We have already seen that the spinal cord can receive sensory input, integrate it, and produce motor output that is totally independent of the brain. The brain also receives this sensory information and uses it to control the motor activity of the spinal reflexes and central pattern generators. How does the brain organize this sensory input and motor output? In many cases, it organizes these functions spatially with neural maps.
In everyday life, we use maps to represent spatial locations. You may use endless ways to construct a map, depending on which features of an area you want to highlight and what sort of transformation you make as you take measurements from the source (the thing being mapped) and place them on the target (the map). Maps of the earth may emphasize topography, the road system, political boundaries, distributions of air temperature and wind direction, population density, or vegetation. A map is a model of a part of the world—and a very limited model at that. The brain also builds maps, most of which represent very selected aspects of our sensory information about the environment or the motor systems controlling our body. These maps can represent spatial qualities of various sensory modalities (e.g., a place in the visual field) or nonspatial qualities (e.g., smell).
The nervous system contains maps of sensory and motor information
Almost all sensory receptors are laid out in planar sheets. In some cases, these receptor sheets are straightforward spatial maps of the sensory environment that they encode. For example, the somatic sensory receptors of the skin literally form a map of the body surface. Similarly, a tiny version of the visual scene is projected onto the mosaic of retinal photoreceptors. The topographies of other sensory receptor sheets represent qualities other than spatial features of the sensory stimuli. For example, the position of a hair cell along the basilar membrane in the cochlea determines the range of sound frequencies to which it will respond. Thus, the sheet of hair cells is a frequency map of sound rather than a map of the location of sounds in space. Olfactory and taste receptors also do not encode stimulus position; instead, because the receptor specificity varies topographically, the receptor sheets may be chemical maps of the types of stimuli. The most interesting thing about sensory receptor maps is that they often project onto many different regions of the CNS. In fact, each sensory surface may be mapped and remapped many times within the brain, the characteristics of each map being unique. In some cases, the brain constructs maps of stimulus features even when these features are not mapped at the level of the receptors themselves. Sound localization is a good example of this property (see the next section). Some neural maps may also combine the features of other neural maps, for example, overlaying visual information with auditory information.
The cerebral cortex has multiple visuotopic maps
Some of the best examples of brain maps are those of the visual fields. Figure 16-8A shows the basic anatomical pathway extending from the retina to the lateral geniculate nucleus of the thalamus and on to the primary visual cortex (area V1). Note that area V1 actually maps the visual thalamus, which in turn maps the retina, the first visuotopic map in the brain. Thus, the V1 map is sometimes referred to as a retinotopic map. Figure 16-8B shows how the visual fields are mapped onto cortical area V1. The first thing to notice is that the left half of the visual field is represented on the right cortex and the upper half of the visual field is represented on the lower portions of the cortex. This orientation is strictly determined by the system's anatomy. For example, all the retinal axons from the left-most halves of both eyes (which are stimulated by light from the right visual hemifield) project to the left half of the brain. Compare the red and blue pathways in Figure 16-8A. During development, each axon must therefore make an unerring decision about which side of the brain to innervate when it reaches the optic chiasm!
FIGURE 16-8 Visual fields and visual maps. A, The right sides of both retinas (which sense the left visual hemifield) project to the left lateral geniculate nucleus (LGN), which in turn projects to the left primary visual cortex (area V1). B, The upper parts of the visual fields project to lower parts of the contralateral visual cortex, and vice versa. Although the fovea represents only a small part of the visual field, its representation is greatly magnified in the primary visual cortex, which reflects the large number of retinal ganglion cells that are devoted to the fovea.
The second thing to notice is that scaling of the visual fields onto the visual cortex—often called the magnification factor—is not constant. In particular, the central region of the visual fields—the fovea—is greatly magnified on the cortical surface. Behavioral importance ultimately determines mapping in the brain. Primates require vision of particularly high resolution in the center of their gaze; photoreceptors and ganglion cells are thus packed as densely as possible into the central retinal region (see p. 363). About half of the primary visual cortex is devoted to input from the relatively small fovea and the retinal area just surrounding it.
