Improvement of Proactive Attitude by Alternating Enhancement of a Sense of Acceptance and Control

When people work collaboratively, they often communicate their task status to each other and point out problems. However, when the collaboration partner is a system or an artificial agent, the collaboration is often not a two-way interaction, but a one-way monitoring and instructional relationship. In this study, we propose a human-agent interaction model that synthetically and progressively encourages one's acceptance of the communication partner's behavior and their control over one's own behavior. By doing so, we aim to reduce users' mental workloads and improve their proactive attitude toward the task and the interaction. Active attitude is the attitude of voluntarily taking the initiative to engage with a subject, such as the task and the communication partner. We conducted an experiment to compare the proposed model with another interaction model that aims to support the decision-making of the participant. The results showed that 1) the proposed model improved participants' proactive attitude to some extent, 2) the proposed model encouraged interaction and improved the participant's sense of acceptance, and 3) participants who applied the proposed model when they interacted with the agent were able to reduce their mental workloads in the task.