Recent physiological studies coupled with intracellular staining have subdivided ON directionally selective (DS) ganglion cells of rabbit retina into two types. One exhibits more “transient” and more “brisk” responses (ON DS-t), and the other has more “sustained’ and more “sluggish” responses (ON DS-s), although both represent the same three preferred directions and show preference for low stimulus velocity, as reported in previous studies of ON DS ganglion cells in rabbit retina. ON DS-s cells have the morphology of ganglion cells previously shown to project to the medial terminal nucleus (MTN) of the accessory optic system, and the MTN-projecting, class IVus1 cells have been well-characterized previously in terms of their dendritic morphology, branching pattern, and stratification. ON DS-t ganglion cells have a distinctly different morphology and exhibit heterotypic coupling to amacrine cells, including axon-bearing amacrine cells, with accompanying synchronous firing, while ON DS-s cells are not coupled. The present study shows that ON DS-t cells are morphologically identical to the previously well-characterized, “orphan” class IIb1 ganglion cell, previously regarded as a member of the “brisk-concentric” category of ganglion cells. Its branching pattern, quantitatively analyzed, is similar to that of the morphological counterparts of X and Y cells, and very different from that of the ON DS-s ganglion cell. Close analysis of the dendritic stratification of class IIb1 ganglion cells together with fiducial cells indicates that they differ from that of the ON DS-s cells. In agreement with one of the three previous studies, class IIb1/ON DS-t cells, unlike class IVus1/ON DS-s ganglion cells, in the main do not co-stratify with starburst amacrine cells. As the present study shows, however, portions of their dendrites do deviate from the main substratum, coming within range of starburst boutons. Parsimony favors DS input from starburst amacrine cells both to ON DS-s and to ON DS-t ganglion cells, given the similarity of their DS responses, but further studies will be required to substantiate the origin of the DS responses of ON DS-t cells. Previously reported OFF DS responses in ON DS-t cells, unmasked by pharmacological agents, and mediated by gap junctions with amacrine cells, suggests an unusual trans-sublaminar organization of directional selectivity in the inner plexiform layer, connecting sublamina a and sublamina b.

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.

I explore here whether linear mechanisms can explain directional selectivity (DS) in simple cells of the cat's striate cortex, a question suggested by a recent upswing in popularity of linear DS models. I chose a simple cell with a space-time inseparable receptive field (RF), i.e. one that shows gradually shifting latency across space, as the RF type most likely to depend on linear mechanisms of DS. However, measured responses of the cell to a moving bar were less modulated, and extended over a larger spatial region than predicted by two different popular “linear” models. They also were more DS in exhibiting a higher ratio of total spikes for the preferred direction. Each of the two models used for comparison has a single “branch” with a single spatiotemporally inseparable linear filter followed by a threshold, hence, a “1-branch” model. Nonlinear interactions between pairs of bars in a 2-bar linear superposition test of the cell also disagreed in time-course with those of the 1-branch models. The only model whose 1-bar and 2-bar predictions matched the measured cell (including a complete “4-branch” motion energy model that matches complex cells) has two branches that differ in phase by about 90 deg, i.e. in quadrature. Each branch has its own threshold that helps define the preceding spatiotemporal unit as a subunit even after the outputs of the two branches are summed. As subunit phases differ by only 90 deg, flashing bar responses of the 2-subunit model are similar to those of the 1-subunit model. Therefore, the number of subunits is hidden from view when testing with a conventional stationary bar. In summary, movement responses and nonlinear interactions between pairs of bars in the measured cell matched those of the 2-subunit model, while they disagreed with the popular 1-subunit model. Thus, multiple nonlinear subunits appear to be necessary for DS, even in simple cortical cells.

