Cerebral Cortex

Pyramidal Cells Are the Most Numerous Neocortical Neurons

Pyramidal cells, the most numerous neurons of the neocortex, are named for their shape ( Fig. 22.3 ). These cells have a conical cell body from which a series of spine-studded dendrites emerge—a long apical dendrite that leaves the “top” of each cell and ascends vertically toward the cortical surface, and a series of basal dendrites that emerge from around the base of the cell and spread out horizontally. Pyramidal cells range in size from less than 10 µm in diameter all the way up to the giant pyramidal cells (Betz cells) of the motor cortex, which are among the largest neurons in the central nervous system (CNS), some measuring more than 100 µm from their base to the beginning of the apical dendrite. Most or all pyramidal cells have long axons that leave the cortex to reach other cortical areas or various subcortical sites, where they make excitatory (glutamate) synapses. The remaining cortical neurons are spoken of collectively as nonpyramidal cells. Many are small (often less than 10 µm), multipolar stellate (or granule ) cells, but a variety of other types and sizes have been described ( Fig. 22.4 ). With few exceptions, nonpyramidal cells have short axons that remain within the cortex. One kind of nonpyramidal cell has spine-covered dendrites, receives inputs from the thalamus, and makes excitatory (glutamate) synapses on nearby neurons; most or all of the others make inhibitory (γ-aminobutyric acid) synapses on their targets. Therefore pyramidal cells are the principal output neurons of the neocortex, and nonpyramidal cells are the principal interneurons.

The dendritic spines of pyramidal cells (see Fig. 1.4E ) are preferential sites of excitatory synaptic contacts and have been the source of considerable interest. They are not merely devices for increasing dendritic surface area, because the area of dendrites located between spines is sparsely populated with synaptic contacts. It has been suggested that dendritic spines may be the sites of synapses that are selectively modified as a result of learning, because small changes in the geometry of a spine can cause relatively large changes in its electrical or diffusional properties and therefore in the efficacy of that synapse. Certain cases of mental retardation are accompanied by faulty development of dendritic spines, including misshapen spines and/or an overabundance of spines, but which is cause and which is effect (if either) is not known. Certainly, however, the most remarkable change that occurs in the cortex after birth is the tremendous expansion of the dendritic trees of its neurons and a parallel increase in the number of dendritic spines. Spines are not unique to cortical pyramidal cells; they are also found on the dendrites of some other neurons, such as Purkinje cells (see Fig. 8.16 ) and many striatal neurons.

Neocortex Has Six Layers

The cells of the neocortex are arranged in a series of six layers, more apparent in some areas than in others. Just as in the case of cerebellar cortex (see Figs. 20.9 and 20.10 ), the most superficial layer is a cell-poor molecular layer. The deepest neocortical layer is the polymorphic (or multiform ) layer, which is populated largely by fusiform-shaped modified pyramidal cells. In between the molecular and polymorphic layers are four layers alternately populated mostly by small cells or mostly by large pyramidal cells. The layers are commonly designated by Roman numerals and by names, as indicated in Fig. 22.5 . Myelin staining reveals vertically oriented bundles of cortical afferents and efferents, as well as horizontal bands through which these fibers and intracortical axons spread. Two particularly prominent horizontal bands are contained in layers IV and V and are called, respectively, the outer and inner bands of Baillarger.

Neocortex does not have the same striking regularity as cerebellar cortex; its six cell layers are not equally prominent everywhere. Areas that give rise to many long axons (e.g., the motor cortex) would be expected to have numerous large pyramidal cells, and this is indeed true ( Fig. 22.6A ). In these areas, nonpyramidal cells appear minor by comparison, and layers II through V are dominated by large pyramidal cells to the extent that individual layers are no longer obvious. Because of the apparent lack of stellate (granule) cells, such cortex is called agranular. In contrast, primary sensory areas project mainly to adjacent cortical areas and do not give rise to many long axons. They have a corresponding dearth of large pyramidal cells; here too, layers II through V look like one continuous layer, but in this case they are dominated by small cells (both pyramidal and nonpyramidal; see Fig. 22.6A ). Such cortex is therefore called granular cortex or koniocortex (from the Greek word konia, meaning “dust,” referring to the numerous tiny cells). There is a continuum of structural types ranging from thick (up to 4.5 mm) agranular cortex to thin (as little as 1.5 mm) granular cortex ( Fig. 22.7 ). The intermediate kinds, in which the six neocortical layers are relatively distinct, are called homotypical cortices (rather than granular and agranular cortices, which are collectively called heterotypical; see Fig. 22.6B ).