Understanding a visual scene requires us to analyze many of its features simultaneously. An object may have shape, color, motion, location, and context, and the brain can usually organize these features to present a seamless interpretation, or image. The details of this process are only now being worked out, but it appears that the task is accomplished with the help of numerous visual areas within the cerebral cortex. Studies of monkey cortex by a variety of electrophysiological and anatomical methods have identified >25 areas that are mainly visual in function, most of which are in the vicinity of area V1. According to recent estimates, humans devote almost half of their neocortex primarily to the processing of visual information. Several features of a visual scene, such as motion, form, and color, are processed in parallel and, to some extent, in separate stages of processing. The neural mechanisms by which these separate features are somehow melded into one image or concept of an object remain unknown, but they depend on strong and reciprocal interconnections between the visual maps in various areas of the brain.
The apparently simple topography of a sensory map looks much more complex and discontinuous when it is examined in detail. Many cortical areas can be described as maps on maps. Such an arrangement is especially striking in the visual system. For example, within area V1 of Old World monkeys and humans, the visuotopic maps of the two eyes remain segregated. In layer IV of the primary visual cortex, this segregation is accomplished by having visual input derived from the left eye alternate every 0.25 to 0.5 mm with visual input from the right. Thus, two sets of information, one from the left eye and one from the right eye, remain separated but adjacent. Viewed edge on, these left-right alternations look like columns (Fig. 16-9A); hence their name: ocular dominance columns, which were identified by David Hubel and Torsten Wiesel, who shared half of the 1981 Nobel Prize in Physiology or Medicine. N16-2 Viewed from the surface of the brain, this alternating left-right array of inputs looks like bands or zebra stripes (see Fig. 16-9B).
FIGURE 16-9 Ocular dominance columns and blobs in the primary visual cortex (area V1). A, Ocular dominance columns are shown as alternating black (right eye) and gray (left eye) structures in layer IV. The alternating light and dark bands are visible in an autoradiograph taken 2 weeks after injecting one eye with 3H-labeled proline and fucose. The 3H label moved from the optic nerve to neurons in the lateral geniculate nucleus and then to the axon terminals in the V1 cortex that are represented in this figure. The blobs are shown as teal-colored pegs in layers II and III. They represent the regular distribution of cytochrome oxidase–rich neurons and are organized in pillar-shaped clusters. B, Cutting the brain parallel to its surface, but between layers III and IV, reveals a polka-dot pattern of blobs in layer II/III and zebra-like stripes in layer IV. (Data from Hubel D: Eye, Brain and Vision. New York, WH Freeman, 1988.)
David Hubel and Torsten Wiesel
David H. Hubel and Torsten N. Wiesel shared the 1981 Nobel Prize in Physiology or Medicine with Roger W. Sperry. Hubel and Wiesel were cited “for their discoveries concerning information processing in the visual system.” For more information visit, http://nobelprize.org/nobel_prizes/medicine/laureates/1981/.
Superimposed on the zebra-stripe ocular dominance pattern in layer IV of the primary visual cortex, but quite distinct from these zebra stripes, layers II and III have structures called blobs. These blobs are visible when the cortex is stained for the mitochondrial enzyme cytochrome oxidase. Viewed edge on, these blobs look like round pegs (see Fig. 16-9). Viewed from the surface of the brain (see Fig. 16-9), the blobs appear as a polka-dot pattern of small dots that are ~0.2 mm in diameter.
Adjacent to the primary visual cortex (V1) is the secondary visual cortex (V2), which has, instead of blobs, a series of thick and thin stripes that are separated by pale interstripes. Some other higher-order visual areas also have striped patterns. Whereas ocular dominance columns demarcate the left and right eyes, blobs and stripes seem to demarcate clusters of neurons that process and channel different types of visual information between areas V1 and V2 and pass them on to other visual regions of the cortex. For example, neurons within the blobs of area V1 seem to be especially attuned to information about color and project to neurons in the thin stripes of V2. Other neurons throughout area V1 are very sensitive to motion but are insensitive to color. They channel their information mainly to neurons of the thick stripes in V2.
