Sato Takayuki
Laboratory for Integrative Neural Systems, RIKEN Brain Science Institute, Saitama, Japan.
Brain Nerve. 2012 Jul;64(7):831-9.
To gain an understanding of object recognition, it is important to identify the neural basis for object recognition. In this review, I have attempted to explain which neural structures are involved in these functions, and how these structures are organized within the brain. Electrophysiological and optical imaging studies suggest that object images are represented by a combination of the columnar activities in the monkey inferior temporal (IT) cortex. Each column represents a visual feature of the object. However, fMRI studies have extracted category-specific activity patches that were formed by neurons responding to the specific object category. These studies suggest that there are 2 functional organization levels, which represent object images and categories in IT cortex. Recently, dense electrophysiological mapping was conducted in a wide range of IT cortex. The results showed that columns that have similar object selectivity are located close to each other, forming functional domains. The selectivity of some domains was related to object category. In a domain found to be selective for faces, columns within the domain showed variability in their selectivity for finer categories, such as human faces or monkey faces. These findings suggest that there are hierarchical functional structures that represent object images and categories. Both structures were embedded in the same IT cortex.
为了理解物体识别，确定物体识别的神经基础很重要。在这篇综述中，我试图解释哪些神经结构参与这些功能，以及这些结构在大脑中是如何组织的。电生理和光学成像研究表明，物体图像由猴子颞下（IT）皮质中的柱状活动组合来表征。每一列代表物体的一个视觉特征。然而，功能磁共振成像（fMRI）研究提取了由对特定物体类别做出反应的神经元形成的类别特异性活动斑块。这些研究表明，在IT皮质中有两个功能组织水平，分别代表物体图像和类别。最近，在广泛的IT皮质区域进行了密集的电生理映射。结果显示，具有相似物体选择性的柱彼此靠近，形成功能域。一些域的选择性与物体类别有关。在一个被发现对面部有选择性的域中，该域内的柱对更精细的类别（如人脸或猴脸）的选择性存在差异。这些发现表明，存在代表物体图像和类别的层次性功能结构。这两种结构都嵌入在同一个IT皮质中。
