Adaptation to Low Signal to Noise Decreases Visual Sensitivity

S. Engel,P. Zhang,M. Bao
DOI: https://doi.org/10.1167/10.7.1126
2010-01-01
Journal of Vision
Abstract:Some neurons in sensory systems will be relatively noisy in a given environment or for a given task. Reducing spiking in such neurons could allow more accurate perception and save limited metabolic resources. Whether the nervous system automatically limits responses of noisy neurons, however, remains unknown. To test this possibility, we measured how subjects' visual sensitivity changed when they adapted to a lowered signal to noise ratio at a specified orientation in the visual environment. Eight subjects viewed the world through an “altered reality” system, comprised of a head mounted gray-scale video camera fed into a laptop computer that drove a head-mounted display (HMD). Vertical information about the world was removed from the video images prior to their display, while keeping overall vertical energy constant. This was performed in real time by randomizing the phases of all vertical Fourier components of the image. Viewing the vertically randomized video images through the HMD, subjects performed everyday tasks, such as playing games, and watching movies in an environment where vertical signals were distracting noise. Prior to and following four hours of adaptation to this environment, contrast detection thresholds were measured for vertical and horizontal sinusoidal patterns (6 deg diameter, 1 cpd, presented 8 deg in the periphery). Following adaptation, vertical thresholds increased by more than 15% relative to horizontal thresholds. A second experiment found a reliable reduction in the apparent contrast of suprathreshold patterns at the noisy orientation following only one hour of adaptation. Sensitivity for simple patterns has been linked to responses of orientation selective neurons in early visual cortex. Exposure to low signal to noise may have caused these neurons to decrease their gain. Such decreases could improve perception in tasks that pool across orientations, and save limited metabolic resources for neurons that signal orientations with higher information content.
What problem does this paper attempt to address?