Date created: | Last Updated:

Category: Project

Description: Exposure to natural environments, compared to built spaces, has been shown to confer important psychological benefits. Furthermore, people generally exhibit preferences for natural landscapes over man-made buildings. However, some forms of architectural design draw inspiration from natural systems and thus exhibit similar sensory qualities as nature itself. Can these natural features be measured, and are buildings which exhibit these features inherently preferred? Here, we examined whether subjective perceptions of naturalness in architectural scenes are driven by objective visual patterns, and we investigated whether natural patterns influence similarity evaluations and preference ratings of architectural scenes. Experiment 1 revealed that low-level spatial and color features of architectural scenes explained over half of the variance in scene naturalness ratings. Scaling and Contrast-related features were especially strong predictors of naturalness for both interior and exterior scenes. In Experiment 2, participants completed an image arrangement task and multidimensional scaling (MDS) analysis was performed on the data to determine the underlying aesthetic dimensions that drove scene similarity judgements. Naturalness ratings explained over half of the variance in MDS Dimension 1 weights, suggesting that people relied on latent perceptions of naturalness to evaluate the similarity of architectural scenes. In the final experiment, participants rated architectural scenes on aesthetic preference. Nature-like patterns of Scaling and Contrast significantly predicted preference ratings. Furthermore, the effects of Scaling and Contrast on preference were mediated by latent perceptions of naturalness, indicating that these architectural patterns are preferred because they evoke intuitive associations with sensory features of nature. Collectively, these studies suggest that people may be innately attuned to naturalistic features of the built environment, and that these features can be measured using computational methods of image analysis.