Scenes are composed of numerous of objects, textures and colors which are arranged in a variety of spatial layouts. Complexity can be understood as a combination of these components. The goal of this study was to uncover the perceptual properties underlying visual complexity of real world scenes. The following question motivated this study: is visual complexity simple enough so that it can be compressed along a unique perceptual dimension? Or is visual complexity better conceptualized as a set of perceptual properties? Twenty participants performed a hierarchical grouping task in which there were presented with 100 scene pictures representing a variety of indoors scenes. Then they were instructed to divide the images into two groups (simple and complex), then split each partition into half, and finally split each quarter into two groups again, creating a final of eight groups. For each partition, participants described the criteria they used. The perceptual properties may be summarized as follows: quantity of objects (first criterion), symmetry and openness (secondary criteria), and a set of additional criteria like cluster organization (from organized to random), compactness (density of objects vs. free space) and brightness. Based on a matrix of distances computed between each image, we ran a multi-dimensional scaling analysis to determine the underlying perceptual dimensions of visual complexity. Overall, the results suggested that visual complexity of real world scenes can be mainly categorized along one dimensional axis, represented by an interconnected composition of features (e.g., quantity, symmetry, compactness).

M.S. was funded by a research assistantship from an NSF-IGERT training grant (DGE-0114378). This research was funded by a NIMH grant (1R03MH068322-1) awarded to A.O.