AECT Handbook of Research

Table of Contents

24: Learning with technology: Using computers as cognitive tools

24.1 Introduction
24.2 Computers as cognitive tools
24.3 Why cognitive tools?
24.4 Overview of the chapter
24.5 Computer programming languages as cognitive tools
24.6 Hypermedia/ Multimedia authoring systems as cognitive tools
24.7 Semantic networking as cognitive tools
24.8 Expert systems as cognitive tools
24.9 Databases as cognitive tools
24.10 Spreadsheets as cognitive tools
24.11 Conclusions
24.12 A final word
Search this Handbook for:

24.8 Expert systems as cognitive tools

24.8. 1 What Are Expert Systems?

Expert systems are computer-based tools that are designed to function as intelligent aids to decision making in all sorts of tasks. Early expert systems, such as MYCIN, were developed to help physicians diagnose bacterial infections with which they were unfamiliar. Prominent expert systems have also been developed to help geologists decide where to drill for oil, fire fighters decide how to extinguish different kinds of fires, computer sales technicians how to configure computer systems, and employees to decide among a large number of company benefits alternatives. Problems whose solution includes recommendations based on a variety of decisions are good candidates for expert systems.

Expert systems have evolved from research in the field of artificial intelligence (AI). Al is a field of computer science and cognitive science that focuses on the development of both hardware innovations and programming techniques that enable machines to perform tasks that are regarded as intelligent when done by people. Intelligence is the capacity to learn, reason, and understand. Artificial means simulated. So, in other words, Al researchers and expert system builders attempt to develop programs that simulate the human capability to reason and to learn. Simulated means only imitating a real object or event. Al programs, including expert systems, may perform functions that resemble human thinking, such as decision making. In reality though, Al programs are just computer programs; they only imitate a human activity within a narrowly defined situation.

An expert system, then, is a computer program that simulates the way human experts solve problems-an artificial decision maker. For example, when we consult an expert (e.g., doctor, lawyer, teacher) about a problem, the expert asks for current information about our condition, searches his or her knowledge base (memory) for existing knowledge that can be related to elements of the current situation, processes the information (thinks), arrives at a decision, and presents his or her solution. Like a human expert, an expert system (computer program) is approached by an individual with a problem. The system queries the individual about the current status of the. problem, searches its own knowledge base (stored previously) for pertinent facts and rules that reflect the knowledge of an expert, processes the information, arrives at a decision, and reports the solution to the user.

Most expert systems consist of several components, including the knowledge base, inference engine, and user interface. The knowledge base consists of facts and rules that are programmed into the system by the designer., For example, an expert system designed to diagnose cars that will not start might include facts and rules such as:

Fact: Battery supplies voltage to ignition.
Fact: Ignition routes voltage to solenoid.
Rule: IF ignition is on,
      AND solenoid is not engaged,
      THEN battery is dead,
      OR ignition switch is faulty.

The expert system inference engine is programmed into the system and acts on the knowledge base and current problem data to generate solutions. It sets a goal and then collects information from the knowledge base in order to yield a solution. When the knowledge base does not contain enough information, the inference engine asks the user to supply the missing information. The inference engine continues to seek information until it is able to reach a solution which the system then presents to the user. The inference engine is the logic unit in the expert system. The part of the expert system that makes it a cognitive tool is the knowledge base. Building the knowledge base requires the designer to articulate the expertise that the system provides, not only in the form of facts but also rules. Identifying the causal relationships and procedural knowledge underlying a knowledge domain necessarily engages designers of expert systems in higher-order thinking.

24.8.2 How Are Expert Systems Used as Cognitive Tools?

When analyzing outcomes from building and using expert-systems, a distinction must be drawn between using an existing expert system rule base to support decision making and building an expert system. The former is the most common application. Although expert systems are primarily used in businesses as advisors that control production processes or in certain professions to assist practitioners in decision making, they also have many applications in education. Chandler (1994) describes the development of an expert-system design to help teachers plan science education lessons. Considerable research has focused on developing expert-system advisors to help teachers identify and classify learning disabled students (cf. Fuchs, 1992). Expert-system advisors have been developed to guide novices through the instructional development process (Tennyson & Christensen, 1991) or to assist students in selecting the correct statistical test (Karake, 1990; Saleem & Azad, 1992). With this type of application, professional knowledge engineers produce expert-system knowledge bases that are accessed by users when they need advice in making decisions (Bossinger & Milheim, 1993). However, simply using existing knowledge bases to get advice does not engage users as deeply as building a knowledge base to reflect their own thinking (Wideman & Owston, 1993). Querying a knowledge base to help solve a problem involves primarily comprehension of the problem and its factors; the application of some predetermined rules for solving the problem is often hidden from the user within the expert system itself

