Neural Mechanisms of three dimensional shape perception in visual cortex

三维物体的感知是稳健和稳定的，尽管零碎和不连续的感官信息。这对于仅基于2D图像的视觉来说尤其如此。视觉系统通过双眼视差和其他单眼深度线索（如运动视差、透视、阴影和遮挡）来推断物体的深度。这些线索都不是经常可用的，而且大多数都是模棱两可的。虽然皮层神经元的视差选择性已被充分记录，但灵长类视觉皮层对单眼深度线索的处理和线索组合的机制尚不清楚。知觉的稳定性表明存在线索不变的皮层深度表征。本研究的目的是确定(a)基本的3D特征是如何在视觉皮层中表现出来的，(b)不同的单眼和双眼线索是如何促进这种表现的，以及(c)来自不同线索的信息是如何组合的，特别是在模糊和冲突的情况下。所提出的方法是，在相同的3D物体和显示条件下，研究(1)人类和猴子受试者的深度感知，以及(2)清醒、注视的猴子皮层V1、V2和V4区域的神经元活动。

Perception of three-dimensional (3D) objects is robust and stable, despite fragmentary and discontinuous sensory information. This is especially true for vision which is based only on 2D images. The visual system infers the depth of objects form binocular disparity and other, monocular, depth cues, such as motion parallax, perspective, shading, and occlusion. None of these cues is constantly available, and most of them are ambiguous. While disparity selectivity of cortical neurons is well documented, little is known about the processing in primate visual cortex of the monocular depth cues and the mechanisms of cue combination. The stability of perception suggests the existence of cortical depth representations that are cue invariant. The aim of this research is to determine (a) how elementary 3D features are represented in the visual cortex, (b) how different monocular and binocular cues contribute to this representation, and (c) how information from different cues is combined, especially in situations of ambiguity and conflict. The proposed approach is to study, for the same 3D objects and display conditions, (1) depth perception in human and monkey subjects, and (2) the neuronal activity in cortical areas V1, V2, and V4, of the awake, fixating monkey.