THE THEORY OF COGNITIVE FIT - Human-Computer Interaction and Management Information Systems: Foundations - page 161

processes problem solvers use during problem solving. Vessey and Galletta (1991) provided an
example of investigating knowledge structures when they varied the sequence of presentation of task
and problem representation in order to determine the importance of spatial and symbolic problem
solving elements to problem solvers' mental representations. Dennis and Carte (1998) and Tuttle
and Kershaw (1998) provide examples of investigating problem-solving processes. Dennis and
Carte assessed the processes used by examining participants' worksheets ex post, while Tuttle and
Kershaw (1998) informed participants of the process that they should use and checked the effec-
tiveness of the manipulation by having participants respond to a set of questions.
Fourth, our recent identification of the importance of dual-task problem solving in the IS arena
and its application to both well-structured and ill-structured tasks suggests that this is potentially
a very fruitful way to look at a number of troublesome issues that have to date proved intractable.
This theoretical formulation seems particularly important in ill-structured tasks where dual-task
interference occurs when the contributing tasks do not manifest cognitive fit.
CONCLUSIONS
This research examined how the theory of cognitive fit has been applied in the area of human-
computer interaction since its introduction in 1991. In particular, the research sought to investigate
whether its promise to serve as one aspect of a general theory of problem solving has been substan-
tiated over that timeframe. In order to do this, we classified relevant studies according to whether
they tested concepts found in the two foundational papers that presented the theory (Vessey, 1991,
1994), whether they applied the theory to new domains of investigation, and whether they used
dimensions of fit other than the spatial-symbolic dichotomy addressed in the foundational studies.
We found that such studies largely support the concepts originally presented, although there are
some areas that need further investigation, particularly the accuracy/time trade-offs that can occur in
response to performance requirements and when solving more complex symbolic problems that
result in strategy change from more effortful analytical processes to more parsimonious perceptual
processes. Under the latter circumstances, when presented with tasks of “medium” complexity, it
appears that problems may be solved more accurately using tables, but more quickly using graphs.
The theory of cognitive fit has been applied quite extensively to two domains in which considerable
amounts of data are required for problem solving, those of human judgment and multi-criteria deci-
sion making using maps. The theory enjoys substantial support in these domains. Finally, the theory
has been applied to dimensions of fit other than in the spatial-symbolic task environment in which it
was formulated. One of the most prevalent forms of this type of fit that we identified was associated
with the complexity of the relationship between the task and the problem representation. This notion
of fit addresses the degree of match between task and problem representation rather than the famil-
iar match-mismatch dichotomy typical of graph/table problem solving.
This paper also presents two recent theoretical developments that promise to significantly open
up the basic area to further investigations. One development separates the problem solver's inter-
nal representation of the problem domain from the mental representation formed to solve the prob-
lem, thus facilitating investigations of other mechanisms underlying fit. The second development
extends the theory of cognitive fit to problem solving that requires two sub-tasks to be addressed
concurrently in arriving at the solution to the problem, a common occurrence in IS problem solv-
ing where, at a minimum, knowledge of both the application and IS domains influence problem
solving. Current research suggests that dual-task problem solving is moderated by the degree of
structure in the task under investigation.