The differences among cortical areas are to some extent more apparent than real. Beneath a square millimeter of any area of mammalian cortex, whether from a hamster or a human, lie approximately the same number of neurons (roughly 100,000). The major exception is the binocular portion of the primary visual cortex of primates, where the neurons are packed more densely. About 80% of the neurons in all cortical areas are pyramidal cells. Therefore 80% of the neurons in granular cortex are very small pyramidal cells. Different cortical areas have different appearances and functions because of the relative sizes of the cell types, the complexities of their dendritic trees, and the patterns of their connections.

Different Neocortical Layers Have Distinctive Connections

Afferents to the cortex come from two general places: other cortical areas and subcortical sites. Afferents from other cortical sites, by far the majority, may arise in the same hemisphere (association fibers) or in the contralateral hemisphere (commissural fibers). The predominant subcortical source of afferents is the thalamus, and its pattern of projections is described in Chapter 16 . Other subcortical sites, such as the locus ceruleus and other chemically coded nuclei, also provide modulatory afferents to the cortex (see Chapter 11 and later in this chapter).

These various types of incoming fibers ramify within the cortex in different patterns ( Fig. 22.8 ). For example, afferents from thalamic relay nuclei end primarily in the middle layers, as in the dense arborizations in layer IV of fibers from sensory relay nuclei ^b

b Because the line of Gennari in striate cortex (see Fig. 17.31 ) represents a particularly large outer band of Baillarger and is located in layer IV, it is often assumed that it represents the massive projection from the lateral geniculate nucleus to the striate cortex. However, cutting all the afferents to the striate cortex does not cause the line of Gennari to degenerate. It exists in sighted and blind individuals. It is thought to be a collection of intracortical axons, although the details of its structure and function are still unknown.

; fibers from other thalamic nuclei and from other cortical areas ascend vertically and terminate diffusely along their course in distinctive patterns (e.g., those from intralaminar nuclei mostly in layer VI, and those from other cortical areas mostly in layers II and III).

Efferents from the cortex, like afferents to it, must be connected either with other cortical areas or with subcortical sites. Efferents to subcortical sites, mentioned in various places throughout this book, descend primarily through the internal capsule. The longest ones continue through the cerebral peduncle, the basal pons, and the medullary pyramids, finally extending all the way to the spinal cord. Others reach an assortment of additional subcortical sites, including the caudate nucleus and putamen, the thalamus, the superior colliculus, the red nucleus, the reticular formation, pontine nuclei, motor neurons of cranial and spinal nerves, and various sensory nuclei of the brainstem and spinal cord. Some corticostriate fibers travel through the external capsule. Just as afferents to the cortex have a distinctive laminar pattern of termination, efferents from the cortex have a laminar pattern of origin. Although there is substantial overlap, layer III is the major source of corticocortical fibers, layer V of corticostriate fibers and fibers to the brainstem and spinal cord, and layer VI of regulatory projections back to the thalamus.

Association Bundles Interconnect Areas Within Each Cerebral Hemisphere

Efferents to ipsilateral cortical areas come in all lengths, from very short ones that never leave the cortex, to U -shaped fibers that dip under one sulcus to reach the next gyrus, to longer association fibers that travel to a different lobe; collectively, they account for a large majority of the axons in the white matter of each hemisphere. The longer fibers collect into reasonably well-defined bundles that can be found by dissection ( Fig. 22.9A ) and now by diffusion tensor imaging ( Fig. 22.9B to D ). The most prominent of these association bundles are the superior longitudinal fasciculus, the superior and inferior occipitofrontal fasciculi, and the cingulum. The superior longitudinal fasciculus (also called the arcuate fasciculus ) sweeps along in a great arc above the insula between the frontal lobe and posterior portions of the hemisphere, where it fans out among the parietal, occipital, and temporal lobes. The superior occipitofrontal fasciculus, as its name implies, runs between the frontal lobe and superior parts of the parietal and occipital lobes. It travels parallel to the corpus callosum and, for much of its course, is located between the corpus callosum and the caudate nucleus. Here it lies adjacent to the subcallosal fasciculus, a pale-staining bundle of fibers on their way from several cortical areas to the caudate nucleus. The inferior occipitofrontal fasciculus passes below the insula between the frontal and occipital lobes, traversing the temporal lobe along the way. Its fibers fan out at both ends of the fasciculus, and those at its anterior end lie adjacent to the uncinate fasciculus (from the Latin uncus, meaning “hook”), an association bundle that hooks around the margin of the lateral sulcus to interconnect the orbital cortex and anterior temporal cortex. Finally, the cingulum courses within the cingulate gyrus and continues around within the parahippocampal gyrus to nearly complete a circle. None of these association bundles is a discrete, point-to-point pathway from one place to another; rather, fibers travel in both directions and enter and leave each pathway all along its course. The inferior occipitofrontal fasciculus provides a good example of this. Few of its fibers actually extend all the way between the occipital and frontal lobes. Rather, those in the posterior part of this fasciculus interconnect occipital and temporal areas and are often considered separately as the inferior longitudinal fasciculus; anterior fibers travel through inferior parts of the extreme capsule to interconnect the superior temporal gyrus, the insula, and orbital and prefrontal cortex.

