Content uploaded by Deborah Boehm-Davis
Author content
All content in this area was uploaded by Deborah Boehm-Davis
Content may be subject to copyright.
Dealing with Interruptions can be Complex, but does Interruption
Complexity Matter: a Mental Resources Approach to Quantifying
Disruptions
David M. Cades1 Nicole Werner1 J. Gregory Trafton2
dcades@gmu.edu nwerner@gmu.edu greg.trafton@nrl.navy.mil
Deborah A. Boehm-Davis1 Christopher A. Monk1
dbdavis@gmu.edu cmonk@gmu.edu
George Mason University1, Naval Research Laboratory2
Past work examining the effects of interruption complexity on primary task performance
has yielded quite mixed results. Some research suggests that more complex interruptions
lead to greater disruption of the primary task, while other studies have shown that
interruption complexity does not directly influence the amount of primary task disruption.
It is our hypothesis that interruption complexity, defined by the number of mental
operators required to complete a task as opposed to an intuitive sense of difficulty, does
affect primary task performance, such that interruptions requiring more mental operators
(more complex) lead to greater disruption than do less complex interruptions. Participants
performed a single primary task in conjunction with either a simple or complex
interruption. The complex interruption required more mental operators to complete than
the simple interruption. Our results showed that it took longer to resume the primary task
following a complex interruption than it did following a simple interruption. These results
suggest that more complex interruptions, as quantified by the number of mental
operations required, do indeed lead to greater primary task disruption.
INTRODUCTION
It is safe to say that interruptions are a part of
each of our lives on a daily basis. However, the
effects of these events are less clear. One body of
research has documented disruptive effects such as
decrements in task completion time (Eyrolle &
Cellier, 2000; Monk, 2004; Trafton, Altmann,
Brock, & Mintz, 2003) and accuracy (Cutrell,
Czerwinski, & Horvitz, 2001; Edwards &
Gronlund, 1998). A few studies, though, have
suggested that interruptions can actually aid
performance in certain contexts (Ratwani, Andrews,
McCurry, Trafton, & Peterson, 2007; Speier,
Vessey, & Valacich, 2003; Ziljstra, Roe, Leonora, &
Krediet, 1999). These findings suggest that specific
aspects of the tasks being performed and the types
of interruptions may play a major role in
determining how disruptive an interruption will be
(or whether it will be disruptive at all). One such
aspect, which has been explored with conflicting
results, is the complexity of the interruption.
Some studies have shown that increased
complexity leads to slower resumption times
(Hodgetts & Jones, 2005, 2006) and decreased
primary task accuracy (Gillie & Broadbent, 1989).
Increased resumption times were attributed to
interference caused by the existence of additional
goals. Decreased task accuracy was attributed to
increased processing and memory loads in the more
complex conditions.
However, other research found that more
complex interruptions did not lead to a decrement in
task performance. Increasing the complexity of
interruptions increased time on task slightly, but not
significantly (Ziljstra et al., 1999). Additionally, no
performance decrement in terms of accuracy was
noted. Eyrolle and Cellier (2000) found that highly
complex interruptions (ones with more items to be
processed) led to slightly higher error rates, but had
little other effects on task performance. Lastly,
Cades, Trafton, Boehm-Davis and Monk (2007)
found that participants were slower at resuming a
primary task when they were interrupted with either
of two difficulty levels of an n-back working
memory task (Lovett, Daily, & Reder, 2000) than
they were when the interruption consisted of a
simpler task (repeating a number aloud that was
read to them by the computer). However, there were
no resumption time differences between the two
difficulty levels of the n-back task.
