Topographic Maps In The Brain Research Paper

1. The Discovery Of Sensory And Motor Maps In The Brain

One of the early advances in the field of neuroscience was the realization that the brain has functionally specialized subdivisions and does not act as a whole or single organ. Instead, the brain consists of a number of interacting and interconnected subdivisions, or organs of the brain, usually termed nuclei for subcortical or brain stem structures and areas or fields for neocortex. Some of the most convincing evidence for the view that functions are localized in specific parts of the brain came from early electrophysiological studies that demonstrated that neocortex has topographic maps of receptors and muscles.

The first signs of the existence of maps in the brain came from careful clinical observations of the impairments of patients with cortical lesions or focal epilepsy. Most notably, in the 1860s the neurologist, John Hughlings Jackson, reasoned that only the existence of orderly maps in the brain would explain the experiences of patients with epilepsy where a sensation or movement would start at some location on the body and then expand systematically to include successively more distant body parts. Compelling experimental evidence for such maps soon came from investigators that exposed the neocortex of dogs and monkeys, and directly stimulated the brain through small electrodes applied to the cortical surface. Remarkably, they found that small, brief currents often evoked movements of some part of the body, especially when the electrodes were placed in the frontal lobe of the brain. While movements could be evoked from many locations in the brain, it soon became obvious that movements could be evoked with the smallest currents in a strip of cortex in the back of the frontal lobe, and that the type of movement depended on the location of the electrode in this strip. In monkeys and in humans, movements of the foot, trunk, hand, and face could be evoked in sequence as the electrode was moved across the strip. Similar organizations were identified in the most posterior portion of frontal cortex in a range of mammalian species, and this cortex became known as primary motor cortex, or M1.

A second systematic representation of body movements was soon identified in cortex along the middle edge of the cerebral hemisphere just ahead of M1, first in human patients undergoing brain surgery and later in other mammals. This motor representation was termed the supplementary motor area, SMA, and it is also sometimes called the second motor area, M2. We now know that monkeys and other primates also have a number of additional motor representations.

The topographic organizations of motor maps were soon revealed in more detail when thin, penetrating microwires or microelectrodes were used instead of surface electrodes. The exposed tip of the microelectrodes could be placed in cortical layer 5, the layer projecting directly to motor neurons in the brain stem and spinal cord, and thus only a small number of the most relevant neurons could be activated. Because the corticospinal projections of the neurons in M1 on the right side of the brain predominantly cross in the lower brain stem to terminate on or near motor neurons projecting to muscles in the left side of the body, neurons in M1 mainly control contralateral body movements. The topographic maps in M1 and other motor areas are largely of contralateral body movements.

Generating small currents to stimulate neurons proved to be technically much easier than recording the small currents generated by neurons as they are activated by sensory stimuli or other neurons. Thus, topographic maps of sensory surfaces were demonstrated later than the first motor maps, initially in the 1930s when the oscilloscope for measuring small currents and good amplifiers came into general use. Clinton Woolsey was in the forefront of such recording, and he soon demonstrated maps of the skin, cochlea, and retina in the cortex of rabbits, cats, and monkeys. A strip of cortex just behind M1 was responsive to light touch and taps on the body surface, and as for M1, the sequence of representation across this strip of cortex was from foot, trunk, hand, and face. This first discovered somatosensory representation became known as S1, or primary somatosensory cortex. A second representation was soon discovered on the lateral margin of S1, and this representation became known as S2. The second representation is activated by the face along its border with S1, while the foot and hand (forepaw) are more lateral. Presently, there is evidence that most mammals have two to three additional somatosensory representations, and monkeys and most likely humans have several more.

