Trust and Reliance on an Automated Target Recognition System for Underwater Mine Detection

Abstract

Automated target recognition (ATR) systems are typically designed to operate using a high sensitivity and a liberal decision criterion to reduce the risk of missing a target. The high number of false alarms that occur as a result of this design tend to lead to a decrease in operator trust and reliance. The purpose of this study was to determine how changing or informing a user of the false alarm rate of an ATR system affects the user’s level of trust and reliance in the system and the user’s performance during an underwater mine detection task. When not informed of the false alarm rate, the number of false alarms made by the system had a significant effect on the participants’ response bias. In addition, when informed of the false alarm rate, the participants had greater trust in the system and a more consistent response bias. These results suggest that informing a user of the false alarm rate of an automated system may positively influence the level of trust and reliance the user has in the aid.