In the rabbit retina, the nuclear dye, 4,6, diarnidino-2-phenylindole (DAPI), selectively labels a third type of amacrine cell, in addition to the previously characterized type a and type b cholinergic amacrine cells. In this study, these “DAPI-3” amacrine cells have been characterized with respect to their somatic distribution, dendritic morphology, and neurotransmitter content by combining intracellular injection of biotinylated tracers with wholemount immunocytochemistry. There are about 100,000 DAPI-3 amacrine cells in total, accounting for 2% of all amacrine cells in the rabbit retina, and their cell density ranges from about 130 cells/mm2 in far-peripheral retina to 770 cells/mm2 in the visual streak. The thin varicose dendrites of the DAPI-3 amacrine cells form a convoluted dendritic tree that is symmetrically bistratified in S1/S2 and S4 of the inner plexiform layer. Tracer coupling shows that the DAPI-3 amacrine cells have a fivefold dendritic-field overlap in each sublamina, with the gaps in the arborization of each cell being occupied by dendrites from neighboring cells. The DAPI-3 amacrine cells consistently show the strongest glycine immunoreactivity in the rabbit retina and they also accumulate exogenous [3H]-glycine to a high level. By contrast, the All amacrine cells, which are the best characterized glycinergic cells in the retina, are amongst the most weakly labelled of the glycine-immunopositive amacrine cells. The DAPI-3 amacrine cells costratify narrowly with the cholinergic amacrine cells and the On-Off direction-selective ganglion cells, suggesting that they may play an important role in movement detection.

We have compared the responses of simple cells to laterally moving sinusoidal gratings and to stationary temporally-modulated gratings. From the amplitudes and temporal phases of the responses to stationary gratings of different spatial phases, it should be possible to predict the preferred direction of movement, the amplitudes of the responses to gratings moving in the preferred and nonpreferred directions and, thence, the degree of directional preference (Reid et al., 1987). The preferred direction can be predicted reliably. However, the magnitude of the directional preference cannot be predicted, since the measured amplitude of the response in the nonpreferred direction of movement is very much less than that predicted by a linear theory. Nonlinearities in the relationship between response amplitude and contrast may contribute to the failure of the predictions, but this contribution is small. We conclude that the magnitude of the directional preference seems to be determined predominantly by nonlinear suppression of the response in the nonpreferred direction of movement.

GABA is a major inhibitory neurotransmitter in the mammalian retina and it acts at many different sites via a variety of postsynaptic receptors. These include GABAA receptors and bicuculline-resistant GABAC receptors. The release of acetylcholine (ACh) is inhibited by GABA and strongly potentiated by GABA antagonists. In addition, GABA appears to mediate the null inhibition which is responsible for the mechanism of directional selectivity in certain ganglion cells. We have used these two well-known examples of GABA inhibition to compare three GABA antagonists and assess the contributions of GABAA and GABAC receptors. All three GABA antagonists stimulated ACh release by as much as ten-fold. By this measure, the ED50s for SR-95531, bicuculline, and picrotoxin were 0.8, 7.0, and 14 μM, respectively. Muscimol, a potent GABAA agonist, blocked the effects of SR-95531 and bicuculline, but not picrotoxin. This indicates that SR-95531 and bicuculline are competitive antagonists at the GABAA receptor, while picrotoxin blocks GABAA responses by acting at a different, nonreceptor site such as the chloride channel. In the presence of a saturating dose of SR-95531 to completely block GABAA receptors, picrotoxin caused a further increase in the release of ACh. This indicates that picrotoxin potentiates ACh release by a mechanism in addition to the block of GABAA responses, possibly by also blocking GABAC receptors, which have been associated with bipolar cells. All three GABA antagonists abolished directionally selective responses from ON/OFF directional-selective (DS) ganglion cells. In this system, the ED50s for SR-95531, bicuculline, and picrotoxin were 0.7 μM, 8 μM, and 94.6 μM, respectively. The results with SR-95531 and bicuculline indicate that GABAA receptors mediate the inhibition responsible for directional selectivity. The addition of picrotoxin to a high dose of SR-95531 caused no further increase in firing rate. The comparatively high dose required for picrotoxin also suggests that GABAC receptors do not contribute to directional selectivity. This in turn suggests that feedforward GABAA inhibition, as opposed to feedback at bipolar terminals, is responsible for the null inhibition underlying directional selectivity.

