Summary: A new study explores how people react to identical faces—a phenomenon with implications for humanoid robots and other humanlike replicas.
Source: Ritsumeikan University
As robotics and computer-generated characters grow ever more lifelike, films like Terminator and Ex Machina imagine machines indistinguishable from humans. If such humanlike robots became widespread, how would we respond when many individuals share the exact same face?
A team of researchers from Kyushu University, Ritsumeikan University, and Kansai University investigated this question in a study published in PLOS ONE. They examined human responses to images in which multiple people display the identical facial appearance—so-called “clone images”—and compared those reactions to images of people with different faces and single-person images.
Advances in robotics and artificial intelligence have already produced highly humanlike androids and digital characters—examples include Geminoid, Saya, and Sophia. Developers hope to overcome the uncanny valley, the psychological discomfort that can arise when a humanoid becomes almost—but not quite—human in appearance. This research considers an adjacent concern: how people perceive and emotionally respond to visual scenarios where facial identity appears duplicated across multiple individuals.
To probe responses, the researchers ran six experiments involving several batches of participants numbering in the hundreds. Each study used carefully edited photographs or illustrations to control facial appearance and context, and participants rated images on subjective eeriness, emotional valence, and perceived realism.
Experiment 1 asked participants to rate a photoshopped image showing six different people sharing the exact same face (a clone image), an image of six people each with a different face (a non-clone image), and an image of a single person. Experiment 2 presented another set of clone and non-clone human images. Experiment 3 repeated this comparison using images of dogs to test whether the effect extended to nonhuman faces.
Experiment 4 investigated identity duplication more directly. In one part, participants rated clone images constructed from two sets of twins; in another, they evaluated images showing clones composed of twins, triplets, quadruplets, and quintuplets. Experiment 5 used clone images of Japanese animation and cartoon characters to test how lower-fidelity, stylized contexts affected responses. Finally, Experiment 6 repeated the clone versus non-clone evaluations for a different image set and added the Disgust Scale Revised to measure individual differences in disgust sensitivity.
Across the studies, a clear pattern emerged: images that showed multiple people with the same face were consistently rated as eerier and less probable than images showing different faces or a single person. The authors named this response the “clone devaluation effect.” In other words, duplication of facial identity produced a negative emotional reaction distinct from the classic uncanny valley.
Lead author Dr. Fumiya Yonemitsu of Kyushu University reports that the effect intensified as the number of identical faces increased—particularly when the number rose from two to four. Notably, the effect did not appear in Experiment 3 with dog faces, where individual animal faces were less distinguishable; this suggests the phenomenon depends on human-specific face-recognition processes.
The research indicates that duplication of identity—encompassing the sense of a unique personality or mind tied to an individual—plays a critical role. Clone faces that implied duplicated identity produced stronger eeriness, as shown in the twin and multiples experiments. Conversely, the effect weakened in contexts perceived as less realistic, such as stylized animation or cartoon images used in Experiment 5.
Experiment 6 linked the perceived improbability and eeriness of clone faces to individual differences in disgust sensitivity, especially the component known as animal-reminder disgust. In short, people who are more prone to this form of disgust tended to find clone faces eerier. Taken together, the findings suggest the clone devaluation effect arises from multiple mechanisms related to face recognition, memory, emotion, and avoidance reactions.
These results reinforce the idea that human faces convey crucial identity information: people typically assume a one-to-one mapping between face and person. When that mapping is violated—when multiple individuals appear to share the same face—observers may misattribute identity and experience increased unease or improbability judgments.
What are the practical implications? The researchers emphasize that as humanoid robots, digital avatars, and cloning technologies advance, designers and policymakers should be aware of psychological responses beyond the uncanny valley. Rapid deployment of humanlike technologies that create repeated or duplicated facial appearances could provoke discomfort or avoidance reactions in everyday users.
Co-author Dr. Akihiko Gobara of Ritsumeikan University notes that while technological progress can create uncomfortable situations, understanding phenomena like the clone devaluation effect can help guide design choices and public communication so people can accept new technologies more smoothly and enjoy their benefits.
About this robotics and psychology research news
Author: Kazuki Kurajo
Source: Ritsumeikan University
Contact: Kazuki Kurajo – Ritsumeikan University
Image: The image is credited to Dr. Fumiya Yonemitsu from Kyushu University
Original Research: The clone devaluation effect: A new uncanny phenomenon concerning facial identity by Fumiya Yonemitsu et al., published open access in PLOS ONE.
Abstract
The clone devaluation effect: A new uncanny phenomenon concerning facial identity
Robotics and related technologies have reached a stage where some robots and digital characters can appear nearly identical to human beings. While efforts focus on overcoming the uncanny valley, another issue arises if humanoid faces are duplicated across multiple individuals. If identical facial appearances become commonplace—through mass production of humanoids or other means—what impressions will such duplication elicit?
Across six studies, we consistently found that images showing multiple people with the same face produced higher ratings of eeriness than images of individuals with different faces. We term this the clone devaluation effect. The effect reflected perceived improbability of facial duplication, depended on the distinguishability of each face and the sense of duplicated identity, and was sensitive to the realism of the context. Additionally, individual differences in disgust sensitivity—especially animal-reminder disgust—predicted the degree of eeriness.
These findings indicate that the clone devaluation effect arises from interacting processes in face recognition, emotion, and memory, and they highlight psychological considerations for the design and acceptance of future humanlike technologies.