This result was explained using a combination of
the Memory for Goals model (Altmann & Trafton,
2002) and an NGOMSL analysis (Kieras & Polson,
1985), The memory for goals model predicts that
the ability to resume depends primarily on
interruption length and the opportunity for rehearsal
during the interruption. Specifically, the greater the
number of mental operators a given task requires,
the less opportunity there is for rehearsal during that
task. In this specific experiment, the NGOMSL
analysis revealed that, even though one of the n-
back tasks seemed harder, both n-back tasks
required a similar number of mental operators. Thus
the opportunity to rehearse was similar in both n-
back conditions. Given that both conditions had the
same interruption length, both should (and did) have
similar resumption times. The further prediction
that the n-back tasks would lead to slower
resumption times than the number repetition
condition, which had the greatest opportunity to
rehearse, was also supported (Cades et al., 2007).
Although all of the previous work cited
attempted to manipulate interruption complexity,
few went so far as to identify the cognitive
mechanisms that made the various tasks more or
less complex. What was clear from these results is
that in order to truly and accurately manipulate
complexity it was essential to first identify (and
quantify) the mechanisms underlying complexity.
Cades et al. (2007) suggested that tasks which differ
in the number of mental operators should be used to
explore how the complexity of an interrupting task
affects people’s ability to resume. This paper aims
to help answer the question of whether a more
complex (as measured by the number of mental
operators) interruption will lead to greater
decrements in the ability to resume the primary task
over and above a simpler (fewer mental operators)
interruption.
EXPERIMENT
In this experiment, participants were required to
perform nine trials of a primary task with either a
simple or a complex interrupting task. The simple
interruption task required participants to decide
which of two two-digit numbers was higher and
respond by clicking on the button corresponding to
that number. The complex interruption was made up
of the simple interruption plus a few additional
steps and was designed to specifically require more
mental steps than the simple interruption. For this
complex interruption, participants first had to
choose the higher of two two-digit numbers (as in
the simple interruption), then add the two digits of
the higher number, determine whether this sum was
odd or even, and finally respond by clicking on
either the odd or even button.
METHOD
Participants
Twenty-four undergraduate students (23 women,
1 man) at George Mason University participated for
course credit. The average age of the participants
was 19. All participants were randomly assigned to
either the simple or complex interruption condition.
Task and Materials
The primary task (see Figure 1) involved
programming a simulated Video Cassette Recorder
(VCR) (Gray, 2000) interface to record a future
television show. The show information was
presented to participants on a 3x5 index card and
each trial ended once all of the information on the
card was programmed into the VCR.
Both interrupting tasks consisted of a pair of
two-digit numbers displayed on the screen below a
set of written instructions. The numbers were
randomly generated with the constraint that both
numbers had the same tens digit (e.g., 25 and 27, 78
and 72, 33 and 31, etc.) This was done to ensure
that all pairs of numbers required similar processing
in order to make the initial high/low judgment
required in both the simple and complex condition
and to ensure that the number of mental steps,
which was our primary manipulation of interest, did
not vary within interruption trials. A new pair of
numbers was displayed every 3.5 seconds during the
interruption. Each interruption lasted for 35 seconds
and participants were interrupted around 3 or 4
times per trial. Interruptions were triggered by a
random number of clicks ranging from fifteen to
twenty-two. This was done to prevent participants
from guessing when they would occur. However, as
some shows required more clicks to program than
others the number of interruptions varied slightly
between participants.
In line with predictions from the Memory for
Goals model (Altmann & Trafton, 2002), it was
hypothesized that both interruption conditions
would disrupt peoples’ ability to resume the primary
task. The complex interruption required completion
of the simple interruption plus additional steps,
which meant that the complex interruption
condition required more mental operations, and
presumably allowed less opportunity for rehearsal.
It was, therefore, also predicted that the complex
interruption condition would be more disruptive to
primary task resumption than the simple
interruption condition.
Design and Procedure
This experiment used a two level (simple or
complex interruptions) between-subjects design.
Prior to the experimental trials, participants were
trained on the VCR alone and the VCR with
whichever interruption corresponded to the
condition to which they were assigned. Participants
then performed nine trials of the VCR task with
interruptions, with each trial consisting of a
different television show to program. The
experiment lasted approximately one hour and
participants experienced somewhere between thirty
and forty 35-second interruptions across the
experiment.
