Central projections of intrinsically photosensitive retinal ganglion cells in the macaque monkey.

Circadian rhythms generated by the suprachiasmatic nucleus (SCN) are entrained to the environmental light/dark cycle via intrinsically photosensitive retinal ganglion cells (ipRGCs) expressing the photopigment melanopsin and the neuropeptide pituitary adenylate cyclase-activating polypeptide (PACAP). The ipRGCs regulate other nonimage-forming visual functions such as the pupillary light reflex, masking behavior, and light-induced melatonin suppression. To evaluate whether PACAP-immunoreactive retinal projections are useful as a marker for central projection of ipRGCs in the monkey brain, we characterized the occurrence of PACAP in melanopsin-expressing ipRGCs and in the retinal target areas in the brain visualized by the anterograde tracer cholera toxin subunit B (CtB) in combination with PACAP staining. In the retina, PACAP and melanopsin were found to be costored in 99% of melanopsin-expressing cells characterized as inner and outer stratifying melanopsin RGCs. Two macaque monkeys were anesthetized and received a unilateral intravitreal injection of CtB. Bilateral retinal projections containing colocalized CtB and PACAP immunostaining were identified in the SCN, the lateral geniculate complex including the pregeniculate nucleus, the pretectal olivary nucleus, the nucleus of the optic tract, the brachium of the superior colliculus, and the superior colliculus. In conclusion, PACAP-immunoreactive projections with colocalized CtB represent retinal projections of ipRGCs in the macaque monkey, supporting previous retrograde tracer studies demonstrating that melanopsin-containing retinal projections reach areas in the primate brain involved in both image- and nonimage-forming visual processing.

Characterization and use of a digital light projector for vision research.

For creating stimuli in the laboratory, digital light projection (DLP) technology has the potential to overcome the low output luminance, lack of pixel independence, and limited chromaticity gamut of the cathode ray tube (CRT). We built a DLP-based stimulator for projecting patterns on the in vitro primate retina. The DLP produces high light levels and has good contrast. Spatial performance was similar to that of a CRT. Temporal performance was limited by the refresh rate (63 Hz). The chromatic gamut was modestly larger than that of a CRT although the primary spectra varied to a small degree with light output and numerical aperture.

Primate horizontal cell dynamics: an analysis of sensitivity regulation in the outer retina.

The human cone visual system maintains sensitivity over a broad range of illumination, from below 1 troland to 1,000,000 trolands. While the cone photoreceptors themselves are an important locus for sensitivity regulation-or light adaptation-the degree to which they contribute in primates remains unclear. To determine the range of sensitivity regulation in the outer retina, the temporal dynamics, neural gain control, and response range compression were measured in second-order neurons, the H1 horizontal cells, of the macaque retina. Situated at the first synapse in the retina, H1 cells receive input from a large population of cones. Lee et al. have previously shown that sensitivity regulation in H1 cells is both cone type-specific and spatially restricted. The sensitivity regulation seen in H1 cells at moderate illuminances thus takes place before the summation of cone signals in these cells, and the data establish the H1 cell as a convenient locus for analyzing cone signals. In the present study, cone-driven responses of primate H1 cells to temporally modulated sine-wave stimuli and to increment pulses were measured at steady levels of 1-1,000 trolands. The H1 cell gave a modulated response to sine-wave stimuli and hyperpolarized to increment pulses with overshoots at stimulus onset and offset. The temporal amplitude sensitivity function was primarily low-pass in shape, with a small degree of low-frequency roll off and a resonance shoulder near 40 Hz. A model incorporating a cascade of first-order filters together with an underdamped second-order filter could describe both temporal sinusoidal and pulse hyperpolarizations. Amplitude sensitivity was estimated from both pulse and sine-wave data as a function of the steady adaptation level. Sensitivity at low light levels (1 troland) showed a slowing in temporal dynamics, indicating time-dependent sensitivity regulation. Sensitivity was reduced at light levels above approximately 10 trolands, reflecting both response range compression and neural gain control. Thus the outer retina is a major locus for sensitivity regulation in primates.

