An important mechanism for the discrimination of direction of motion in the retina is a spatially asymmetric inhibition. This inhibition has been postulated to operate either as a subtraction, like in difference-of-Gaussians' models, or as a division, like in shunting-inhibition models of directional selectivity. The latter, but not the former, is nonlinear. This raises the question of whether the inhibitory mechanism involved in directional selectivity is nonlinear. To investigate this issue, we studied the linearity of the contrast dependence of the extracellularly recorded responses to apparent motions in ON-OFF directionally selective ganglion cells of the rabbit retina. The results show that the inhibition underlying directional selectivity is nonlinear and fits shunting-inhibition models well. Other biophysical mechanisms that might account for the type of nonlinearity observed in the data are also considered.

The vertebrate retina processes visual information in parallel neural pathways known as the ON and OFF pathways. These pathways encode increments and decrements of light independently as excitatory responses. We examined the photopic spectral response of ON and OFF mechanisms in goldfish by measuring the sensitivity of optic nerve responses to the onset and termination of stimuli of various wavelengths. Using various adapting backgrounds, we found that the ON and OFF responses have different spectral sensitivities. The weighting of the cone inputs to the responses was estimated by an algebraic summation model. This model suggests that for the ON response, input from S-cones is stronger and more independent than for the OFF response, and M- and L-cones show stronger antagonism in the ON response than in the OFF response. The OFF response probably receives input from all cone types, but spectral antagonism is weak and its dominant input is from L-cones.

We propose a network model that accounts for the kinetic depth in structure from motion phenomena. Using plausible neural mechanisms, the model accounts for (1) fluctuations in perception when viewing a simple kinetic depth stimulus, (2) disambiguation of this stimulus with stereoscopic information, and (3) subsequent bias of the percept of this stimulus following stereoscopic adaptation. The model comprises two levels: a layer of monocular directionally selective motion detectors that provide input to a second layer of disparity- selective and direction-selective binocular mechanisms. The network of facilitatory and inhibitory connections between binocular mechanisms gives rise to fluctuations in network activity that mimic the fluctuations in perception of kinetic depth in the absence of disparity information. The results of a psychophysical experiment are consistent with the nature of the proposed interactions.

Gamma-aminobutyric acid (GABA) is an inhibitory neurotransmitter used by many neurons of the mammalian retina. To identify the synaptic targets of these cells, we undertook an immunohistochemical study with a monoclonal antibody that recognizes the GABAA receptors (62−3G1, generously donated by A. de Blas). This antibody labels the somata of at least one group of amacrine cells in the inner nuclear layer. It also labels two groups of somata in the ganglion cell layer: one small and the other much larger. The small cells are likely to be displaced amacrine cells based on their size, although some could be gamma ganglion cells. The much larger receptor-positive cells are clearly ganglion cells, based both on their size and the antibody labeling of the initial portion of their axon. In the peripheral retina, the size of these large somata suggests that many are beta ganglion cells. However, at any point across the retina the density of these cells never exceeded 50% of the density of beta cells as a whole.

The antibody also labels a dense plexus of processes that extends throughout the inner plexiform layer (IPL), with a marked concentration in the inner third of the layer. This is the portion of the IPL in which the rod bipolar cells terminate. It is difficult to recognize processes of individual cells in the IPL, so retinae were dissociated. The rod bipolar cells were identified by protein kinase C immunoreactivity (Negishi et al., 1988; Karschin & Wässle, 1990). They were not labeled by the GABAA receptor antibody. This is surprising in light of tight-seal, whole cell voltage-clamp recordings that have shown that the rod bipolars express functional GABAA receptors. One possible explanation is that the antibody recognizes only a subset of the GABAA receptors.

The response amplitude of cat striate cortical cells is usually reduced after exposure to high-contrast stimuli. The temporal characteristics and contrast sensitivity of this phenomenon were explored by stimulating cortical cells with drifting gratings in which contrast sequentially incremented and decremented in stepwise fashion over time. All responses showed a clear hysteresis, in which contrast gain dropped on average 0.36 log unit and then returned to baseline values within 60 s. Noticeable gain adjustments were seen in as little as 3 s and with peak contrasts as low as 3%. Contrast adaptation was absent in responses from LGN cells. Adaptation was found to depend on temporal frequency of stimulation, with greater and more rapid adaptation at higher temporal frequencies. Two different tests showed that the mechanism controlling response reduction was influenced primarily by stimulus contrast rather than response amplitude. These results support the existence of a rapid and sensitive cortically based system that normalizes the output of cortical cells as a function of local mean contrast. Control of the adaptation appears to arise at least in part across a population of cells, which is consistent with the idea that the gain control serves to limit the information converging from many cells onto subsequent processing areas.