When interrupted, the VCR screen disappeared
and the interruption screen with the instructions and
numbers was presented for the duration of the
interruption. After the 35 seconds, the interruption
screen disappeared and participants were returned to
the VCR task. This pattern continued until the
completion of each trial, at which time the VCR
program was reset.
Measures
Each mouse click was time-stamped and
recorded for all participants. Inter-action intervals
were calculated as the average time between clicks
on the primary task. Disruption was quantified using
a special inter-action interval called the resumption
lag (Altmann & Trafton, 2002), which was the time
between the end of the interruption and the first
action, or mouse click, back on the primary task.
Comparing this measure across experimental
conditions has been shown to accurately assess the
amount of disruption caused by a given interruption
(Altmann & Trafton, 2007; Monk, Boehm-Davis, &
Trafton, 2004; Ratwani et al., 2007).
RESULTS AND DISCUSSION
As a manipulation check, inter-action intervals
were compared to resumption lags across condition
(Figure 2). A repeated measures ANOVA
confirmed, that interruptions were indeed
disruptive, with resumptions lags (M = 3118.29, SE
= 134.83) significantly longer than inter-action
intervals (M = 875.81, SE = 29.04) collapsed across
Figure 1: The VCR Interface
Figure 2: Average click times by type across
conditions (Error bars are standard error of the
mean)
condition (F(1, 22) = 308.92, p < .001, MSE =
195,339.98, η2 = .93).
If the complexity of the interruption is an
important aspect in determining that interruption’s
disruptiveness, then we would expect the simple
interruption condition to lead to faster resumption
times than the complex interruption, supporting
earlier findings, which showed greater disruption
following more complex interruptions (Cades et al.,
2007; Gillie & Broadbent, 1989; Hodgetts & Jones,
2005, 2006). Alternatively, if interruption
complexity turned out to be either unimportant or
not central to the disruptive mechanism of
interruptions, then we would not expect to see
differences between the simple and complex
interruption conditions (Eyrolle & Cellier, 2000;
Ziljstra et al., 1999). A omnibus ANOVA revealed a
main effect of interruption complexity. Resumption
lags following a complex interruption (M =
3461.92, SE = 234.35) were significantly slower
than resumption lags following a simple
interruption (M = 2774.66, SE = 147.03) (F(1, 22) =
6.17, p < .05, MSE = 459217.27, η2 = .22) (Figure
3).
These results support the Memory for Goals
model by showing that increasing the number of
mental operators during the interruption and
reducing the opportunity for rehearsal leads to
increases in the disruptiveness of the interruption.
GENERAL DISCUSSION
In both this experiment and Cades et al. (2007),
greater interruption complexity led to greater
primary task disruption, when complexity was
evaluated in terms of number of mental operators.
As previous research has shown, simply because a
task seems more complex does not mean that it
necessarily requires additional mental operators
(Cades et al., 2007). In the case of findings (past,
current, and future) which claim that interruption
complexity does not influence disruptiveness, it will
be important to examine those tasks to see if the
more complex tasks actually require more mental
resources or if they simply seem harder. Using more
quantitative evaluation techniques will surely aid in
our abilities to predict the disruptive effects of
various types of interruptions.
Although these results offer strong support for
the Memory for Goals model’s predictions about
interruption complexity, it is also interesting to note
the relatively small effect size (Cohen, 1992). It
suggests that, although the difference between the
simple and complex conditions is most likely not
Figure 3: Resumption lags by interruption
difficulty (Error bars are standard error of the
mean)
due to chance, the actual magnitude of the effect is
moderate at best. Given the conflicting nature of the
previous research examining the relationship
between the complexity of an interruption and its
disruptiveness, it is not surprising that the effect
size is relatively small.
These data show that while the complexity of the
interruption is an important part in understanding its
disruptiveness, and that more complex interruptions
are more disruptive, this one aspect cannot fully
explain what causes different interruptions to be
more or less disruptive. Rather, this is just one
aspect of many that play a role in people’s abilities
to perform tasks with interruptions.
