Amacrine cells represent the most diverse class of retinal neuron, comprising dozens of distinct cell types. Each type exhibits a unique morphology and generates specific visual computations through its synapses with a subset of excitatory interneurons (bipolar cells), other amacrine cells, and output neurons (ganglion cells). Here, we review the intrinsic and network properties that underlie the function of the most common amacrine cell in the mammalian retina, the AII amacrine cell. The AII connects rod and cone photoreceptor pathways, forming an essential link in the circuit for rod-mediated (scotopic) vision. As such, the AII has become known as the rod–amacrine cell. We, however, now understand that AII function extends to cone-mediated (photopic) vision, and AII function in scotopic and photopic conditions utilizes the same underlying circuit: AIIs are electrically coupled to each other and to the terminals of some types of ON cone bipolar cells. The direction of signal flow, however, varies with illumination. Under photopic conditions, the AII network constitutes a crossover inhibition pathway that allows ON signals to inhibit OFF ganglion cells and contributes to motion sensitivity in certain ganglion cell types. We discuss how the AII’s combination of intrinsic and network properties accounts for its unique role in visual processing.

Thirty years ago virtually everything known about primate color vision derived from psychophysical studies of normal and color-defective humans and from physiological investigations of the visual system of the macaque monkey, the most popular of human surrogates for this purpose. The years since have witnessed much progress toward the goal of understanding this remarkable feature of primate vision. Among many advances, investigations focused on naturally occurring variations in color vision in a wide range of nonhuman primate species have proven to be particularly valuable. Results from such studies have been central to our expanding understanding of the interrelationships between opsin genes, cone photopigments, neural organization, and color vision. This work is also yielding valuable insights into the evolution of color vision.

The retinal ganglion cells (RGCs) of the primate form at least two classes—M and P—that differ fundamentally in their functional properties. M cells have temporal-frequency response characteristics distinct from P cells (Benardete et al., 1992; Lee et al., 1994). In this paper, we elaborate on the temporal-frequency responses of M cells and focus in detail on the contrast gain control (Shapley & Victor, 1979a,b). Earlier data showed that the temporal-frequency response of M cells is altered by the level of stimulus contrast (Benardete et al., 1992). Higher contrast shifts the peak of the frequency-response curve to higher temporal frequency and produces a phase advance. In this paper, by fitting the data to a linear filter model, the effect of contrast on the temporal-frequency response is subsumed into a change in a single parameter in the model. Furthermore, the model fits are used to predict the response of M cells to steps of contrast, and these predictions demonstrate the dynamic effect of contrast on the M cells' response. We also present new data concerning the spatial organization of the contrast gain control in the primate and show that the signal that controls the contrast gain must come from a broadly distributed network of small subunits in the surround of the M-cell receptive field.

A neural network model is developed to explain how visual thalamocortical interactions give rise to boundary percepts such as illusory contours and surface percepts such as filled-in brightnesses. Top-down feedback interactions are needed in addition to bottom-up feed-forward interactions to simulate these data. One feedback loop is modeled between lateral geniculate nucleus (LGN) and cortical area V1, and another within cortical areas V1 and V2. The first feedback loop realizes a matching process which enhances LGN cell activities that are consistent with those of active cortical cells, and suppresses LGN activities that are not. This corticogeniculate feedback, being endstopped and oriented, also enhances LGN ON cell activations at the ends of thin dark lines, thereby leading to enhanced cortical brightness percepts when the lines group into closed illusory contours. The second feedback loop generates boundary representations, including illusory contours, that coherently bind distributed cortical features together. Brightness percepts form within the surface representations through a diffusive filling-in process that is contained by resistive gating signals from the boundary representations. The model is used to simulate illusory contours and surface brightnesses induced by Ehrenstein disks, Kanizsa squares, Glass patterns, and cafe wall patterns in single contrast, reverse contrast, and mixed contrast configurations. These examples illustrate how boundary and surface mechanisms can generate percepts that are highly context-sensitive, including how illusory contours can be amodally recognized without being seen, how model simple cells in V1 respond preferentially to luminance discontinuities using inputs from both LGN ON and OFF cells, how model bipole cells in V2 with two colinear receptive fields can help to complete curved illusory contours, how short-range simple cell groupings and long-range bipole cell groupings can sometimes generate different outcomes, and how model double-opponent, filling-in and boundary segmentation mechanisms in V4 interact to generate surface brightness percepts in which filling-in of enhanced brightness and darkness can occur before the net brightness distribution is computed by double-opponent interactions.