Here we examine further the basis for directional selectivity (DS) in simple cells of the cat's striate cortex. We use a distinctly different input stimulus and different analysis of the output signal for the same type of space-time inseparable receptive field (RF) that was measured with flashing bars in the companion paper (Emerson, 1997). As in the companion paper, we have mimicked a popular “linear” model with a single “branch” that consists of a linear spatiotemporal filter followed by a soft threshold, which should mimic any simple cells that have a single subunit. Counterphase sinusoidal measurements of such a configuration always generate elliptically shaped polar plots of amplitude versus temporal phase, often pinched along the minor axis because of a high threshold. However, for many spatiotemporal frequencies, such polar phase plots, as measured in simple cells by others, show a consistent rotational phase skew. Here we apply counterphase analysis to the same 2-branch model as developed in the companion paper. We show that the model accounts for the skew as the summation of signals from linear filters separated spatially and temporally by approximately 90 deg (i.e. in spatiotemporal phase quadrature), each separated from the output stage by a soft-threshold nonlinearity. We also prove conclusively that such skew cannot be generated by a single-subunit configuration. This demonstration supports the proposed two-subunit structure for DS simple cells, such as in the example from the companion paper, which has strong linear contributions from its inseparable RF. The presence of at least two nonlinear subunits appears to be an obligatory concomitant of DS in all visual cortical cells. The primary function of these subunits may be to enhance the strength of responses to images moving in the preferred direction, as in complex cells. However, subunits may also aid in identifying the moving object through overcoming, at least partially, the phase-concealing properties of the neuron's threshold by generating a steady signal that effectively decreases the threshold for the preferred direction.

In almost every study of the linearity of spatiotemporal summation in simple cells of the cat's visual cortex, there have been systematic mismatches between the experimental observations and the predictions of the linear theory. These mismatches have generally been explained by supposing that the initial spatiotemporal summation stage is strictly linear, but that the following output stage of the simple cell is subject to some contrast-dependent nonlinearity. Two main models of the output nonlinearity have been proposed: the threshold model (e.g. Tolhurst & Dean, 1987) and the contrast-normalization model (e.g. Heeger, 1992a, b). In this paper, the two models are fitted rigorously to a variety of previously published neurophysiological data, in order to determine whether one model is a better explanation of the data. We reexamine data on the interaction between two bar stimuli presented in different parts of the receptive field; on the relationship between the receptive-field map and the inverse Fourier transform of the spatial-frequency tuning curve; on the dependence of response amplitude and phase on the spatial phase of stationary gratings; on the relationships between the responses to moving and modulated gratings; and on the suppressive action of gratings moving in a neuron's nonpreferred direction. In many situations, the predictions of the two models are similar, but the contrast-normalization model usually fits the data slightly better than the threshold model, and it is easier to apply the equations of the normalization model. More importantly, the normalization model is naturally able to account very well for the details and subtlety of the results in experiments where the total contrast energy of the stimuli changes; some of these phenomena are completely beyond the scope of the threshold model. Rigorous application of the models' equations has revealed some situations where neither model fits quite well enough, and we must suppose, therefore, that there are some subtle nonlinearities still to be characterized.

Biological visual systems can detect positional changes that are finer than these systems' acuity to sine-wave gratings, a property known as hyperacuity. Some systems can even detect changes finer that the photoreceptor spacing. We report here that rabbit's directionally selective ganglion cells not only detect positional changes in the hyperacuity range, but also discriminate the direction of their motion. Our experiments show that directional selectivity occurs for edges of light moving as little as 1.1 μm (26” of visual angle) across the retina. This distance corresponds to a hyperacuity, since the acuity to sine-wave gratings of rabbit's On-Off DS ganglion cells is about 125 μm (50′). In addition, this distance is smaller than the minimal spacing between rabbit photoreceptors (1.9 μm or 46”), as estimated from cell-density studies (Young & Vaney, 1991). Such a hyperacuity suggests low-noise high-gain signal transmission from photoreceptors to ganglion cells and that directional selectivity can arise in small portions of retinal dendritic processes.

The excitatory receptive-field centers of On-Off directionally selective (DS) ganglioncells of the rabbit retina correspond closely to the lateral extent of their dendritic arborizations. Some investigators have hypothesized from this that theories for directionalselectivity that entail a lateral spread of excitation from outside the ganglion cell dendritic tree, such as from starburst amacrine cells, are therefore untenable. We show herethat significant motion facilitation is conducted from well outside the classical excitatory receptive-field center (and, therefore, dendritic arborization) of On-Off DS ganglioncells for preferred-direction, but not null-direction moving stimuli. These results are consistent with a role in directional selectivity for cells with processes lying beyond the On-Off ganglion cell's excitatory receptive-field center. These results also highlight the fundamental distinction in retinal ganglion cell receptive-field organization between classical excitatory mechanisms and those that facilitate other excitation without producing directly observable excitation by themselves.

