Clap Twice for Trust: How Framing Shapes Perceived Trustworthiness in VR-mediated Industrial Human-Robot Interaction

Matouš Jelínek, Pat Treusch

Keywords: Human-Robot Interaction, Framing, Perceived Trustworthiness, Trust Calibration, Overtrust, Mental Models, Industrial Robots

TL;DR: This study investigates how the verbal framing of an industrial robot - either as a low-capability prototype or a high-capability advanced technology - shapes per- ceived trustworthiness of the system.

Abstract: Trust is a crucial element in industrial human-robot interaction, yet little is known about how initial system descriptions shape perceived trustworthiness of the system, not only before use. Trans- parency about a system’s capabilities is crucial for reaching appro- priate reliance and perceived trustworthiness. This paper under- lines the importance of what users are told about a system before their first contact. This information shapes their expectations about the robot and early reliance. Thus, this study investigates how the verbal framing of an industrial robot—either as a low-capability prototype or a high-capability advanced technology—shapes per- ceived trustworthiness of the system. Utilizing a VR simulation of a poultry packing robot, participants (N=21) interacted with an iden- tical system. Notably, participants working with the high-capability framing rated the system as significantly more competent, showed higher reliability ratings, and also tended to express trusting be- havior sooner than those interacting with the system framed as a low-capability prototype. The results show that communicated competences affect perceived trustworthiness and trusting behavior in a simulated, VR-mediated industrial human-robot interaction. Based on those findings, we argue for the relevance of proper fram- ing, as, e.g., improper framing poses a risk of a miscalibration when system performance does not match its framing. Framing, thus, we argue, is an under-valued dimension of evaluating trustworthiness in (industrial) HRI.

Submission Number: 2