Visual Neuroscience

The study of the biological basis of sensation and perception is a fascinating areas of study for two reasons. The first reason is that the pace of discovery is breathtaking. It seems that every few weeks a new set of findings challenges our conceptions of how the brain processes sensory information. The second reason is interrelated with the first. In many ways, the understanding of how the brain processes sensory and perceptual information is used as a basis for understanding how the cerebral cortex, as a whole, operates. The cerebral cortex is where it is believed that our most advanced functions reside. The most extreme version of this approach is seen in Zeki's (1993) Vision of the Brain. Zeki unapologetically uses the knowledge of the architecture and function of the visual areas of the cortex to develop some general ideas about cerebral cortex. While most researchers will not go quite so far, it is certain that some general insights about cerebral cortex will come out of the study of the sensory cortices. A brief review of the current knowledge of the visual system and how it has changed in the last 30 years will illustrate these points.

The idea that the brain operates with many parallel elements is the major emerging theme in our present understanding of the neural bases of the sensory systems; different pathways and different modules of the sensory systems operate to extract unique features of the information in the sensory stimulus. Throughout the sensory systems, there is evidence of parallel (that is simultaneous) operations, both with parallel pathways and parallel targets for sensory information.

Sensory information processing was first viewed as predominantly serial (that is sequential), based upon the groundbreaking work of Hubel and Wiesel (1962, 1968). However, even in these early studies researchers began to find evidence of parallel processing. For example, Hubel and Wiesel uncovered the existence of cortical columns. In a cortical column, all cells process the same feature of the environment from the same location on the retina. For example, one column processes oriented stimuli of a certain width from a certain location on the retina; the next column processes stimuli with a slightly different orientation. Each column is organized in a parallel fashion processing information simultaneously.

Figure 1. An illustration of the difference between serial and parallel processing. Serial processing is in stages and parallel processing is different processes proceeding at the same time.

Another important early finding that foreshadowed todayís emphasis on parallel processing was the study of cat retinal ganglion cells by Christina Enroth-Cugell and John Robson (1966). Enroth-Cugell and Robson identified two types of cells in the feline retina, named X and Y cells, that responded very differently to specific features of light stimuli. X cells tend to respond consistently throughout the time that the stimulus was presented, had relatively small receptive fields, and required fairly high contrast. Y cells, in contrast, tend to respond primarily to stimulus onsets and offsets and do not respond well to a stimulus that does not change over time (see Figure 2). Also, the receptive fields of y cells tend to be larger and require less contrast. These findings, which have been replicated and extended to primates, clearly indicate that there is something fundamentally parallel in the processing of visual information. Studies by Hubel and Wiesel (1968) found a similar functional segregation in the cortex between their simple and complex cells although at the time they interpreted these cells as sequentially linked.

Figure 2. The response patterns of X versus Y cells.

At the lateral geniculate nucleus (LGN) in the thalamus, for example, the visual system is still divided into two main pathways. The LGN is a composed of six layers in the primate. Two of the layers have relatively large cell bodies and are called magnocellar layers while the other four layers have relatively small cell bodies and are called parvocellular layers (Lennie, Trevarthen, Van Essen and Waessle, 1990). The -cellular suffix is often dropped and the two types of cells are called magno and parvo or even M and P. The magno and parvo cells show very different response patterns. In fact, the response patterns are very similar to the X and Y ganglion cells seen in the cat retina. In addition, the parvo pathways carry color information. There is a tendency to apply these distinctions, based initially upon anatomy of the LGN, to the retinal ganglion cells. There is a lot of functional similarity between the magno and Y systems and the parvo and X systems. For example, the magno and Y systems have larger receptive fields and have transient response patterns. Similarly both the parvo and X systems have smaller receptive fields and more sustained response patterns. The Y cells even have relatively large cell bodies just like the Magno cells of the LGN and X cells have relatively small cell bodies like the Parvo cells of the LGN. The overlap of functions and anatomy of the retinal and LGN systems allows this distinction to be nicely applied. See Table 1 for a summary of the distinction between x/parvo and y/magno systems.

Table 1
A Summary of the X/Parvo and Y/Magno systems of the retina, LGN
___________________________________________________________

Response pattern Sustained Transient

Receptive Field Size Smaller Larger

Contrast Sensitivity Relatively Low Relatively High

Color Sensitivity Yes No

The first stop for visual information in the cortex is the striate cortex (V1) in the occipital lobe at the very back of the brain. A little geometry of the cortex is helpful. First, all parts of the neocortex are composed of six layers and each of these layers are involved in the same type of function regardless of where that layer is on the cortex. For example, the fourth layer from the surface of the brain, Layer 4, always receives input from sensory systems (see Kolb and Whishaw, 1996, for a more detailed review of the organization of the cortex). In the sensory areas of the brain, such as the Visual cortex, this layer is very thick. This layer stains very darkly, which gives this first part of the visual cortex its name of striate cortex (See Figure 3).

