A new study delves into the intriguing psychological phenomenon known as the "Medusa effect," revealing a persistent bias in how individuals perceive others within nested images. When presented with a photograph of a person holding another photograph, observers tend to attribute less cognitive and emotional capacity to the individual depicted in the inner picture. This intriguing finding, published in the journal Cognition, highlights how deeply ingrained structural presentation can override the physical details of a human face.
Psychologists utilize the concept of mind perception to understand how individuals assess the inner lives of others. This theory posits that people intuitively evaluate the mental capabilities of various entities along two key dimensions: agency and experience. Agency refers to the capacity for conscious thought, planning, and independent action, while experience encompasses the ability to perceive the environment and feel emotions. Typically, living humans are afforded high levels of both agency and experience, occupying the apex of social perception. In contrast, animals, robots, and two-dimensional representations like photographs are generally ascribed significantly lower mental qualities.
Prior research has already established a decline in mind perception as visual abstraction increases, coining this phenomenon the Medusa effect. Specifically, a person shown directly in a photograph is perceived as more real and possessing greater mental capacity compared to a person in a picture within a picture. The initial photograph represents a primary level of abstraction, while a photo embedded within it introduces a secondary, more abstract layer.
To further investigate the cognitive mechanisms underlying this psychological bias, a research team led by Jing Han at Kyushu University undertook a comprehensive study. Their primary objective was to determine whether manipulating facial information processing could mitigate or eliminate the Medusa effect. They designed a series of experiments incorporating culturally diverse photographs, synthetic media, and physical obstructions to disrupt typical visual processing.
The initial experiment involved Japanese participants who were shown culturally adapted images featuring Asian models. These participants viewed images where a primary person held a portrait of a secondary, nested individual. They were asked to assign scores for agency, experience, and realness on a scale of zero to ten. The results confirmed that the Medusa effect was present even with these culturally adapted images, mirroring findings from previous studies involving Western populations.
Subsequent experiments explored the impact of disrupting visual processing pathways. One test specifically targeted holistic visual processing by inverting the photographs vertically. While flipping faces significantly impaired overall mind attribution scores for all individuals in the pictures, the participants consistently rated the people in the nested photographs lower than the primary subjects holding them. This suggests that the Medusa effect is robust even when the brain's ability to process faces holistically is compromised.
Further investigations focused on feature processing by strategically occluding parts of the face. In three consecutive experiments, models were photographed wearing surgical masks, dark sunglasses, or both. These accessories, which obscure crucial visual cues for emotion and inner mental states, drastically reduced the general perception of mentality across the board. However, the relative difference in mind perception persisted: the nested subjects were invariably judged as possessing less mental capacity than the directly photographed subjects, indicating the resilience of the Medusa effect.
The researchers also examined the influence of authenticity and artificial intelligence on the Medusa effect. Utilizing image generation software, they created completely artificial scenes featuring synthetic individuals holding pictures of other synthetic people. Participants evaluated these images without knowledge of their AI origin. Intriguingly, observers instinctively attributed less mental capacity to these synthetic subjects compared to real humans from previous trials. Even within these artificial constructs, the psychological gap remained, with the artificial mind in the primary photo being rated higher than the artificial mind in the nested photo.
The final experiment involved spatial scrambling, where the internal facial features of the models were unnaturally rearranged. This manipulation severely impaired the ability to interpret the stimuli as coherent social agents, leading to the lowest mind perception scores in the entire study. Despite evaluating these highly distorted faces, observers still demonstrated the Medusa effect, rating the nested scrambled faces lower than the primary scrambled faces. This further underscores the remarkable robustness of the phenomenon, suggesting it operates largely independently of readily recognizable facial information.
The consistent findings across these diverse manipulations indicate that the Medusa effect is an exceptionally robust phenomenon, resistant to basic perceptual disruptions. Researchers propose that this effect may be linked to Construal Level Theory, which suggests that psychological distance—whether spatial, temporal, or hypothetical—promotes more abstract mental representations. A nested photograph, by creating psychological distance, makes the individual appear more remote to the observer. Another possibility is that the Medusa effect reflects a deeper categorical sorting process, wherein an image embedded within another is unconsciously treated more as a decorative element than as a sentient human being.
While the study provides significant insights, the researchers acknowledge certain limitations that warrant future investigation. The current study primarily utilized neutral faces and upper torsos, thus limiting the influence of full body posture, which can convey substantial social and emotional information. Future research should explore how altering body language or introducing animated and robotic figures might affect the impact of nested abstraction on social judgments. Additionally, examining individual differences in visual processing speed and accuracy could illuminate why some individuals are more prone to this visual bias than others. The study, titled "Robust Medusa effect across facial manipulations," was a collaborative effort by Jing Han, Kyoshiro Sasaki, Fumiya Yonemitsu, Kaito Takashima, and Yuki Yamada.