Hegdé Jay, Felleman Daniel J
Department of Psychology, University of Minnesota, Minneapolis, Minnesota 55455, USA.
Neuroscientist. 2007 Oct;13(5):416-21. doi: 10.1177/1073858407305201.
The primate visual system has been shown to be organized into an anatomical hierarchy by the application of a few principled criteria. It has been widely assumed that cortical visual processing is also hierarchical, with the anatomical hierarchy providing a defined substrate for clear levels of hierarchical function. A large body of empirical evidence seemed to support this assumption, including the general observations that functional properties of visual neurons grow progressively more complex at progressively higher levels of the anatomical hierarchy. However, a growing body of evidence, including recent direct experimental comparisons of functional properties at two or more levels of the anatomical hierarchy, indicates that visual processing neither is hierarchical nor parallels the anatomical hierarchy. Recent results also indicate that some of the pathways of visual information flow are not hierarchical, so that the anatomical hierarchy cannot be taken as a strict flowchart of visual information either. Thus, while the sustaining strength of the notion of hierarchical processing may be that it is rather simple, its fatal flaw is that it is overly simplistic.
通过应用一些有原则的标准,灵长类动物的视觉系统已被证明是按照解剖学层次组织起来的。人们普遍认为,皮层视觉处理也是分层的,解剖学层次为明确的分层功能水平提供了确定的基础。大量的经验证据似乎支持这一假设,包括一般观察结果,即视觉神经元的功能特性在解剖学层次结构中逐渐升高的水平上逐渐变得更加复杂。然而,越来越多的证据,包括最近对解剖学层次结构中两个或更多水平的功能特性进行的直接实验比较,表明视觉处理既不是分层的,也不与解剖学层次结构平行。最近的结果还表明,一些视觉信息流的途径不是分层的,因此解剖学层次结构也不能被视为视觉信息的严格流程图。因此,虽然分层处理概念的持久优势可能在于它相当简单,但其致命缺陷在于它过于简单化。