Subcellular pathways through VGluT3-expressing mouse amacrine cells provide locally tuned object-motion-selective signals in the retina.

VGluT3-expressing mouse retinal amacrine cells (VG3s) respond to small-object motion and connect to multiple types of bipolar cells (inputs) and retinal ganglion cells (RGCs, outputs). Because these input and output connections are intermixed on the same dendrites, making sense of VG3 circuitry requires comparing the distribution of synapses across their arbors to the subcellular flow of signals. Here, we combine subcellular calcium imaging and electron microscopic connectomic reconstruction to analyze how VG3s integrate and transmit visual information. VG3s receive inputs from all nearby bipolar cell types but exhibit a strong preference for the fast type 3a bipolar cells. By comparing input distributions to VG3 dendrite responses, we show that VG3 dendrites have a short functional length constant that likely depends on inhibitory shunting. This model predicts that RGCs that extend dendrites into the middle layers of the inner plexiform encounter VG3 dendrites whose responses vary according to the local bipolar cell response type.

Avoidance of axonal stimulation with sinusoidal epiretinal stimulation.

Objective.Neuromodulation, particularly electrical stimulation, necessitates high spatial resolution to achieve artificial vision with high acuity. In epiretinal implants, this is hindered by the undesired activation of distal axons. Here, we investigate focal and axonal activation of retinal ganglion cells (RGCs) in epiretinal configuration for different sinusoidal stimulation frequencies.Approach.RGC responses to epiretinal sinusoidal stimulation at frequencies between 40 and 100 Hz were tested inex-vivophotoreceptor degenerated (rd10) isolated retinae. Experiments were conducted using a high-density CMOS-based microelectrode array, which allows to localize RGC cell bodies and axons at high spatial resolution.Main results.We report current and charge density thresholds for focal and distal axon activation at stimulation frequencies of 40, 60, 80, and 100 Hz for an electrode size with an effective area of 0.01 mm2. Activation of distal axons is avoided up to a stimulation amplitude of 0.23µA (corresponding to 17.3µC cm-2) at 40 Hz and up to a stimulation amplitude of 0.28µA (14.8µC cm-2) at 60 Hz. The threshold ratio between focal and axonal activation increases from 1.1 for 100 Hz up to 1.6 for 60 Hz, while at 40 Hz stimulation frequency, almost no axonal responses were detected in the tested intensity range. With the use of synaptic blockers, we demonstrate the underlying direct activation mechanism of the ganglion cells. Finally, using high-resolution electrical imaging and label-free electrophysiological axon tracking, we demonstrate the extent of activation in axon bundles.Significance.Our results can be exploited to define a spatially selective stimulation strategy avoiding axonal activation in future retinal implants, thereby solving one of the major limitations of artificial vision. The results may be extended to other fields of neuroprosthetics to achieve selective focal electrical stimulation.

Intercellular communication atlas reveals Oprm1 as a neuroprotective factor for retinal ganglion cells.

Previous studies of neuronal survival have primarily focused on identifying intrinsic mechanisms controlling the process. This study explored how intercellular communication contributes to retinal ganglion cell (RGC) survival following optic nerve crush based on single-cell RNA-seq analysis. We observed transcriptomic changes in retinal cells in response to the injury, with astrocytes and Müller glia having the most interactions with RGCs. By comparing RGC subclasses characterized by distinct resilience to cell death, we found that the high-survival RGCs tend to have more ligand-receptor interactions with neighboring cells. We identified 47 interactions stronger in high-survival RGCs, likely mediating neuroprotective effects. We validated one identified target, the µ-opioid receptor (Oprm1), to be neuroprotective in three retinal injury models. Although the endogenous Oprm1 is preferentially expressed in intrinsically photosensitive RGCs, its neuroprotective effect can be transferred to other subclasses by pan-RGC overexpression of Oprm1. Lastly, manipulating the Oprm1 activity improved visual functions in mice.

Non-image-forming vision as measured through ipRGC-mediated pupil constriction is not modulated by covert visual attention.

