Inconsistency in Behaviors of Virtual Agents and Robots: Case Studies on its Influences into Dialogues with Humans

Mutual interaction among contents and ways in utterances, nonverbal behaviors such as postures and gestures, and contexts is an important factor in human communication. Bateson (1972) argued, in his theory of double-bind situations, that inconsistency in human communication can cause mental illness, by combination with power relations forcing the inconsistency. Along Bateson’s theory, Aoki (1993) experimentally validated influences of inconsistency between utterance contents and ways into human cognition and memorization. Krauss, Chen, and Chawla (1996) surveyed research results on relationships between hand gestures, comprehension of contents in speech, and speech production. These studies suggest the importance of consistency on agents’ and robots’ forms of information presentation in dialogues with humans. However, there have been few studies that straightforwardly investigated this possibility in humanagent and human-robot interaction research. As an attempt for this investigation, some psychological experiments were conducted in Japan, by using simple animated characters (Ohnishi and Nomura, 2004; Nomura and Okuyama, 2007) and a small-sized humanoid robot (Nomura and Nakamura, 2010).