Search

The central part of the retina, which includes the fovea, is substantially overrepresented in the topographic map of the retina in the striate cortex. We tested whether this simply reflects the uneven distribution of ganglion cells in the retina in accordance with the “principle” of peripheral scaling, or whether there is additional expansion of the fovea and adjacent retina in the retinocortical projection. Wheatgerm agglutinated horseradish peroxidase was injected into the striate cortex of three rhesus macaque monkeys so as to surround the representation of the fovea at a mean eccentricity of 8.6°, and the retinae were processed histochemically to stain the retrogradely and transneuronally labelled ganglion cells which projected topographically to the injection sites. This enabled regions of the striate cortex to be related precisely to corresponding regions of the dorsal lateral geniculate nucleus and retina. Mathematical models of the distribution of ganglion cells in the retina, clipped, three-dimensional computer reconstructions of the striate cortex and lateral geniculate nucleus, and counts of neurons in the latter, were used to calculate the proportion of neurons allocated to the marked perifoveal region at each stage of projection. This was used to calculate the relative allocation of neurons to the representation of the fovea and surrounding retina among the different stages of the visual pathway. The values obtained showed that the cortical representation of the perifovea was expanded two to three times more than could be accounted for on the basis of ganglion cell topography in the retina, and that the expansion occurred both between the retina and the thalamus, and between the thalamus and the cortex. These results are inconsistent with the idea that peripheral scaling is a general principle of sensory representation in the cortex. They could also explain why many visual thresholds, including hyperacuities, cannot be accounted for by peripheral factors such as ganglion cell density.

We devised mathematical models of the topography of ganglion cells in the retina of macaque monkeys. The models consisted of a sum-of-three exponentials function fitted to measurements of ganglion cell density made on the nasal horizontal meridian, combined with known anisotropies across the horizontal and vertical meridians by means of elliptic interpolation to provide a full description of their density across the whole of the retinal surface. Integration using standard numerical techniques allowed the number of ganglion cells in arbitrary regions of the retina to be estimated. The topography of actual and effective total ganglion cell populations, and of primate alpha and gamma retinal ganglion cells, was modelled on previously published data. The models were used to test the hypothesis that the retinal projection to the striate cortex in macaque monkeys is peripherally scaled (i.e. merely reflects the eccentricity-dependent variation in density of ganglion cells in the retina) by comparing the cumulative proportion of ganglion cells with the cumulative proportion of cortical area as a function of eccentricity in the visual field. Discrepancies between the two curves indicated that the fovea and immediately surrounding retina are overrepresented in the striate cortex (i.e. there is more cortex per ganglion cell in and near the fovea than in the periphery), and the fact that the discrepancies persisted out to 25-50 degrees of eccentricity showed that the overrepresentation cannot be explained by the lateral displacement of foveal ganglion cells.

We placed pellets of horseradish peroxidase in the superior colliculus of four macaque monkeys and retrogradely labelled the retinal ganglion cells of both eyes. The ratio of labelled cells in the contralateral nasal retina and the ipsilateral temporal retina was no different from the ratio found after implants in the optic nerve, which label the entire afferent pathway. Our finding therefore invalidates the proposal that prominent differences in the properties of “blindsight” in monocular nasal and temporal visual fields arise from differences in the projection from the nasal and temporal retina to the midbrain. We also measured the size of the soma and dendritic field of the labelled ganglion cells (mostly gamma cells) and compared them with those of alpha and beta cells that project to the dorsal lateral geniculate nucleus. Soma size was very close to that of beta cells at all eccentricities but was much smaller than that of alpha cells. Dendritic field size was significantly larger than that of beta cells but was smaller than that of alpha cells. The number of primary dendrites was counted for cells labelled from the midbrain and in samples of alpha and beta cells labelled from the optic nerve. At eccentricities of 3–7 mm there was a consistent and prominent difference between beta and gamma cells. The results show that at intermediate eccentricities even ganglion cells whose distal dendrites are too poorly labelled to reveal their morphological class can never the less be categorized as alpha, beta or gamma by using a combination of soma size and number of primary dendrites. This is particularly useful when attempting to classify retinal ganglion cells following microinjections into selected target nuclei of optic axons.