In brightness, the pupil constricts, while in darkness, the pupil dilates; this is known as the pupillary light response (PLR). The PLR is driven by all photoreceptors: rods and cones, which contribute to image-forming vision, and intrinsically photosensitive retinal ganglion cells (ipRGCs), which mainly contribute to non-image-forming vision. Rods and cones cause immediate pupil constriction upon light exposure, whereas ipRGCs cause sustained constriction throughout light exposure. Recent studies have shown that covert attention modulated the initial PLR; however, it remains unclear whether the same holds for the sustained PLR. We tested this by leveraging ipRGCs' responsiveness to blue light, causing the most prominent sustained constriction. While replicating previous studies by showing that pupils constricted more when either directly looking at, or covertly attending to, bright as compared to dim stimuli (with the same color), we also found that the pupil constricted more when directly looking at blue as compared to red stimuli (with the same luminosity). Crucially, however, in two high-powered studies (n = 60), we did not find any pupil-size difference when covertly attending to blue as compared to red stimuli. This suggests that ipRGC-mediated pupil constriction, and possibly non-image-forming vision more generally, is not modulated by covert attention.

Neutrophil-inflicted vasculature damage suppresses immune-mediated optic nerve regeneration.

In adult mammals, injured retinal ganglion cells (RGCs) fail to spontaneously regrow severed axons, resulting in permanent visual deficits. Robust axon growth, however, is observed after intra-ocular injection of particulate ß-glucan isolated from yeast. Blood-borne myeloid cells rapidly respond to ß-glucan, releasing numerous pro-regenerative factors. Unfortunately, the pro-regenerative effects are undermined by retinal damage inflicted by an overactive immune system. Here, we demonstrate that protection of the inflamed vasculature promotes immune-mediated RGC regeneration. In the absence of microglia, leakiness of the blood-retina barrier increases, pro-inflammatory neutrophils are elevated, and RGC regeneration is reduced. Functional ablation of the complement receptor 3 (CD11b/integrin-αM), but not the complement components C1q-/- or C3-/-, reduces ocular inflammation, protects the blood-retina barrier, and enhances RGC regeneration. Selective targeting of neutrophils with anti-Ly6G does not increase axogenic neutrophils but protects the blood-retina barrier and enhances RGC regeneration. Together, these findings reveal that protection of the inflamed vasculature promotes neuronal regeneration.

Are ipRGCs involved in human color vision? Hints from physiology, psychophysics, and natural image statistics.

Human photoreceptors consist of cones, rods, and melanopsin-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs). First studied in circadian regulation and pupillary control, ipRGCs project to a variety of brain centers suggesting a broader involvement beyond non-visual functions. IpRGC responses are stable, long-lasting, and with a particular codification of photoreceptor signals. In comparison with the transient and adaptive nature of cone and rod signals, ipRGCs' signaling might provide an ecological advantage to different attributes of color vision. Previous studies have indicated melanopsin's influence on visual responses yet its contribution to color perception in humans remains debated. We summarized evidence and hypotheses (from physiology, psychophysics, and natural image statistics) about direct and indirect involvement of ipRGCs in human color vision, by first briefly assessing the current knowledge about the role of melanopsin and ipRGCs in vision and codification of spectral signals. We then approached the question about melanopsin activation eliciting a color percept, discussing studies using the silent substitution method. Finally, we explore various avenues through which ipRGCs might impact color perception indirectly, such as through involvement in peripheral color matching, post-receptoral pathways, color constancy, long-term chromatic adaptation, and chromatic induction. While there is consensus about the role of ipRGCs in brightness perception, confirming its direct contribution to human color perception requires further investigation. We proposed potential approaches for future research, emphasizing the need for empirical validation and methodological thoroughness to elucidate the exact role of ipRGCs in human color vision.

Assessment of the relationship between structural and functional tests in patients with idiopathic intracranial hypertension.

PURPOSE: To assess the relationship between structural and functional tests in mild and moderate idiopathic intracranial hypertension (IIH). METHODS: Patients with mild and moderate IIH and a control group were enrolled. Best-corrected visual acuity (BCVA), macular ganglion cell layer (MGCL) thickness, peripapillary retinal nerve fiber layer (pp RNFL) thickness, perimetric mean deviation (MD), and photopic negative responses (PhNR) of the electroretinogram were recorded. The associations between structural (pp RNFL and MGCL thickness) and functional (PhNR amplitude, MD and BCVA) parameters were assessed. RESULTS: 154 eyes from 78 subjects (74 eyes from IIH patients and 80 eyes from healthy subjects) were included in this comparative observational study. The MGCL thickness, VA, pp RNFL, and PhNR base-to-trough (BT) amplitude were significantly worse in moderate IIH. The BCVA and MD were associated with MGCL thickness only in moderate IIH. The relationship between MD and MGCL thickness started when MD fell below -5.7 dB. CONCLUSIONS: The association between functional and structural parameters varies between mild and moderate IIH. The MD and MGCL thickness outperformed in assessing disease severity in mild and moderate IIH, respectively. The association between MD and MGCL thickness could be considered in IIH severity categorization.

