Effects of false alarms and miscues of decision support systems on human–machine system performance: a study with airport security screeners

Abstract

Decision support systems such as explosives detection systems for cabin baggage (EDSCB) at airport security checkpoints help screeners detect bombs by highlighting areas in X-ray images that might contain explosives. However, these systems are not perfect and can produce false alarms (i.e. alarm when no target is present) and miscues (i.e. a non-target is cued but the actual target is located elsewhere in the image). This study investigated the consequences of such automation errors in 112 professional airport security screeners who were supported by a simulated EDSCB with realistic X-ray images of cabin baggage. They had to detect bombs, guns, and knives under one of three experimental conditions: miscue prone, false alarm prone, or multiple failures (false alarms and miscues). Results showed that screeners missed more knives when the EDSCB provided miscues. We conclude that on-screen alarm resolution of EDSCB alarms in primary screening has the disadvantage that miscues can result in missing prohibited articles at airport security checkpoints. To avoid this problem, automated decision or clear instructions to screeners should be considered. Practitioner Statement: Airport security screeners inspect X-ray images of cabin baggage through visual search and decision making with the help of explosives detection system for cabin baggage screening (EDSCB). The present experiment addresses whether EDSCB miscues affect operator performance and whether miscues are a problem when conducting EDSCB on-screen alarm resolution.