Redundant Automation Monitoring: Four Eyes Don't See More Than Two, if Everyone Turns a Blind Eye.

Abstract

In safety-critical and highly automated environments, more than one person typically monitors the system in order to increase reliability. We investigate whether the anticipated advantage of redundant automation monitoring is lost due to social loafing and whether individual performance feedback can mitigate this effect. In two experiments, participants worked on a multitasking paradigm in which one task was the monitoring and cross-checking of an automation. Participants worked either alone or with a team partner on this task. The redundant group was further subdivided. One subgroup was instructed that only team performance would be evaluated, whereas the other subgroup expected to receive individual performance feedback after the experiment. Compared to participants working alone, those who worked collectively but did not expect individual feedback performed significantly less cross-checks and found 25% fewer automation failures. Due to this social loafing effect, even the combined team performance did not surpass the performance of participants working alone. However, when participants expected individual performance feedback, their monitoring behavior and failure detection performance was similar to participants working alone and a team advantage became apparent. Social loafing in redundant automation monitoring can negate the expected gain, if individual performance feedback is not provided. These findings may motivate safety experts to evaluate whether their implementation of human redundancy is vulnerable to social loafing effects.