Stephen Sebastian

Constrained sampling experiments reveal principles of detection in natural scenes

Significance The visibility of a target object may be affected by the specific properties of the background scene at and near the target’s location, and by how uncertain the observer is (from one occasion to the next) about the values of the background and target properties. An experimental technique was used to measure how several background properties, and uncertainty, affect human detection thresholds for target objects in natural scenes. The thresholds varied in a highly lawful fashion—multidimensional Weber’s law—that is predicted directly from the statistical structure of natural scenes. The results suggest that the neural gain control mechanisms underlying multidimensional Weber’s law evolved because they are optimal for detection in natural scenes under conditions of high uncertainty. A fundamental everyday visual task is to detect target objects within a background scene. Using relatively simple stimuli, vision science has identified several major factors that affect detection thresholds, including the luminance of the background, the contrast of the background, the spatial similarity of the background to the target, and uncertainty due to random variations in the properties of the background and in the amplitude of the target. Here we use an experimental approach based on constrained sampling from multidimensional histograms of natural stimuli, together with a theoretical analysis based on signal detection theory, to discover how these factors affect detection in natural scenes. We sorted a large collection of natural image backgrounds into multidimensional histograms, where each bin corresponds to a particular luminance, contrast, and similarity. Detection thresholds were measured for a subset of bins spanning the space, where a natural background was randomly sampled from a bin on each trial. In low-uncertainty conditions, both the background bin and the amplitude of the target were fixed, and, in high-uncertainty conditions, they varied randomly on each trial. We found that thresholds increase approximately linearly along all three dimensions and that detection accuracy is unaffected by background bin and target amplitude uncertainty. The results are predicted from first principles by a normalized matched-template detector, where the dynamic normalizing gain factor follows directly from the statistical properties of the natural backgrounds. The results provide an explanation for classic laws of psychophysics and their underlying neural mechanisms.