Categorization difficulty is associated with negative evaluation in the “ uncanny valley ” phenomenon

Human observers often experience strongly negative impressions of human-like objects falling within a particular range of visual similarity to real humans ('uncanny valley' phenomenon). We hypothesized that negative impressions in the uncanny valley phenomenon are related to a difficulty in object categorization. We produced stimulus images by morphing each two of real, stuffed and cartoon human face images (Experiment 1). Observers were asked to categorize each of these images as either category and evaluated the likability of it. The results revealed that the longest latency, the highest ambiguity in categorization, and the lowest likability score co-occurred at consistent morphing rates. Similar results were obtained even when we employed stimulus images that were created by morphing each two of real, stuffed and cartoon dog images (Experiment 2). However, the effect of categorization difficulty on evaluation was weak when two real human faces were morphed (Experiment 3). These results suggest that the difficulty in categorizing an object as either of dissimilar categories is linked to negative evaluation regardless of whether the object is human-related or not. Introduction People often experience an eerie impression or even a sense of revulsion on seeing elaborately designed human-like agents, such as dolls, animated characters, computer-game characters, virtual reality avatars, or robots. Mori (1970) postulated that the likability of a robot gradually increased up to a certain level as its appearance became more human-like, but beyond a certain level of similarity observers suddenly perceived the face as eerie or disgusting. Finally, as the robot’s appearance reached the maximum possible degree of similarity, and the perceived strangeness reverted to likability. Mori (1970) referred to this as the “uncanny valley” phenomenon, with reference to this precipitous fall in likability. Various explanations have been proposed to explain the uncanny valley phenomenon. A previous study suggested that the elicitation of the uncanny valley phenomenon may be related to a fear of mortality, where negative evaluation occurs as a result of an observer doubting their own identity as a living human being on seeing an artificial human-like agent (MacDorman & Ishiguro, 2006). Likewise, an observer may associate disassembled or incomplete human-like agents with scenes of a battlefield containing wounded human bodies after conflict (MacDorman & Ishiguro, 2006). Similarly, MacDorman and Ishiguro attempted to explain the uncanny valley phenomenon in terms of a disgust reaction associated with threat (such as pathogen) avoidance, based on the notion that our cognitive system has evolved a feeling of disgust as a mechanism to avoid infection by harmful bacteria or viruses (Rozin & Fallon, 1987). As illustrated by these hypotheses, researchers have interpreted the uncanny valley phenomenon mainly from an evolutionary point of view. However, it is currently unclear what type of cognitive processing underlies the uncanny valley phenomenon, and how this processing is related to evolutionary factors. In the present study, we examined a new hypothesis that categorization difficulty is an important factor in the generation of the uncanny valley phenomenon. The uncanny valley occurs when the appearance of a nonhuman agent becomes more human-like (Mori, 1970). As increasing similarity entails the sharing of more visual features (Humphreys, Riddoch, & Quinlan, 1988), it consequently becomes increasingly difficult for observers to categorize agents as human or nonhuman. This difficulty in categorization increases the processing load involved. In addition, previous studies have suggested that relatively high processing fluency is associated with positive impressions (Kuchinke, Trapp, Jacobs, & Leder, 2009; Reber, Schwarz, & Winkielman, 2004; Reber, Winkielman, & Schwarz, 1998). For example, Reber et al. (1998) reported that priming images, which were subliminally presented before target images, increased positive affective judgment on the target images. In turn, it is natural to assume that relatively low processing fluency This is a preprint version of the article that has