A precise retinotopic map of primate striate cortex generated from the representation of angioscotomas

Abstract

Shadows cast by retinal blood vessels are represented in striate cortex of the squirrel monkey. Their pattern was exploited to generate a true retinotopic map of V1. For calibration, retinal landmarks were projected onto a tangent screen to measure their visual field location. Next, the retina was warped onto striate cortex, distorting it as necessary to match each retinal vessel to its cortical representation. Maps from four hemispheres of two normal adult squirrel monkeys were created and used to derive expressions for cortical magnification factor (M). A mean map was produced by averaging the individual maps. To address the controversial issue of whether the ratio of retinal ganglion cell (RGC) density to M is constant at all eccentricities, we stained a retinal whole mount from one of the two monkeys for Nissl substance. A ganglion cell density map was compiled by sampling the concentration of cells at 171 retinal points. Allowance was made for displaced amacrine cells and for the centripetal displacement of RGCs from central photoreceptors. After these corrections the V1 surface area and RGC density were compared at each eccentricity. The cortical representation of the macula was found to be amplified, even beyond the magnification expected from its high density of RGCs. For example, the central 4 degrees of visual field were allotted 27% of the surface area of V1 but were supplied by only 12% of RGCs. We conclude that, in monkey striate cortex, more tissue is allocated per ganglion cell for the analysis of information emanating from the macula as compared with the peripheral retina.

Fig. 1.

Monkey P. A, Montage of retinal photographs. B, Flat-mount of cortex, stained for CO after enucleation of the left eye, showing the pattern of ocular dominance columns. The dark lines emerging from the blind spot representation correspond to the cortical representations of retinal blood vessels. Note the CO pattern of alternating pale-thin-pale-thick stripes in area V2. Narrowing of V2 helps to pinpoint the foveal representation. The V1/V2 border corresponds to the vertical meridian of the visual field. C, Diagram of blood vessels inA, showing in color those represented in the cortex. D, Drawing of angioscotoma representations visible in the right cortex above, numbered and color-coded to match retinal vessels in C. The white vessel representation outlined with dots was generated by a vessel in the temporal retina of the right eye (not illustrated). BS, Blind spot; MC, monocular crescent; V2, second visual area.

Fig. 2.

Monkey P. To generate a retinotopic map, we superimposed a ring and ray pattern on the left nasal retina. Correspondence points, marked with yellow squares, were placed at locations that could be identified unambiguously in the cortex and the retina. In this example 64 correspondence points and their intercalated segments were warped from the retina onto the cortex to prepare the retinotopic map. Yellow squares in the central 4° were derived fromCowey (1964).

Fig. 3.

Mean retinotopic map compiled by averaging maps from the left and the (reflected) right striate cortex of Monkey P and Monkey Q [see Adams and Horton (2002), their Fig. 7]. Eccentricity is denoted by the vertical lines representing 1, 2, 4, 8, 16, 24, 32, 50, and 72°. Polar angle is indicated by the horizontal lines from 90 to 270° in 10° intervals. All of the polar rays converge on the foveal representation at the left edge of the map. Error bars indicate ± SEM in the x- and y-axes. The central 4° are based on data from Cowey (1964).

Fig. 4.

A, Cortical areal magnification factor (M_a) as a function of eccentricity. The dotted line represents the equationM_a = a(b+ E)^c, with cconstrained to –2. The heavy solid curve is the best fit of the equation with an unconstrained exponent. The complex log function fits the data well only for the first 8°. Beyond this eccentricity the decline in M_a is greater than that predicted by a conformal map. The thin solid line shows thatM_a for the macaque is extremely similar.B, Linear magnification along isopolar rays (M_p) as a function of eccentricity. The dotted line represents the equationM_p = a(b+ E)^c, with cconstrained to –1. It is close to the best fit with an unconstrained exponent (solid line). C, Linear magnification along isoeccentricity rings (M_e) versus eccentricity. The dotted curve (c = –1) deviates from the best fit (solid curve; c = –1.85) beyond 8°. Error bars indicate ± SEM of four hemispheres.

Fig. 5.