Morphology of wide-field bistratified and diffuse human retinal ganglion cells.

To study the detailed morphology of human retinal ganglion cells, we used intracellular injection of horseradish peroxidase and Neurobiotin to label over 1,000 cells in an in vitro, wholemount preparation of the human retina. This study reports on the morphology of 119 wide-field bistratified and 42 diffuse ganglion cells. Cells were analyzed quantitatively on the basis of dendritic-field size, soma size, and the extent of dendritic branching. Bistratified cells were similar in dendritic-field diameter (mean +/- S.D. = 682 +/- 130 microm) and soma diameter (mean +/- S.D. = 18 +/- 3.3 microm) but showed a broad distribution in the extent of dendritic branching (mean +/- S.D. branch point number = 67 +/- 32; range = 15-167). Differences in the extent of branching and in dendritic morphology and the pattern of branching suggest that the human retina may contain at least three types of wide-field bistratified cells. Diffuse ganglion cells comprised a largely homogeneous group whose dendrites ramified throughout the inner plexiform layer. The diffuse cells had similar dendritic-field diameters (mean +/- S.D. = 486 +/- 113 microm), soma diameters (mean +/- S.D. = 16 +/- 2.3 microm), and branch points numbers (mean +/- s.D. = 92 +/- 32). The majority had densely branched dendritic trees and thin, very spiny dendrites with many short, fine, twig-like thorny processes. Five of the diffuse cells had much more sparsely branched dendritic trees (<50 branch points) and less spiny dendrites, suggesting that there are possibly two types of diffuse ganglion cells in human retina. Although the presence of a diversity of large bistratified and diffuse ganglion cells has been observed in a variety of mammalian retinas, little is known about the number of cell types, their physiological properties, or their central projections. Some of the human wide-field bistratified cells in the present study, however, show morphological similarities to monkey large bistratified cells that are known to project to the superior colliculus.

The mosaic of horizontal cells in the macaque monkey retina: with a comment on biplexiform ganglion cells.

To further characterize the H1 and H2 horizontal cell populations in macaque monkey retinae, cells were injected with the tracer Neurobiotin following intracellular recordings. Tracer coupling between cells of the same type revealed all H1 or H2 cells in small patches around the injected cell. The mosaics of their cell bodies and the tiling of the retina with their dendrites were analyzed. Morphological differences between the H1 and H2 cells observable in Neurobiotin-labeled patches made it possible to recognize H1 and H2 cells in retinae immunolabeled for the calcium-binding proteins parvalbumin and calbindin, and thus to study their relative spatial densities across the retina. These data, together with the intracellularly stained patches, show that H1 cells outnumber H2 cells at all eccentricities. There is, however, a change in the relative proportions of H1 and H2 cells with eccentricity: close to the fovea the ratio of H1 to H2 cells is approximately 4 to 1, in midperipheral retina approximately 3 to 1, and in peripheral retina approximately 2 to 1. In both the Neurobiotin-stained and the immunostained retinae, about 3-5% of the H2 cells were obviously misplaced into the ganglion cell layer. Several features of the morphology of the misplaced H2 cells suggest that they represent the so-called "biplexiform ganglion cells" previously described in Golgi studies of primate retina.

Parallel pathways for spectral coding in primate retina.

The primate retina is an exciting focus in neuroscience, where recent data from molecular genetics, adaptive optics, anatomy, and physiology, together with measures of human visual performance, are converging to provide new insights into the retinal origins of color vision. Trichromatic color vision begins when the image is sampled by short- (S), middle- (M) and long- (L) wavelength-sensitive cone photoreceptors. Diverse retinal cell types combine the cone signals to create separate luminance, red-green, and blue-yellow pathways. Each pathway is associated with distinctive retinal architectures. Thus a blue-yellow pathway originates in a bistratified ganglion cell type and associated interneurons that combine excitation from S cones and inhibition from L and M cones. By contrast, a red-green pathway, in which signals from L and M cones are opposed, is associated with the specialized anatomy of the primate fovea, in which the "midget" ganglion cells receive dominant excitatory input from a single L or M cone.

