Multiple scales of representation in V1

点击翻译按钮获取中文摘要

 DESCRIPTION (provided by applicant): Primary visual cortex (V1) is likely the best studied sensory cortical area, and is a model for understanding broad principles of cortical processing. Similarly, orientation in V1 is likely one of the simplest and best studied cortical sensory features. Orientation is used as a model for understanding more complex feature processing in other cortical areas, and oriented V1-like receptive fields play an important role in successful computational models of vision. Yet even something as basic as the map of orientation on V1 is inadequately understood. We propose a multidisciplinary series of theoretical and empirical studies to characterize orientation selectivity from the scale of columns to that of retinotopic maps. We will test the hypothesis that coarse-scale biases in orientation preferences are fundamental to understanding the link between orientation-selective neural activity in V1 and orientation perception. We will distinguish and separately measure 3 different processes (stimulus vignetting, cardinal/radial gain fields, and asymmetric surround suppression) that might contribute to coarse-scale orientation biases. Doing so will enable us to characterize the cardinal/radial gain fields in V1 (i.e., the intrinsic representation of the stimulus orientation), independent of the edge effects (from stimulus vignetting and surround suppression), and determine the extent to which the gain field predict a perceptual phenomenon called the oblique effect. We will settle the controversy about fMRI decoding of stimulus orientation by quantifying the relative contribu- tions of fine- (i.e., columnar) vs. intermediate- (i.e., vascular pooling) v. coarse (i.e., stimulus vignetting, asymmetric surround suppression, cardinal/radial gain fields) scale biases in orientation preferences. Resolving the source of the orientation preferences in fMRI measurements from V1 will guide the interpretation of thousands of studies based on multivariate statistical analyses in other brain areas. We will develop and implement a model of the responses of an entire population of neurons in V1, that will enable simulations of all variety of methods: single- and multi-unit firing rates, calcium imaging, optical imaging, and fMRI responses, implemented so that it can be run on any stimulus image, including the stimulus aperture. Amongst other applications, the model will be used to establish the extent to which stimulus vignetting is a confound in vision and visual neuroscience research; almost every study in our field utilizes a stimulus aperture, but ignores the potential impact of the aperture. Variou disorders have been associated with differences in the topography and functional architecture of visual cortex, and/or with differences in visual sensitivity across the visual field. The experimental protocols that we propose will be readily applicable to patient populations. Consequently, the experimental protocols and theoretical principles that we propose will be widely applicable in basic as well as translational research.