A, Inouye's map, published in 1909, appears to show an ideal complex logarithmic mapping of the visual field onto the cortex, although he called it an “area true” map. It appears as a grid, with horizontal lines corresponding to isopolar lines and vertical lines corresponding to isoeccentricity rings (eccentricities are underlined). The map is physically impossible, because the center of gaze (0°) is represented by a line, not a point. B, Modified conformal map, after Schwartz (1984), bringing the isopolar rays and isoeccentricity rings to a point at the fovea. In the periphery they remain orthogonal. In reality (Fig.3), the isopolar rays converge peripherally, and the isoeccentricity rings become curved and shortened.

Fig. 6.

A, A field of uniform squares projected onto the visual cortex (B), showing the distortion produced by global changes in cortical magnification (M_a) and by local anisotropy. The anisotropy index (M_p/M_e) is depicted by the color scale and ranges from 2.8 to 0.5 (mean, 1.2; SD, 0.4). It tends to be greatest along the vertical meridia and in the periphery. Consequently, squares near the vertical meridia (for example, see asterisk) are elongated along the isopolar direction, whereas those along the horizontal meridian (see dagger) are elongated along the isoeccentricity direction.

Fig. 7.

A,M_p, graphed as a function of polar angle between each pair of isoeccentricity rings, declines at the horizontal meridian (180°). Graphed points represent the mean of four hemispheres ± SEM. B,M_a, plotted for each compartment in the visual field map, remains approximately constant between each pair of isoeccentricity rings. The decline in M_pis offset by an increase in M_e. This effect can be appreciated in the cortical map (Fig. 3). Near the vertical meridia the compartments are rectangular, whereas near the horizontal meridian they are more square. Their area along any isoeccentricity belt, however, is essentially independent of polar angle.

Fig. 8.

A, Single section cut tangentially through a flat-mount of the right visual cortex of Monkey Q, stained for CO to show the patches in the upper layers. This hemisphere was one of four used to compile the retinotopic map in Figure 3. The patches are distributed evenly throughout the cortex. MC, Monocular crescent.B, Fourier-filtered and thresholded image of the CO section in A, prepared for analysis of patch density and back-projection onto the visual field in Figure 10.

Fig. 9.

Graph showing patch density for the cortex and its projection onto the visual field (y-axis, left) as a function of eccentricity. The reciprocal, patch domain, measured in millimeters squared of cortex per patch or degrees squared in the visual field, is shown also (y-axis, right). The number of patches allotted to each square degree in the visual field parallels M_a (Fig.4A), because patch density is quite constant in the cortex.

Fig. 10.

A, Back-projection of Figure8A onto the visual field, showing how patch domain varies as a function of eccentricity. Each patches covers a wide domain in the periphery, but centrally the domains are so small that the image must be magnified selectively (right). B, Back-projection of Figure 8B onto the visual field.

Fig. 11.

A, Nissl-stained whole mount of the left retina of Monkey P. Arrows show boundary between retina (blue) and pars plana (dark brown) in peripheral temporal retina. Lettered boxes are shown at higher power in Figure 12. B, Superimposed fundus montage showing the appearance of the retinain vivo. When we scaled the picture to match the location of the optic disc, fovea, and blood vessels (arrows) in the whole mount, it was possible to transfer the eccentricity of retinal landmarks measured in vivo onto the whole mount to determine the location of the central isoeccentricity rings. The dashed circle, tangential to the edge of the temporal retina, corresponds to the limit of binocular visual field (72°). C, Ring and ray pattern, superimposed on the whole mount, using empirical measurements for 0, 1, 2, 4, 8, 16, 24, and 72° rings and the Drasdo and Fowler (1974) schematic eye for the 32 and 50° rings. Red arrows show how the location of the peripheral rays was warped onto the retina to compensate for relieving cuts. D, Final registration of the visual field and the retinal whole mount. The yellow dots mark the location of 162 counting windows in which ganglion cell density was measured. Their size accurately represents the area of retina that was examined (see Materials and Methods). E, Enlargement of the central 4°.

Fig. 12.

Representative fields of Nissl-stained ganglion cells from locations shown in Figure 11A. The appearance of the ganglion cell layer depends critically on eccentricity. At all eccentricities the ganglion cells (open arrows) can be recognized on the basis of their relative size, large nucleus, and prominent dense nucleoli. A, A 0.5° field, at the edge of the foveal pit (lower half), where ganglion cells are small and form a monolayer. A few microglia, identifiable by their densely stained irregularly shaped nuclei, are indicated with filled arrows.B, A 6° field just beyond the peak density of ganglion cells. The ganglion cells are small and packed four to six cells deep. It was necessary to focus up and down through the ganglion cell layer to count them accurately. Outside the fovea the blood vessels perfuse the ganglion cell layer. Their endothelial cells (arrowheads) could be recognized easily and excluded. C, A 20° field showing that the ganglion cells are larger, less numerous, and organized into a single layer. D, A 61° field showing the largest cells, scattered widely in the ganglion cell layer.

