S. Wong. (2017). The Uncanny Valley Effect: Implications on Robotics and A.I. Development.
Summary
The term ‘uncanny valley’ describes the negative responses of humans to agents with slightly imperfect human likeness. These negative responses entail feelings of eeriness and discomfort. In robotics, there is a correlation between the increased human-likeness and familiarity ratings by human observers. The uncanny valley effect lends its name from the drop in the positive correlation when the robot reaches a degree of realism that the robot has an almost perfectly human-like appearance. Hence a ‘valley’ is created in the correlation between human-likeness and familiarity.
Evolutionary approach of explaining the uncanny valley effect
An evolutionary approach suggests that the effect finds its origins in the human mechanism for indicating health of others. A near perfect human appearance may suggest the presence of diseases or death. This indicator may subconsciously have us feel disgusted to prevent us from getting close so that we don’t contract a bacteria.
Attractiveness servers as an indicator of fertility and health. Aversions to unattractive features are often enhanced when viewing humanistic androids. Therefore the mate selection mechanism might evoke the uncanny valley effect when looking at a near perfectly human-like robot and considering it is a possible mate.
Cognitive neuroscience approach of explaining the uncanny valley effect
Measurements of brain activity indicate that humans have a sensitive preference for real faces of humans. This preference is specific to strictly human facial stimuli, but not animal faces or cartoons. This suggests that we often experience uncanny feelings towards realistic robots because they specifically remind us of living humans.
A specific facial stimuli that humans are sensitive to are the eyes of humanoids as they provide a lot of social cues and emotional information. Failure to recognize ‘life’ within eye gazes of unrealistic robots can therefore promote fear and signifies unpredictable or incomprehensible behavior.
Another suggestion is that the discomfort caused by realistic, artificial human forms is not a result of their appearance, but what they signify. Because these robots give a vague sense of identity that lies between human and non-human, thus the perception of such an almost perfectly human-like robot might cause uncanny experiences related to cognitive dissonance.
One approach that would avoid the uncanny valley effect would be to deliberately design robots and other appliances that do not resemble humans at all, while having it show humanistic features that accentuates its liveliness. The robot may be anthropomorphic, i.e. having similarities to the human appearance, as long as it does not approach a near perforce human appearance.