The zebrafish has recently assumed a central position in the study of vertebrate development. Numerous studies of other fish have shown that their central nervous systems, and especially their visual systems, continue to add new neurons throughout life, which is probably related to their abilities to regenerate axons and whole nervous tissue. Retinal neurogenesis had not been examined in adult zebrafish, and two reports concluded that the optic tectum ceased neurogenesis early in life, so the question arose whether the zebrafish was anomalous in this regard. We labeled embryonic (24- and 48-h postfertilization) and adult zebrafish with the thymidine analog, bromo-deoxyuridine, and, after short and long survivals, examined the retina and brain for labeled cells. They were abundant in both the optic tectum and the retina. Although the rate of retinal growth slows considerably between embryonic and adult stages, the patterns of neurogenesis in both the embryo and the adult are similar to those described in other fish, so these “fish-specific” features of general interest can justifiably be studied in zebrafish.

The orientation selectivity, ocular dominance, and binocular disparity tuning of 272 cells in areas 17 and 18 of barbiturate-anesthetized, paralyzed cats were studied with automated, quantitative techniques. Disparity was varied along the axis orthogonal to each cell's best orientation. Binocular correspondence was established by means of a reference electrode positioned at the boundary of lamina A and Al in the area centralis representation of the lateral geniculate nucleus. Measures were derived that expressed each cell's disparity sensitivity and best disparity and the shape and slope of its tuning curve. Cells were found that corresponded to categories described by previous authors (“disparity-insensitive,” “tuned excitatory,” “near,” and “far” cells), but many others had intermediate response patterns, or patterns that were difficult to categorize. Quantitative analysis suggested that the various types belong to a continuum.

No relationship could be established between a cell's best orientation and its ocular dominance or any aspect of its disparity tuning. There was no relationship between a cell's ocular dominance and its sensitivity to disparity. Ocular dominance and best disparity were related. As reported by others, cells with best disparities close to zero (the fixation plane) tended to have balanced ocularity, while cells with best disparities in the near or far range had a broad distribution of ocular dominance. Among cells with receptive fields near the vertical meridian, those preferring far disparities tended to be dominated by the contralateral eye, and those preferring near disparities by the ipsilateral eye. It is suggested that this relationship follows from the geometry of near and far images and the pattern of decussation in the visual pathway. There was a significant grouping of cells with similar best disparities along tangential electrode tracks. We believe that this grouping is due to the columnar organization for ocular dominance and the relationship between ocular dominance and best disparity. No evidence was found for a columnar segregation of disparity-sensitive and disparity-insensitive cells.

Myopia is increasing in prevalence world-wide, nearing epidemic proportions in some populations. This has led to expanded research efforts to understand how ocular growth and refractive errors are regulated. Eye growth is sensitive to visual experience, and is altered by both form deprivation and optical defocus. In these cases, the primary targets of growth regulation are the choroidal and scleral layers of the eye that demarcate the boundary of the posterior vitreous chamber. Of significance to this review are observations of local growth modulation that imply that the neural retina itself must be the source of growth-regulating signals. Thus the retinal pigment epithelium (RPE), interposed between the retina and the choroid, is likely to play a critical role in relaying retinal growth signals to the choroid and sclera. This review describes the ion transporters and signal receptors found in the chick RPE and their possible roles in visually driven changes in eye growth. We focus on the effects of four signaling molecules, otherwise implicated in eye growth changes (dopamine, acetylcholine, vasoactive intestinal peptide (VIP), and glucagon), on RPE physiology, including fluid transport. A model for RPE-mediated growth regulation is proposed.