Effect of eccentric fixation on the steady-state pattern electroretinogram.

PURPOSE: The steady-state pattern electroretinogram (ssPERG) is used to assess retinal ganglion cell function in a variety of research contexts and diagnostic applications. In certain groups of patients or study participants, stable central fixation of the stimulus is not guaranteed. The present study aimed at assessing the effects of misfixation on the ssPERG response to checkerboard reversal stimuli. METHODS: Using two check sizes (0.8° and 15°), we compared ssPERG responses for several amounts of fixation deviation, ranging from 0° to 19° horizontally and from 0° to 14° diagonally. The stimulus area extended to 15° eccentricity, stimulus reversal rate was 15/s. RESULTS: Up to around 7° eccentricity, there was no sizable effect of fixation deviation under most conditions. Effects were somewhat larger for nasal than for temporal deviation, in particular for small checks. Diagonal deviation was associated with a response to luminance onset/offset at 7.5 Hz (subharmonic of the reversal rate), most prominently when the interior of a large check was fixated. CONCLUSION: Generally, moderate inaccuracies of fixation do not have a sizable effect on ssPERG amplitude. However, with large checks, the luminance response has to be considered.

Retinal origin of orientation but not direction selective maps in the superior colliculus.

Neurons in the mouse superior colliculus ("colliculus") are arranged in ordered spatial maps. While orientation-selective (OS) neurons form a concentric map aligned to the center of vision, direction-selective (DS) neurons are arranged in patches with changing preferences across the visual field. It remains unclear whether these maps are a consequence of feedforward input from the retina or local computations in the colliculus. To determine whether these maps originate in the retina, we mapped the local and global distribution of OS and DS retinal ganglion cell axon boutons using in vivo two-photon calcium imaging. We found that OS boutons formed patches that matched the distribution of OS neurons within the colliculus. DS boutons displayed fewer regional specializations, better reflecting the organization of DS neurons in the retina. Both eyes convey similar orientation but different DS inputs to the colliculus, as shown in recordings from retinal explants. These data demonstrate that orientation and direction maps within the colliculus are independent, where orientation maps are likely inherited from the retina, but direction maps require additional computations.

Efficacy and specificity of melanopsin reporters for retinal ganglion cells.

Intrinsically photosensitive retinal ganglion cells (ipRGCs) are specialized retinal output neurons that mediate behavioral, neuroendocrine, and developmental responses to environmental light. There are diverse molecular strategies for marking ipRGCs, especially in mice, making them among the best characterized retinal ganglion cells (RGCs). With the development of more sensitive reporters, new subtypes of ipRGCs have emerged. We therefore tested high-sensitivity reporter systems to see whether we could reveal yet more. Substantial confusion remains about which of the available methods, if any, label all and only ipRGCs. Here, we compared many different methods for labeling of ipRGCs, including anti-melanopsin immunofluorescence, Opn4-GFP BAC transgenic mice, and Opn4cre mice crossed with three different Cre-specific reporters (Z/EG, Ai9, and Ai14) or injected with Cre-dependent (DIO) AAV2. We show that Opn4cre mice, when crossed with sensitive Cre-reporter mice, label numerous ganglion cell types that lack intrinsic photosensitivity. Though other methods label ipRGCs specifically, they do not label the entire population of ipRGCs. We conclude that no existing method labels all and only ipRGCs. We assess the appropriateness of each reporter for particular applications and integrate findings across reporters to estimate that the overall abundance of ipRGCs among mouse RGCs may approach 11%.

The Impact of Aging on the Function of Retinal Ganglion Cells.

