The visual pathway. Each structure, shown as a box, consists of millions of cells, aggregated into sheets. Each receives inputs from one or more structures at lower levels in the path and each sends its output to several structures at higher levels. The path has been traced only for four or five stages beyond the primary visual cortex.
An experimental plan for recording from the visual pathway. The animal, usually a macaque monkey, faces a screen onto which we project a stimulus. We record by inserting a microelectrode into some part of the pathway, in this case, the primary visual cortex. (The brain in this diagram is from a human, but a monkey brain is very similar.)
these are called dendrites. The entire nerve cell—the cell body, axon, and dendrites—is enclosed in the cell membrane.
The cell body and dendrites receive information from other nerve cells; the axon transmits information from the nerve cell to other nerve cells.
The axon can be anywhere from less than a millimeter to a meter or more in length; the dendrites are mostly in the millimeter range. Near the point where it ends, an axon usually splits into many branches, whose terminal parts come very close to but do not quite touch the cell bodies or dendrites of other nerve cells. At these regions, called synapses, information is conveyed from one nerve cell, the presynaptic cell, to the next, the postsynaptic cell.
The signals in a nerve begin at a point on the axon close to where it joins the cell body; they travel along the axon away from the cell body, finally invading the terminal branches. At a terminal, the information is transferred across the synapse to the next cell or cells by a process called chemical transmission, which we take up in Chapter 2.
Far from being all the same, nerve cells come in many different types. Although we see some overlap between types, on the whole the distinctiveness is what is impressive. No one knows how many types exist in the brain, but it is certainly over one hundred and could be over one thousand. No two nerve cells are identical. We can regard two cells of the same class as resembling each other about as closely as two oak or two maple trees do and regard two different classes as differing in much the same way as maples differ from oaks or even from dandelions. You should not view classes of cells as rigid divisions:
whether you are a splitter or a lumper will determine whether you think of the retina and the cerebral cortex as each containing fifty types of cells or each half a dozen (see the examples on the this).
The connections within and between cells or groups of cells in the brain are usually not obvious, and it has taken centuries to work out the most prominent pathways. Because several bundles of fibers often streak through each other in dense meshworks, we need special methods to reveal each bundle separately. Any piece of brain we choose to examine can be packed to an incredible degree with cell bodies, dendrites, and axons, with little space between. As a result, methods of staining cells that can resolve and reveal the organization of a more loosely packed structure, such as the liver or kidney, produce only a dense black smear in the brain. But neuroanatomists have devised powerful new ways of revealing both the separate cells in a single structure and the connections between different structures.
As you might expect, neurons having similar or related functions are often interconnected. Richly interconnected cells are often grouped together in the nervous system, for the obvious reason that short axons are more efficient:
they are cheaper to make, take up less room, and get their messages to their destinations faster. The brain therefore contains hundreds of aggregations of cells, which may take the form of balls or of stacks of layered plates. The cerebral cortex is an example of a single gigantic plate of cells, two millimeters thick and a square foot or so in area. Short connections can run between the neurons within a given structure, or large numbers of long fibers that form cables, or tracts, can run from one structure to another. The balls or plates are often connected in serial order as pathways (see the illustration on this page).
A good example of such a serially connected system is the visual pathway.
The retina of each eye consists of a plate having three layers of cells, one of which contains the light-sensitive receptor cells, or rods and cones. As we saw earlier, each eye contains over 125 million receptors. The two retinas send their output to two peanut-size nests of cells deep within the brain, the lateral geniculate bodies. These structures in turn send their fibers to the visual part of the cerebral cortex. More specifically, they go to the striate cortex, or primary visual cortex. From there, after being passed from layer to layer through several sets of synaptically connected cells, the information is sent to several neighboring higher visual areas; each of these sends its output to several others (see the illustration on this page). Each of these cortical areas contains three or four synaptic stages, just as the retina did. The lobe of the brain farthest to the rear, the occipital lobe, contains at least a dozen of these visual areas (each about the size of a postage stamp), and many more seem to be housed in the parietal and temporal lobes just in front of that. Here, however, our knowledge of the path becomes vague.
Our main goal in this book will be to understand why all these chains of neuronal structures exist, how they work, and what they do. We want to know what kind of visual information travels along a trunk of fibers, and how the information is modified in each region—retina, lateral geniculate body, and the various levels of cortex. We attack the problem by using the microelectrode, the single most important tool in the modern era of neurophysiology.
We insert the microelectrode (usually a fine insulated wire) into whatever structure we wish to study—for example, the lateral geniculate body—so that its tip comes close enough to a cell to pick up its electrical signals. We attempt to influence those signals by shining spots or patterns of light on the animal's retina.
Because the lateral geniculate body receives its main input from the retina, each cell in the geniculate will receive connections from rods and cones—not directly but by way of intermediate retinal cells. As you will see in Chapter 3, the population of rods and cones that feed into a given cell in the visual pathway are not scattered about all over the retina but are clustered into a small area. This area of the retina is called the receptive field of the cell. So our first step, in shining the light here and there on the retina, is to find the cell's receptive field. Once we have defined the receptive field's boundaries, we can begin to vary the shape, size, color, and rate of movement of the stimulus—to learn what kinds of visual stimuli cause the cell to respond best.
