Title :
Toward avatar models to enhance performance and engagement in educational games
Author :
Dominic Kao;D. Fox Harrell
Author_Institution :
Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
Abstract :
This paper presents work toward better understanding the roles that avatars can play in supporting learning in educational games. Specifically, the paper presents results of empirical studies on the impact of avatar type on learner/player performance and engagement. These results constitute work establishing baseline understandings to inform our longer term goal of developing models that use dynamic avatars to best support learners in educational games. Our aim is motivated by a convergence of research in the social sciences establishing that identity plays an important role in learning. Of note, aspects of social identity (e.g., race, ethnicity, and gender) have been shown to impact student performance [1] via triggering stereotypes [2]. Recently, performance and engagement studies in our educational game for Science, Technology, Engineering and Mathematics (STEM) learning suggest these same phenomena can be activated through virtual avatars [3], [4]. Here, we present results of a comparative study between avatars in the likeness of players and avatars as geometric shapes. In our STEM learning game, results show that players that had selected and used a shape avatar had significantly higher performance than players that had customized and used a likeness avatar. Players using the shape avatar also had significantly higher self-reported engagement, despite having lower self-reported affect towards the avatar.
Keywords :
"Avatars","Games","Shape","Pragmatics","Machine learning algorithms","Predictive models","Computational modeling"
Conference_Titel :
Computational Intelligence and Games (CIG), 2015 IEEE Conference on
Electronic_ISBN :
2325-4289
DOI :
10.1109/CIG.2015.7317959