Visual Feature Encoding Ganglion Cell Response Transience Is Determined by the Summation of Converging Parallel Signals

Alma Ganczer University of Pécs, Szentágothai Research Centre Gergely Szarka University of Pécs, Szentágothai Research Centre Márton Balogh University of Pécs, Szentágothai Research Centre Ádám Jonatán Tengölics University of Pécs, Szentágothai Research Centre Tamás Kovács-Öller University of Pécs, Szentágothai Research Centre Béla Völgyi (  volgyi01@gamma.ttk.pte.hu ) University of Pécs, Szentágothai Research Centre


INTRODUCTION
Information gathered from the visual field travels through parallel intraretinal pathways and converges onto retinal ganglion cells (RGCs) that in turn summate and encrypt incoming signals into action potential trains prior to transmitting towards the brain. Light-evoked RGC responses have been characterized by their polarity (ON, OFF, and ON-OFF), sensitivity to various stimuli and kinetics. Based on response speed RGCs can be sorted into either brisk or sluggish categories, whereas the pattern can be a maintained spiking (sustained) or a brisk spike burst (transient). Both aspects of RGC response kinetics are likely important in terms of signal efficiency on postsynaptic neuronal targets in higher visual centers 1, 2, 3 . The transient/sustained dichotomy has been documented in a variety of vertebrate species, including cold-blooded animals, primates and non-primate mammals as well 4,5,6,7,8,9,10,11,12,13,14 . Since all photoreceptors generate sustained responses upon illumination 15 a sustainedto-transient response transformation must occur along the retinal signal flow. Previous work in the salamander and rabbit retinas suggested that response transience is determined by the kinetics of the postsynaptic glutamate receptors (mGluR 6 , AMPA, Kainate) at the site of the very first retinal contact, the photoreceptor-to-bipolar cell synapse 16,17 . In contrast, a recent study in the mouse retina presented a thorough analysis of the various possible sites of retinal circuits that may participate in determining response transience 18 . Although the conclusion of this latter study was that different RGC types use diverse mechanisms to produce sustained or transient light responses, the presented data clearly showed that outer retinal postsynaptic receptor (AMPA/kainate receptor) desensitization had only a minor effect on response transience. The discrepancy in these previous studies can be attributed to species differences (salamander vs. mammals), the difference in examined cell types (bipolar cells vs. RGCs), and the selection of examined RGC subtypes (random vs. targeted). In addition, in most of these relevant studies patch-clamp EPSC recording was the method of choice, which does not allow for the direct observation of RGC spiking, the real output signal of the retina.
To this end we utilized extracellular RGC spike recordings to examine the kinetics of the real retinal output conveyed towards visual brain centers. We show a line of evidence supporting the view that RGC response transience is largely independent of outer retinal signal kinetics, including (i) the existence of both transient and sustained ON RGC responses contrary to a single underlying mGluR 6 receptor in cone bipolar cell dendrites; (ii) despite the difference in ON and OFF signaling glutamate receptors in bipolar cell dendrites the frequency distribution for ON and OFF RGC photopic response transience are rather similar; (iii) the existence of both transient and sustained ON and OFF cell responses under scotopic light levels, in which condition the primary rod pathway provides the sole signal conveying path. On the other hand, RGC response transience was altered considerably when stimulus intensity was altered and thereby an input dominance switch from scotopic through mesopic to photopic conditions occurred. In addition, a similar input dominance switch was also achieved pharmacologically by blocking GABAergic inhibitory and gap junction mediated excitatory inputs to RGCs. Therefore, the above data indicate that the RGC response kinetics are determined by inner retinal interactions. While direct excitatory bipolar cell inputs are summated by RGCs, inner retinal microcircuits serve as temporal filters to augment or moderate certain response components to achieve better visual performance.