Center surround receptive field structure of cone bipolar cells in primate retina.

In non-mammalian vertebrates, retinal bipolar cells show center-surround receptive field organization. In mammals, recordings from bipolar cells are rare and have not revealed a clear surround. Here we report center-surround receptive fields of identified cone bipolar cells in the macaque monkey retina. In the peripheral retina, cone bipolar cell nuclei were labeled in vitro with diamidino-phenylindole (DAPI), targeted for recording under microscopic control, and anatomically identified by intracellular staining. Identified cells included 'diffuse' bipolar cells, which contact multiple cones, and 'midget' bipolar cells, which contact a single cone. Responses to flickering spots and annuli revealed a clear surround: both hyperpolarizing (OFF) and depolarizing (ON) cells responded with reversed polarity to annular stimuli. Center and surround dimensions were calculated for 12 bipolar cells from the spatial frequency response to drifting, sinusoidal luminance modulated gratings. The frequency response was bandpass and well fit by a difference of Gaussians receptive field model. Center diameters were all two to three times larger than known dendritic tree diameters for both diffuse and midget bipolar cells in the retinal periphery. In one instance intracellular staining revealed tracer spread between a recorded cell and its nearest neighbors, suggesting that homotypic electrical coupling may contribute to receptive field center size. Surrounds were around ten times larger in diameter than centers and in most cases the ratio of center to surround strength was approximately 1. We suggest that the center-surround receptive fields of the major primate ganglion cell types are established at the bipolar cell, probably by the circuitry of the outer retina.

Interindividual and topographical variation of L:M cone ratios in monkey retinas.

We analyzed the ratio of L:M cone photopigment mRNA in the retinas of Old World monkeys, using the method of rapid polymerase chain reaction-single-strand conformation polymorphism. The L:M cone pigment mRNA ratio in whole retina ranged from 0.6 to 7.0, with a mean of approximately 1.6 (standard deviation, +/- 0.56; n = 26). There was no change in this ratio with eccentricity up to 9 mm (approximately 45 degrees), though the ratio was approximately 30% greater in temporal than in nasal retina. The mRNA ratios are in good agreement with the L:M cone ratio in these same retinas, inferred from electrophysiological recordings of cone signal gain in horizontal cell interneurons. The correlation between mRNA ratios and physiological cone gain ratio supports the conclusion that both measures reflect the relative number of L and M cones.

Physiology of L- and M-cone inputs to H1 horizontal cells in the primate retina.

In the primate retina, H1 horizontal cells form an electrically coupled network and receive convergent input from long- (L-) and middle- (M-) wavelength-sensitive cones. Using an in vitro preparation of the intact retina to record the light-evoked voltage responses of H1 cells, we systematically varied the L- and M-cone stimulus contrast and measured the relative L- and M-cone input strength for 137 cells across 33 retinas from three Old World species (Macaca nemestrina, M. fascicularis, and Papio anubis). We found that the L- and the M-cone inputs were summed by the H1 cell in proportion to the stimulus cone contrast, which yielded a measure of what we term L- and M-cone contrast gain. The proportion of L-cone contrast gain was highly variable, ranging from 25% to 90% [mean +/- standard deviation, (60 +/- 14)%]. This variability was accounted for by retinal location within an individual, with the temporal retina showing a consistently higher percentage of L-cone gain, and by large overall variation across individuals, with the mean percentage of L-cone gain ranging from 32% to 80%. We hypothesize that the relative L- and M-cone contrast gain is determined simply by the relative number of L and M cones in the H1 cell's receptive field and that the variability in L- and M-cone contrast gain reflects a corresponding variability in the mosaic of L and M cones.

