Effect of Descriptive Information and Experience on Automation Reliance

Abstract

OBJECTIVE: The present research addresses the issue of reliance on decision support systems for the long-term (DSSLT), which help users develop decision-making strategies and long-term planning. It is argued that providing information about a system's future performance in an experiential manner, as compared with a descriptive manner, encourages users to increase their reliance level. BACKGROUND: Establishing appropriate reliance on DSSLT is contingent on the system developer's ability to provide users with information about the system's future performance. METHOD: A sequence of three studies contrasts the effect on automation reliance of providing descriptive information versus experience for DSSLT with two different positive expected values of recommendations. RESULTS: Study I demonstrated that when automation reliance was determined solely on the basis of description, it was relatively low, but it increased significantly when a decision was made after experience with 50 training simulations. Participants were able to learn to increase their automation reliance levels when they encountered the same type of recommendation again. Study 2 showed that the absence of preliminary descriptive information did not affect the automation reliance levels obtained after experience. Study 3 demonstrated that participants were able to generalize their learning about increasing reliance levels to new recommendations. CONCLUSION: Using experience rather than description to give users information about future performance in DSSLT can help increase automation reliance levels. APPLICATIONS: Implications for designing DSSLT and decision support systems in general are discussed.