Aging is a major risk factor for retinal neurodegenerative diseases. Aged mammalian retinal ganglion cells (RGCs) lack the ability to regenerate axons after injury. Rodent models suggest that older age increases the vulnerability of RGCs to injury and impairs RGC function as well as their functional recovery. Molecular changes - including decreased circulating levels of brain-derived neurotrophic factor (BDNF) - might contribute to impaired RGC dendritic extension during aging. Moreover, age-related mitochondrial dysfunction plays a major role in aging processes, as it leads to reduced adenosine triphosphate and increased generation of reactive oxygen species. Autophagy activity is necessary for the maintenance of cellular homeostasis and decreases with aging in the central nervous system. During aging, vascular insufficiency may lead to impaired oxygen and nutrient supply to RGCs. Microglial cells undergo morphological changes and functional impairment with aging, which might compromise retinal homeostasis and promote an inflammatory environment. Addressing these age-related changes by means of a low-energy diet, exercise, and neurotrophic factors might prevent age-related functional impairment of RGCs. This review focuses on the current understanding of aging RGCs and key players modulating those underlying mechanisms.

Parallel pathways carrying direction-and orientation-selective retinal signals to layer 4 of the mouse visual cortex.

Parallel visual pathways from the retina to the primary visual cortex (V1) via the lateral geniculate nucleus are common to many mammalian species, including mice, carnivores, and primates. However, it remains unclear which visual features present in both retina and V1 may be inherited from parallel pathways versus extracted by V1 circuits in the mouse. Here, using calcium imaging and rabies circuit tracing, we explore the relationships between tuning of layer 4 (L4) V1 neurons and their retinal ganglion cell (RGC) inputs. We find that subpopulations of L4 V1 neurons differ in their tuning for direction, orientation, spatial frequency, temporal frequency, and speed. Furthermore, we find that direction-tuned L4 V1 neurons receive input from direction-selective RGCs, whereas orientation-tuned L4 V1 neurons receive input from orientation-selective RGCs. These results suggest that direction and orientation tuning of V1 neurons may be partly inherited from parallel pathways originating in the retina.

Contribution of the eye and of opn4xa function to circadian photoentrainment in the diurnal zebrafish.

The eye is instrumental for controlling circadian rhythms in mice and human. Here, we address the conservation of this function in the zebrafish, a diurnal vertebrate. Using lakritz (lak) mutant larvae, which lack retinal ganglion cells (RGCs), we show that while a functional eye contributes to masking, it is largely dispensable for the establishment of circadian rhythms of locomotor activity. Furthermore, the eye is dispensable for the induction of a phase delay following a pulse of white light at CT 16 but contributes to the induction of a phase advance upon a pulse of white light at CT21. Melanopsin photopigments are important mediators of photoentrainment, as shown in nocturnal mammals. One of the zebrafish melanopsin genes, opn4xa, is expressed in RGCs but also in photosensitive projection neurons in the pineal gland. Pineal opn4xa+ projection neurons function in a LIGHT ON manner in contrast to other projection neurons which function in a LIGHT OFF mode. We generated an opn4xa mutant in which the pineal LIGHT ON response is impaired. This mutation has no effect on masking and circadian rhythms of locomotor activity, or for the induction of phase shifts, but slightly modifies period length when larvae are subjected to constant light. Finally, analysis of opn4xa;lak double mutant larvae did not reveal redundancy between the function of the eye and opn4xa in the pineal for the control of phase shifts after light pulses. Our results support the idea that the eye is not the sole mediator of light influences on circadian rhythms of locomotor activity and highlight differences in the circadian system and photoentrainment of behaviour between different animal models.

Sensitivity and specificity of the uniform field electroretinogram in glaucoma detection in comparison to the pattern electroretinogram.

