Social Agents for Learning in Virtual Environments

Several serious games have been proposed to practice communication strategies in formal contexts. Intelligent virtual agents (IVA) can be used to show the player the effects of a conversational move. In this paper we discuss the key role of using social context for the virtual agents in these serious games. Social practices are exploited to bundle social interactions into standard packages and as a basis to model the deliberation processes of IVAs. We describe a social practice oriented IVA architecture used in the implementation of a serious game for the practicing of communication in medical interviews.