An eletrical potential recorded from the cornea, the a-wave of the ERG, is evaluated as a measure of human photoreceptor activity by comparing its behavior to a model derived from in vitro recordings from rod photoreceptors. The leading edge of the ERG exhibits both the linear and nonlinear behavior perdicted by this model. The capability for recording the electrical activity of humans photoreceptors in in vivo opens new avenues for assessing normal and abnormal receptor activity in humans. Furthermore, the quantitative model of the receptor response can be used to isolate the inner retinal contribution, Granit's PII, to the gross ERG. Based on this analysis, the practice of using the trough-to-peak amplitude of the b-wave as a proxy for the amplitude of the inner nuclear layer activity is evaluated.

Relay cells of the lateral geniculate nucleus, like those of other thalamic nuclei, manifest two distinct response modes, and these represent two very different forms of relay of information to cortex. When relatively hyperpolarized, these relay cells respond with a low threshold Ca2+ spike that triggers a brief burst of conventional action potentials. These cells switch to tonic mode when depolarized, since the low threshold Ca2+ spike, being voltage dependent, is inactivated at depolarized levels. In this mode they relay information with much more fidelity. This switch can occur under the influence of afferents from the visual cortex or parabrachial region of the brain stem. It has been previously suggested that the tonic mode is characteristic of the waking state while the burst mode signals an interruption of the geniculate relay during sleep. This review surveys the key properties of these two response modes and discusses the implications of new evidence that the burst mode may also occur in the waking animal.

Mosaics of neurons are usually quantified by methods based on nearest-neighbor distance (NND). The commonest indicator of regularity has been the ratio of the mean NND to the standard deviation, here termed the ‘conformity ratio.’ However, an accurate baseline value of this ratio for bounded random samples has never been determined; nor was its sampling distribution known, making it impossible to test its significance. Instead, significance was assessed from goodness-of-fit to a Rayleigh distribution, or from another ratio, that of the observed mean NND to an expected mean predicted by theory, termed the dispersion index. Neither approach allows for boundary effects that are common in experimental mosaics. Equally common are ‘missing’ neurons, whose effects on the statistics have not been studied. To address these deficiencies, random patterns and real neuronal mosaics were analyzed statistically. Ns independent random-point samples of size Np were generated for 13 Np values between 25 and 6400, where Ns × Np ≥ 144,000. Samples were generated with rectangular boundaries of aspect ratio 1:1, 1:5, and 1:10 to examine the influence of sample geometry. NND distributions, conformity ratios, and dispersion indices were computed for the resulting 45,997 independent random patterns. From these, empirical sampling distributions and critical values were determined. NND distributions for small-to-medium, bounded, random populations were shown to differ significantly from Rayleigh distributions. Thus, goodness-of-fit tests are invalid for most experimental mosaics. Charts are presented from which the significance of conformity ratios or dispersion indices can be read directly. The conformity ratio reacts conservatively to extremes of sample geometry, and provides a useful and safe test. The dispersion index is nonconservative, making its use problematic. Tests based on the theoretical distribution of the dispersion index are unreliable for all but the largest samples. Random deletions were also made from 33 real retinal ganglion cell mosaics. The mean NND, conformity ratio, and dispersion index were determined for each original mosaic and 36 independent samples at each of nine sampling levels, retaining between 90% and 10% of the original population. An exclusion radius, based on a spatial autocorrelogram, was also calculated for each of these 10,725 mosaic samples. The mean NND was moderately insensitive to undersampling, rising smoothly. The exclusion radius was remarkably insensitive. The conformity ratio and dispersion index fell steeply, sometimes failing to reach significance while half of the cells still remained. For the same 33 original mosaics, linear regression showed the exclusion radius to be 62 ± 3% of the mean NND.