We examined contrast, direction of motion, and concentration dependencies of the effects of GABAergic and cholinergic antagonists, and anticholinesterases on responses to movement of On—Off directionally selective (DS) ganglion cells of the rabbit's retina. The drugs tested were curare and hexamethonium bromide (cholinergic antagonists), physostigmine (anticholinesterase), and picrotoxin (GABAergic antagonist). They all reduced the cells' directional selectivity, while maintaining their preferred-null axis. However, cholinergic antagonists did not block directional selectivity completely even at saturating concentrations. The failure to eliminate directional selectivity was probably not due to an incomplete blockade of cholinergic receptors. In a extension of a Masland and Ames (1976) experiment, saturating concentrations of antagonists blocked the effects of exogenous acetylcholine or nicotine applied during synaptic blockade. Consequently, a noncholinergic pathway may be sufficient to account for at least some directional selectivity. This putative pathway interacts with the cholinergic pathway before spike generation, since physostigmine eliminated directional selectivity at contrasts lower than those saturating responses. This elimination apparently resulted from cholinergic-induced saturation, since reduction of contrast restored directional selectivity. Under picrotoxin, directional selectivity was lost in 33% of the cells regardless of contrast. However, 47% maintained their preferred direction despite saturating concentrations of picrotoxin, and 20% reversed the preferred and null directions. Therefore, models based solely on a GABAergic implementation of Barlow and Levick's asymmetric-inhibition model or solely on a cholinergic implementation of asymmetric-excitation models are not complete models of directional selectivity in the rabbit. We propose an alternate model for this retinal property.

We examined the effects of picrotoxin and pentylenetetrazol (PTZ) on the responses to motions of ON-OFF directionally selective (DS) ganglion cells of the turtle's retina. These drugs are antagonists of the inhibitory neurotransmitter GABA. For continuous motions, picrotoxin markedly reduced the overall directionality of the cells. In 21% of the cells, directional selectivity was lost regardless of speed and contrast. However, other cells maintained their preferred direction despite saturating concentrations of picrotoxin. And in most cells, loss, maintenance, or even reversal of preferred and null directions could occur as speed and contrast were modulated. In 50% of the cells, reversal of preferred and null directions occurred at some condition of visual stimuli. However, picrotoxin did not tend to alter the preferred-null axis for directional selectivity. For apparent motions, picrotoxin made motion facilitation, which normally occurs exclusively in preferred-direction responses, to become erratic and often occur during null-direction motions. Finally, PTZ had effects similar to picrotoxin but with less potency. The results in this paper indicated that models of directional selectivity based solely on a GABAergic implementation of Barlow and Levick's asymmetric-inhibition model do not apply to the turtle retina. Alternative models may comprise multiple directional mechanisms and/or a symmetric inhibitory one, but not asymmetric facilitation.

Previous tests of the linearity of spatiotemporal summation in cat simple cells have compared the responses to moving sinusoidal gratings and to gratings whose contrast was modulated sinusoidally in time. In particular, since a moving grating can be expressed as a sum of modulated gratings, the response to a moving grating should be predictable (assuming linearity) from the responses to modulated gratings. However, these simple linear predictions have shown varying degrees of failure (e.g. Reid et al., 1987, 1991), depending on the directional selectivity of the neurons (Tolhurst & Dean, 1991). We demonstrate here that the failures of these linear predictions are, in fact, explained by the contrast-normalization model of Heeger (1993). We concentrate on the ratio of the measured to predicted moving grating responses. In the context of the contrast-normalization model, calculating this ratio turns out to be particularly appropriate, since the ratio is independent of the precise details of the linear front-end mechanisms ultimately responsible for directional selectivity. Hence, the contrast-normalization model can be compared quantitatively with this ratio measure, by varying only one free parameter. When account is taken both of the expansive output nonlinearity and of contrast normalization, the directional selectivity of simple cells seems to be dependent only on linear spatiotemporal filtering.

Bistratified (BS) ganglion cells have long been recognized in vertebrate retina. Thirty years ago, it became clear that bistratification allows the integration of ON and OFF retinal pathways to produce contrast-independent responses in ganglion cells. Best studied is the type 1 bistratified (BS1) ganglion cell of rabbit retina, the physiologically well-characterized ON-OFF directionally selective (DS) ganglion cell, which is co-stratified with the two types of starburst amacrine (SA) cells in sublaminae a and b of the inner plexiform layer (IPL). DS responses have recently been documented in the latter. In this report, BS1 cells are further studied and are used as “fiducials” to characterize a second type of BS ganglion cell. An example of a possible third type is shown to be distinct from examples of BS1 and BS2 cells. All three have two distinct, narrowly stratified arborizations, one in sublamina a and one in sublamina b. All have similar dimensions, except for their dendritic trees, differing also in branching pattern. BS1 cells have compact, regular, highly branched trees; BS2 cells have significantly larger, more sparsely branched, irregular, radiate trees; the proposed BS3 type is intermediate in field size, and its branching pattern is different from the first two. BS2 and BS3 cells are co-stratified, branching nearer to the margins of the IPL, out of range of SA cells. In a previous report by others, illustrating the morphology of intracellularly stained ganglion cells, one example each of both “orientation-selective” ganglion cells and “uniformity detectors” resembles the BS2 cell. A rationale is presented for correlating BS2 cells with uniformity detectors.

