Peoples Biased Decisions to Trust and Cooperate with Agents that Express Emotions
Abstract
Research in the behavioral sciences shows that emotion expressions impact people's decisions to trust and cooperate with others in situations where self and collective interests collide. Building on such findings, computer scientists have shown that emotion expressions in agents can also impact people's decision making. However, recent findings in neuroeconomics reveal that people systematically show different behavior and brain activation patterns in decision making tasks with computers, when compared to humans. These findings suggest a bias people might have with respect to autonomous agents and, in particular, agents that express emotions. To clarify this, the paper presents a novel experiment where participants engaged in the iterated prisoner's dilemma, for clear financial stakes, with counterparts, either agents or humans, that showed facial displays of emotion that were compatible with a cooperative (e.g., smile after mutual cooperation) or competitive (e.g., smile after exploiting the participant) goal orientation. The results showed that participants cooperated, as expected, more with cooperative than competitive counterparts but, also revealed that people trusted and cooperated more with a human that showed cooperative displays than an agent that showed the exact same displays. We discuss implications of such a bias for trust and cooperation in human-agent interaction.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2018
- Accession Number
- AD1159390
Entities
People
- Celso M de Melo
- Jonathan Gratch
- Peter J. Carnevale
Organizations
- University of Southern California