The responses of simple cells (recorded from within the striate visual cortex) were measured as a function of the contrast and the frequency of sine-wave grating patterns in order to explore the effect of contrast on the spatial and temporal phase transfer functions and on the spatiotemporal receptive field. In general, as the contrast increased, the phase of the response advanced by approximately 45 ms (approximately one-quarter of a cycle for frequencies near 5 Hz), although the exact value varied from cell to cell. The dynamics of this phase-advance were similar to the dynamics of the amplitude: the amplitude and the phase increased in an accelerating fashion at lower contrasts and then saturated at higher contrasts. Further, the gain for both the amplitude and the phase appeared to be governed by the magnitude of the contrast rather than the magnitude of the response. For the spatial phase transfer function, variations in contrast had little or no systematic effect; all of the phase responses clustered around a single straight line, with a common slope and intercept. This implies that the phase-advance was not due to a change in the spatial properties of the neuron; it also implies that the phase-advance was not systematically related to the magnitude of the response amplitude. On the other hand, for the temporal phase transfer function, the phase responses fell on five straight lines, related to the five steps in contrast. As the contrast increased, the phase responses advanced such that both the slope and the intercept were affected. This implies that the phase-advance was a result of contrast-induced changes in both the response latency and the shape/symmetry of the temporal receptive field.

In both anesthetized and behaving macaques, we examined the responses of neurons in the inferior temporal cortex (IT) to repeated presentation of a visual stimulus. In anesthetized animals, the responsiveness of IT neurons decreased with repeated stimulus presentation at interstimulus intervals (ISIs) of 2–12 s but not at 20 s. Responsiveness recovered after a 5-min period of no stimulus presentation. The response decrement was similar in anesthetized and awake animals at a 2-s ISI, but at a 6-s ISI, response decrement in the awake animal was much less.

The rod-driven, AII amacrine cells in the mammalian retina maintain homologous gap junctions with one another as well as heterologous gap junctions with on-cone bipolar cells. We used background illumination to study whether changes in the adaptational state of the retina affected the permeabilities of these two sets of gap junctions. To access changes in permeability, we injected single AII amacrine cells with the biotinylated tracer, Neurobiotin, and measured the extent of tracer coupling to neighboring AII cells and neighboring cone bipolar cells. We also measured the center-receptive field size of All cells to assess concomitant changes in electrical coupling. Our results indicate that in well dark-adapted retinas, AII cells form relatively small networks averaging 20 amacrine cells and covering about 75 μm. The size of these networks matched closely to the size of AII cell on-center receptive fields. However, over most of their operating range, AII cells formed dramatically larger networks, averaging 326 amacrine cells, which corresponded to an increased receptive-field size. As the retina was light adapted beyond the operating range of the AII cells, they uncoupled to form networks comparable in size to those seen in well dark-adapted retinas. Our results, then, indicate that the adaptational state of the retina has a profound effect on the extent of electrical coupling between AII amacrine cells. Although we observed light-induced changes in the number of tracer-coupled cone bipolar cells, these appeared to be an epiphenomenon of changes in homologous coupling between AII amacrine cells. Therefore, in contrast to the robust changes in AII–AII coupling produced by background illumination, our data provided no evidence of a light-induced modulation of coupling between AII cells and on-cone bipolar cells.

Three approaches to study the function of mammalian rod bipolar cells are described. Extracellular recordings from the intact cat eye under light- and dark-adapted conditions showed that in dark-adapted retina all light responses can be blocked by 2-amino-4-phosphonobutyrate (APB). Immunocytochemical staining with an antibody against protein kinase C (PKC) labeled rod bipolar cells in all mammalian retinae tested. When rat retinae were dissociated, PKC immunoreactivity was also found in isolated bipolar cells and could be used for their identification as rod bipolars. Patch-clamp recordings were performed from such dissociated rod bipolar cells and their responses to APB were measured. APB closed a nonselective cation channel in the cell membrane. The actions of GABA and glycine were also tested and both opened chloride channels in dissociated rod bipolar cells. These results suggest that rod bipolar cells are depolarized by a light stimulus and that GABA as well as glycine modulate their light responses.

