Do the Eyes Have It? A Review of Using Eye Tracking for Automation Trust Measurement

Abstract

ObjectiveWe conducted a literature review investigating the validity of eye tracking metrics appropriately representing trust in automation.BackgroundAs researchers grow interested in measuring trust in automation, there has been a need to find a reliable and accurate measurement tool. Many articles have measured automation trust using eye tracking, assuming that as trust increases, visual attention from eye tracking metrics decreases. Eye tracking is an attractive potential measure for its nonintrusive and objective nature.MethodIn this systematic literature review, we looked at studies that have tested the relationship between eye tracking and trust to determine its validity and reliability.ResultsAcross 22 articles that investigated the relationship between trust and eye tracking, only about half found a negative significant relationship, whereas the other half found no relationship at all.ConclusionThe relationship between automation trust and eye tracking is inconsistent and unreliable. A wide variety of trust and eye tracking metrics were used, but only about half of the papers found any kind of relationship. The relationship did not appear robust enough to prevail when different eye tracking and trust metrics were applied in various study designs.ApplicationAn effective eye tracking-trust relationship would be useful in various applications (e.g., autonomous driving). Further, this relationship is crucial when there is a clear distinction between attention allocated to automated components of a system (e.g., car display) and unrelated displays to allow for an easy separation of a location associated with high trust versus low trust.

Keywords: autonomous driving; eye movements; physiological measurement; tracking; trust in automation; usability/acceptance measurement and research.