TITLE:
Autonomous Adaptive Agent with Intrinsic Motivation for Sustainable HAI*
AUTHORS:
Takayuki Nozawa, Toshiyuki Kondo
KEYWORDS:
Human-Agent Interaction (HAI), Intrinsic Motivation, Reinforcement Learning, Functional Near-Infrared Spectroscopy (Fnirs)
JOURNAL NAME:
Journal of Intelligent Learning Systems and Applications,
Vol.2 No.4,
December
14,
2010
ABSTRACT: For most applications of human-agent interaction (HAI) research, maintaining the user’s interest and continuation of interaction are the issues of primary importance. To achieve sustainable HAI, we proposed a new model of intrinsically motivated adaptive agent, which learns about the human partner and behaves to satisfy its intrinsic motivation. Simulation of interaction with several types of other agents demonstrated how the model seeks new relationships with the partner and avoids situations which are not learnable. To investigate effectiveness of the model, we conducted a comparative HAI experiment with a simple interaction setting. The results showed that the model was effective in inducing subjective impressions of higher enjoyability, charm, and sustainability. Information theoretic analysis of the interaction suggested that a balanced information transfer between the agent and human partner would be important. The participants’ brain activity measured by functional near-infrared spectroscopy (fNIRS) indicated higher variability of activity at the dorsolateral prefrontal cortex during the interaction with the proposed agent. These results suggest that the intrinsically motivated adaptive agent successfully maintained the participants’ interest, by affecting their attention level.