To improve the usefulness of the ERG in identifying the sites and mechanisms of adaptation, development, and disease processes, a theoretical framework based upon Granit's analysis of the ERG was evaluated. The framework assumes that the ERG is the sum of two potentials, one, P3, generated by the receptors and the other, P2, generated by the cells of the INL. Hood and Birch (1990a, b) demonstrated that the leading edge of the a-wave can be quantitatively described by a model used to describe the response from single rod receptors. This model provides P3(t), a theoretical receptor response as a function of time, for any given flash intensity. The ERGs from normal observers and patients with retinal diseases were analyzed in this framework, first by deriving P2 by computer subtracting the predicted P3(t) responses. This analysis was successful and a computational model of the ERG was then derived. The model of P2(t) was constructed with linear filters and a static nonlinearity and using P3(t) as the input. The ERG for any given flash intensity is then P3(t) + P2(t) The model describes (1) the change both in implicit times and in trough-to-peak b–wave amplitudes with flash intensity for the normal, dark-adapted observers; and (2) the changes in b–wave implicit times and amplitudes for three patients with retinal diseases.

Among the implications drawn from these analyses were as follows: (1) The fits of the Naka-Rushton equation to trough-to-peak b–wave amplitudes must be interpreted with great care. (2) When the INL is affected by retinal disease, the b–wave may be a very poor reflection of INL activity. (3) The implicit time of the b–wave can provide a measure of receptor sensitivity.

Intracellular injections of Neurobiotin were used to determine whether the major ganglion cell classes of the macaque monkey retina, the magnocellular-projecting parasol, and the parvocellular-projecting midget cells showed evidence of cellular coupling similar to that recently described for cat retinal ganglion cells. Ganglion cells were labeled with the fluorescent dye acridine orange in an in vitro, isolated retina preparation and were selectively targeted for intracellular injection under direct microscopic control. The macaque midget cells, like the beta cells of the cat's retina, showed no evidence of tracer coupling when injected with Neurobiotin. By contrast, Neurobiotin-filled parasol cells, like cat alpha cells, showed a distinct pattern of tracer coupling to each other (homotypic coupling) and to amacrine cells (heterotypic coupling).

In instances of homotypic coupling, the injected parasol cell was surrounded by a regular array of 3–6 neighboring parasol cells. The somata and proximal dendrites of these tracer-coupled cells were lightly labeled and appeared to costratify with the injected cell. Analysis of the nearest-neighbor distances for the parasol cell clusters showed that dendritic-field overlap remained constant as dendritic-field size increased from 100–400 μm in diameter.

At least two amacrine cell types showed tracer coupling to parasol cells. One amacrine type had a small soma and thin, sparsely branching dendrites that extended for 1–2 mm in the inner plexiform layer. A second amacrine type had a relatively large soma, thick main dendrites, and distinct, axon-like processes that extended for at least 2–3 mm in the inner plexiform layer. The main dendrites of the large amacrine cells were closely apposed to the dendrites of parasol cells and may be the site of Neurobiotin transfer between the two cell types. We suggest that the tracer coupling between neighboring parasol cells takes place indirectly via the dendrites of the large amacrine cells and provides a mechanism, absent in midget cells, for increasing parasol cell receptive-field size and luminance contrast sensitivity.

A specific morphological class of ganglion cell, the alpha cell, was first defined in cat retina. Alpha cells have since been found in a wide range of mammalian retinae, including several orders of placental and marsupial mammals. Characteristically, they have the largest somata and a large dendritic field with a typical branching pattern. They occur as inner and outer stratifying subpopulations, presumably corresponding to ON-center and OFF-center receptive fields. In all species, alpha cells account for less than 10% of the ganglion cells, their somata are regularly spaced, and their dendritic fields evenly and economically cover the retina in a mosaic-like fashion. The morphology of alpha cells and many features, both of single cells and of the population, are conserved across species with different habitats and life-styles. This suggests that alpha cells are a consistent obligatory ganglion cell type in every mammalian retina and probably subserve some fundamental task(s) in visual performance.

