ABSTRACT
The visual system is one of the most thoroughly studied sensory regions in the brain due to its importance to humans. (The standard reference for neuroscience is [10] and in particular, chapter 27, but see also chapter 11 by Bressloff and Cowan in this volume.) Indeed, unlike many other mammals, our understanding of the outside world is through our eyes. The transformation of this limited range in the electromagnetic spectrum into a conscious image is complex and involves many different regions of the brain. The inputs to the visual system start at the eye. Light enters the eye where it excites photoreceptors. These receptors connect to several layers of neurons within the retina ultimately exciting the retinal ganglion cells. Retinal ganglion cells respond best to spots of light. They are organized topographically so that nearby ganglion cells respond to nearby points of light in the visual field. The ganglion cells send axons into a region of the brain called the lateral geniculate nucleus (LGN). Images presented on the left-hand side end up projecting to the right LGN (see figure 12.1(A)) and vice versa. Topography is maintained in the LGN as well; that is, nearby ganglion cells project to nearby neurons in the LGN. The inputs remain segregated in different layers of the LGN according to whether they come from the left or right eye. Like ganglion cells, LGN neurons respond best to spots of light. LGN neurons then project to layer 4 of the visual cortex. Topography is maintained in the cortex but, additionally, there are new forms of organization in the projections from the cortex to the LGN. Inputs from the left and right eyes project to layer 4C of the cortex in a regular periodic pattern; see figures 12.1(B) and 12.3. Unlike LGN neurons
which respond best to spots of light, cortical neurons are excited by oriented bars. Figure 12.1(C) shows the generally accepted mechanism. A bar of light excites an array of LGN neurons which are connected to a cortical neuron. If the light bar is of the appropriate orientation (in this case horizontal), then the cortical neuron will fire at a maximal rate. Thus, the simple diagrams in figures 12.1(B) and (C) suggest that if one records the activity in a cortical neuron, then whether it fires or not depends on the properties of the stimulus: e.g. the position in the visual field of the stimulus, the occularity (left or right eye), or the orientation. These tuned cortical neurons are not randomly dispersed across the visual area. Rather, their properties form well-ordered maps. There are at least three well-studied maps in the mammalian visual system:
(i) the topographic map, (ii) the orientation map, and (iii) the occular dominance map. In this chapter, we will take the topographic map for granted and assume that it already exists. Each neuron in the cortex responds maximally to a particular orientation and thus, one can project these preferred orientations on the twodimensional surface of the primary visual cortex. It is experimentally possible to see this map; for example, figure 12.2 shows the orientation map from the tree shrew. We point out several aspects of the map: (i) there are ‘singularities’ at spatially periodic intervals and (ii) there are linear regions where the orientation varies smoothly. The precise structure of the map is different in every animal so that it is not ‘hardwired’ genetically but rather arises during development of the visual system.
