The following is a summary of “Sensitivity to naturalistic texture relies primarily on high spatial frequencies,” published in the February 2023 issue of Ophthalmology by Lieber, et al.
Information at many spatial scales can be found in natural photographs. Although early visual mechanisms may divide multiscale images into multiple spatial frequency channels, mid-level visual processes are still needed to fully comprehend how to interpret these channels’ outputs. To separate these mid-level mechanisms, researchers recently created a texture discrimination task that used artificial, multi-scale, “naturalistic” textures.
They demonstrated that perceptual sensitivity to naturalistic structure depends heavily on characteristics at high object spatial frequencies (measured in cycles/image) using three experimental manipulations (image blur, picture rescaling, and eccentric looking). Because of this, sensitivity is dependent on a texture acuity limit, a characteristic of the visual system that establishes the maximum retinal spatial frequency (measured in cycles/degrees) at which viewers can identify realistic details. Naturalistic picture characteristics at high object spatial frequencies transmit more task-relevant information than those at low object spatial frequencies, according to the examination of the texture images using a model observer analysis.
In other words, rather than being a feature of the visual system, the dependency of sensitivity on high object spatial frequencies is a trait of the texture images. As a result, they discovered that human observers’ efficiency in extracting naturalistic information was consistent across all object spatial frequencies. Furthermore, they concluded that regardless of the retinal and object spatial frequency of all picture characteristics below the texture acuity limit, the mid-level systems underlying perceptual sensitivity successfully harvested information from them.