The same recording techniques allowed the discovery of topographic representations of the cochlea and retina. In initial studies, different parts of the cochlea were electrically stimulated while recording electrodes were variously placed in auditory cortex. Later brief tones of different frequencies were used selectively to activate receptors along the long, narrow, coiled row of receptors in the cochlea. Both methods revealed a primary auditory area, A1, where the cochlea is represented from base (low tones) to apex (high tones) in a sequential manner. Inputs from both ears activate auditory cortex, but the stronger input is from the contralateral cochlea. Soon, a second auditory area, A2, was proposed, but this second area, originally described in cats, was later found to include parts of other auditory areas. A redefined A2 was described with a crude representation of tones that parallels A1 in representation of tones. Possibly because of early confusion over the nature of A2, the term is not commonly applied to an auditory area except in cats. Gradually, more topographic maps of the cochlea have been described in cats and monkeys, and monkeys may have as many as 10 areas with at least a crude representation of tones.

The organizations of visual areas were first explored by positioning flashing lights in various parts of the visual field and later by presenting bars and spots of light on a screen in front of the eye. Most of visual cortex can be stimulated by either eye, because both eyes project to the lateral geniculate nucleus of the visual thalamus of each cerebral hemisphere. However, the inputs to each nucleus are only from the portions of retina that are devoted to the contralateral half of visual space, the contralateral half of the visual field. In the visual cortex of the right cerebral hemisphere, neurons are activated by stimuli in the left visual field via the nasal half, or more, of the retina of the left eye and the temporal half, or less, of the right eye. Recordings from visual cortex at the posterior end of the cerebral hemisphere revealed a systematic representation of the contralateral half of the visual field, that was termed primary visual cortex, V1.

In all mammals, V1 corresponds to the classically defined architectonic subdivision of cortex, striate cortex or area 17 of Brodmann. In V1 of cats and rabbits, stimuli along the vertical midline (the zero vertical meridian) of the visual field activate neurons along the lateral border; stimuli progressively further into the lower visual quadrant activate neurons located progressively further anterior, while upper quadrant stimuli activate more posterior neurons. Stimuli in peripheral temporal vision activate more medially located recording sites in V1. In monkeys, V1 is rotated so that the border representing the zero vertical meridian is mediolateral in orientation, and much of V1 is on medial and ventral surfaces of the posterior end of the cerebral hemisphere, as well as within a brain fissure of the midline surface, a fissure that is found in all primates. Soon after the discovery of V1, recordings in cortex lateral to V1 in cats and rabbits revealed a second representation of the visual field, V2. This representation occupied a narrow strip of cortex, sometimes identified as area 18. V2 parallels V1 in representations of lower and upper visual quadrants along a common representation of the zero vertical meridian, but the outer border of V2 corresponds to the zero horizontal meridian. Thus, V2 is a ‘split’ or second-order transformation of the visual field. More recently, a number of additional representations of the visual field have been discovered, especially in well-studied cats and monkeys.

2. Brain Maps Today

After many years of intense study, a number of conclusions about brain maps seem justified. First, brain maps are often associated with anatomically distinct subdivisions of the brain. Recently, a number of histochemical methods have been developed to help reveal these subdivisions. Furthermore, some of these procedures reveal internal patterns within some cortical areas and subcortical nuclei that precisely conform to the representational topography of these structures. Thus, some brain maps have visible morphological correlates of their physiological maps. Most notably, there are maps of skin receptors that have visible subdivisions for each digit, pads of the palm, and often separate parts of the body. The maps contain isomorphs of the receptor sheet.

Second, maps vary from fairly precisely reflecting the order of the receptor sheet or muscle arrangement, to only roughly doing so. Thus, maps range from being highly topographical to being only roughly topographic. In general, the more precise maps are found early in sensory processing hierarchies, and less precise maps occur later. Thus, the brain stem nuclei that directly receive peripheral nerve afferents have the most detailed maps, while subsequent relays of this information to higher brain stem and thalamic nuclei may degrade this order, with further changes occurring in primary sensory cortex and beyond. Most higherorder representations are only globally topographic.

Third, even the gross topography of brain maps may be arranged in various ways. While primary visual cortex topographically represents the contralateral half of the visual field, the superior colliculus of the midbrain represents both the contralateral half of the visual field and much of the ipsilateral half of the visual field, including all that is seen by the contralateral eye. Only primates appear to have a superior colliculus that represents only the contralateral half of the visual field. Another example is the split representation of the visual hemifield along the zero horizontal meridian in V2. The reasons for the different types of topographic representations are not clear. However since there can be different types of representations of the same receptor surface in the same individual, the arrangement of the receptor sheet is not the sole factor in the formation of maps.

