Investigating the effects of physical and virtual embodiment in task-oriented and conversational contexts

Abstract

Studies comparing physically embodied robots with virtually embodied screen characters (e.g. Powers et al., 2007. Jung and Lee, 2004.) have resulted in unsimilar findings with respect to subjective (users’ evaluations) as well as objective (e.g. task performance of the users) measurements. The comparability of these results is mainly impeded by the use of different robots, a variety of virtual embodiments (video recording, computer simulation, animated characters, etc.) and different interaction scenarios. To overcome this problem, an experimental study was conducted in which the embodiment of an artificial entity was varied systematically as well as the type of interaction using a 2×2 between subjects design (N=83). Participants interacted with either a robot or a virtual representation of this robot (on a screen) in a task-oriented or a persuasive–conversational scenario. The results revealed that participants perceived the robot as more competent than the virtual character in the task-oriented scenario, but the opposite was true for the persuasive–conversational scenario. Furthermore, participants in the task-oriented scenario felt better after the interaction than participants who had a persuasive–conversational interaction with the artificial entity, regardless of its embodiment. No statistically significant differences between the experimental conditions emerged with respect to objective measures (persuasion and task performance). Various explanations for these findings are discussed and implications for the application of robots and virtual characters are derived.