Maps of somatic sensory information magnify some parts of the body more than others
One of the most famous depictions of a neural map came from studies of the human somatosensory cortex by Penfield and colleagues. Penfield stimulated small sites on the cortical surface in locally anesthetized but conscious patients during neurosurgical procedures; from their verbal descriptions of the position of their sensations, he drew a homunculus, a little person representing the somatotopy—mapping of the body surface—of the primary somatic sensory cortex (Fig. 16-10A). The basic features of Penfield's map have been confirmed with other methods, including recording from neurons while the body surface is stimulated and modern brain-imaging methods, such as positron emission tomography and functional magnetic resonance imaging. The human somatotopic map resembles a trapeze artist hanging upside down—the legs are hooked over the top of the postcentral gyrus and dangle into the medial cortex between the hemispheres, and the trunk, upper limbs, and head are draped over the lateral aspect of the postcentral gyrus.
FIGURE 16-10 Somatosensory and motor maps. A, The plane of section runs through the postcentral gyrus of the cerebral cortex, shown as a blue band on the image of the brain. B, The plane of section runs through the precentral gyrus of the cerebral cortex, shown as a violet band on the image of the brain. (Data from Penfield W, Rasmussen T: The Cerebral Cortex of Man. New York, Macmillan, 1952.)
Two interesting features should be noticed about the somatotopic map in Figure 16-10A. First, mapping of the body surface is not always continuous. For example, the representation of the hand separates those of the head and face. Second, the map is not scaled like the human body. Instead, it looks like a cartoon character: the mouth, tongue, and fingers are very large, whereas the trunk, arms, and legs are tiny. As was the case for mapping of the visual fields onto the visual cortex, it is clear in Penfield's map that the magnification factor for the body surface is not a constant but varies for different parts of the body. Fingertips are magnified on the cortex much more than the tips of the toes. The relative size of cortex that is devoted to each body part is correlated with the density of sensory input received from that part, and 1 mm2 of fingertip skin has many more sensory endings than a similar patch on the buttocks. Size on the map is also related to the importance of the sensory input from that part of the body; information from the tip of the tongue is more useful than that from the elbow. The mouth representation is probably large because tactile sensations are important in the production of speech, and the lips and tongue are one of the last lines of defense in deciding whether a morsel is a potential piece of food or poison.
The importance of each body part differs among species, and indeed, some species have body parts that others do not. For example, the sensory nerves from the facial whisker follicles of rodents have a huge representation on the cortex, whereas the digits of the paws receive relatively little. Rodent behavior explains this paradox. Most are nocturnal, and to navigate they actively sweep their whiskers about as they move. By touching their local environment, they can sense shapes, textures, and movement with remarkable acuity. For a rat or mouse, seeing things with its eyes is often less important than “seeing” things with its whiskers.
As we have already seen for the visual system, other sensory systems usually map their information numerous times. Maps may be carried through many anatomical levels. The somatotopic maps in the cortex begin with the primary somatic sensory axons (see Table 12-1) that enter the spinal cord or the brainstem, each at the spinal segment appropriate to the site of the information that it carries. The sensory axons synapse on second-order neurons, and these cells project their axons into various nuclei of the thalamus and form synapses. Thalamic relay neurons in turn send their axons into the neocortex. The topographical order of the body surface (i.e., somatotopy) is maintained at each anatomical stage, and somatotopic maps are located within the spinal cord, the brainstem, and the thalamus as well as in the somatosensory cortex. Within the cortex, the somatic sensory system has several maps of the body, each unique and each concerned with different types of somatotopic information. Multiple maps are the rule in the brain.
