Conditions under which “Neither-Good-Nor-Bad” Information Works Effectively on User’s Trust in Agents

• Published in: International Journal of Affective Engineering Vol. 20; no. 4; pp. 199 - 208
• Main Authors: YOKOYAMA, Shinku, IMAI, Jun-ichi
• Format: Journal Article
• Language: English
• Published: Tokyo Japan Society of Kansei Engineering 2021; Japan Science and Technology Agency
• Subjects: Human-agent interaction; Trustworthiness; Trustworthiness of agents; “Neither-Good-Nor-Bad” information
• ISSN: 2187-5413; 2187-5413
• DOI: 10.5057/ijae.IJAE-D-20-00036

The effect of the provision of “Neither-Good-Nor-Bad” (NGNB) information on the trustworthiness of agents has been reported. However, in this previous study, the size of the profits generated by the NGNB information was fixed at 50% of the maximum profit, even though this ratio can vary in practical situations. Furthermore, in this experiment, all the consequences were disclosed to the users, including those that they had not chosen. In practical situations, we are often unaware of consequences that we have not chosen. In this study, we focus on these two factors and investigate the effects of them on the trustworthiness of the agents. Experimental results reveal that the agent that reliably provides the NGNB information tends to gain a greater degree of users’ trust in a situation where it is harder to obtain large profits, and when consequences that they had not chosen were not disclosed to them.