REFERENCES
Altmann, E. M., & Trafton, J. G. (2002). Memory for goals:
An activation-based model. Cognitive Science, 26,
39-83.
Altmann, E. M., & Trafton, J. G. (2007). Timecourse of
recovery from task interruption: Data and a model.
Psychonomics Bulletin and Review.
Cades, D. M., Trafton, J. G., Boehm-Davis, D. A., & Monk, C.
A. (2007). Does the difficulty of an interruption affect
our ability to resume? Paper presented at the 51st
Annual Human Factors and Ergonomics Society
Conference, Baltimore, Maryland.
Cohen, J. (1992). A Power Primer. Psychological Bulletin,
112(1), 155-159.
Cutrell, E., Czerwinski, M., & Horvitz, E. (2001). Notification,
disruption, and memory: Effects of messaging
interruptions on memory and performance. Paper
presented at the Human-Computer Interaction-
INTERACT 2001 Conference Proceedings,
Amsterdam.
Edwards, M. B., & Gronlund, S. D. (1998). Task interruption
and its effects on memory. Memory, 6(6), 665-687.
Eyrolle, H., & Cellier, J.-M. (2000). The effects of
interruptions in work activity: field and laboratory
results. Applied Ergonomics, 31(5), 537-543.
Gillie, T., & Broadbent, D. (1989). What makes interruptions
disruptive? A study of length, similarity, and
complexity. Psychological Research, 50, 243-250.
Gray, W. D. (2000). The nature and processing of errors in
interactive behavior. Cognitive Science, 24(2), 205-
248.
Hodgetts, H. M., & Jones, D. M. (2005). Interrupting problem
solving: Effects of interruption position and
complexity. Paper presented at the Past Reflections,
Future Directions: 40th Australian Psychological
Society Annual Conference, Melbourne, Australia.
Hodgetts, H. M., & Jones, D. M. (2006). Interruption of the
tower of London task: Support for a goal-activation
approach. Journal of Experimental Psychology-
General, 135(1), 103-115.
Kieras, D., & Polson, P. G. (1985). An Approach to the
Formal Analysis of User Complexity. International
Journal of Man-Machine Studies, 22(4), 365-394.
Lovett, M. C., Daily, L. Z., & Reder, L. M. (2000). A source
of activation theory of working memory: Cross-task
prediction of performance in ACT-R. Journal of
Cognitive Systems Research, 1, 99-118.
Monk, C. A. (2004). The effect of frequent versus infrequent
interruptions on primary task resumption. Paper
presented at the Proceedings of the Human Factors
and Ergonomics Society 48th Annual Meeting.
Monk, C. A., Boehm-Davis, D. A., & Trafton, J. G. (2004).
Recovering from interruptions: Implications for
driver distraction research\. Human Factors, 46(4),
650-663.
Ratwani, R. M., Andrews, A. E., McCurry, M., Trafton, J. G.,
& Peterson, M. S. (2007). Using peripheral
processing and spatial memory to facilitate task
resumption. Paper presented at the 51st Annual
Meeting of the Human Factors and Ergonomics
Society, Santa Monica.
Speier, C., Vessey, I., & Valacich, J. S. (2003). The effects of
interruptions, task complexity, and information
presentation on computer-supported decision-making
performance. Decision Sciences, 34(4), 771-797.
Trafton, J. G., Altmann, E. M., Brock, D. P., & Mintz, F. E.
(2003). Preparing to resume an interrupted task:
effects of prospective goal encoding and retrospective
rehearsal. International Journal of Human-Computer
Studies, 58(5), 583-603.
Ziljstra, F. R. H., Roe, R. A., Leonora, A. B., & Krediet, I.
(1999). Temporal factors in mental work: Effects of
interrupted activities. Journal of Occupational and
Organizational Psychology, 72(2), 164-185.
ACKNOWLEDGEMENTS
We thank Dr. Raja Parasuraman for comments on an earlier
version of this project, Dr. Patrick E. McKnight for his help
with statistical and methodological concerns, and members of
the interruptions research group at George Mason University
for their feedback and support.
