Would you Trust a Robot that Distrusts you?

Abstract

Trust is an important antecedent to successful human-robot collaboration and is conceptualized as a reciprocal process. The current study evaluated whether a robot could influence participants' trust levels through the disclosure of its own trust. We conducted a 3 (trust disclosure) × 2 (trust repair) between-subjects experiment in which N = 194 participants sorted boxes together with a humanoid robot that either disclosed distrust, high trust, or no trust information (control condition) towards the participant in a virtual warehouse environment. Additionally, participants either received an apology or did not (control) in response to a subsequent robotic error. Our analyses so far reveal that the disclosure of distrust from a robot towards a human indeed negatively impacts human trust in the robot. However, the disclosure of high trust did not significantly differ from no disclosure. Our findings contribute to a better understanding of reciprocal trust calibration and highlight the importance to consider trust as a reciprocal interaction. Moreover, our findings illustrate that disclosing distrust by a robot might serve as a promising strategy to proactively mitigate overtrust.