Trust Calibration in Older Adults’ Human–Robot Interaction: A Dual-Path Model of Cognitive and Emotional Trust

As service robots are increasingly deployed in eldercare settings, trust calibration following robot errors has become a core issue affecting the effectiveness of human-robot collaboration. Focusing on older adult users, this study systematically investigates the joint influence mechanism of error situations and explanation strategies on trust calibration through a 2 (error severity: high vs. low) × 2 (transparency: high vs. low) mixed-design experiment, and further tests a cognitive–emotional dual-path model of trust calibration. Using video-based scenario simulation combined with scale-based measurement, the study was conducted with 80 older adults. Results show that high-severity errors significantly undermine both cognitive trust and emotional trust, with a more pronounced detrimental effect on cognitive trust. Transparency interventions primarily facilitate trust repair through the emotional trust pathway, with no significant direct effect on cognitive trust. The study further reveals that transparency is more effective in repairing trust under low-severity error conditions, and that individuals' technology anxiety significantly attenuates the repair efficacy of transparency on the cognitive pathway. These findings not only provide preliminary empirical evidence for a dual-path dissociation pattern in older adults' trust repair—characterized by emotional priority and cognitive lag—but also offer an important empirical basis and design directions for developing age-appropriate service robots with context-awareness, emotional interaction, and personalized adaptation capabilities.