PURPOSE: To determine the ability of the photopic negative response (PhNR) of the uniform field electroretinogram (UF-ERG) to identify early glaucomatous changes in comparison to the checkerboard and bar stimuli of the pattern electroretinogram (PERG). METHODS: Forty-nine glaucoma patients were classified into two groups: glaucoma-suspect (23 eyes) and early to moderate glaucoma (30 eyes), based on their clinical examination and the results of standard automated perimetry. Thirty patients (30 eyes) with intraocular pressures (IOP) of 21 mmHg or less, with no history of reported high IOP, were included as controls. PERG and UF-ERG recordings were obtained on a Diagnosys D-341 Attaché-Envoy System. Visual field testing was done only for glaucoma-suspect and glaucoma patients. RESULTS: All three tests (PERG bar stimulus, PERG checkerboard stimulus and PhNR) displayed significantly prolonged peak times for glaucoma and glaucoma-suspect patients, with delays ranging from 7.8 to 14.8%, depending on the test. The PERG bar stimulus also showed a significantly lower N95 amplitude for both glaucoma groups (with reductions of 26.0% and 33.0% for glaucoma-suspect and glaucoma groups, respectively). The PERG checkerboard N95 amplitude component had high sensitivity for detecting glaucoma patients but a low specificity (97% and 37%, respectively; AUC = 0.61). Overall, the PhNR peak time showed the highest sensitivity and specificity (77% and 90%, respectively; AUC = 0.87). CONCLUSIONS: PERG bar stimuli and the PhNR of the UF-ERG can be used in the clinical setting to detect glaucoma-related changes in glaucoma-suspect and glaucoma patients. However, our data confirm that the PhNR peak time has the best combined sensitivity and specificity.

Neural activity of retinal ganglion cells under continuous, dynamically-modulated high frequency electrical stimulation.

Objective.Current retinal prosthetics are limited in their ability to precisely control firing patterns of functionally distinct retinal ganglion cell (RGC) types. The aim of this study was to characterise RGC responses to continuous, kilohertz-frequency-varying stimulation to assess its utility in controlling RGC activity.Approach.We usedin vitropatch-clamp experiments to assess electrically-evoked ON and OFF RGC responses to frequency-varying pulse train sequences. In each sequence, the stimulation amplitude was kept constant while the stimulation frequency (0.5-10 kHz) was changed every 40 ms, in either a linearly increasing, linearly decreasing or randomised manner. The stimulation amplitude across sequences was increased from 10 to 300µA.Main results.We found that continuous stimulation without rest periods caused complex and irreproducible stimulus-response relationships, primarily due to strong stimulus-induced response adaptation and influence of the preceding stimulus frequency on the response to a subsequent stimulus. In addition, ON and OFF populations showed different sensitivities to continuous, frequency-varying pulse trains, with OFF cells generally exhibiting more dependency on frequency changes within a sequence. Finally, the ability to maintain spiking behaviour to continuous stimulation in RGCs significantly reduced over longer stimulation durations irrespective of the frequency order.Significance.This study represents an important step in advancing and understanding the utility of continuous frequency modulation in controlling functionally distinct RGCs. Our results indicate that continuous, kHz-frequency-varying stimulation sequences provide very limited control of RGC firing patterns due to inter-dependency between adjacent frequencies and generally, different RGC types do not display different frequency preferences under such stimulation conditions. For future stimulation strategies using kHz frequencies, careful consideration must be given to design appropriate pauses in stimulation, stimulation frequency order and the length of continuous stimulation duration.

Effective treatment of optic neuropathies by intraocular delivery of MSC-sEVs through augmenting the G-CSF-macrophage pathway.

Optic neuropathies, characterized by injury of retinal ganglion cell (RGC) axons of the optic nerve, cause incurable blindness worldwide. Mesenchymal stem cell-derived small extracellular vesicles (MSC-sEVs) represent a promising "cell-free" therapy for regenerative medicine; however, the therapeutic effect on neural restoration fluctuates, and the underlying mechanism is poorly understood. Here, we illustrated that intraocular administration of MSC-sEVs promoted both RGC survival and axon regeneration in an optic nerve crush mouse model. Mechanistically, MSC-sEVs primarily targeted retinal mural cells to release high levels of colony-stimulating factor 3 (G-CSF) that recruited a neural restorative population of Ly6Clow monocytes/monocyte-derived macrophages (Mo/MΦ). Intravitreal administration of G-CSF, a clinically proven agent for treating neutropenia, or donor Ly6Clow Mo/MΦ markedly improved neurological outcomes in vivo. Together, our data define a unique mechanism of MSC-sEV-induced G-CSF-to-Ly6Clow Mo/MΦ signaling in repairing optic nerve injury and highlight local delivery of MSC-sEVs, G-CSF, and Ly6Clow Mo/MΦ as therapeutic paradigms for the treatment of optic neuropathies.

The cognitive impact of light: illuminating ipRGC circuit mechanisms.