Fourth, topographic cortical maps are commonly and perhaps always modular in internal organization. Primary visual cortex of most primates segregates inputs related to the right or left eye with so-called ocular dominance columns or bands, separates neurons according to preference for stimulus orientation and direction of movement in so-called orientation columns, and isolates different classes of neurons within and without a dot-like honeycomb pattern of neuron groups that express high amounts of the oxidative enzyme, cytochrome oxidase. The second visual area, V2, of monkeys and other primates has three types of band-like modules that are arranged in repeating sets of three along the length of V2. Thus, V2 actually consists of three fragmented, interdigitated maps of the contralateral visual field. Locally, there are discontinuities and repetitions in the map, but globally a topographic pattern is maintained. Primary somatosensory cortex, area 3b, isolates small patches of cortex devoted to slowly adapting cutaneous afferents from patches activated by rapidly adapting afferents. Auditory maps are unusual in that the receptor array is a row rather than a sheet, and thus a representation of these receptors covers only one dimension of a two dimensional surface of a cortical area. Thus, one dimension is of tone frequencies and the other dimension consists of the same frequency contours. Nevertheless, bands of neurons with similar binaural response properties are segregated from bands of neurons with different binaural properties. Motor maps are also quite modular and fragmented, with variable adjacencies and arrangements of modules of neurons for given movements. Thus, each type of module may be adjacent to a number of different types of other modules, possibly allowing more types of interactions for cooperative movements.

Fifth, there are maps that do not reflect the order of the receptor sheets, but are orderly in some other way. As these maps appear to be derived or computed from topographic maps, they have been called computational maps. As an often given example, a map of auditory space for the systematic location of sounds in space exists in the midbrain of birds and probably other vertebrates. This map does not reflect any arrangement of auditory receptors, but it is constructed from information derived from the differences in how the same sound source in different locations stimulates the two ears. Most higher-order cortical maps appear to reflect both topographic and computational components.

Sixth, species of vertebrates and especially mammals vary in the number of brain maps that they have. Mammals are impressive in that many have greatly expanded sheets of neocortex, providing more space for cortical maps. Most mammals resemble the first mammals in having small brains with relatively little neocortex. These mammals only have a few maps in cortex, usually S1, S2, and two or three other somatosensory maps, V1, V2 and one or more other visual maps, A1 and perhaps a few other auditory maps, and possibly but not always M1 and M2. In contrast, monkeys have a lot of neocortex and they have over 30 visual representations, 12 or more auditory representations, 10 or more somatosensory representations, and six or more motor representations. Possibly humans, with vastly more neocortex, have even more cortical maps. More complex information processing seems to depend on having more maps.

3. The Development Of Maps

Sensory inputs to brain stem nuclei form orderly arrangements during development, and these nuclei form orderly connections with other nuclei, which in turn project to cortical areas in an orderly fashion and so on. The rough order in at least the early stations of this sequential process seems to depend on intrinsic factors in the brain. More specifically, it has long been hypothesized that afferents are chemically labeled according to some gradient, and target neurons are also chemically labeled with a comparable gradient across the target structure. Thus, the neurons find chemical matches and a topographic map is created. However, chemical or molecular matching is not the only factor in the development of brain maps, since there is much evidence that the development of such maps can be altered by changes in the sensory environment that induce changes in neural activity patterns. The details of sensory maps appear to be fine tuned by an adjustment and sorting of local connections in largely adaptive ways in response to variations in patterns of sensory activation. The claim is that neurons that fire together wire together. Coactive neurons with overlapping connections maintain and strengthen their synapses on shared target neurons, while neurons that do not respond at the same time tend to weaken and lose synapses on their shared target neurons. Thus, neurons with uncorrelated activity patterns tend to segregate their terminations so as to activate separate groups of neurons in the next structure. This allows maps to adjust and compensate for losses or additions of inputs, and maps in a sequence to adjust to each other.

