Role Of Visual Cortex In Depth Perception

National Eye Institute

Investigators

Linked publications, trials & patents

Abstract

A fundamental problem faced by the brain is posed by the fact that the activity of individual neurons is noisy - an identical stimulus presented many times will elicit different response rates. This stimulus-independent spike-rate variability tends to be weakly correlated between neuronal pairs in sensory cortex. It is frequently suggested that these noise correlations reflect the stochastic nature of the afferent pathways that encode sensory input. Importantly, these correlations can place hard limits on the fidelity of sensory encoding. When combining the outputs of many neurons, noise that is independent across neurons makes negligible contribution. However, noise that is shared by neurons remains in the combined output. The extent to which correlated noise limits the brains ability to represent information about the external world depends on the exact structure of the correlations. These correlations also determine how reliably a single neuron can predict upcoming choices of an animal performing a threshold discrimination. However, this work all assumes that the source of these correlations is in the sensory afferents, which means they are not known to the rest of the brain. If some part of the noise correlation in sensory populations is generated by centrally generated signals, then a decoder with access to these signals could achieve much higher performance that current theoretical work suggests. We have previously produced evidence that much of this correlation is generated by top-down signals, generated within the brain, in response to the demands of a particular task. From the standpoint of information processing, this result is very puzzling. The brain appears to add correlated noise that is detrimental to performance. Although this loss can be offset by incorporating knowledge of these top-down signals, it is unclear what benefit they might confer. Here we examine the possibility that these top-down signals support perceptual stability. The central innovation was to study noise correlations while animals performed a threshold task with a bistable stimulus. This consisted of moving dots on a screen that reproduced the image of a rotating three-dimensional cylinder. While this stimulus produces a compelling sensation of a three dimensional cylinder, the direction of rotation is not specified by the stimulus (if no binocular disparity is present). As a result, subjects perceive different directions at different times, with occasional spontaneous reversals in the percept. Because the precept is stable for many seconds, it is hard to explain the percept at any one moment as the result of feedforward noise (which has a short half-life). If appropriate binocular disparity is given to the moving dots, the cylinder rotation is rendered unambiguous. Binocular disparity can then be used to train animals to report the perceived direction of rotation of a cylinder stimulus. In animals trained to perform this task, we recorded from populations of neurons in area MT, where neurons are selective for the direction of rotation when defined with disparity. During presentation of the ambiguous stimulus, we found strikingly high noise correlations (more than double the largest previously reported), with a mean spike count correlation of 0.4. This paints a radically different picture of population activity during perceptual decision making, which whole populations changing their firing rates in a coordinated fashion to follow change in perceptual state. This striking difference from results with previous tasks suggest that top-down signals in sensory cortex play a role in supporting perceptual stability.