Abstract
Crowding impairs recognition, so might it impair beauty as well? We operationally defined perceived beauty as rated enjoyment. 120 online users rated their enjoyment of 360 images (120 abstract art, 120 figurative art, and 120 photos of natural or man-made environments) from Forsythe et al. (2011). In each trial, an observer fixated a cross at the center of the screen, saw an image presented for 100 ms, and rated their enjoyment of the image on a 5-point Likert scale. Each observer saw each image once. The image was displayed at one of six sizes (width 0.4, 1, 2, 4, 6, or 8 deg) at one of two eccentricities (foveal or peripheral). The peripheral eccentricity, randomly right or left of fixation, was roughly 8 deg, assuming that participants viewed their monitors from 1.5 feet (46 cm) as instructed, and that the monitors had 150 ppi (Forsythe et al., 2011). For each image, each of the 12 conditions (6 sizes × 2 eccentricities) received ratings from 10 observers. We estimated crowding distance as 1/3 eccentricity (Bouma, 1970; Kurzawski et al., 2021). Crowding between components of the image depends on eccentricity, image size, and image content. To estimate the influence of image content on crowding, a judge estimated, for each image, the key spacing (in pixels): the minimum distance separating groups of features that are essential for semantic recognition of the image. We estimated the degree of crowding of each trial by calculating the crowding ratio: the ratio of crowding distance (1/3 radial eccentricity) divided by the key spacing (in radian and at the image size on that trial), which is very little influenced by the viewing distance. Across observers, sizes, and eccentricities, beauty is negatively correlated with estimated crowding, r(4318) = −0.48, p < 0.001. Crowding kills beauty.