Dow B M
Fed Proc. 1976 Jan;35(1):54-9.
Despite repeated attempts by several laboratories to discover increasingly complex "feature detector" neurons in higher visual centers of mammals, there has been little convincing evidence for the existence of such neurons. What had been reported instead is that many neurons in higher centers show less rather than more specificity when compared with cells in areas 17, 18 and 19. Studies in mammalian retina have revealed multiple processing systems apparently operating in parallel at the level of the ganglion cells. Striate cortex receives at least two different kinds of visual input from the lateral geniculate, and sends at least two different parallel outputs to other brain regions. Within striate cortex there is some segregation of different functional cell types in separate layers. The accumulated evidence suggests the existence of parallel visual processing mechanisms beginning in mammalian retina and extending through striate cortex to higher cortical centers. The notion of separate processing systems for the detection of different features in the visual world recalls earlier work by Lettvin, Maturana, McCulloch and Pitts concerning visual processing in the frog retina and optic tectum.
尽管多个实验室多次尝试在哺乳动物的高级视觉中枢中发现越来越复杂的“特征探测器”神经元,但几乎没有令人信服的证据证明此类神经元的存在。相反,所报道的情况是,与17区、18区和19区的细胞相比,高级中枢中的许多神经元表现出的特异性更低而非更高。对哺乳动物视网膜的研究表明,在神经节细胞水平上,多个处理系统显然是并行运作的。纹状皮质从外侧膝状体接收至少两种不同类型的视觉输入,并向其他脑区发送至少两种不同的并行输出。在纹状皮质内,不同功能的细胞类型在不同层中存在一定程度的分离。积累的证据表明,从哺乳动物视网膜开始,经纹状皮质延伸至更高皮质中枢,存在并行视觉处理机制。视觉世界中用于检测不同特征的单独处理系统这一概念,让人想起莱特文、马图拉纳、麦卡洛克和皮茨早期关于青蛙视网膜和视顶盖视觉处理的研究。