Fig. 13.

Single 3 μm plastic cross section of squirrel monkey retina, immunoreacted for GABA, to identify displaced amacrine cells in the ganglion cell layer. It was cut from the piece of retinal tissue shaded in the whole mount schematic (inset). The tissue was folded with the nerve fiber layer facing outward before embedding. Rectangles marked with letters are shown at higher power to the right. A DIC oil immersion 40× objective was used to visualize unstained and immunostained cells. A, Parafoveal retina containing a high density of ganglion cells, which are unstained. A few GABA-positive cells (arrowheads), stained by the peroxidase reaction used to identify displaced amacrine cells, are present in the ganglion cell layer right next to the inner plexiform layer. Amacrine cells are also present in the inner nuclear layer (long arrows).B, Retina at 8–10°, showing a greater number of displaced amacrine cells. Note that the nerve fiber and inner plexiform layers are much thicker than near the fovea. C, Retina at ∼25°. The ganglion cell layer is reduced to single cell thickness and contains ∼25% displaced amacrine cells. The three GABA-positive bands in the inner plexiform layer are particularly prominent at this eccentricity. D, Retina at 52°, showing a similar proportion of amacrine cells to those present at 25°. The overall intensity of GABA immunoreactivity declined in peripheral retina. NFL, Nerve fiber layer; GCL, ganglion cell layer; BV, blood vessel; IPL, inner plexiform layer; INL, inner nuclear layer.

Fig. 14.

Plot of true ganglion cell and displaced amacrine cell density as a function of retinal eccentricity. Cell densities in the ganglion cell layer were sampled at the points shown in Figure 11,D and E; displaced amacrine cell densities were calculated from their proportion in the ganglion cell layer, which was determined from analysis of GABA-immunostained sections (see Fig. 13). Displaced amacrine cell densities were subtracted from cell densities in the ganglion cell layer to derive ganglion cell densities at each eccentricity. This “true” ganglion cell density (per mm²) was averaged over 360° of polar angle. It also is plotted in units of cells/degrees squared, using the retinal magnification factor derived from the schematic eye of Drasdo and Fowler (1974). Error bars indicate ± SEM.

Fig. 15.

A, In vitrophotograph of the right macula of Monkey Q, showing measurement of the centrifugal displacement from photoreceptors to their target ganglion cells. After enucleation the retina was photographed, embedded in plastic, and sectioned at 1 μm along the vertical meridian through the fovea (white line; divisions = 1°). B, Lower portion of a 1 μm cross section, including the fovea and three major vessels marked with red, yellow, and green arrows. These same three vessels are indicated in the retinal photograph with colored arrows. The boxed region is shown in the next panel. C, Higher magnification view of the central 1–3°. To illustrate how the centrifugal displacement of ganglion cells from their photoreceptors was measured, we have drawn a schematic example of three cells in the chain. The axon from a single cone at 1° follows a long radial course in the Henle fiber layer (solid line). It terminates on a bipolar cell at 2.8°, which projects to a ganglion cell at 3.1° (dashed line). Measurements of the displacement at 1, 2, 4, and 8° were made independently by both authors. These agreed within 5% and therefore were averaged.

Fig. 16.

Eccentricity plot of photoreceptors versus corresponding ganglion cells. The solid line represents zero relative displacement. In the macula the ganglion cells are displaced centrifugally relative to their photoreceptors, accounting for the deviation of points from the solid line. By 16° the ganglion cells are centered over the photoreceptors that supply them.

Fig. 17.

Plot of effective ganglion cell density after correction for eccentric displacement in the retina, showing a huge peak at the fovea. The dashed line shows the unshifted ganglion cell density, replotted from Figure 14.

Fig. 18.

Log–log plot showing the cumulative percentage of ganglion cells and cortical surface area from the fovea to each eccentricity. Note that at central eccentricities the percentage of cortical surface area exceeds by twofold the percentage of retinal ganglion cells. Therefore, relative to its number of ganglion cells, the central visual field is “over-represented” in striate cortex.