Effects of blinded and nonblinded sequential human redundancy on inspection effort and inspection outcome in low prevalence visual search

Abstract

Human redundancy is often used in safety–critical domains to help protect against errors. For example, mammograms are read by two radiologists, or the dose of a drug is calculated by two physicians who perform the task redundantly one after the other. However, the expected reliability gain may be compromised by social loafing (SL), i.e., a reduction of individual effort caused by the team setting. In two laboratory studies, we therefore investigated whether different forms of human redundancy cause SL and impact task outcome. In each experiment, about 20 participants inspected images alone for rare targets, while 40 participants worked in teams of two, either performing the first or second inspection. We measured effort using inspection time and inspection area, and measured task outcome using the number of detected targets and false alarms. While there was no transfer of target marks in experiment 1 (blinded redundancy), the second inspectors in experiment 2 saw marks set during the first inspection (nonblinded redundancy). In experiment 1, data revealed no indications of SL but increased time, area covered, detections and false alarms for team-based relative to solo performance. In experiment 2, teams clearly adapted their inspection intensity. The second inspectors searched the images significantly shorter than the first inspectors. While detected targets did not differ between the two redundant groups, the second inspectors were found to produce significantly fewer false alarms than the first inspectors. However, the advantage of detected targets and area coverage still favored the teams. Thus, the principle of redundancy was effective in both forms of sequential redundancy, even if it led to different behaviors and perceptions at the individual level.