[Paper Review] This robot stinks! Differences between perceived mistreatment of robot and computer partners

Zachary Carlson, Louise Lemmon+3 more|arXiv (Cornell University)|Nov 1, 2017

Social Robot Interaction and HRI14 references3 citations

TL;DR

This study investigates how people perceive mistreatment of robot versus computer partners in a collaborative setting. Using a between-subjects experiment with 80 participants, it found that observers felt significantly more sympathy and perceived greater emotional capacity in a humanoid robot (NAO) than in a computer agent when subjected to verbal aggression, indicating that robot embodiment intensifies perceived victimhood and emotional response to mistreatment compared to computers.

ABSTRACT

Robots (and computers) are increasingly being used in scenarios where they interact socially with people. How people react to these agents is telling about the perceived animacy of such agents. Mistreatment of robots (or computers) by co-workers might provoke such telling reactions. The purpose of this study was to discover if people perceived mistreatment directed towards a robot any differently than toward a computer. This will provide some understanding of how people perceive robots in collaborative social settings. We conducted a between-subjects study with 80 participants. Participants worked cooperatively with either a robot or a computer which acted as the "recorder" for the group. A confederate either acted aggressively or neutrally towards the "recorder." We hypothesized that people would not socially accept mistreatment towards an agent that they felt was intelligent and similar to themselves; that participants would perceive the robot as more similar in appearance and emotional capability to themselves than a computer; and would observe more mistreatment. The final results supported our hypothesis; the participants observed mistreatment in the robot, but not the computer. Participants felt significantly more sympathetic towards the robot and also believed that it was much more emotionally capable.

Motivation & Objective

• To examine whether people perceive mistreatment of a robot differently than mistreatment of a computer in a social, collaborative context.
• To investigate whether the physical embodiment of a robot influences observers' perceptions of emotional capacity and sympathy toward the agent.
• To explore the social and psychological implications of mistreating robots in human-robot interaction (HRI) environments.
• To compare observer reactions to verbal aggression toward a robot versus a computer, controlling for functional role and task context.
• To contribute to understanding the moral and social boundaries of human-robot relationships in cooperative work settings.

Proposed method

• Conducted a between-subjects experimental study with 80 participants divided into four conditions: robot aggressive (RA), robot neutral (RN), computer aggressive (CA), and computer neutral (CN).
• Participants collaborated with a partner (either a NAO robot or a computer) in a simulated data-entry task, with a confederate acting aggressively or neutrally toward the partner.
• Measured observer reactions using a 12-item survey assessing perceived emotional capacity, sympathy, and similarity to self.
• Used ANOVA and post-hoc tests to compare mean responses across conditions, with significance set at p < .05.
• Controlled for potential confounds such as voice differences, gender distribution, and technical issues with the robot (e.g., malfunctioning eyes).
• Analyzed data for gender differences and technical failure impacts, finding no significant effects on outcomes.

Experimental results

Research questions

• RQ1 Do observers perceive mistreatment of a robot as more socially unacceptable than mistreatment of a computer in a cooperative task?
• RQ2 Does the humanoid embodiment of a robot increase observers’ perceived emotional capacity and sympathy compared to a computer agent?
• RQ3 Are there measurable differences in how people react to verbal aggression toward a robot versus a computer, despite identical behavioral stimuli?
• RQ4 To what extent does the robot’s physical form influence perceptions of its capacity to feel emotional harm?
• RQ5 How do observer perceptions of mistreatment differ when the agent is a robot versus a computer, particularly in terms of empathy and perceived autonomy?

Key findings

• Participants perceived the robot as significantly more emotionally capable than the computer, with a p-value of .001 for the emotional capacity comparison between robot and computer conditions.
• Observers expressed significantly higher levels of sympathy toward the robot when it was mistreated, compared to the computer, supporting the hypothesis of increased emotional response to robot mistreatment.
• The robot condition (RA) showed a statistically significant increase in perceived emotional capacity and sympathy compared to the computer aggressive (CA) condition, with effect sizes indicating strong practical significance.
• Despite technical issues with the robot’s eyes during some sessions, no significant difference in participant responses was found between sessions with and without the malfunction, suggesting robustness of the core findings.
• No significant gender differences were found in observer responses, indicating that the results were consistent across male and female participants.
• The control conditions (RN and CN) showed no significant differences in most measures, except for familiarity, which was expected due to the novelty of the robot.