RESULTS mGlur 6 Receptor Signaling Serves both Transient and Sustained ON RGC Responses
Based on our casual RGC recordings we often observed apparent discrepancies between response decay obtained for spike trains and EPSCs of the same cell (Supplemental Figure 1). Since the retinal output to the brain is delivered in the form of spike trains it is crucial to utilize RGC spike recordings in order to evaluate if (and how) temporal features of light responses affect vision. To this end we generated peristimulus time histograms (PSTHs) and determined corresponding PSTHτ values upon extracellular spike recordings according to Ganczer et al. 20 . Once generated, PSTHτ values were used to assay RGC response length (decay). Briefly, upon spike PSTHs the amplitude (A -peak frequency) and delay (D -time to peak) were determined, PSTHτ values were calculated as the time required for A to drop to 1/e*A (Supplemental Figure 1a). When both EPSCτ and PSTHτ values were determined for sample RGCs, they often showed significant discrepancies (Supplemental Figure 1b, c), thus further supporting our view that spike recordings are appropriate approaches to examine the topic of RGC response transience.
One goal of this work was to examine if, as it has been claimed previously 16,17 , transience values of RGC responses are determined in the outer retina or whether inner retinal mechanisms play a role as well. To test this, we first recorded ON-center RGC responses. In darkness, photoreceptors continuously release glutamate and upon light exposure this release decreases (or stops). Glutamate, once released into the synaptic cleft, binds to mGluR 6 receptors 21 in the postsynaptic membrane of ON bipolar cells, initiating an intracellular cascade that involves the G (o) protein-mediated inactivation of adenylate cyclase and the closure of TRPM 1 bound nonspecific cation channels 22 . Taken together, light onset pauses glutamate release from photoreceptors, ultimately depolarizing ON center bipolar cells. Although 5-6 ON bipolar cell types and corresponding ON signaling streams are known to exist in the mammalian retina 23,24 , the initial signal at the photoreceptor/bipolar cell synapse is generated by a single postsynaptic glutamate receptor type, the mGluR 6 for all ON bipolar cells . If bipolar   cell glutamate receptors play a major role in shaping response transience, then the kinetics of   ON-center light responses should be very similar for all ON bipolar cells and their   Representative perievent raster diagrams show that individual RGCs provide light-evoked spiking responses upon full-field illumination that are rather similar across trials (four consecutive trials for each recorded cell). However, RGCs display a great variety in terms of their response length (or decay -expressed as the PSTHτ value in this work) for both the ON (cells 1 and 2) and OFF (cells 3 and 4) subpopulations. The white bar below the recordings represents the timing of the on-and offset of the stimulus in this and in all other figure panels of this paper. Rh*/rod/sec) and corresponding PSTHτ values were determined. To test if our stimuli activated mostly the primary but not the secondary and/or tertiary rod signaling routes we blocked mGluR 6 glutamate-mediated signaling to OFF RGCs by using the agonist L-2-amino-4phosphonobutyric acid (APB 50 μM). This pharmacological blockade eliminated OFF RGC responses (Supplemental Figure 2) thus proving that the dominant signaling route was the primary rod pathway under these conditions. We found that APB sensitive low-scotopic RGC responses, though appeared somewhat more delayed and sustained (scotopic -mean: 0.155 s, SD 0.134; photopic -mean:   . PSTHτ values of this RGC changed non-monotonously during this experiment when the stimulus intensity was gradually increased (see also panel c). PSTHs clearly show a peak of very sensitive signal component (red arrow) evoked by weak, scotopic stimuli. This sensitive response component appears relative delayed when it is compared to the less sensitive but brisk signal component (light blue arrow). These two signal components differ in their delays but appear similar in response decay, therefore PSTHτ values are shifted towards the sustained range when the two signals are summated (mesopic conditions -2 nd , 3 rd and 4 th panels), whereas remain transient when only one signal is present (scotopic condition -1 st panel) or dominates over the other component (photopic conditions -5 th . 6 th and 7 th panels; see also panel c). b. Representative OFF RGC light-evoked rasters (left column) and PSTHs (right column) recorded as a response to varying intensity stimuli (intensity values are reflected in the right top corner of each -panel). PSTHτ values of this RGC clearly changed during this experiment as the stimulus intensity was gradually increased. Similar to the cell in panel a this OFF RGC showed a very sensitive but rather delated response peak (red arrow) and a faster but less sensitive (light blue arrow) peak. The two signal components differed in their delays and sensitivities and a slight alteration in PSTHτ values occurred as a result of the summation of components (mostly in mesopic conditions -middle panels). While the distinction of response components can clearly be differentiated for the ON RGC in a, this OFF cell (and most examined RGCs) showed a less obvious and less separable summation of incoming signals. c and d. Diagrams show that, similar to cells shown in panels a and b (values of these cells appear in black and red in the diagrams), most recorded RGCs displayed stimulus strength driven changes of PSTHτ values (grey curves). e. Diagram shows minimum/maximum PSTHτ value pairs for the recorded cells during the course of the stimulus intensity recording paradigm. The examined RGCs showed ~18-73% PSTHτ changes during the course of this experiment.

