# Improving of Robotic Virtual Agent's errors that are accepted by reaction and human's preference

**Authors:** Takahiro Tsumura, Seiji Yamada  
**arXiv:** [2304.00247](https://arxiv.org/abs/2304.00247)  
**Date:** 2023-04-01  
**Summary page:** https://www.summarizepaper.com/en/paper/2304.00247v2/

## Key points

- Study focuses on enhancing the relationship between humans and anthropomorphic agents through empathy
- Investigates impact of agent reaction and human preference on human empathy and acceptance of agent's mistakes
- Experiment involved intentional mistake made by agent during task with a human
- Four-condition, three-factor mixed design employed to identify significant factors for designing empathetic robotic agents
- Results showed that neither agent reaction nor human preference significantly affected empathy towards the agent
- Robotic agent's mistakes during the task were allowed by these factors
- Empathy for the agent decreased among participants when an error was made
- Insights provided into controlling impressions of robotic virtual agents' behaviors in society
- Understanding how humans perceive actions of anthropomorphic agents is crucial for positive relationships between humans and robots
- Designers can refine robotic agents based on factors like agent reaction and human preference to create more meaningful interactions

## Summary

In the study titled "Improving of Robotic Virtual Agent's errors that are accepted by reaction and human's preference," authors Takahiro Tsumura and Seiji Yamada explore ways to enhance the relationship between humans and anthropomorphic agents by focusing on empathy. The researchers investigate the impact of agent reaction and human preference on human empathy and acceptance of the agent's mistakes. The experiment involved a task between an agent and a human, where the agent intentionally made a mistake. To identify significant factors for designing a robotic agent that can promote human empathy, the study employed a four-condition, three-factor mixed design. The factors included agent reaction, selected agent's body color based on human preference, and pre- and post-task evaluations. Contrary to expectations, the results revealed that neither agent reaction nor human preference significantly affected empathy towards the agent. However, these factors did allow the robotic agent to make mistakes during the task. Notably, when the agent made an error, empathy for the agent decreased among participants. This study provides valuable insights into controlling impressions of robotic virtual agents' behaviors in society. As these agents become increasingly prevalent in various domains, understanding how humans perceive their actions is crucial for fostering positive relationships between humans and anthropomorphic agents. By considering factors such as agent reaction and human preference, designers can refine robotic agents to better align with users' empathetic responses to create more meaningful interactions between people and robots.