The cerebral cortex has a motor map that is adjacent to and well aligned with the somatosensory map
Neural maps are not limited to sensory systems; they also appear regularly in brain structures that are considered to have primarily motor functions. Studies done in the 1860s by Fritsch and Hitzig showed that stimulation of particular parts of the cerebral cortex evokes specific muscle contractions in dogs. Penfield and colleagues generated maps of the primary motor cortex in humans (see Fig. 16-10B) by microstimulating and observing the evoked movements. They noted an orderly relationship between the site of cortical stimulation and the body part that moved. Penfield's motor maps look remarkably like his somatosensory maps, which lie in the adjacent cortical gyrus (see Fig. 16-10A). Note that the sensory and motor maps are adjacent and similar in basic layout (legs represented medially and head laterally), and both have a striking magnification of the head and hand regions. Not surprisingly, there are myriad axonal interconnections between the primary motor and primary somatosensory areas. However, functional magnetic resonance imaging of the human motor cortex shows that the motor map for hand movements is not nearly as simple and somatotopic as Penfield's drawings might imply. Movements of individual fingers or the wrist that are initiated by the individual activate specific and widely distributed regions of motor cortex, but these regions also overlap one another. Rather than following an obvious somatotopic progression, it instead appears that neurons in the arm area of the motor cortex form distributed and cooperative networks that control collections of arm muscles. Other regions of the motor cortex also have a distributed organization when they are examined on a fine scale, although Penfield's somatotopic maps still suffice to describe the gross organization of the motor cortex.
In other parts of the brain, motor and sensory functions may even occupy the same tissue, and precise alignment of the motor and sensory maps is usually the case. For example, a paired midbrain structure called the superior colliculus receives direct retinotopic connections from the retina as well as input from the visual cortex. Accordingly, a spot of light in the visual field activates a particular patch of neurons in the colliculus. The same patch of collicular neurons can also command, through other brainstem connections, eye and head movements that bring the image of the light spot into the center of the visual field so that it is imaged onto the fovea. The motor map for orientation of the eyes is in precise register with the visual response map. In addition, the superior colliculus has maps of both auditory and somatosensory information superimposed on its visual and motor maps; the four aligned maps work in concert to represent points in polysensory space and help control an animal's orienting responses to prominent stimuli (Fig. 16-11).
FIGURE 16-11 Polysensory space in the superior colliculus. A, The representation of visual space projected onto the right superior colliculus of a cat. Note that visual space is divided into nasal versus temporal space and superior versus inferior space. B and C, Comparable auditory and somatosensory maps, respectively. D, Superimposition of the preceding three maps. Note the approximate correspondence among the visual (red), auditory (green), and somatosensory (blue) maps. The motor map for orienting the eyes (not shown) is in almost perfect register with the visual map in A. (Data from Stein BE, Wallace MT, Meredith MA: Neural mechanisms mediating attention and orientation to multisensory cues. In Gazzaniga M [ed]: The Cognitive Neurosciences. Cambridge, MA, MIT Press, 1995.)
Sensory and motor maps are fuzzy and plastic
We have described a sample of the sensory and motor maps in the brain, but we are left to wonder just why neural maps are so ubiquitous, elaborate, and varied. What is the advantage of mapping neural functions in an orderly way? You could imagine other arrangements: spatial information might be widely scattered about on a neural structure, much as the bytes of one large digital file may be scattered across the array of memory elements in a computer. Various explanations may be proposed for the phenomenon of orderly mapping in the nervous system, although most remain speculations. Maps may be the most efficient way of generating nearest-neighbor relationships between neurons that must be interconnected for proper function. For example, the collicular neurons that participate in sensing stimuli 10 degrees up and 20 degrees to the left and other collicular neurons that command eye movements toward that point undoubtedly need to be strongly interconnected. Orderly collicular mapping enforces togetherness for those cells and minimizes the length of axons necessary to interconnect them. In addition, if brain structures are arranged topographically, neighboring neurons will be most likely to become activated synchronously. Neighboring neurons are very likely to be interconnected in structures such as the cortex, and their synchronous activity serves to reinforce the strength of their interconnections because of the inherent rules governing synaptic plasticity (see pp. 328–333).
