions surrounding the interruption (Trafton, Altmann, Brock, & Mintz, 2003;
Ratwani, McCurry, & Trafton, 2008; Monk, Trafton, & Boehm-Davis, 2008) and most of
these studies have been conducted in controlled laboratory environments (Eyrolle & Cel-
lier, 2000; Oulasvirta & Saariluoma, 2004; Gillie & Broadbent, 1989). The purposes of the
following studies are to expand the understanding of the disruptive e ects of interruptions
to include how interruptions impact the overall quality of the task being performed. Addi-
tionally, this research investigates how both quality and previously identi ed local e ects
generalize and can be understood in real-world environments, speci cally the classroom
and the
ightdeck. Previous theoretical frameworks used to explain interrupted task per-
formance in the lab (Altmann & Trafton, 2002, 2007; Oulasvirta & Saariluoma, 2004, 2006)
were combined with theories of decision making (Brunswik, 1952, 1955; Adelman, Miller,
Henderson, & Schoelles, 2003; Hogarth, 1987; Simon, 1955) to provide a more complete
picture of how interruptions a ect task quality in the real world. Two classroom studies
and one
ightdeck study were conducted to further our understanding of how interruptions
actually a ect performance in these environments at both a quality and local level. The
rst classroom experiment showed that certain tasks can be performed with interruptions
and distractions without showing a decrement in overall quality, although they might take
longer. This study also showed that expertise was not necessary to avoid the disruptive
e ects of interruptions. The second classroom study showed that quality decrements due
to interruptions are positively correlated with greater cognitive resource demand and that
local disruptions have little in
uence on the overall quality of performance. The
ightdeck
experiment showed that su cient expertise can help mitigate the disruptive e ects of in-
terruptions and that tools designed to facilitate resumption are generally liked by pilots,
but that the information on them needs to be carefully designed. Taken as a whole, the
results of these experiments suggest that theories of interrupted task performance need to
be broadened and augmented with theories from other domains in order to provide a more
complete view of how interruptions a ect task performance in terms of both local metrics
and more global measures such as quality.