Some general rules about the construction principles of ganglion cell classes are inferred from the alpha cells, stressing the importance of population parameters for the definition of a class. The principle, that a functionally and morphologically homogeneous population should have a regular arrangement and a complete and even coverage of the retina to perform its part in image processing at each retinal location, is especially evident across species and across ganglion cell types.

We have studied the responses of MT neurons to moving gratings, spatially modulated in luminance and chromaticity. Most MT neurons responded briskly and with high contrast sensitivity to targets whose luminance was modulated, with or without added chromatic contrast. When luminance modulation was removed and only chromatic stimulation was used, the responses of all MT neurons were attenuated. Most were completely unresponsive to stimulation with targets whose modulation fell within a “null” plane in color space; these null planes varied from neuron to neuron, but all lay close to the plane of constant photometric luminance. For about a third of the neurons, there was no color direction in which responses were completely abolished; almost all of these neurons had a definite minimum response for chromatic modulation near the isoluminant plane. MT neurons that responded to isoluminant targets did so inconsistently and with poor contrast sensitivity, so that only intensely modulated targets were effective. Whereas the best thresholds of MT neurons for luminance targets are close to behavioral contrast threshold, the thresholds for isoluminant targets lie considerably above behavioral contrast threshold. Therefore, although some MT neurons do give responses to isoluminant targets, they are unlikely to be the source of the chromatic motion signals revealed behaviorally.

The normal postnatal development, the influence of age, and the effects of visual deprivation on the dopamine system in the retina of rhesus monkeys were examined. The lowest level of retinal dopamine was found at birth. By 3–4 weeks of age, the dopamine concentration had more than doubled. This level remained relatively constant in the retinas of older infants and of adult monkeys up to 34 yr of age. The level of the dopamine metabolite 3,4-dihydroxyphenylacetic acid (DOPAC) and the activity of tyrosine hydroxylase did not significantly change as a function of age during the postnatal life span.

Monocular occlusion of newborn or infant monkeys for 1–15 months with opaque contact lenses resulted in decreases in the retinal concentrations of dopamine and DOPAC relative to the concentrations in the same animals' unoccluded eyes. Occlusion also resulted in a lower level of tyrosine hydroxylase activity in the retina. Monocular eyelid suture from birth to 15 months of age resulted in less consistent alterations of retinal dopamine and DOPAC levels; tyrosine hydroxylase activity, however, was consistently reduced by lid suture. Thus, dopamine synthesis and metabolism, and the ontogenetic increase of the retinal dopamine level of rhesus monkey are reduced by light deprivation.

Behavioral and electrophysiological studies of insects demonstrate both spectrally independent and chromatically dependent behaviors and interneurons. This account describes the neuroanatomical identification of two parallel retinotopic subsystems, one supplying descending channels to spectrally independent neck and flight motor circuits, the other supplying polychromatic channels to neuropils associated with leg motor circuits in the thoracic ganglia. In the compound eye, two classes of photoreceptors contribute to each of several thousand sampling units. High-sensitivity, chromatically uniform short-axon photoreceptors (R1-R6) supply the lamina's external plexiform layer and are presynaptic to L1, L2 efferents. These project in parallel with a second system of trichromatic long-axon receptors and the L3 efferent. Both pathways supply columns of the medulla, equal in number to ommatidia. Golgi and cobalt-silver impregnation demonstrates that neurons from the medulla diverge to two deeper regions, the lobula plate and lobula, the former a thin tectum of neuropil dorsal to the more substantial lobula. Layer relationships between medulla neurons and their afferent supply suggest that the lobula plate and lobula are each supplied by one or the other, but not both, of the two parallel subsystems. Independence of the two parallel pathways is suggested by ablation of the photoreceptor layer leading to selective degeneration of the motion-sensitive lobula plate neuropil. In addition, octets of small-field neurons associated with the R1-R6/L1, L2 pathway give rise to synaptic complexes with motion-sensitive neurons of the lobula plate. A variety of behavioral and electrophysiological studies provide supporting evidence that certain insects possess parallel visual pathways comparable to the magnocellular and parvocellular subsystems of primates.