Cognition: Mental Representations, Problem Solving, and Decision Making - Foundations for Designing User-Centered Systems - page 188

models perform the task (Ritter et al. 1998 ), mirroring how behavior changes with
expertise in the task (Larkin et al. 1980 ).
Within human factors a body of work has developed that looks at decision
making in the wild—naturalistic decision making—like fire-fighting, and search
and rescue. Work on recognition-primed decision making (RPDM, Klein 1997 ),
argues that experts do not do problem solving, but that they recognize the situation
which directly leads them to the correct actions to take. This approach is consistent
with previous work in psychology, and with Rasmussen's work on levels of
expertise (Rasmussen 1983 ).
The work on naturalistic decision making encourages designers to provide
information to users to help them make decisions. They should also provide the
information in a way that supports recognition of the correct decisions, and
explicitly acknowledges that experts may move more quickly to a solution using
cues that relate the current situation to previously encountered situations.
6.4.6 Summary of Decision Making with Implications
for System Design
Many decision-support systems try to support the process of formalizing the
information contributing to the solution/decision. However, human problem
solving is not always based on logic and rationality in terms of using the infor-
mation to the full. This makes good decision support systems helpful, but difficult
to design because often the problem is not with the quality of the information
being presented to users.
Users do not make decisions the way you might think they will in several ways.
They will take more time the more choices they have. With good design, these
choices can help them and the choices can be made relatively quickly. With more
choices, or worse design, the time to choose might become large. Users will be
able to make more accurate and rapid choices if the choices (as described) match
their mental models and the names of the tasks they are attempting to perform. So,
provide time, reduce choices where appropriate, and have the choices match the
mental model and terms that the user has.
The decisions users make will be based on their mental models, and they will
often have different mental models than you. The cognitive architecture they can
bring to solve the problem will limit their ability to make rational choices. They
will often make choices consistent with previous choices because these choices
confirm their previous choices. They will base their decisions on what they can
retrieve from memory, and what they can retrieve first and easiest from memory is
not always the most useful information for making their decisions. So, if your
users often have to modify their behavior, you might wish to make repetitive
behavior less easy. If they choices are likely to be consistent, then options like
''Apply to all?'' are helpful.