Abstract
Despite the large variability that exists across nature (e.g. forests, deserts, mountains), natural scenes share many statistical properties. Firstly, they are similar in their photometric properties since they each contain a unique distribution of luminance intensities across spatial and temporal frequencies known as the 1/fα amplitude spectrum (α ≈ 1). Secondly, natural scenes are similar in their geometric properties as they each contain a similar density of structure across spatial and temporal scales—a property which classifies them as fractal (e.g. how the branching pattern of rivers and trees are similar irrespective of scale). Recent research suggests that the visual system is preferentially tuned to natural geometry over photometry. However, so far research has been restricted to the spatial domain. It is currently unclear whether this tuning extends to the temporal domain (e.g. how waves roll into the shore). Here, we use a psychophysics task (4AFC) to measure discrimination sensitivity (N = 90) to synthetic noise movies that varied across three movie types—greyscale, thresholded, and edges. Each movie type shared the same geometric properties (measured using fractal D), but differed substantially in their photometric properties (measured α). We observe a characteristic dependency on geometry across movie types where sensitivity peaks for stimuli with natural geometry despite large differences in their photometric properties in both space and time. This preferential tuning may not be surprising given the stability of structure in natural scenes irrespective of scene illumination (e.g. the structural properties of a tree do not change whether it is morning or night despite large changes in illumination/photometric properties across the two time points). Whilst only measured here at the behavioural level, our findings may infer that the neural processes underlying this tuning may have evolved to be sensitive to the most stable signal in our natural environment—structure.