Human information processing: Mental models and expert system interface applied across task domains

Expert Systems (ES) are being incorporated into an ever-larger number of integrated computer systems. The effects of ES interface on human information processing and the ability of the human-computer interaction to be influenced by the ES inference explanations are essential components of the human computer system interaction.; Previous studies have focused on the use of diagnostic tasks to examine human cognitive processes. In this study, the user mental model and the effects of ES inference explanations will be assessed across a number of task domains.; The results of the study have shown that an individual's cognitive preference (left dominant tactics versus right dominant tactics) does not effect their ability to interact with an expert system inference explanation. Individuals with poor mental model formation will not perform up to the level of achievement of individuals with well formed mental models, confirming the findings of Rook and Donnell (1993) over a broader set of task domains. Additional research findings demonstrate that graphic/visual mental model formation can improve subject's performance over textual presentation and achieve a higher number of correct answers.; Research findings in the area of information presentation once mental models were formed, showed that there was significant observable difference in subject performance between graphic versus textual inference explanation. After analysis of subject's experiment Test Run Time, Total Correct Answers, and Number of Matched Critical Nodes the results varied by dependent variable with no clear conclusion. The “Total Correct Answers” dependent variable does provide the best measure of overall performance; with these criteria, the textual inference explanation subjects out performed the graphic inference explanation subjects.; An analysis of the Inference explanation over task domains, which include Planning, Scheduling, Interpretation and Data Driven showed that there was no significant difference in the subject's performance over the various domains. The analysis did confirm the previous finding that the type of learning presented for mental model formation does significantly influence the subject's performance. Therefore, this confirmed one of the experimental hypotheses, mental model preference will not be affected by the task domain in which the subject is operating.