Rapid Selectivity to Natural Images Across Layers of Primate V1

Abstract

Visual systems are thought to have adapted to the statistical properties of natural scenes. However, the extent to which visual neurons respond selectively to natural images, and the stage at which that selectivity emerges remains unclear. To address these questions, we recorded the visual activity of neurons in macaque V1 using high-density electrode arrays (Neuropixels), and compared neuronal responses to images presented at three levels of naturalness. We found that within 60 ms of stimulus onset, neurons in all cortical layers, including input layers 4C alpha and beta, responded more vigorously to natural images than to statistically matched naturalistic texture and noise images. The result remained when residual variations in the local image statistics were factored out. V1 neurons also showed high population and lifetime sparseness for natural images. Across the population of V1 neurons, sensitivity to natural images exceeded the sensitivity to other image categories. The results reveal a rapid and pervasive preference for natural images is present at the earliest stages of cortical processing.