The inversion effect on the cubic humanness-uncanniness relation in humanlike agents.

Area of Science:

• Cognitive Psychology
• Human-Computer Interaction
• Robotics

Background:

• The uncanny valley describes a non-linear relationship between an artificial entity's human likeness and its aesthetic appeal.
• This phenomenon is often linked to specialized configural processing, heightening sensitivity to human norm deviations.

Purpose of the Study:

• To investigate the uncanny valley effect in computer-generated android and human faces.
• To examine the role of dynamic facial expressions and their synchrony in modulating uncanniness.
• To determine if configural processing influences the perception of eeriness in humanlike artificial entities.

Main Methods:

• Presentation of computer-generated android and human faces with dynamic angry and happy expressions.
• Varying degrees of facial expression synchrony were used.
• Stimuli were presented in upright and inverted orientations.
• Participants rated faces on eeriness, strangeness, and human likeness.

Main Results:

• A sigmoidal relationship was observed between human likeness and uncanniness for upright facial expressions.
• A linear relationship was found for inverted faces.
• Configural processing was supported as a moderator of human likeness's effect on uncanniness.

Conclusions:

• The study supports the role of configural processing in moderating the uncanny valley effect.
• The findings extend the understanding of configural processing to dynamic facial expressions in artificial entities.
• While not a classic uncanny valley, the results highlight the nuanced perception of humanlikeness in artificial agents.