Horizontal cells reveal cone type-specific adaptation in primate retina.

The human cone visual system maintains contrast sensitivity over a wide range of ambient illumination, a property known as light adaptation. The first stage in light adaptation is believed to take place at the first neural step in vision, within the long, middle, and short wavelength sensitive cone photoreceptors. To determine the properties of adaptation in primate outer retina, we measured cone signals in second-order interneurons, the horizontal cells, of the macaque monkey. Horizontal cells provide a unique site for studying early adaptational mechanisms; they are but one synapse away from the photoreceptors, and each horizontal cell receives excitatory inputs from many cones. Light adaptation occurred over the entire range of light levels evaluated, a luminance range of 15-1,850 trolands. Adaptation was demonstrated to be independent in each cone type and to be spatially restricted. Thus, in primates, a major source of sensitivity regulation occurs before summation of cone signals in the horizontal cell.

Primate retina: cell types, circuits and color opponency.

The link between morphology and physiology for some of the cell types of the macaque monkey retina is reviewed with emphasis on understanding the neural mechanism for spectral opponency in the light response of ganglion cells. An in vitro preparation of the retina is used in which morphologically identified cell types are selectively targeted for intracellular recording and staining under microscopic control. The goal is to trace the physiological signals from the long (L), middle (M) and short-wavelength sensitive (S) cones to identified cell types that participate in opponent and non-opponent signal pathways. Heterochromatic modulation photometry and silent substitution are used to characterize L-, M- or S-cone inputs to the receptive fields of distinct horizontal cell, bipolar cell, ganglion cell and amacrine cell types. The majority of the retinal cell types await detailed analysis, and knowledge of the mechanisms of opponency remains incomplete. However results thus far have established: (1) Horizontal cell interneurons make preferential connections with the three cone types, but cannot provide a basis for spectral opponency in the circuitry of the outer retina. (2) A morphologically distinctive bistratified ganglion cell type transmits a blue-ON yellow-OFF spectral opponent signal to the parvocellular division of lateral geniculate nucleus. The morphology of this ganglion cell type suggests a simple synaptic mechanism for blue yellow opponency via converging input from an S-cone connecting ON-bipolar cell and an L - M cone connecting OFF-bipolar cell. (3) Midget ganglion cells, whose axons project to the parvocellular layers of the lateral geniculate nucleus and are assumed to be the origin of red/green opponent signals, show a non-opponent, achromatic physiology when recorded in the retinal periphery the underlying circuitry for red green opponency thus remains controversial, and (4) recent recordings from identified bipolar and amacrine cells in macaque suggest that a more complete accounting of opponent circuitry is a realistic goal.

Morphology of wide-field, monostratified ganglion cells of the human retina.

To determine the number of wide-field, monostratified ganglion cell classes present in the human retina, we analyzed a large sample of ganglion cells by intracellular staining in an in vitro, whole-mount preparation of the retina. Over 1000 cells were labeled by horseradish peroxidase or Neurobiotin; some 200 cells had wide dendritic trees narrowly or broadly stratified within either the inner (ON) or outer (OFF) portion of the inner plexiform layer. Based on dendritic-field size and the pattern and extent of dendritic branching, we have distinguished six wide-field cell groups. The giant very sparse ganglion cells included both inner and outer stratifying cells and were unique both for their extremely large dendritic field (mean diameter = 1077 microm) and extremely sparsely branched dendrites. Four of the cell groups had similarly large dendritic fields, ranging in mean diameter from 737 to 791 microm, but differed in the pattern and extent of dendritic branching, with the number of dendritic branch points ranging from a mean of 33 to 129. Of these four groups, the large very sparse group and the large dense group included both inner and outer stratifying cells, while the large sparse and large moderate groups consisted of inner stratifying cells only. The thorny monostratified ganglion cells were distinct from the other cells in having medium size dendritic fields (mean diameter = 517 microm) and moderately branched, inner stratifying dendritic trees with many thin, spiny, twig-like branchlets. All six groups had medium-size cell bodies, with mean soma diameters ranging from 17 to 21 microm. Though the physiological properties and central projections of human wide-field, monostratified ganglion cells are not known, some of the cells resemble macaque ganglion cells known to project to the lateral geniculate nucleus, the pretectum, or the superior colliculus.

