Jia Gu¹ (), Dirk B. Walther¹, Claudia Damiano¹; ¹University of Toronto

Many factors shape how humans judge something as beautiful or threatening. Research has extensively explored the relationship between symmetry and perceived beauty—demonstrating a consistent preference for symmetrical faces—while findings in other visual contexts, like art and landscapes, are less pronounced. Additionally, the relationship between symmetry and threat perception remains unclear. Given the link between threat judgments and felt expectancy, symmetry may influence threat perception by affecting the predictability of visual environments. In understanding the role of visual features on perception of beauty and threat, this study investigated whether the level of symmetry and complexity correlates with pleasure and threat ratings of natural scenes. To explore the underlying mechanisms of such influence on human perception, we included tasks to measure expectancy and memorability of each scene. We hypothesized that more symmetrical images would be more predictable, leading to reduced threat perception. Participants viewed 132 natural scenes with either the left or right half covered for 4 seconds and reported how well they could anticipate the covered half. Next, participants completed an old-new recognition test. Finally, they rated the pleasure and threat of each scene from the first task. We found a negative correlation between symmetry and complexity, r (130) = -.25, p = .003. We also observed that participants predicted images with more symmetric arrangements better as we hypothesized, r (130) = .48, p < .001. However, the analysis revealed that pleasure and threat ratings were not significantly associated with symmetry, memorability, or expectancy. Instead, visual complexity of natural scenes was correlated with threat (r (130) = .30, p < .001) and beauty (r (130) = .64, p < .001) ratings. These findings suggest that while visual complexity is related to symmetry, only complexity affects beauty and threat perception of natural scenes.

Acknowledgements: This work was supported by the University of Toronto Undergraduate Research Fund to J. G.