We do not have to shine our light directly into the retina. It is usually easier and more natural to project our stimuli onto a screen a few meters away from the animal. The eye then produces on the retina a well-focused image of the screen and the stimulus. We can now go ahead and determine the position, on the screen, of the receptive field's projection. If we wish, we can think of the receptive field as the part of the animal's visual world—in this case, the screen—
that is seen by the cell we are recording from.
We soon learn that cells can be choosy, and usually are. It may take some time and groping before we succeed in finding a stimulus that produces a really vigorous response from the cell. At first we may have difficulty even finding the receptive field on the screen, although at early stages, such as in the geniculate, we may locate it easily. Cells in the geniculate are choosy as to the size of a spot they will respond to or as to whether it is black on a white background or white on black. At higher levels in the brain, an edge (the line produced by a light-dark boundary) may be required to evoke a response from some cells, in which case the cells are likely to be fussy about the orientation of the edge—
whether it is vertical, horizontal, or oblique. It may be important whether the stimulus is stationary or moves across the retina (or screen), or whether it is colored or white. If both eyes are looking at the screen, the exact screen distance may be crucial. Different cells, even within the same structure, may differ greatly in the stimuli to which they respond. We learn everything we can think to ask about a cell, and then move the electrode forward a fraction of a millimeter to the next cell, where we start testing all over again.
From any one structure, we typically record from hundreds of cells, in experiments that take hours or days. Sooner or later we begin to form a general idea of what the cells in that structure have in common, and the ways in which they differ. Since each of these structures has millions of cells, we can sample only a small fraction of the population, but luckily there are not millions of kinds of cells, and sooner or later we stop finding new varieties. When we are satisfied, we take a deep breath and go on to the next level—going, for example, from the lateral geniculate body to the striate cortex—and there we repeat the whole procedure. The behavior of cells at the next stage will usually be more complicated than the behavior of cells at the previous level: the difference can be slight or it can be dramatic. By comparing successive levels, we begin to understand what each level is contributing to the analysis of our visual world—what operation each structure is performing on the information it receives so that it can extract from the environment information that is biologically useful to the animal.
By now, the striate cortex has been thoroughly studied in many laboratories. We have far less knowledge about the next cortical area, visual area 2, but there, too, we are beginning to get a fair understanding of what the cells are doing. The same is true of a third area, the middle temporal (MT) area, to which both the striate cortex and visual area 2 connect. From there on, however, our knowledge becomes rapidly more sketchy: in two or three regions we have only a vague idea of the kinds of information that are handled—things such as color or recognition of complex objects such as faces—and after that, for the dozen or so areas that we can be sure are primarily visual, we know practically nothing. But the strategy is clearly paying off, to judge from the rate at which our understanding is increasing. In the chapters to come, I will fill out some of the details of this picture for levels up to and including the striate cortex. In Chapter 2, I describe roughly how impulses and synapses work and give a few examples of neural pathways in order to illustrate some general principles ofneuronal organization. From then on I will concentrate on vision, first on the anatomy and physiology of the retina, then on the physiology of the striate cortex and its anatomy. I next describe the remarkable geometric cortical patterns that result from the fact that cells with similar functions tend to aggregate together. Then will come several special topics:
mechanisms for color perception and depth perception, the function of the fibers that connect the two hemispheres (the corpus callosum), and, finally, the influence of early experience on the visual system. Some parts of the story, such as the sections dealing with the nerve impulse and with color vision, will necessarily be slightly more technical than others. In those cases, I can only hope that you will adhere to the wise advice: "When in perplexity, read on!"
Left: The cerebellar Purkinje cell, shown in a drawing by Santiago Ramon y Cajal, presents an extreme in neuronal specialization. The dense dendritic arborization is not bushlike in shape, but is flat, in the plane of the paper, like a cedar frond.
Through the holelike spaces in this arborization pass millions of tiny axons, which run like telegraph wires perpendicular to the plane of the paper. The Purkinje cell's axon gives off a few initial branches close
to the cell body and then descends to cell clusters deep in the cerebellum some centimeters away, where it breaks up into numerous terminal branches. At life size, the total height of the cell (cell body plus dendrites) is about 1 millimeter. Middle:
Ramon y Cajal made this drawing of a pyramidal cell in the cerebral cortex stained by the Golgi method. At life size, the total height of this drawing would be about 1 millimeter. Only a part of the main axon
(a) is shown: after giving off two branches (c), it might continue out of the picture for a distance of centimeters—even meters—
before ending in a dense bush of branches.
The cell body is the small black blob.
Rif;ht: This drawing by Jennifer Lund shows a cortical cell that would be classed as "stellate". The dark blob in the center is the cell body. Both axons (fine) and dendrites (coarse) branch and extend up and down for a distance of 1 millimeter.
This Golgi stain, in a drawing by Ramon y Cajal, shows a few cells in the upper layers of cerebral cortex in a onc-month-old human baby. Only a tiny fraction of a percent of the cells in the area have stained.