Two Classes of Problem-Solving Situations in Managing Computer Systems

摘要

Previous studies in complex domains such as power plants and air traffic control have described a wide range of behaviours not addressed by traditional psychological studies. This led to a field study of problem solving in anaesthesiology, using various methods of collecting behavioural data directly from the field while anaesthesiologists were doing their job. The methods used include direct observation, interviewing, and on-line verbal protocol recording with thinking aloud verbalisation. This paper describes findings from the field study and presents an analysis of how problem solving situations arise during the management of anaesthesia, a task similar to process control. Two classes of problem solving situations are identified: (1) managing the patient during non-critical but stressful and long lasting incidents-problem solving is characterised by competing goals and multiple hypotheses, and the memory load while keeping track of what is in the 'pipeline'; (2) deploying preventive and preparatory strategies-the problem solving is characterised by feedforward control of system (patient) status in the face of anticipated inputs to the patient, preparation of preconditions of necessary and contingency procedures, and anticipation of troublesome situations to be avoided. A categorisation scheme is proposed to classify problem solving situations according to how they arise: incident-induced and strategy-induced.