There are many types of evidence that maps adjust to sensory changes during development. Usually, this evidence comes from some sort of experiment where some sensory input is removed or altered. Normal inputs expand to activate neurons formerly devoted to the removed or weakened inputs. Other evidence comes from the variations observed in maps in nature. Rats and mice have morphological maps of the whiskers of the face in S1. Each whisker activates neurons in a small oval of cortical tissue that resembles a barrel, and is commonly called a cortical barrel. Rats and mice usually have the same number of whiskers, and there is one barrel in S1 for each whisker. Likewise there is one barrel-like structure for each whisker for the termination of afferents from the face in the brain stem, and in the ventroposterior nucleus of the thalamus. Thus, maps at three levels of the somatosensory system precisely match the arrangement and number of whiskers on the side of the face. This match could be due to innate developmental mechanisms, such as the matching of connections between levels as a result of a chemical code, but this appears to be unlikely.

Mice have been found that differ in the number of whiskers by having one or two more or less than the normal number, and in these mice the maps at all three levels match exactly the increase or decrease in whiskers. Thus, somehow the receptors in the skin are informing the central maps about their organizational state. In a similar way, the star-nosed mole has 11 sensory rays or protrusions from each side of its nose, and processed sections of somatosensory cortex reveal 11 corresponding bands of tissue in S1, one for each ray. Eleven bands are also seen in S2, the ventroposterior nucleus, and the somatosensory brain stem. Moles with 10 or 12 rays have been found, and 10 or 12 bands are then seen in all four brain structures in these moles. This matching of maps from one level to the next and with the peripheral receptor sheet would be most easily accomplished by a system that uses natural activity patterns to fine-tune its maps.

4. Why Ha E Topographic Maps?

Much of the brain is devoted to topographic maps suggesting that they are a useful way of organizing brain tissue. One cannot argue that such maps are simply preserving the order of afferents from the periphery, since afferents become mixed and disorderly, but sort themselves out again as they terminate. Instead, topographic maps appear to be adaptive in that they provide a substrate for making the most common types of computations in the nervous system over short, local connections. Neurons in topographic representations are well arranged to make center-surround comparisons between stimuli anywhere on the receptive field sheet, and stimuli in the immediate surround. This type of comparison provides biologically useful information. Neurons rapidly determine if a stimulus at some location is brighter, greener, harder, louder, or moving differently than stimuli in the immediate surround. While it may be possible to make such computations in nontopographic structures and systems, neurons would need to be massively interconnected and over larger distances. This would add unreasonable bulk to the brain, and would be metabolically costly. In complex brains with large amounts of neocortex, it is probably more efficient to have multiple interconnected maps, rather than just making maps larger. Large maps are most suitable for extremely local comparisons of stimuli, while more global comparisons can be mediated over short connections in small maps. Thus, it seems to be more productive to subdivide tasks, and process information in parallel over sequences of interrelated small maps, and use long connections to coordinate the activities of maps within and across processing sequences. Maps vary in size in the same individual because large maps are better at fine-grain local comparisons while small maps are better at more global comparisons.

Bibliography:

1. Kaas J H 1997 Topographic maps are fundamental to sensory processing. Brain Research Bulletin 44: 107–12
2. Kaas J H 2000 Organizing principles of sensory representations. In: Bock G R, Cardew G (eds.) Evolutionary Developmental Biology of the Cerebral Cortex. Wiley, New York, pp. 188–205
3. Knudsen E I, duLac S, Esterly S D 1987 Computational maps in the brain. Annual Review of Neuroscience 10: 41–65
4. Penfield W, Boldrey E 1937 Somatic motor and sensory representation in the cerebral cortex as studied by electrical stimulation. Brain 60: 389–443
5. Udin S B, Fawcett J W 1988 Formation of topographic maps. Annual Review of Neuroscience 11: 289–327
6. Woolsey C N, Marshall W H, Bard P 1942 Representation of cutaneous tactile sensibility in cerebral cortex of monkeys as identified by evoked potentials. Bulletin of Johns Hopkins Hospital 70: 399–441