Processes in Diagnostic Reasoning: Information Use in Causal Explanations

This paper presents examples of the processes people use in generating qualitative solutions to complex diagnostic problems. We developed a high fidelity model of the electrical power system for the International Space Station, and presented scenarios of off-nominal, or fault, situations. The model interface provides rich information about functional organization of the power system, including system topology and graphs of variables changing over time. We presented two versions, with system information organized hierarchically or displayed in a single level. Novices, who were unfamiliar with the system to be diagnosed but technically sophisticated, were asked to study the scenarios and diagnose the fault situations. The particular scenario reported here was designed to be difficult, violate users ’ expectations, and require ‘thinking outside the box. ’ Users chose to view quantitative information frequently as part of developing qualitative, causal explanations. We found sophisticated reasoning processes and frequently correct explanations despite the difficulty of the task. Design successes and weaknesses are discussed. Keywords[trouble-shooting, diagnostic reasoning, problem solving, human computer interaction