Building Resilience in Human–Robot Collaboration: Affective and Cognitive Feedback from Robot for Human-Initiated Failure Handling

Paper Title

Affective and Cognitive Feedback from a Robot for Human-attributed Failure Handling

Publication Info

• Topic area: Human–robot collaboration (HRC) and failure handling in collaborative tasks.
• Keywords: Human–robot collaboration, human-attributed failure, affective feedback, cognitive feedback, teamwork quality, copresence, intimacy, socio-relational dynamics, collaborative robots, failure recovery.

Background and Problem

• Problem / challenge: Limited understanding of how robots should respond to human-attributed failures in collaborative tasks, as most research focuses on robot-initiated failures.
• Significance: Effective failure handling by robots can improve collaboration quality, maintain trust, and support socio-relational dynamics, making robots more effective teammates.
• Motivation and related work: Previous studies have explored failure handling in human–human collaboration (HHC) and robot-initiated failures in HRC but have not systematically examined robot responses to human-attributed failures. This paper addresses this gap by investigating the roles of affective and cognitive feedback in such scenarios.

Solution

• Proposed approach: A controlled experiment to evaluate the effects of affective and cognitive feedback on collaboration experience after human-attributed failures in HRC and HHC.
• Novelty:
  1. Empirical evidence showing that both affective and cognitive feedback improve collaboration experience after human-attributed failures.
  2. Demonstration of complementary roles of affective (emotional reassurance) and cognitive (task guidance) feedback in failure handling.
  3. Cross-context comparison revealing that humans apply similar social expectations to robots and human collaborators.
  4. Design implications for creating socially adaptive robots that integrate affective and cognitive feedback.
• Procedure and key techniques:
  • Participants (N = 60) performed a block-stacking task with either a humanoid robot (NAO) or a human collaborator.
  • Four scenarios were tested: success, failure with affective feedback (FAF), failure with cognitive feedback (FCF), and failure with no feedback (FNF).
  • Feedback was standardized across collaborators and evaluated using teamwork quality, copresence, and intimacy measures, complemented by qualitative post-interviews.

Results

• Concrete findings:
  • Affective feedback yielded the highest ratings for copresence and intimacy, while cognitive feedback improved teamwork quality.
  • Both feedback types significantly outperformed the no-feedback condition in all measures.
  • Participants attributed human-attributed failures to themselves in 76.67% of cases, validating the experimental manipulation.
• Advantage over baselines:
  • FAF and FCF significantly improved collaboration experience compared to FNF, with FAF providing stronger socio-relational benefits and FCF enhancing task-focused coordination.
• Experiments / evaluation:
  • Mixed factorial design (4 × 2) with within-subjects manipulation of feedback scenarios and between-subjects assignment of collaborator type (robot or human).
  • Quantitative measures (teamwork quality, copresence, intimacy) and qualitative post-interviews analyzed for insights.
• Limitations and future work:
  • Limited to humanoid robots; findings may not generalize to non-anthropomorphic systems.
  • Focused on two feedback types; future work should explore multimodal and personalized feedback strategies.
  • Conducted in controlled settings with scripted failures; real-world applicability requires testing in dynamic environments.
  • Sample lacked professional experience with robots; future studies should include diverse participant groups and cultural contexts.

Summary

This study demonstrates that both affective and cognitive feedback enhance collaboration experience after human-attributed failures in HRC and HHC. Affective feedback primarily supports emotional recovery and socio-relational dynamics, while cognitive feedback aids task-focused adaptation. The findings reveal that humans apply shared social expectations to robots and human collaborators, emphasizing the importance of socially intelligent behaviors in robotic systems. These results offer actionable insights for designing collaborative robots that integrate affective and cognitive cues to sustain effective teamwork, especially in failure-prone scenarios. Future research should explore adaptive feedback strategies in real-world, high-stakes environments.