Neocortex Also Has a Columnar Organization

Even though the cortex is horizontally laminated, there is also a vertical organization (“vertical” meaning perpendicular to the surface). Apical dendrites of pyramidal cells have vertical courses, as do afferents to the cortex and the axons of some intracortical cells (see Fig. 22.8 ); even the cell bodies of cortical neurons often look as though they are arranged in vertical columns (see Fig. 22.5 ). Physiological and anatomical studies have shown that this is not just an illusion. If an electrode is slowly advanced through somatosensory cortex along a path perpendicular to the cortical surface, all the cells encountered respond with about the same latency to the same type of stimulus delivered to about the same region of the body. Similarly, all the cells along a vertical path through visual cortex respond best to bars or edges with the same orientation in about the same part of the visual field (see Fig. 17.35 ); if the electrode is moved 50 µm or so across the surface of the cortex, cells with a different preferred stimulus orientation are encountered. Furthermore, most cells along such a vertical path respond better to stimulation of one eye than to stimulation of the other eye; cells in a nearby vertical region may have not only a different preferred stimulus orientation but also a different preferred eye. The picture that has emerged is one in which the cortex is organized into vertical columns, each 50 to 500 µm wide, in which some parameter (e.g., stimulus orientation) is constant for all cells.

This kind of columnar organization most likely reflects a general strategy used in the construction of neocortex. Anatomical tracing techniques make it possible to visualize the columns (see Fig. 17.35 ), and there are indications that columnar organization is widespread. For example, in at least some cortical areas, afferents from the thalamus, from other ipsilateral cortical areas, and from contralateral cortical areas end in vertical columns separated by columns that do not receive that particular kind of input. Throughout the neocortex, the basic building blocks appear to be “minicolumns” about 50 µm in diameter containing about 100 neurons; dozens of minicolumns, linked by horizontal intracortical connections, make up larger functional modules such as the hypercolumns in visual cortex (see Fig. 17.35 ).

Different Neocortical Areas Have Subtly Different Structures

Seeing that various cortical areas are structurally distinct from one another in fairly obvious ways (e.g., granular vs. agranular cortex; [see Fig. 22.6A ] or striate vs. extrastriate cortex [see Fig. 17.31 ]), a number of anatomists have sought to map the cortex in terms of these differences and often of considerably more subtle differences. One mapping system, published in the early 1900s, whose terminology remains in widespread use is that devised by Korbinian Brodmann, who divided the neocortex of each hemisphere into 44 areas ( Fig. 22.10 ). The boundaries between many of these areas are not precise, as they often grade into each other by degrees. In addition, as noted previously, the correlation of functions with specific anatomical areas is not nearly as precise as was once hoped. Nevertheless, many of the areas described by Brodmann correspond remarkably well to areas defined by other measures of connection or function, and many of the numbers proposed by him are still commonly used for reference purposes ( Table 22.1 ).

Although each of us has roughly the same total amount of cerebral cortex, there are surprisingly large variations in the sizes of particular areas. The areas of visual, somatosensory, and motor cortex may vary by a factor of 2 to 3 among typical individuals. Because the total neocortical area is much more constant than this, someone with a larger-than-average visual cortex presumably has other areas that are smaller than average. Whether these differences in area are correlated with differences among individuals in various skills and functional capacities is not known. ^c

c Mark Twain apparently alluded to this possibility: “I never could keep a promise. I do not blame myself for this weakness, because the fault must lie in my physical organization. It is likely that such a liberal amount of space was given to the organ which enables me to make promises that the organ which should enable me to keep them was crowded out.” (From Twain M: The innocents abroad, New York, 1869, Charles L Webster. Recounted in Harvey PH, Krebs JR: Science 249:140, 1990.)