Kota Sasaki¹ (), Fumiya Yonemitsu¹, Atsunori Ariga¹; ¹Chuo University

We feel strong eeriness for non-human objects that are highly similar to humans (a phenomenon known as uncanny valley). Previous research has accounted for this phenomenon based on animacy perception (e.g., avoidance to threat or anxiety for mortality) or cognitive processes irrespective of animacy (categorization difficulty hypothesis). However, whether animacy perception is responsible for uncanny valley to be elicited is unclear at present. This is because all the previous studies used objects related to animacy as stimuli (e.g., faces). Therefore, the present study examined whether uncanny valley occurs for non-animacy objects, using geometric figures (square, circle, and triangle). In Experiment 1, we made stimulus figures by systematically morphing an original geometric figure (e.g., circle) into another figure (e.g., triangle). Participants first categorized which original figure (e.g., circle or triangle) the morphed figure was perceived as and then evaluated its likability with a 7-point Likert scale. As the result, we found the uncanny-valley-like phenomenon with non-animacy objects; that is, the figures that took longer times to be categorized (difficult-to-categorize objects) were rated as being less likeable as compared to the figures that were quickly categorized (easy-to-categorize objects). Our results were in favor with the categorization difficulty hypothesis, suggesting that animacy perception is not necessary for the occurrence of uncanny valley. In the following experiments, we hypothesized that the categorization difficulty deteriorates perceptual and cognitive fluency of the morphed figures, resulting in the lower evaluation for them. To assess this hypothesis, we independently manipulated the perceptual (Experiment 2) and cognitive (Experiment 3) fluency of to-be-evaluated figures and investigated how the uncanny-valley-like phenomenon is modulated in response to the processing fluency. Based on these results, we discuss not only the underlying mechanism(s) for the uncanny valley phenomenon but also how to overcome the phenomenon in industry.

Acknowledgements: This research was financially supported by JSPS KAKENHI Grant Numbers JP21J01431 to FY and JP 22K03210 to AA.