The spatial and temporal interactions in the receptive fields of On-Off directionally selective (DS) ganglion cells endow them with directional selectivity. Using a variety of stimuli, such as sinusoidal gratings, we show that these interactions make directional selectivity of the DS ganglion cell robust with respect to stimulus parameters such as contrast, speed, spatial frequency, and extent of motion. Moreover, unlike the directional selectivity of striate-cortex cells, On-Off DS ganglion cells display directional selectivity to motions not oriented perpendicularly to the contour of the objects. We argue that these cells may achieve such high robustness by combining multiple mechanisms of directional selectivity.

Acetylcholine (ACh) enhances the preferred direction responses of directionally selective ganglion cells (DS GCs; Ariel & Daw, 1982; Ariel & Adolph, 1985) through the activation of nicotinic acetylcholine receptors (nAChRs; Ariel & Daw, 1982; Massey et al., 1997; Kittila & Massey, 1997). DS GCs appear to express at least two types of nAChRs, those that are sensitive to the partially subtype-specific antagonist methyllycaconitine (MLA), and those that are MLA-insensitive (Reed et al., 2002). Our purpose was to confirm the expression of α7 nAChRs by DS GCs and to assess the contributions of other nAChR subtypes to DS GC responses. Using choline as a nAChR partially subtype-specific agonist, we found that the majority of DS GCs demonstrated responses to choline while under synaptic blockade. The blockade or reduction of choline-induced responses by bath application of nanomolar (nM) concentrations of MLA provided direct evidence that the choline responses were mediated by α7 nAChRs. Because choline is a partial agonist for α3β4 nAChRs (Alkondon et al., 1997), the residual choline responses are consistent with mediation by α3β4 nAChRs. Additionally, a subset of DS GCs responded to nicotine but not to choline, indicating the expression of a third nAChR subtype. The pharmacological results were supported by single cell reverse transcription polymerase chain reaction (RT-PCR) and immunohistochemistry experiments. The expression of α7 and specific non-α7 nAChR subtypes was correlated with the preferred direction. This indicates the possibility of differential responses to ACh depending on the direction of movement. This is the first description of differential expression of multiple nAChR subtypes by DS GCs.

Synchronous spiking has been postulated to be a meta-signal in visual cortex and other CNS loci that tags neuronal spike responses to a single entity. In retina, however, synchronized spikes have been postulated to arise via mechanisms that would largely preclude their carrying such a code. One such mechanism is gap junction coupling, in which synchronous spikes would be a by-product of lateral signal sharing. Synchronous spikes have also been postulated to arise from common-source inputs to retinal ganglion cells having overlapping receptive fields, and thus code for stimulus location in the overlap area. On–Off directionally selective ganglion cells of the rabbit retina exhibit a highly precise tiling pattern in which gap junction coupling occurs between some neighboring, same-preferred-direction cells. Depending on how correlated spikes arise, and for what purpose, one could postulate that synchronized spikes in this system (1) always arise in some subset of same-direction cells because of gap junctions, but never in non-same-preferred-directional cells; (2) never arise in same-directional cells because their receptive fields do not overlap, but arise only in different-directional cells whose receptive fields overlap, as a code for location in the overlap region; or (3) arise in a stimulus-dependent manner for both same- and different-preferred-direction cells for a function similar to that postulated for neurons in visual cortex. Simultaneous, extracellular recordings were obtained from neighboring On–Off directionally selective (DS) ganglion cells having the same and different preferred directions in an isolated rabbit retinal preparation. Stimulation by large flashing spots elicited responses from DS ganglion-cell pairs that typically showed little synchronous firing. Movement of extended bars, however, often produced synchronous spikes in cells having similar or orthogonal preferred directions. Surprisingly, correlated firing could occur for the opposite contrast polarity edges of moving stimuli when the leading edge of a sweeping bar excited the receptive field of one cell as its trailing edge stimulated another. Pharmacological manipulations showed that the spike synchronization is enhanced by excitatory cholinergic amacrine-cell inputs, and reduced by inhibitory GABAergic inputs, in a motion-specific manner. One possible interpretation is that this synchronous firing could be a signal to higher centers that the outputs of the two DS ganglion cells should be “bound” together as responding to a contour of a common object.

