How human-automation interaction experiences, trust propensity and dynamic trust affect drivers’ physiological responses in conditionally automated driving: Moderated moderated-mediation analyses

Although the human-automation interaction experiences (i.e., the takeover request [TOR] lead time and system usage time) have been shown to play an important role in the physiological responses of humans driving automated vehicles (AVs), little is known about the mediating and moderating mechanisms underlying these relations under different trust propensity levels. To fill this gap, we conducted an experimental study in a simulator with 34 licensed drivers involving a series of takeover events. The self-reported trust ratings, drivers’ galvanic skin response (GSR), heart rate (HR), and monitoring ratio were collected and analyzed. The results show that, after drivers experience TORs, dynamic trust mediates the effects of the human-automation interaction experiences on the GSR and monitoring ratio during the following non-takeover periods. Additionally, high trust propensity weakens the impact of the TOR lead time on dynamic trust, while long system usage time strengthens the relations between dynamic trust and the GSR, HR. Moreover, the TOR lead time moderates the relation between dynamic trust and the monitoring ratio. Specifically, this significant relation only exists during non-driving-related tasks after experiencing the 8 s and +∞ s TOR lead time but not after experiencing the 4 s TOR lead time. These findings could provide additional support for the design of the adaptive vehicle-driver interface and onboard soothing technology in conditionally AVs to calibrate trust and improve driving experiences.