Social believability of human-driven embodied conversational avatars in shared virtual worlds: the impact of the adaptation gap on users' experience of interacting with VR avatars

The research investigates socially believable interactions of human-driven Embodied Conversational Avatars visiting shared 3D virtual communities. Many social VR platforms allow users to use two kinds of avatar control mechanisms: motion-tracked or automated-gesturing. This potentially creates a discrepancy in the user's experience and impacts immersion when exposed to different avatars. This thesis adapts Komatsu et al.'s Adaptation Gap concept to measure the difference in participants' expectations and perceptions of both avatars, testing the assumption that the difference in avatar control type has an impact on user experience. The results support the argument that avatar control type matters (p < .05) and so does the context (or scenario) in which the interaction takes place (p < .05). The difference between user expectations and their perceptions is correlated (p < .05) with the believability of the avatars, which is known to have an impact on their interaction.