Uncanny worlds: viewer perception of facial expression and the uncanny valley in virtual characters

Realistic, human-like characters are increasingly being used in virtual worlds for entertainment in animation and video games, and in healthcare and education for therapeutic and elearning applications. However, despite technological advances directed towards increasing the realism and human-likeness of such characters, increased realism does not imply increased acceptance. In 1970, Mori proposed the Uncanny Valley theory to explain why, as robots become increasingly human-like, humans became increasingly averse to them. This was due to a perceived mismatch in a robot’s human-like appearance with their behaviour. Recent empirical work has sought to use Mori's theory to explain the lack of empathy between humans and characters with a human-like appearance in virtual worlds. Since 2009, a research project at Bolton University has explored how aspects of a character’s facial expression and speech may exaggerate viewer perception of the Uncanny Valley in characters. The findings of this research project will be presented thus far that lead to a new theory that aberrant facial expression of emotion in a character may trigger possible psychological causes that underpin the existence of the Uncanny Valley. This paper will also present how male and female 3D realistic, human-like heads developed at The Abertay University will be used as stimuli in future experiments. This future work will help to define an explanation as to the cause of the Uncanny Valley, based on a perception of a lack of empathy in a character and psychological and neuroscientific literature on the purpose of facial expression in human social interaction.