RGC Response Transience is Altered Due to Signal Summation
We reported previously that most examined RGCs showed some intensity-dependent alteration of PSTHτ values 20  intermingling signals upon mid-range illumination conditions provided less obvious combinations of a fast peak and a slow shoulder component (see Figure 3b). However, the general observation of this test was that RGCs endured a considerable stimulus intensitydependent alteration in their response transience. The difference between the minimum and maximum PSTHτ values across the examined intensity range varied between 18.3% and 73.2% (mean 41.5 +/-16.5 SD) for sample RGCs. These results, therefore, indicate that signal summation in fact is a significant factor to determine response transience for most cells.

RGC Response Transience is Altered by the Perturbation of Lateral Signaling
In the previous section we showed that an illumination strength-dependent dominance shift of summated scotopic and photopic signals can underly RGC response transience changes. We posit here that a similar summation of parallel signals can determine response transience even under the same stimulating conditions. The morphological substrate for this hypothesis is provided by previous studies showing that many RGC subtypes receive excitatory inputs from 2 or more bipolar cell subtypes (see Discussion). Unfortunately, parallel photopic pathways utilize the same postsynaptic receptor in the photoreceptor-to-bipolar cell synapse (mGluR 6 receptor in ON cone bipolar cell dendrites, AMPA and kainate receptors in OFF cone bipolar cells) for all vertical retinal pathways thus direct testing of their signal summation is not feasible via pharmacology. However, we can take advantage of the fact that certain vertical pathways target RGCs directly via excitatory bipolar cells, whereas others contact them indirectly through intermediary inhibitory (mostly GABAergic) amacrine cells. In addition, many RGCs also display gap junction coupling to their RGC and/or amacrine cell neighbors 26,27,28,29,30,31,32,33 , thereby diversifying the potentially summable incoming signals.
Since both GABA and gap junction mediated signals can be diminished pharmacologically while glutamate-driven bipolar cell signaling is intact, we carried out experiments in which one of the indirect signaling streams was blocked pharmacologically thereby isolating signal components carried to RGC targets via parallel streams.
To interfere with amacrine cell inhibition the nonspecific GABA a /GABA c receptor antagonist picrotoxin (PTX; in a concentration of 50 μM) was utilized and changes in response

The Distribution of PSTHτ Values Reveal a Wide Range of RGC Response Transience
The transient/sustained division of RGC light responses has been widely utilized to characterize and classify RGC subtypes and is thought to be strongly related to visual function.
Transient, burst-like responses likely transmit information about 'fast-paced' and dynamic aspects of the visual field, including direction and movement whereas sustained responses provide a continuous feed of information on static aspects of the view. Therefore, transient and sustained RGC responses encode dissimilar but equally important facets of visual information.
This transient/sustained dichotomy has been documented in various vertebrates including cold-blooded animals, primates and non-primate mammals 4,5,6,7,8,9,10,11,12,13,14,18 . However, many of the studies examined this issue based on slow-wave EPSC recordings that, as we demonstrated here, occasionally deviate from RGC spike response transience, the parameter that is inherently associated with the visual code. In addition, the most elaborate studies in this topic were restricted to only a few RGC subtypes 18 , most of which were the non-image-forming melanopsin expressing ipRGCs, were conducted in cold-blooded vertebrates 16 and/or carried out in retinal slice preparations 16,17 where lateral connections were compromised. Therefore, we feel that the present reexamination of RGC response transience and related RGC spike coding is strongly justified.
We reported previously 20

