Effects of Explanations by Robots on Trust Repair in Human-Robot Collaborations

Trust can be undermined when robots deviate from human expectations in human-robot collaborations (HRC), whereas proactive trust repair strategies can be employed by robots to mitigate the negative impacts of trust violations. Drawing from the attribution theory, the current study investigated the effects of four types of explanation strategies on trust repair, namely internal-low integrity attribution, internal-low ability attribution, external attribution, and no repair. This study involved 149 university students in an online, between-subjects experiment scenario to simulate a situation in which a robot violates integrity-based trust in HRC. Participants’ trust in the robot was measured before and after the trust violation across four time points. The results showed that external attribution outperforms internal-low ability attribution, internal-low integrity attribution, and no repair to restore trust. The explanatory strategies that induce individuals to attribute low integrity have the most negative impact on trust repair.