Victor J D
Department of Neurology and Neuroscience, Weill Medical College of Cornell University, New York, NY 10021, USA.
Network. 1999 Nov;10(4):R1-66.
The early stages of visual processing provide excellent models for the study of how information is represented in, and processed by, the activity of neurons. The fact that the retina contains both non-spiking and spiking neurons leads us to frame questions about neural coding in a general fashion, rather than in a manner specific either to point processes or continuous signals. In particular, we ask about the role of the statistical structure of the response, the extent to which the neural representation is 'literal', and how information content can be estimated from laboratory data. The broad theme that emerges from a review of experimental data is that each stage of visual processing is accompanied by new features, including adaptive filtering, feedback, rectification and spike generation. These dynamical elements allow an increasingly rich set of strategies for the representation and processing of visual information at retinal and thalamic levels.
视觉处理的早期阶段为研究信息如何在神经元活动中被表征以及如何被处理提供了极佳的模型。视网膜中同时包含非脉冲发放神经元和脉冲发放神经元这一事实,引导我们以一种通用的方式来构建关于神经编码的问题,而非局限于点过程或连续信号的特定方式。具体而言,我们会探讨反应的统计结构的作用、神经表征“如实反映”的程度,以及如何从实验室数据中估计信息内容。对实验数据进行回顾后浮现出的一个广泛主题是,视觉处理的每个阶段都伴随着新的特征,包括自适应滤波、反馈、整流和脉冲生成。这些动态元素为视网膜和丘脑层面视觉信息的表征和处理提供了一套日益丰富的策略。