Sensitivity and dynamics of rod signals in H1 horizontal cells of the macaque monkey retina.

We measured the sensitivity, temporal frequency response, latency, and receptive field diameter of rod input to the H1 horizontal cell type in an in vitro preparation of the macaque retina. The H1 cell has both a cone-connected dendritic tree and a long axon-like process that terminates in a rod-connected arbor. We recorded from the H1 cell body where rod signals were distinguished by sensitivity to short wavelength light after dark adaptation. Receptive fields of rod vs. cone mediated responses were coextensive, indicating that the rod signal is transmitted via rod-cone gap junctions. Sensitivity of the H1 cell rod signal was approximately 1 log unit higher than that of the cone signal. Below cone threshold rod signals were temporally low-pass, with a cutoff frequency below 10 Hz. Rod signals became faster and more transient with increasing light levels. We conclude that the H1 cell rod signal is not sensitive in the low scotopic range and, by comparison with the rod signal recorded directly in cones (Schneeweis & Schnapf (1995) Science, 268, 1053-1056), signal transmission across the cone-H1 synapse does not significantly filter the temporal properties of the rod signal.

Morphology of human retinal ganglion cells with intraretinal axon collaterals.

Ganglion cells with intraretinal axon collaterals have been described in monkey (Usai et al., 1991), cat (Dacey, 1985), and turtle (Gardiner & Dacey, 1988) retina. Using intracellular injection of horseradish peroxidase and Neurobiotin in in vitro whole-mount preparations of human retina, we filled over 1000 ganglion cells, 19 of which had intraretinal axon collaterals and wide-field, spiny dendritic trees stratifying in the inner half of the inner plexiform layer. The axons were smooth and thin (approximately 2 microm) and gave off thin (<1 microm), bouton-studded terminal collaterals that extended vertically to terminate in the outer half of the inner plexiform layer. Terminal collaterals were typically 3-300 microm in length, though sometimes as long as 700 microm, and were present in clusters, or as single branched or unbranched varicose processes with round or somewhat flattened lobular terminal boutons 1-2 microm in diameter. Some cells had a single axon whereas other cells had a primary axon that gave rise to 2-4 axon branches. Axons were located either in the optic fiber layer or just beneath it in the ganglion cell layer, or near the border of the ganglion cell layer and the inner plexiform layer. This study shows that in the human retina, intraretinal axon collaterals are associated with a morphologically distinct ganglion cell type. The synaptic connections and functional role of these cells are not yet known. Since distinct ganglion cell types with intraretinal axon collaterals have also been found in monkey, cat, and turtle, this cell type may be common to all vertebrate retinas.

Physiology of the A1 amacrine: a spiking, axon-bearing interneuron of the macaque monkey retina.

