This project investigates how a robot's designated social role—specifically its priority—influences human attribution of blame and trust during coordination failures in multi-robot environments.
Important
DEMO STUDY DISCLAIMER: This is a "ready to run" study proposal. The results described in this report are projected based on comparable studies in Human-Robot Interaction (HRI) literature rather than newly collected empirical data.
As autonomous robots integrate into complex environments like hospitals, coordination failures are inevitable. This research explores how functional labels (e.g., "Medicine Delivery" vs. "Trash Disposal") serve as psychological frames that significantly alter how humans interpret and rate technical errors.
- H1: The Competence Cliff – High-priority robots (Medicine Delivery) will experience a significantly larger decline in perceived competence following a failure compared to low-priority robots (Trash Disposal) due to the violation of higher initial performance expectations.
- H2: Hierarchical Blame – Participants will attribute more responsibility for coordination failures to low-priority robots, reflecting a social bias where lower-status agents are expected to yield.
- H3: Role Rationalization – Qualitative justifications will cite the "urgency" of the high-priority robot as a mitigating factor, while the low-priority robot will be blamed due to a perceived lack of consequence.
The study utilizes a between-subjects experimental design (
- Condition 1 (Low Priority Failure): Both robots meet in a corridor; the trash robot fails to yield, causing a failure.
- Condition 2 (High Priority Failure): Both robots meet in a corridor; the medicine robot fails to yield, causing a failure.
- MDMT: Multi-Dimensional Measure of Trust (Reliability and Ethical trust).
- ROSAS: Robotic Social Attributes Scale (Competence, Warmth, and Discomfort).
- Blame Attribution: Analysis of open-ended responsibility descriptions.
The project applies Expectancy-Disconfirmation Theory to HRI. High-priority roles establish a high baseline of reliability; when this is violated, the resulting drop in perceived competence is significantly steeper than that of a low-priority agent.
- Empirical Collection: Transitioning from the proposed design to live data collection with the target sample of 128 participants.
- Physical Prototyping: Replacing animated stimuli with physical robots in controlled hospital mock-ups or VR environments.
- Expert Perspectives: Expanding the participant pool to include healthcare professionals to see how domain expertise influences blame attribution.
Priyesh Vashistha Queen's University, Kingston, Canada
Student Number: 2049905
The author thanks Dr. Pan for their guidance and supervision throughout this project as the Principal Investigator.