The ultrastructural features and synaptic contacts of two amacrine cell types with corticotropin-releasing factor-like immunoreactivity in the turtle retina were examined using electron immunocytochemistry. Type A cells were found only in the visual streak and had elongated dendritic arborizations that ran parallel to the visual streak. These cells arborized primarily in stratum 1 and near the border of strata 2 and 3 of the inner plexiform layer, with some processes extending into stratum 5. Type B cells were found only ventral to the visual streak and arborized primarily in a wide band in strata 4 and 5, with sparse dendritic arborizations in stratum 1.

There was a diffuse cytoplasmic reaction product within each cell type; however, large labeled vesicles were rarely observed. Type A amacrine cells received many conventional synaptic contacts from amacrine cells in stratum 1 and at the border of strata 2 and 3, but only a small number of contacts in stratum 5. Bipolar synaptic contacts onto type A amacrine cells were observed in strata 1 and at the border of strata 2 and 3. The only positively identified synaptic outputs of type A cells were conventional synapses onto amacrine cells in strata 1 and at the border of 2 and 3. Type B amacrine cells received synaptic contacts from amacrine cells in strata 1 and 5, and bipolar cell synaptic input in stratum 5. They made conventional synapses onto amacrine cells in strata 1 and 5, and onto bipolar cells in stratum 5. We also found conventional synaptic contacts between unlabeled amacrine cells and type B amacrine cells outside of the primary layers of stratification. In addition, there were specialized junctions observed between type A cell profiles in stratum 1 and between type B cell profiles in stratum 5. The unique regional distributions of the type A and B cells, as well as their differences in synaptic connectivity, suggested that these amacrine cells play distinct physiological roles although they contain the same neuropeptide.

Unilateral partial ablation of the superior colliculus in the hamster results in a compression of the retinotopic map onto the remaining tectal fragment. In a previous electrophysiological study (Pallas & Finlay, 1989a), we demonstrated that receptive-field properties of single tectal units (including receptive-field size) remain unchanged, despite the increased afferent/target convergence ratios in the compressed tecta. The present study was done to investigate the mechanism that produces increased convergence from retina to tectum at the population level while maintaining apparent stability of convergence at the single neuron level. We injected comparable quantities of horseradish peroxidase into the tecta of normal adult hamsters and adult hamsters that had received neonatal partial tectal ablations of varying magnitude. We then compared the area of retina backfilled from the injection and the number and density of labeled retinal ganglion cells within it to the size of the remaining tectal fragment.

As expected from earlier anatomical (Jhaveri & Schneider, 1974) and physiological (Finlay et al., 1979a; Pallas & Finlay, 1989a) studies demonstrating compression of the retinotectal projection, we found that the area of retina labeled from a single tectal injection site increases linearly with decreasing tectal fragment size. However, for fragment sizes down to 30% of normal, total number of retinal ganglion cells projecting to the injection site remains in or above the normal range. For large lesions (less than 30% of tectum remaining), total number of labeled retinal ganglion cells declines from normal, despite the fact that a larger absolute area of retina is represented on each unit of tectum under these conditions. Comparison of retinal ganglion cell density with tectal fragment size shows an initial decline with decreasing fragment size, which becomes sharper with very large lesions (small tectal fragments).

The maintenance of the normal number of retinal ganglion cells innervating each patch of tectum could be accomplished by an elimination of the tectal collaterals of some retinal ganglion cells. Our results suggest that, in addition to collateral elimination, reduction in the size of ganglion cell arbors is occurring, since the peak density of backfilled ganglion cells declines less rapidly than backfilled retinal area increases, especially for small lesions. However, arbor reduction and collateral elimination must occur in such a way that individual tectal cells represent the same amount of visual space as normal.

Thus, collateral elimination and arbor reduction are two mechanisms that operate to maintain afferent/target convergence ratios (and thus receptive-field properties) over large variations in afferent availability. This compensation may occur through an activity-dependent stabilization mechanism that does not change its selectivity even when excess afferents are available. For very large lesion sizes, receptive-field size and innervating ganglion cell number and density are not preserved, thus demonstrating a limit to the afferent/target matching mechanism. The same ontogenetic mechanisms might provide a buffer for normal variations in afferent populations, and could help to align topographic maps with differing numbers of afferents.

Reversible inactivation by cooling of the transcallosal projecting neurons in areas 17 and 18 of one hemisphere bring about complex changes in the spontaneous and evoked activity of neurons in the callosal receiving zone of the opposite hemisphere. These changes include increase and decreases in evoked and spotaneous activities. Overall, 90% of neurons in alyers II and III, 50% in layer IV, and 100% in layers V and VI were affected by the block of transcallosal input. The complexity of the changes was greatest in layers II and III, which are the major callosal recipient layers. The results indicate that many excitatory and inhibitory circuits are under the direct control of transcallosal fibers in the normally fuctioning brain.