Expert systems can 'also function as cognitive tools (Kommers, Jonassen & Mayes, 1992). Trollip, Lippert, Starfield, and Smith (1992) believe that the development of expert systems results in deeper understanding because they provide an intellectual environment that demands the refinement of domain knowledge, supports problem solving, and monitors the acquisition of knowledge. Building expert systems requires the developer to model explicitly the knowledge of the expert (Starfield, Smith & Bleloch, 1990). This entails identifying declarative knowledge (facts and concepts), structural knowledge (the knowledge of the interrelationships of ideas in memory), and procedural knowledge (how to apply the former). In fact, building expert systems is one of the few formalisms for depicting procedural knowledge. Psychologists usually represent procedural knowledge as a series of IF-THEN rules (Gagne, 1985); such a representation mode is obviously well suited to expert-system codification. As learners identify the IF-THEN structure of a domain, they will tend to understand the nature of decision- making tasks better, and this deeper understanding should make subsequent practice opportunities more meaningful. This is not to suggest that the mere development of an expert system necessarily leads learners to acquire the compiled procedural knowledge of a domain. Students could correctly identify many of the IF-THEN rules involved in flying an airplane, but actually acquiring the procedural expertise to fly would still require extended practice opportunities in realistic performance settings.

When expert systems are used as cognitive tools, the roles of teachers and students change dramatically. Students as knowledge engineers assume a more active role in acquiring prerequisite knowledge and focusing and directing interactions with the teacher, who assumes the role of expert (Morrelli, 1990). This frees the teacher from having to motivate students and allows them to respond as an expert to student probing concerning the more demanding and interesting aspects of various problems. Students must analyze the knowledge domain (identifying outcomes, factors, and values for those factors) and then synthesize rules and rule sequences. Morrelli argues that interaction between active, self-directed learners and a supportive, articulate teacher is an excellent model for learning science. We agree.

24.8.3 What Research Supports the Use of Expert Systems as Cognitive Tools?

Much of the research with the use of expert systems has focused on teachers and students as users of predefined rule bases. For instance, students who used an expert system to select the most appropriate statistical analysis procedure were more accurate in their selections and also retained the information better than students who used traditional computer-assisted instruction (Marcoulides, 1988). Grabinger and Pollock (1989) used expert systems to direct students to evaluate their own projects. Students who generated their own feedback with the help of expert systems produced a greater number of criteria in subsequent exercises and favored the method to teacher-only feedback. As described earlier, using expert systems supplants (provides or substitutes knowledge that is not known) thinking and therefore does not necessarily engage users in thinking critically about the content they are studying.

The use of expert systems as cognitive tools is relatively recent. Trollip and Lippert (1987) found that the analysis of subject matter required to develop expert systems is so deep and so incisive that learners develop a greater comprehension of their subject matter. They reported that building expert system rule bases engages learners in analytical reasoning, elaboration strategies such as synthesis, and metacognition. Lippert (1988, 1989), among the early advocates of expert systems as cognitive tools, argued that asking students to construct small rule bases is a valuable method for teaching problem solving and knowledge structuring for students from sixth grade to adults. Not only do learners solve problems, they also engage in metacognitive reflection on their problem solving while constructing rule bases (Trollip & Lippert, 1988). Developing the knowledge base requires learners to isolate facts, variables, and rules about the relationships between content in a domain. Developing rule bases as a cognitive tool represents a constructivist application of expert systems (Jonassen, Wilson, Wang & Grabinger, 1993).

A small body of research has validated the use of expert systems as cognitive tools. Lai (1992) found that when nursing students developed medical expert systems, they developed enhanced reasoning skills and acquired a deeper understanding of the subject domain. Lippert (1988) described the development of rule bases to solve problems about forces by six freshmen physics students who used an expert-system shell to create questions, decisions, rules, and explanations pertaining to classical projectile motion. The students developed more refined, domain-specific knowledge due to greater degrees of elaboration during encoding and greater quantity of material processed in an explicit, coherent context, and therefore in greater semantic depth (Lippert & Finley, 1988). Students identified factors such as kind of force acting on an object (e.g., gravitational or centripetal), motion of the object (e.g., free fall, circular, or sliding), velocity of the object, and so on. The decisions that students defined were based on the laws that affect the motion and the formulas that should be applied. Students reported meaningful learning from evaluating their own thought processes, more enthusiasm for learning, and the learning of content that they were not expected to master.

Knox-Quinn (1992) reported that MBA (masters of business administration) students who developed knowledge bases on tax laws in an accounting course were consistently engaged in higher-order thinking, such as classifying information, breaking down content, organizing information, and integrating and elaborating information. All of the students who developed rule bases showed substantial gains in the quantity and quality of declarative and procedural knowledge and improved their problem-solving strategies. Students who built expert systems reasoned similarly to experts.

Like most cognitive tools, die research base on expert systems is very limited. However, with the increased interest in constructivist applications of expert systems and other computer tools, the research base should grow dramatically. We predict that future research will continue to verify the cognitive and metacognitive effects of learners functioning as knowledge engineers.

Updated August 3, 2001
Copyright © 2001
The Association for Educational Communications and Technology

1800 North Stonelake Drive, Suite 2
Bloomington, IN 47404

877.677.AECT (toll-free)

AECT Home Membership Information Conferences & Events AECT Publications Post and Search Job Listings