Intelligent virtual agents have been developed to study, assess and teach a variety of human interpersonal skills. Here we examine the impact of an agent's perspective-taking sophistication on human negotiators. Good perspective-takers can discover creative solutions that benefit both parties, but many have difficulty with this skill. In particular, novices focus on explicit goal-statements (e.g., "I want apples more than bananas") but discount goal-relevant information implicit in the opponent's offers. Many human-agent negotiation agents similarly ignore implicit information. We examined the influence of implicit information on human negotiators by independently enhancing agents in two ways: do agents communicate implicit information and do they attend to implicit information communicated by users. We find that communicating implicit information seems to confuse user's perspective-taking ability, yet paradoxically, helps lead them to better outcomes. In contrast, an agent that attends to user's implicit communications shows better perspective-taking but fails to translate this into better outcomes. These results emphasize the challenges associated with implicit information. We discuss how these results impact the design of negotiation agents for applications, analysis and pedagogy.
CITATION STYLE
Johnson, E., & Gratch, J. (2020). The Impact of Implicit Information Exchange in Human-agent Negotiations. In Proceedings of the 20th ACM International Conference on Intelligent Virtual Agents, IVA 2020. Association for Computing Machinery, Inc. https://doi.org/10.1145/3383652.3423895
Mendeley helps you to discover research relevant for your work.