Ever-present in our environments, light entrains circadian rhythms over long timescales, influencing daily activity patterns, health and performance. Increasing evidence indicates that light also acts independently of the circadian system to directly impact physiology and behaviour, including cognition. Exposure to light stimulates brain areas involved in cognition and appears to improve a broad range of cognitive functions. However, the extent of these effects and their mechanisms are unknown. Intrinsically photosensitive retinal ganglion cells (ipRGCs) have emerged as the primary conduit through which light impacts non-image-forming behaviours and are a prime candidate for mediating the direct effects of light on cognition. Here, we review the current state of understanding of these effects in humans and mice, and the tools available to uncover circuit-level and photoreceptor-specific mechanisms. We also address current barriers to progress in this area. Current and future efforts to unravel the circuits through which light influences cognitive functions may inform the tailoring of lighting landscapes to optimize health and cognitive function.

Synaptic Origins of the Complex Receptive Field Structure in Primate Smooth Monostratified Retinal Ganglion Cells.

Considerable progress has been made in studying the receptive fields of the most common primate retinal ganglion cell (RGC) types, such as parasol RGCs. Much less is known about the rarer primate RGC types and the circuitry that gives rise to noncanonical receptive field structures. The goal of this study was to analyze synaptic inputs to smooth monostratified RGCs to determine the origins of their complex spatial receptive fields, which contain isolated regions of high sensitivity called "hotspots." Interestingly, smooth monostratified RGCs co-stratify with the well-studied parasol RGCs and are thus constrained to receiving input from bipolar and amacrine cells with processes sharing the same layer, raising the question of how their functional differences originate. Through 3D reconstructions of circuitry and synapses onto ON smooth monostratified and ON parasol RGCs from central macaque retina, we identified four distinct sampling strategies employed by smooth and parasol RGCs to extract diverse response properties from co-stratifying bipolar and amacrine cells. The two RGC types differed in the proportion of amacrine cell input, relative contributions of co-stratifying bipolar cell types, amount of synaptic input per bipolar cell, and spatial distribution of bipolar cell synapses. Our results indicate that the smooth RGC's complex receptive field structure arises through spatial asymmetries in excitatory bipolar cell input which formed several discrete clusters comparable with physiologically measured hotspots. Taken together, our results demonstrate how the striking differences between ON parasol and ON smooth monostratified RGCs arise from distinct strategies for sampling a common set of synaptic inputs.

An actor-model framework for visual sensory encoding.

A fundamental challenge in neuroengineering is determining a proper artificial input to a sensory system that yields the desired perception. In neuroprosthetics, this process is known as artificial sensory encoding, and it holds a crucial role in prosthetic devices restoring sensory perception in individuals with disabilities. For example, in visual prostheses, one key aspect of artificial image encoding is to downsample images captured by a camera to a size matching the number of inputs and resolution of the prosthesis. Here, we show that downsampling an image using the inherent computation of the retinal network yields better performance compared to learning-free downsampling methods. We have validated a learning-based approach (actor-model framework) that exploits the signal transformation from photoreceptors to retinal ganglion cells measured in explanted mouse retinas. The actor-model framework generates downsampled images eliciting a neuronal response in-silico and ex-vivo with higher neuronal reliability than the one produced by a learning-free approach. During the learning process, the actor network learns to optimize contrast and the kernel's weights. This methodological approach might guide future artificial image encoding strategies for visual prostheses. Ultimately, this framework could be applicable for encoding strategies in other sensory prostheses such as cochlear or limb.

A presynaptic source drives differing levels of surround suppression in two mouse retinal ganglion cell types.

In early sensory systems, cell-type diversity generally increases from the periphery into the brain, resulting in a greater heterogeneity of responses to the same stimuli. Surround suppression is a canonical visual computation that begins within the retina and is found at varying levels across retinal ganglion cell types. Our results show that heterogeneity in the level of surround suppression occurs subcellularly at bipolar cell synapses. Using single-cell electrophysiology and serial block-face scanning electron microscopy, we show that two retinal ganglion cell types exhibit very different levels of surround suppression even though they receive input from the same bipolar cell types. This divergence of the bipolar cell signal occurs through synapse-specific regulation by amacrine cells at the scale of tens of microns. These findings indicate that each synapse of a single bipolar cell can carry a unique visual signal, expanding the number of possible functional channels at the earliest stages of visual processing.