Brain Surface

Inside of the Brain

Figure 3. An illustration of the layers of the cortex as might be see in V1. Layer 4 is where the input from the LGN occurs and is the thickest layer here. It stains darkly giving this region one of its names, the striate cortext.

Within the striate cortex (V1) the parvo and magno pathways are now segregated into three separate units that function independently: blobs, interblobs, and Layer 4b (Wong-Riley, 1979). Blobs receive inputs from both the magno and parvo system and seem to play an important role in the processing of color. Interblobs receive inputs from only the parvo system and seem to process fine patterns in the stimulus. Layer 4b is a subpart of one of the six layers of the neocortex. Its input is solely from the magno system and these cells seem to respond to motion and very low contrast. Thus, the two parallel pathways have divided now into three pathways. These pathways from the LGN to the striate cortex are summarized in Figure 4 .

Figure 4. The relationship of the layers in the LGN to the different functional parts of V1, the striate cortex.

In the primate, five visual areas have been identified and given the clever names of V1, V2, V3, V4 and V5. Each section is anatomically distinct and is thought to perform different functions relevant to our perception of the world. A simplified view of how these regions are connected is shown in Figure 5. The current evidence suggests that V3 is involved with the processing of form, V4 with color constancy, and V5 complex motion processing. Some of these findings have been supported by the study of lesions in the occipital cortex. For example, a person with a lesion of the human region analogous to primate V4 on the left side of the brain, would report having no color vision in the right half of their visual world (Zeki, 1993).

Figure 5. A simplified diagram of the interconnections between different regions of the cortex.

In addition to all of the connections from V1 and V2 to V3, V4 and V5, each of these regions connects back to V1 and V2. These seemingly backward or reentrant connections are not well understood but one possible role for them is to index the processing in V3, V4 and V5 to the more precise visual maps found in V1 and V2. Although each of the visual regions has a map of the visual world, they are not nearly as precise and detailed as those found in V1 and V2. In other words, their receptive fields are much larger which allows for poorer localization of a stimulus or object. For example, although V4 locates a color region in space, the receptive field is rather large and might not indicate whether this color belongs to the cup or the book on the table in front of you. These reentrant connections, feeding back to the precise maps of V1 and V2 may provide a mechanism that allows the visual system to assign the color precisely to the appropriate location and object, say a part of the cover of the book in front of you.

Our present understanding of the visual system is very different from the serial model Hubel and Wiesel initially proposed. There are separate functional modules that operate relatively independently; information, instead of flowing in one direction, now flows both directions. Thus, later levels do not simply receive information and send it forward, but are in an intimate two-way communication with other modules.

This emerging and modular and parallel view of the visual cortex has helped us better understand some terrible but fortunately rare disorders of the visual system. They are achromatopsia and akinetopsia. Achromatopsia refers to a selective loss of color vision, usually as a result of stroke (Zeki, 1990). It is important to note that the loss seems to be selective to color vision. These patients can read, recognize patterns, and respond to motion. In fact, current research indicates that little if any other visual function is lost, with the exception of occasional but temporary loss of form vision immediately following the injury (Sacks & Wasserman, 1987). These patients describe the world as being very drab and gray. It might be like permanently viewing the world on an old black and white TV set. Apparently this is not a very bright TV set either. With this modular view of the brain, it is possible to understand how a patient might lose color vision only. Loss of the human equivalent of V4 would take out color vision selectively. PET scans on these human patients that allow doctors to examine the living functioning brain, find hat the lesions in achromatopsia is found in V4 (Sacks & Wasserman, 1987).

Akinetopsia is basically the same type of syndrome, only the person looses the perception of motion (Zihl, von Cramon, & Mai, 1983). The patient can only see objects as stationary. They might be perceived in different places at different times but they are not perceived to be moving. At each place they are perceived to be still. Imagine the danger of crossing the street with this condition. This syndrome is associated with damage to visual cortex V5, mentioned above. A with achromatopsia, other visual functions are spared. The person sees in color and can read, for example. Again, our developing modular view of the brain allows these syndromes to be clearly understood.

Where does the study of the biological basis of the brain go from here? Probably in two or three directions. One direction certainly will be the further breaking down of the different sensory regions. Subregions for each of the primary visual areas will be discovered for each of the visual areas discovered so far and new regions will probably be discovered. Some have already been found. For example, while we have a sense for the processing of form, color, and motion, we do not yet know clearly how the brain functions to process depth and texture. A second direction for future research will be how information from these regions is used by the rest of the brain. The neural pathways to the parietal and temporal lobes have been described but little is known about how sensory information is integrated with more cognitive functions. A third possible path would be to better understand how these separate modules lead to unified perceptions. While color can be selectively lost, color is not usually perceived separately from the object. Somehow, that fact, and other integrative perceptual experiences, needs to be a part of our understanding of the brain.

The discovery of the extent of parallel processing in sensory systems has greatly changed how we conceptualize sensory information processing. This work has been coupled with a grown in the use of the computer to understand the operation of the visual system.