“But You Shouldn’t Blame Me”: A Cross-National Comparison of the Effects of Performance Failures and Trust Repairs in Human–Robot Interactions

摘要

To enhance human–robot collaboration and bolster sustainable human–robot relationships, researchers and designers have extensively investigated the effectiveness of trust repair strategies following robotic failures. Focusing on sociocultural factors in repair effectiveness, the current study examined cross-national differences in responses to three categories of robotic failures (logic, semantic, and syntax) and four types of trust repair strategies (internal-attribution apology, external-attribution apology, denial, and no repair). We compared trust repair effectiveness across a U.S. and South Korean sample, given their distinct history of technology usage and sociocultural norms regarding trust repairs. Additionally, the study explored four dimensions of blame attribution (locus, stability, external controllability, and personal controllability) as potential mediating mechanisms. Analyses of 743 observations from 330 American participants and 1,027 observations from 368 Korean participants underscored the importance of culturally adaptive robotics. Findings revealed that robotic failures across all categories had unexpectedly more negative impacts on Korean participants, accompanied by less benevolent attribution. While general repair outcomes were similar across nations, syntax failures highlighted notable differences. For Korean participants, internal-attribution apologies significantly outperformed denial, aligning with sociocultural norms. However, external-attribution apologies did not yield better outcomes for Koreans compared to Americans, diverging from interpersonal trust repair patterns. These results highlight a design challenge posed by two positive mediators of trust attitude, external and personal controllability.