Scholarlink

<p>A recent study found that people sometimes reject overly generous offers from humans because they imagine “phantom costs” in the absence of a clear rationale for such generosity. This study explores whether similar perceptions emerge when interacting with a robot (HRI). To test this, a human or robot agent—presented as either screen-embodied or physically-embodied—offered participants a cookie with or without $2. Participants then decided whether to accept the offer and justified their choice. Results showed that participants perceived phantom costs and rejected the offer more often when money was also offered. Overall, participants accepted the offers more often when made by a robot than a human. When the agent was a human, acceptance rates were higher in the physically-embodied condition than in the screen-embodied condition. No significant difference was found across these two conditions when the agent was a robot. These findings suggest that people can treat robots (and humans) as social agents with hidden intentions, especially when explanations provided about an action seem insufficient. This provides insights into how robots should be designed to facilitate HRI. While embodiment appears to influence the perception of phantom costs, further research is needed to fully understand these dynamics and their implications for robot design and HRI research.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Too Good to Be True: People Reject Free Gifts from Robots Because They Infer Bad Intentions

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月30日

查看全文
  • Benjamin Lebrun,
  • Andrew Vonasch,
  • Christoph Bartneck

摘要

A recent study found that people sometimes reject overly generous offers from humans because they imagine “phantom costs” in the absence of a clear rationale for such generosity. This study explores whether similar perceptions emerge when interacting with a robot (HRI). To test this, a human or robot agent—presented as either screen-embodied or physically-embodied—offered participants a cookie with or without $2. Participants then decided whether to accept the offer and justified their choice. Results showed that participants perceived phantom costs and rejected the offer more often when money was also offered. Overall, participants accepted the offers more often when made by a robot than a human. When the agent was a human, acceptance rates were higher in the physically-embodied condition than in the screen-embodied condition. No significant difference was found across these two conditions when the agent was a robot. These findings suggest that people can treat robots (and humans) as social agents with hidden intentions, especially when explanations provided about an action seem insufficient. This provides insights into how robots should be designed to facilitate HRI. While embodiment appears to influence the perception of phantom costs, further research is needed to fully understand these dynamics and their implications for robot design and HRI research.