Character-based Interfaces Adapting to Users’ Autonomic Nervous System Activity

This paper explores some possibilities of using real-time processing of users’ physiological signals in order to improve the interaction with character-based interfaces. Specifically, interfaces that adapt to or reflect a user’s affective state as derived from the user’s autonomic nervous system activity will be discussed. The ‘Emotion Mirror’ is an example of a system where the user’s emotions are reflected back to the user, and hence allows to train emotion management and regulation. The ‘Emphatic Companion’ is an extension of the previous system employing a decision-theoretic embodied agent that adapts its behavior depending the recognized emotional state of the user, e.g. by giving support and encouragement. A simple virtual job interview scenario has been implemented to illustrate the potential of considering physiological data of users in real-time. key words: Affective Computing, adaptive user interfaces, realtime emotion recognition, autonomic nervous system (ANS), bio-signals, life-like characters, decision networks, web-based environments.