Cognition: Memory, Attention, and Learning - Foundations for Designing User-Centered Systems - page 138

2. Using closed loop control: performance is mediated using feedback on any
actions carried out, so conscious monitoring of behavior is required. Only one
activity can be carried out at one time. Learning how to drive, deliberate
walking on a rock field, editing a manuscript are examples, or learning how to
play a game by mapping commands to a controller.
Very few tasks can be performed completely using open loop control. Slips of
action typically occur when open loop control is being used instead of closed loop
control, such as:
• When users overlook information that affects behavior. For example, users that
do not note the mode of a word-processor or the location of the insertion point
and input and modify text in ways that they did not intend.
• When users continue performing a familiar activity even though they intended
to do something different. For example, clicking on the ''Yes'' button in
response to the prompt ''Do you really want to delete this file?''
• When users fail to correctly discriminate between relevant objects in the
world—performing an action on unintended objects. For example, trying to
click on an image icon on a web-based form.
Although it is not clear that we can design to always pre-empt these slips, we
can predict the kinds of circumstances when they occur. We can also recognize the
power of open loop behavior that can foil ''Do you really want to…?'' dialogues.
We will return to discuss slips of action and other types of erroneous behavior in
Chap. 10 .
5.3.5 Interruptions
Interruptions are becoming increasingly common in interfaces as systems become
more advanced and can be given tasks to perform asynchronously. There are also
other causes of interruptions, including colleagues, phones, and email systems that
beep when you have new messages.
In many ways interruptions can be seen as a secondary task, and hence the work
in this area is effectively a subset of the work on dual task performance, where the
user is trying to do two tasks at once. Interruptions are effectively the secondary
task, which is much less important, generally not regarded as part of the user's
main task, and is not under control of the user.
Interruptions do appear to cause real decrements in performance (Bailey and
Konstan 2001 ; McFarlane 1999 ) and users do not like them because they lead to
changes in affect (Bailey and Konstan 2001 ). McFarlane ( 1997 ) has attempted to
create a taxonomy of interruption types and their implications for design. His
results indicate that the choice of how to deal with interruptions will depend on the
importance and types of the main task and of the interruption. Dealing with