Library

Organization of Primary Visual Cortex

      In 1782 a young medical student named Francesco Gennari described a thin white stripe of myelin running through the gray matter of the occipital lobe. This stria, visible to the naked eye in fresh or fixed specimens, prompted him to suggest that the cortex might be subdivided into discrete anatomical regions. This suggestion was revolutionary, because anatomists had previously assumed that the cortex was a uniform sheet of tissue, lacking any internal subdivisions. Today, of course, it is a well accepted notion that the cerebral cortex is partitioned into dozens of distinct areas controlling all aspects of human behavior. It was Gennari's discovery which opened the door for our modern view of cortical function. Ironically, in identifying the first cortical area in the brain, Gennari had no inkling that he had stumbled upon the primary visual cortex. More than a century was to elapse before it was finally proven by Henschen, a Swedish neuropathologist, that the stria of Gennari is coextensive with the primary visual cortex. The primary visual cortex is often called the "striate cortex", recalling the prominent white stria found by Gennari.

      The representation of vision in the cerebral cortex was explored early in this century by the clinical examination of soldiers wounded in battle. Inouye constructed the first detailed retinotopic map of primary visual cortex by matching visual field deficits with the trajectory of missiles penetrating the occipital lobes. We have prepared a revised version of these maps (Fig. 6) with the aid of magnetic resonance, a high-resolution imaging technique that allows direct correlation of occipital lobe lesions with visual field defects in patients:

Fig. 6) Retinotopic map of the human striate cortex. Upper right shows left ocipital lobe, with most of striate cortex buried in the calcarine fissure. Upper left shows the fissure opened, with distance (eccentricity) from the fovea (center of gaze) marked in degrees. The horizontal meridian (HM) runs roughly along the base of the fissure. Lower left shows the map, removed from the calcarine fissure and flattened artificially. Dots depict occipital pole; the central 1o is located on the exposed lateral convexity, although this varies from person to person. Note the immense magnification of central vision. Dark oval = blind spot, stippled zone = monocular crescent. From: Horton JC. Hoyt WF. Arch Ophthalmology, 1991.

      The upper and lower visual quadrants are represented in the lower and upper calcarine banks respectively, separated by the horizontal meridian along the base of the calcarine fissure. The fovea, a specialized region within the retina specialized for best acuity, is represented at the occipital pole where primary visual cortex usually extends about one centimeter onto the lateral convexity of the hemisphere. The vertical meridian corresponds to the perimeter of primary visual cortex located along the exposed medial surface of the occipital lobe. Most of primary visual cortex is actually buried within the depth of the calcarine fissure. The clearest view of the visual field map can be obtained by schematically unfolding and flattening visual cortex to create an artificial planar surface. The primary visual cortex contains a topographic but highly distorted representation of the contralateral hemifield of vision.

      The most striking feature of the visual field map is the enormous fraction of visual cortex assigned to the representation of central vision. About 55% of the surface area of primary visual cortex is devoted to the representation of the central 10° of vision. The cortical "magnification factor" -- the millimeters of cortex representing one degree of visual field -- has a ratio of more than 40:1 between the fovea (0° eccentricity) and the periphery (60° eccentricity). The temporal crescent representation (stippled area, Figure 1, lower panels constitutes only about 5% of the total surface area of primary visual cortex. The representation of central vision is highly magnified compared with peripheral vision, so that the cortical area devoted to the central 1° of visual field roughly equals the cortical area allotted to the entire monocular temporal crescent.

      We have compiled a map of the visual field representation in striate cortex using a novel approach. We discovered that the shadows cast by retinal blood vessels (known as angioscotomas) are represented in striate cortex of the squirrel monkey (Adams DL, Horton JC. Science, 2002). We systematically identified corresponding points between retinal vessels and their cortical representations. We then warped the retina and its blood vessels directly onto the cortex, using these corresponding points as a guide. Analysis of the resulting map showed that the macula (the central 10 degrees of the retina) is over-represented in striate cortex, even if one allows for its high concentration of ganglion cells. For further details, see: Adams DL, Horton JC. J. Neuroscience, 2003.

      The relatively magnified representation of the macula in primary visual cortex furnishes an important clue to how the cerebral cortex analyzes sensory information. In the retina, 250 μm of tissue equals about 1 degree for all points in the visual field. This must remain nearly constant because the eye is engaged in processing an optical image of the visual environment. The steep gradient in visual acuity, from 20/20 centrally to 20/400 peripherally, is achieved by variation in the density of cells in various layers. For example, in central retina the ganglion cells are stacked 6-8 cells deep, declining to a broken monolayer in peripheral retina. Free of any optical constraints, the cerebral cortex handles the richer flow of visual information emanating from the central retina in a different fashion. The cortical sheet is essentially uniform in thickness throughout primary visual cortex but far more tissue is allocated for the analysis of central vision. In the visual cortex, the magnification factor -- rather than the cell density -- varies with eccentricity in the visual field representation. A similar strategy is employed by the somatosensory cortex to represent the most densely innervated regions of the body surface, as exemplified by the exaggerated size of the face, tongue, and hands of the homunculus.