We characterized the light response, morphology, and receptive-field structure of a distinctive amacrine cell type (Dacey, 1989), termed here the A1 amacrine, by applying intracellular recording and staining methods to the macaque monkey retina in vitro. A1 cells show two morphologically distinct components: a highly branched and spiny dendritic tree, and a more sparsely branched axon-like tree that arises from one or more hillock-like structures near the soma and extends for several millimeters beyond the dendritic tree. Intracellular injection of Neurobiotin reveals an extensive and complex pattern of tracer coupling to neighboring A1 amacrine cells, to two other amacrine cell types, and to a single ganglion cell type. The A1 amacrine is an ON-OFF cell, showing a large (10-20 mV) transient depolarization at both onset and offset of a photopic, luminance modulated stimulus. A burst of fast, large-amplitude (approximately 60 mV) action potentials is associated with the depolarizations at both the ON and OFF phase of the response. No evidence was found for an inhibitory receptive-field surround. The spatial extent of the ON-OFF response was mapped by measuring the strength of the spike discharge and/or the amplitude of the depolarizing slow potential as a function of the position of a bar or spot of light within the receptive field. Receptive fields derived from the slow potential and associated spike discharge corresponded in size and shape. Thus, the amplitude of the slow potential above spike threshold was well encoded as spike frequency. The diameter of the receptive field determined from the spike discharge was approximately 10% larger than the spiny dendritic field. The correspondence in size between the spiking receptive field and the spiny dendritic tree suggests that light driven signals are conducted to the soma from the dendritic tree but not from the axon-like arbor. The function of the axon-like component is unknown but we speculate that it serves a classical output function, transmitting spikes distally from initiation sites near the soma.

Horizontal cells of the primate retina: cone specificity without spectral opponency.

The chromatic dimensions of human color vision have a neural basis in the retina. Ganglion cells, the output neurons of the retina, exhibit spectral opponency; they are excited by some wavelengths and inhibited by others. The hypothesis that the opponent circuitry emerges from selective connections between horizontal cell interneurons and cone photoreceptors sensitive to long, middle, and short wavelengths (L-, M-, and S-cones) was tested by physiologically and anatomically characterizing cone connections of horizontal cell mosaics in macaque monkeys. H1 horizontal cells received input only from L- and M-cones, whereas H2 horizontal cells received a strong input from S-cones and a weaker input from L- and M-cones. All cone inputs were the same sign, and both horizontal cell types lacked opponency. Despite cone type selectivity, the horizontal cell cannot be the locus of an opponent transformation in primates, including humans.

Circuitry for color coding in the primate retina.

Human color vision starts with the signals from three cone photoreceptor types, maximally sensitive to long (L-cone), middle (M-cone), and short (S-cone) wavelengths. Within the retina these signals combine in an antagonistic way to form red-green and blue-yellow spectral opponent pathways. In the classical model this antagonism is thought to arise from the convergence of cone type-specific excitatory and inhibitory inputs to retinal ganglion cells. The circuitry for spectral opponency is now being investigated using an in vitro preparation of the macaque monkey retina. Intracellular recording and staining has shown that blue-ON/yellow-OFF opponent responses arise from a distinctive bistratified ganglion cell type. Surprisingly, this cone opponency appears to arise by dual excitatory cone bipolar cell inputs: an ON bipolar cell that contacts only S-cones and an OFF bipolar cell that contacts L- and M-cones. Red-green spectral opponency has long been linked to the midget ganglion cells, but an underlying mechanism remains unclear. For example, receptive field mapping argues for segregation of L-and M-cone signals to the midget cell center and surround, but horizontal cell interneurons, believed to generate the inhibitory surround, lack opponency and cannot contribute selective L- or M-cone input to the midget cell surround. The solution to this color puzzle no doubt lies in the great diversity of cell types in the primate retina that still await discovery and analysis.

The 'blue-on' opponent pathway in primate retina originates from a distinct bistratified ganglion cell type.

Colour vision in humans and Old World monkeys begins with the differential activation of three types of cone photoreceptor which are maximally sensitive to short (S), medium (M) and long (L) wavelengths. Signals from the three cone types are relayed to the retinal ganglion cells via cone-specific bipolar cell types. Colour-coding ganglion cells fall into two major physiological classes: the red-green opponent cells, which receive antagonistic input from M- and L-sensitive cones, and the blue-yellow opponent cells, which receive input from S-sensitive cones, opposed by combined M- and L-cone input. The neural mechanisms producing colour opponency are not understood. It has been assumed that both kinds of opponent signals are transmitted to the lateral geniculate nucleus by one type of ganglion cell, the midget cell. We now report that a distinct non-midget ganglion cell type, the small bistratified cell, corresponds to the physiological type that receives excitatory input from S cones, the 'blue-on' cell. Our results thus demonstrate an anatomically distinct pathway that conveys S-cone signals to the brain. The morphology of the blue-on cell also suggests a novel hypothesis for the retinal circuitry underlying the blue-yellow opponent response.

