“Sorry, it was my fault”: Repairing trust in human-robot interactions

The current study develops a three-fold categorization (i.e., logic, semantic, and syntax failures) for technical failures that are commonly observed in human-robot interactions (HRI), and investigates it along with four trust repair strategies: internal-attribution apology, external-attribution apology, denial, and no repair. The 743 observations conducted through an online experiment reveal some nuances in participants' perceprions of competence- and integrity-based trust violations, which may reflect ontological differences between humans and machines. The analysis also shows significant main effects of failure types and repair methods on HRI-based trust.