      As discussed in the preceding paragraph, the central degree of vision has a hugely magnified representation in striate cortex. If the same amount of tissue were devoted to the representation of every single degree of the visual field, an enormous amount of cerebral cortex would be required. In fact, in such an imaginary brain, the entire surface of the cerebral cortex would be occupied with the processing of visual information (as matters stand already, more than a third of the cerebral cortex is already devoted to vision). Thus, one can view the evolution of a circumscribed high acuity foveal zone as an economy strategy by nature -- both at the retinal and the cortical level.

      Our constant visual goal in life is to project and maintain targets of interest upon the fovea. Foveation requires six extraocular muscles to move each eye, systems for smooth pursuit, systems to abruptly shift gaze (saccades), and systems to stabilize gaze during head and body movements (the vestibulo-ocular and opto-kinetic reflexes). These systems are governed by an elaborate circuitry housed to a large extent at the level of the brainstem. Familiarity with these systems and their basic anatomy is of immense clinical importance because a wide variety of disease processes attack the brainstem or the cranial nerves innervating the eye muscles.

Ocular Dominance Columns

      Axon terminals of cells within individual laminae of the lateral geniculate body terminate predominately in layer IVc, the main input layer to striate cortex. Axon terminals serving the right eye and left eye are not randomly distributed, but rather, they are segregated into a system of alternating parallel stripes called ocular dominance columns. In monkeys, the ocular dominance columns have been revealed by injecting one eye with tritiated proline, a radioactive tracer(Fig. 7). They can also be labeled by removing one eye and then staining the cortex for cytochrome oxidase, a mitochondrial enzyme. This method exploits the fact that levels of metabolic activity diminish in columns formerly driven by the missing eye. In flattened sections from subjects with a history of monocular enucleation, a mosaic of alternating light and dark stripes emerges in layer IVc. The light columns correspond to the ocular dominance columns of the enucleated eye:

Fig. 7) Flattened left striate cortex from a normal macaque monkey. A) Drawing of the ocular dominance columns. BS = blind spot representation, MC = monocular crescent representation. B) Montage of sections processed for cytochrome oxidase after enucleation of the right eye, showing alternating dark and pale stripes, corresponding to the ocular dominance columns in layer IVc. C) Montage compiled from alternate sections processed for autoradiography after tritiated proline injection into the vitreous of the left eye. Note the perfect match between the dark columns labeled by their metabolic activity in (B) and the light columns revealed by transneuronal axon tracing in (C).From Sincich LC, Horton JC. Visual Neuroscience, 2002

      Although ocular dominance columns are present in all macaques and humans tested so far, their expression in some species is inconsistent. For example, in the squirrel monkey, ocular dominance columns are present in some animals and not in others. Even more surprisingly, in some animals they are present in only part of the cortex. These observations have confounded efforts to determine the function of ocular dominance columns. For more details see: Adams DL, Horton JC. Nature Neuroscience, 2003.

Cytochrome Oxidase Patches

      Striate cortex contains a regular array of patches containing increased activity of the metabolic enzyme, cytochrome oxidase. This array is most obvious in sections cut parallel to the pial surface (Fig 8.) (Horton JC. Phil. Trans. R. Soc. (Lond.) B, 1984). In three dimensions, these patches form cylinders that extend through the cortex from the white matter to the brain surface (except in layers IVc and IVa, where they are absent).

Fig. 8) Tangential section from a normal macaque monkey, cut through layers 2/3, showing patches of increased cytochrome oxidase activity , each about 250 μm by 150 μm, in striate cortex. The patches are organized into rows that are in register with the ocular dominance columns in layer 4c. In the right drawing, the patches end at the V1/V2 border (dashed line).

      Cytochrome oxidase patches may constitute the basic modules of striate cortex. In macaques and humans, they form rows in register with the ocular dominance columns. Curiously, this special relationship between cytochrome oxidase patches and ocular dominance columns is absent in the squirrel monkey (Horton JC, Hocking DR. J. Neuroscience, 1996).

      Cytochrome oxidase patches in layers 2/3 receive direct input from the lateral geniculate body (Horton JC. Phil. Trans. R. Soc. (Lond.) B, 1984), organized into separate projections serving the left eye and the right eye (Horton JC, Hocking DR. J. Neuroscience, 1996). This input originates from a class of cells in the lateral geniculate body called the konio cells.