Physiology, morphology and spatial densities of identified ganglion cell types in primate retina.

The use of in vitro preparations of primate retina provides new perspectives on the mosaic organization and physiological properties of three ganglion cell types that project to the lateral geniculate nucleus: the parasol, midget and small bistratified cells. Dendritic field sizes and coverage for the three types suggest that their relative densities vary with eccentricity. Of the total ganglion cells in the human fovea, midget cells constitute about 90%, parasol cells about 5%, and small bistratified cells about 1%. In the periphery, midget cells make up about 40-45%, parasol cells about 20% and small bistratified cells about 10% of the total. Thus from peripheral to central retina the number of midget ganglion cells progressively increases relative to the parasol and small bistratified types. Physiological properties of these cells have recently been studied in macaque (Macaca nemestrina) retina by combining intracellular recording and dye injection. As expected, parasol cells, projecting to geniculate magnocellular layers, give phasic, non-opponent light responses. Midget cells, which project to geniculate parvocellular layers, show opponent responses sensitive to only mid and long wavelengths; no evidence of short-wavelength-sensitive cone (S-cone) input to any midget ganglion cell has been found. However, the small bistratified cells, which also project to the parvocellular geniculate layers, give a strong blue-ON response to stimuli designed to modulate S-cones. Thus, S-cone and medium- or long-wavelength-sensitive cone opponent signals arise from morphologically distinct ganglion cell types that project in parallel to the lateral geniculate nucleus.

The mosaic of midget ganglion cells in the human retina.

To study their detailed morphology, ganglion cells of the human retina were stained by intracellular tracer injection, in an in vitro, whole-mount preparation. This report focuses on the dendritic morphology and mosaic organization of the major, presumed color-opponent, ganglion cell class, the midget cells. Midget cells in the central retina were recognized by their extremely small dendritic trees, approximately 5-10 microns in diameter. Between 2 and 6 mm eccentricity, midget cells showed a steep, 10-fold increase in dendritic field size, followed by a more shallow, three- to fourfold increase in the retinal periphery, attaining a maximum diameter of approximately 225 microns. Despite large local variation in dendritic field size, midget cells formed one morphologically distinctive class at all retinal eccentricities. Two midget cell types were distinguished by their dendritic stratification in either the inner or outer portion of the inner plexiform layer (IPL), and presumably correspond to ON- and OFF-center cells respectively. The mosaic organization of the midget cells was examined by intracellularly filling neighboring cells in small patches of retina. For both the inner and outer midget populations, adjacent dendritic trees apposed one another but did not overlap, establishing a coverage of no greater than 1. The two mosaics differed in spatial scale, however: the outer midget cells showed smaller dendritic fields and higher cell density than the inner midget cells. An outer:inner cell density ratio of 1.7:1 was found in the retinal periphery. An estimate of total midget cell density suggested that the proportion of midget cells increases from about 45% of total ganglion cell density in the retinal periphery to about 95% in the central retina. Nyquist frequencies calculated from midget cell spacing closely match a recent measure of human achromatic spatial acuity (Anderson et al., 1991), from approximately 6 degrees to 55 degrees eccentricity. Outside the central retina, midget cell dendrites arborized in clusters within the overall dendritic field. With increasing eccentricity, the dendritic clusters increased in number and remained small (approximately 10-20 microns diameter) relative to the size of the dendritic field. Because neighboring midget cell dendritic trees do not overlap, the mosaic as a whole showed a pattern of clusters and holes. We hypothesize that midget cell dendritic trees may contact individual axon terminals of some midget bipolar cells and avoid contacting others, providing a basis for the formation of cone-specific connections in the IPL.