Implications of Cognitive Theory for Knowledge Acquisition
Sallie E. Gordon
Knowledge acquisition for expert systems is a subclass of any instructional situation whereby knowledge must be externalized from a human expert and transferred to one or more "systems." These systems have historically been other people, as in education and training. However, with the advent of artificial intelligence, the systems are more frequently becoming computers. There are many issues that cut across all types of knowledge acquisition. However, two key questions are, (a) What is the nature of the knowledge and/or skills used by an expert? and (b) What are the implications of the nature of expertise for methods of transferring the knowledge or skill?
In knowledge acquisition for expert systems, the traditional approach has been to develop elicitation methods that are derived from, or strongly related to, the knowledge base being developed. However, a better means of developing an artificial knowledge base might be first to address the question of the basic nature of expert knowledge and skill, and then to develop techniques appropriate for each type of knowledge. We have learned from educational research that the nature of knowledge has important implications for the development of effective methods of instruction (e.g., Anderson, 1989; Dillon & Sternberg, 1986; Gagne & Briggs, 1979; Glaser, 1984, 1989; Resnick, 1981; Scandura, 1977). One might assume that this holds for transfer to artificial as well as biological systems.
The goal of this chapter is to present theory and research from cognitive psychology relevant to the two questions posed above. I will address the first question by describing a particular viewpoint whereby development of expertise revolves around changes in declarative and procedural knowledge, and then by presenting empirical evidence relevant to this view. I will address the second question by outlining some implications of the theoretical viewpoint presented for designing and choosing expert system knowledge acquisition methods.
Expertise has become a popular research topic in recent years, resulting in a growing body of literature on this subject. Some researchers have focused on the content and form of expert knowledge structures. Much of this research has consisted of identifying differences between novices and experts. (See Chi, Glaser, & 2, 1988, for reviews of some of this work.) For example, research has shown that experts have knowledge structures that are more detailed as well as more organized than those of novices ( Chase, 1983; Chi, Glaser, & Rees, 1982; Larkin, 1979; Reif & Heller, 1982); that novices tend to organize and perceive problems at a more concrete level whereas experts rely more on abstract concepts ( Chi, Glaser,