The correlation between cholinergic sensitivity and the level of stratification for ganglion cells was examined in the rabbit retina. As examples, we have used ON or OFF α ganglion cells and ON/OFF directionally selective (DS) ganglion cells. Nicotine, a cholinergic agonist, depolarized ON/OFF DS ganglion cells and greatly enhanced their firing rates but it had modest excitatory effects on ON or OFF α ganglion cells. As previously reported, we conclude that DS ganglion cells are the most sensitive to cholinergic drugs. Confocal imaging showed that ON/OFF DS ganglion cells ramify precisely at the level of the cholinergic amacrine cell dendrites, and co-fasciculate with the cholinergic matrix of starburst amacrine cells. However, neither ON or OFF α ganglion cells have more than a chance association with the cholinergic matrix. Z-axis reconstruction showed that OFF α ganglion cells stratify just below the cholinergic band in sublamina a while ON α ganglion cells stratify just below cholinergic b. The latter is at the same level as the terminals of calbindin bipolar cells. Thus, the calbindin bipolar cell appears to be a prime candidate to provide the bipolar cell input to ON α ganglion cells in the rabbit retina. We conclude that the precise level of stratification is correlated with the strength of cholinergic input. Alpha ganglion cells receive a weak cholinergic input and they are narrowly stratified just below the cholinergic bands.

Based on comparative anatomical studies and electrophysiological experiments, we have identified a conserved subset of neurons in the lamina, medulla, and lobula of dipterous insects that are involved in retinotopic visual motion direction selectivity. Working from the photoreceptors inward, this neuronal subset includes lamina amacrine (α) cells, lamina monopolar (L2) cells, the basket T-cell (T1 or β), the transmedullary cell Tm1, and the T5 bushy T-cell. Two GABA-immunoreactive neurons, the transmedullary cell Tm9 and a local interneuron at the level of T5 dendrites, are also implicated in the motion computation. We suggest that these neurons comprise the small-field elementary motion detector circuits the outputs of which are integrated by wide-field lobula plate tangential cells. We show that a computational model based on the available data about these neurons is consistent with existing models of biological elementary motion detection, and present a comparable version of the Hassenstein-Reichardt (HR) correlation model. Further, by using the model to synthesize a generic tangential cell, we show that it can account for the responses of lobula plate tangential cells to a wide range of transient stimuli, including responses which cannot be predicted using the HR model. This computational model of elementary motion detection is the first which derives specifically from the functional organization of a subset of retinotopic neurons supplying the lobula plate. A key prediction of this model is that elementary motion detector circuits respond quite differently to small-field transient stimulation than do spatially integrated motion processing neurons as observed in the lobula plate. In addition, this model suggests that the retinotopic motion information provided to wide-field motion-sensitive cells in the lobula is derived from a less refined stage of processing than motion inputs to the lobula plate.

Directionally selective (DS) ganglion cells of rabbit retina are of two principal types. ON DS ganglion cells prefer low velocity in one of three directions of movement and project axons to the accessory optic system (AOS), whereas ON–OFF DS ganglion cells prefer higher velocity in one of four directions and project to tectum and thalamus. Each has a distinct, recognizable dendritic morphology, based upon the correlation of form, physiology, and central projections. In previous Golgi studies, ON and ON–OFF DS cells were found to be partly co-stratified, and ON–OFF DS cells were found to co-stratify with starburst amacrine (SA) cells, the cholinergic amacrine cells of the retina, which also contain elevated levels of GABA. SA cells are radially symmetrical, have synaptic boutons in a distal annular zone of its dendritic tree, are presynaptic primarily to ganglion cell dendrites, co-stratify with ON–OFF DS ganglion cells, and contain the neurotransmitters shown pharmacologically to be involved in DS responses. For these reasons, SA cells are thought to play a role in the DS mechanism. Several models of this mechanism have utilized SA cell dendritic geometry in a centrifugal, radial format to impose directional inputs on DS ganglion cells.