Modifying Fault Diagnosis Strategies

Abstract

This study investigated the ability of trained diagnosticians to alter their strategies in situations where either time or testing must be kept to a minimum. Four subjects, extensively trained in previous experiments on fault diagnosis, participated in time bonus and testing efficiency bonus conditions. They isolated faults in networks of 25 nodes connected by links. In the time bonus condition, subjects were rewarded for the number of problems completed within a time limit. In the test bonus condition, subjects were rewarded for minimizing the number of tests used to diagnose a fault. Total diagnosis time was lower in the time bonus condition, and the number of tests was lower in the test bonus condition. In the time bonus condition, subjects tended to make initial tests more rapidly than they did in the test bonus condition, and the location of initial tests was relatively “shallow” in the network. The between-subject variability in diagnosis time was larger in the test bonus condition than in the time bonus condition. There was no significant difference for between-subject variability in testing efficiency. These results suggest two important conclusions regarding fault diagnosis. First, trained diagnosticians can rapidly adapt their diagnosis strategies to changes in their problem domain. Second, explicit payoffs can, in some cases, reduce the individual differences in fault diagnosis performance. This appears to be partly due to strategies becoming more similar under explicit payoffs. It is suggested that in future studies the use of explicit payoffs should be considered.