Embodied and Embedded Intelligence: Actor Agents on Virtual Stages

The paper was delivered to an audience of approximately 20. Comments were directed to the symposium group with nothing specific to our paper. The symposium group went out for lunch following the presentation and plans were made to submit our papers as part of a special issue to one of three journals.In this presentation, we will argue that Actor Agents, a type of Animated Pedagogical Agent with embodied conversational intelligence, are well positioned to act in virtual environments with embedded intelligence and that, together, actor agents on virtual stages can greatly extend the scope of simulation pedagogy. Simulations, as defined by Sauve, Renaud, Kaufman, & Marquis (2007) may be argued as the quintessential adaptive learning system. According to Sauve et al. (2007), a key feature of a true simulation is that it behave as a dynamic model of reality which responds to user initiated actions and perceptions. Users can control the reality of the simulation in ways that reflect their learning preferences and existing knowledge. Ideally, simulations are a perfect reflection of individual learning trajectories. Users can interact with Actors and within their stage to create a dynamic model of reality that directly meets educational goals and objectives. We will review the origins of Actor Agents within the broader Animated Pedagogical Agent field highlighting the importance of a conversational interface and the sought-after persona effect (e.g. Payr, 2005) Although the promise of APAs has been limited, we believe that Actor Agents have a special attraction as a pedagogical tool when conversational abilities are added as embodied intelligence and personas are selected to maximize engagement and motivation. Our work with Freudbot (Heller & Procter, 2009) shows promise and may yet reveal the illusive persona effect hypothesized to underlie learning gains. An equally important development for effective simulations has been the growth of virtual worlds, like Second Life. In this part of the presentation we will review some of the some of the key features of virtual worlds, focusing on Second Life, and how these features can give rise to embedded intelligence to create a type of holodeck or virtual stage for Actor Agents. A good simulation necessarily requires some sense of immersion in order for learning to be effective (Gutiérrez , et al., 2007). Equally important in the simulation is a capacity for the virtual world to respond intelligently to user behaviour. Finally, we hope to demonstrate the synergy that may be created if specialized spaces were developed for actors to act within. Intelligent objects that can collaborate with an actor agent and interact dynamically with a user in a well structured simulated environment can greatly extend the scope of simulation pedagogy to professions that routinely deal with people of any age, gender, or cultural status. Our theoretical framework is eclectic although we are strongly rooted in cognitive theory and most closely aligned with Mayer’s theory of multimedia learning. We also align with framework of situated cognition and argue that past work on APAs may have downplayed the importance of situational context. Our methods of inquiry are a combination of observational and experimental approaches. Our data sources include conversational logs and self report measures