Providing Evidence of a Multiple-Process Model of Trust in Automation

This study focuses on the effects of human responses to computer automation aids. Previous research has shown that different types of automation errors (false alarms and misses) affect human trust in different ways. False alarms tend to negatively affect operator compliance, whereas misses tend to negatively affect operator reliance. Participants were asked to determine whether an enemy target was present or absent in a series of images, a task similar to what a UAV operator might be asked to perform. A diagnostic aid provided recommendations before participants viewed each image. Reliability and type of automation error were manipulated in order to provide data to determine which of four theoretical models is most accurate. Analyses provided conclusive evidence that a multiple-process theory of operator trust is the only model which accurately explains behavior outcomes in this type of situation. A discussion of theoretical and practical implications of this finding is included.