Toward an understanding of the neural processing for 3D shape perception
Toward an understanding of the neural processing for 3D shape perception
复制标题
DOI：
10.1016/j.neuropsychologia.2004.11.003
复制
发表时间：
2005-12
期刊：
影响因子：
2.6
通讯作者：
H. Sakata;K. Tsutsui;M. Taira
中科院分区：
文献类型：
--
作者：
H. Sakata;K. Tsutsui;M. Taira
在这篇文章中，我们讨论了空间视觉的主要问题，即大脑如何在心理物理学和神经生理学的基础上表示现实世界中物体的3D形状。在心理物理学中，吉布森发现纹理梯度和宽度梯度以及双目视差的梯度是表面方向深度的主要线索。Marr提出，基于曲面的表示是迈向3D形状表示的主要步骤。在我们对猴顶叶皮质的神经生理学研究中，我们在AIP区发现了视觉占优势的神经元，在物体的3D形状上具有选择性，在尾侧顶内(CIP)区也发现了表面定向选择(SOS)神经元。在带有视差梯度的随机点立体图中，SOS神经元对表面深度方向有选择性的反应。许多SOS神经元也选择性地对纹理梯度和线性透视做出反应，这些线索的组合增强了它们的反应。我们还在CIP区发现了轴向选择(AOS)神经元，选择性地对细长物体纵轴的方向做出深度反应。我们在这里提供的初步数据表明，一些偏爱中等厚度的AOS神经元是形状选择性的，它们可能区分表面曲率。这些数据表明，CIP区及其周围的神经元可能具有代表3D形状的能力。
In this article we address the major issue of space vision, how the brain represents the 3D shape of objects in the real world, on the basis of psychophysics and neurophysiology. In psychophysics, Gibson found texture gradients and width gradients, as well as the gradient of binocular disparity, as the major cues for surface orientation in depth. Marr proposed that a surface-based representation is the main step towards 3D shape representation. In our neurophysiological studies of the monkey parietal cortex, we have found visual-dominant neurons in area AIP with selectivity in 3D shape of the objects, and also surface-orientation-selective (SOS) neurons in the caudal intraparietal (CIP) area. SOS neurons responded selectively to surface orientation in depth presented in random dot stereograms with a disparity gradient. Many of the SOS neurons responded selectively also to texture gradients and linear perspective, and their responses were enhanced by the combination of these cues. We also found axis-orientation-selective (AOS) neurons in area CIP, responding selectively to the orientation of the longitudinal axis of elongated objects in depth. We present preliminary data here to demonstrate that some of AOS neurons that prefer intermediate thickness are shape-selective, and they are likely to discriminate surface curvature. These data suggest that neurons in and around area CIP may have the capacity to represent 3D shape.