Mental Models and the Acquisition of Expert Knowledge
Robert M. Schumacher and Mary P. Czerwinski
There is a well-known fable about six learned blind men and an elephant. In this fable, each blind man encounters the elephant from a different perspective and develops his own opinion as to the nature of this beast: One grabs his ear and thinks it is a fan, another feels his tail and believes it to be a rope, and so on. The fable concludes with each blind man steadfastly holding his own narrow opinion. Among the many metaphorical elephants in cognitive science, one that stands out is the concept mental model. In this paper we examine this elephant by exploring the multiplicity of claims made about mental models. We do not pretend to be any less blind than other researchers who have tackled these problems, nor do we claim to have the only answers to many of the questions we raise. What we hope is that by taking a fresh look at mental models research we can provide some unification of concepts and point to promising methodologies and applications. We hope that by examining the issues surrounding mental models we raise the awareness of those who would try to use mental models in construction of knowledgebased systems (see Davis, 1982).
We have organized this chapter in the following way. First we discuss definitions of mental models, followed by a detailed discussion of several properties of mental models. Next we consider four sets of research that, we believe, provide interesting evidence for mental models. Last, we present a theoretical account of how mental models are acquired.
The concept of a mental model is one that many cognitive scientists and human factors specialists use in explaining certain behaviors, but hardly one that everyone would accept. As a working definition we will define a mental model as a collection of knowledge about a physical device, system, or process. Some efforts have even been made to develop a logical proof of the existence of mental models; for instance, Conant and Ashby ( 1970) argue that the brain must "model" the environment in order to survive. The two main sources of research on mental models come from cognitive science and human factors psychology. In cognitive science, researchers have used the mental model construct to explain such things as how people learn to use in-fix notation calculators (e.g., Halasz & Moran, 1983; Young, 1983); operate simulated devices (e.g., Kieras & Bovair, 1984; Schumacher & Gentner, 1988); and understand physical processes (e.g., McCloskey, 1983). In human factors, mental models have been used to describe how people learn computer systems (e.g., Carroll & Olson, 1988; Rumelhart & Norman, 1981); comprehend programming concepts (e.g., Cooke & Schvaneveldt, 1988; Mayer & Bayman, 1981); interact with everyday devices ( Norman, 1988); and perform complex process control tasks (e.g., Moray, 1986, 1988; Veldehuysen & Stassen, 1977; Wickens & Kessel, 1979, 1980).
-61-