An additional advantage of mapping is that it may simplify establishment of the proper connections between neurons during development. For example, it is easier for an axon from neuron A to find neuron B if distances are short. Maps may thus make it easier to establish interconnections precisely among the neurons that represent the three sensory maps and one motor map in the superior colliculus. Another advantage of maps may be to facilitate the effectiveness of inhibitory connections. Perception of the edge of a stimulus (edge detection) is heightened by lateral connections that suppress the activity of neurons representing the space slightly away from the edge. If sensory areas are mapped, it is a simple matter to arrange the inhibitory connections onto nearby neurons and thereby construct an edge-detector circuit.
It is worth clarifying several general points about neural maps. “The map is not the territory,” as the philosopher Alfred Korzybski pointed out. In other words, all maps, including neural maps, are abstract representations. They are also distorted by the shortcomings of particular experimental measurements. A problem with neural maps is that different experimenters, using different methods, may sometimes generate quite different maps of the same part of the brain. As more and better-refined methods become available, our understanding of these maps is evolving. Moreover, the brain itself muddies its maps. Maps of sensory space onto a brain area are not point-to-point representations. On the contrary, a point in sensory space (e.g., a spot of light) activates a relatively large group of neurons in a sensory region of the brain. However, such activation of many neurons is not due to errors of connectivity; the spatial dissemination of activity is part of the mechanism used to encode and to process information. The strength of activation is most intense within the center of the activated neuronal group, but the population of more weakly activated neurons may encompass a large portion of an entire brain. This diversity in strength of activation means that a point in sensory space is unlikely to be encoded by the activity of a single neuron; instead it is represented by the distributed activity in a large population of neurons. Such a distributed code has computational advantages, and some redundancy also guards against errors, damage, and loss of information.
Finally, maps may change with time. All sensory and motor maps are clearly dynamic and can be reorganized rapidly and substantially as a function of development, behavioral state, training, or damage to the brain or periphery. Such changes are referred to as plasticity. Figure 16-12 illustrates two examples of dramatic changes in neocortical mapping, one sensory and one motor, after damage to peripheral nerves. In both cases, severing a peripheral nerve causes the part of the map that normally relates to the body part served by this severed nerve to become remapped to another body part. Although the mechanisms of these reorganizations are only partially known, they probably reflect the same types of processes that underlie our ability to learn sensorimotor skills with practice and to adjust and improve after neural damage from trauma or stroke.
FIGURE 16-12 Plasticity of maps. A, The first panel on the right, labeled “Normal organization,” shows the somatotopic organization of the right hand in the left somatosensory cortex of the monkey brain. The colors correspond to different regions of the hand (viewed from the palm side, except for portions labeled “Dorsum”). The second panel shows (in gray) the territory that is deprived of input by sectioning the median nerve. The third panel shows that the cortical map is greatly changed several months after nerve section. The nerve was not allowed to regrow, but the previously deprived cortical region now responds to the dorsal skin of D3, D2, and D1. Notice that responses to regions P1, P2, and T have disappeared; region I has encroached; and regions H and P3 have suddenly appeared at a second location. B, The first panel on the right, labeled “Normal organization,” shows the somatotopic organization of the left motor cortex (M1) of the rat brain. The colors correspond to the muscles that control different regions of the body. The second panel shows (in gray) the territory that normally provides motor output to the facial nerve, which has been severed. The third panel shows that, after several weeks, the deprived cortical territory is now remapped. Notice that the deprived territory that once evoked whisker movements now evokes eye, eyelid, and forelimb movements. FL, additional representation of forelimb; N, neck area. (A, Data from Kaas JH: The reorganization of sensory and motor maps in adult mammals. In Gazzaniga M [ed]: The Cognitive Neurosciences. Cambridge, MA, MIT Press, 1995; B, data from Sanes J, Suner S, Donoghue JP: Dynamic organization of primary motor cortex output to target muscles in adult rats: Long-term patterns of reorganization following motor or mixed peripheral nerve lesions. Exp Brain Res 79:479–491, 1990.)