Possible Origin of Response Transience
In addition to the postsynaptic glutamate receptor on bipolar cell dendrites 16,17,34,35,36 , a variety of mechanisms have been proposed to affect RGC transience, including direct inhibitory amacrine cell input to bipolar cells 7 , specific membrane characteristics 37 and differences in the synaptic reuptake of glutamate 38,39 . A recent study in the mouse retina presented a thorough analysis of the various sites of the retinal hyper-circuitry that may participate in determining response transience 18 . Although the conclusion of this latter study was that different RGC types use diverse mechanisms to produce sustained or transient light responses, the presented data clearly showed that outer retinal postsynaptic receptor (AMPA/kainate receptor) desensitization had only a minor effect on response transience. This is in an agreement with other studies showing that exclusively the kainite receptors transmit glutamatergic signals to OFF bipolar cells in the mouse and primate retinas 40,41,42,43,44,45,46 and generate transient (mouse type 2, 3a), sustained (mouse type 1, 4) and intermediary (mouse type 3b) responses. This somewhat contradicts the immunohistochemistry data 47,48,49,50 and functional work established in the ground squirrel 17 that argues for an equal contribution of AMPA receptor signaling for certain OFF bipolar cells and it is unclear if this evident discrepancy is due to species differences or some other factors. Regardless of the origin of this discrepancy, however, Zhao and colleagues showed that RGCs response transience is rather determined by the balance of transient/sustained bipolar cell inputs and RGC resting membrane potential as well as the presence/lack of inhibition 18,50,52 . This latter study, however, examined mostly EPSC recordings in only a subset of the RGCs, therefore it provided limited information for the entire population. Our examination here utilized RGC spikes (events that encode visual signals) and the random sampling for the analyses thus they provide a broader perspective for the topic at hand.
Besides the above listed inconsistencies, our findings here also challenge the classical view that originates RGC response transience from the differential kinetics of bipolar cell glutamate receptors. First, the broad, unimodal peak of ON RGC PSTHτ frequency histograms indicate that the same mGluR 6  would provide a greater variability for OFF RGC responses when they are compared to their ON counterparts. However, we found that the range and the frequency distribution profiles for OFF PSTHτ values were rather similar to those of ON RGCs. Therefore, converging evidence of this work support, the previous conclusion that the postsynaptic receptor(s) in bipolar cell dendrites are not a major factor in determining RGC response transience 18 . Therefore, we conclude that while bipolar cell response transience may in fact is determined by the expressed glutamate receptors located in the photoreceptor/bipolar cell synapse, RGCs do not simply inherit this feature but rather perform considerable signal transformation (summation, filtering and digitizing) before generating the RGC spiking output (see the same phenomenon  . These two inputs have dissimilar delays (due to differential bipolar cell signaling and/or different location of synapses over the RGC dendritic arbor) and therefore the summation of the responses results in an intermediate or sustained RGC spiking response. b. This RGC receives excitatory inputs from two sources, from a transient bipolar cell (light blue EPSC) and from a gap junction coupled amacrine cell (purple depolarization). If the dynamics of these two inputs differ their summation will induce intermediate and/or sustained RGC spiking. c. This RGC receives excitation from a bipolar cell (light blue EPSC) and delayed inhibition (red IPSC) from an amacrine cell resulting in a transient RGC response. d. An RGC that receives excitation from a bipolar cell (light blue EPSC) and inhibition (red IPSC) from an amacrine cell. In this scenario the two inputs have about the same delays therefore the excitation will be truncated and the RGC output is an intermediate/sustained spiking.
We showed an example where low threshold (sensitive) scotopic and high threshold photopic signals were summated. As the dominance of the two input components changed throughout the stimulus strength sequence the response delay and decay were altered as well.
As a result, the spike train became more sustained upon mesopic stimulation when none of the signals dominated over the other. In this latter case response components were easily separated based on their dissimilar light sensitivities, however, a similar signal summation should occur for two or more parallel conveyed photopic signals to the same RGC targets as well. In this scheme a certain RGC subtype receives inputs from two or more signaling streams (bipolar cells). This in fact is the case for ON alpha RGCs of the mouse retina that receive a mixture of inputs from type 6, 7 and 8 bipolar cells 54 . In addition, sustained OFF alpha cells have been shown to be postsynaptic to type 2 bipolar and tGluMI cells whereas transient OFF alpha cells are targeted by type 3a and type 4 bipolar cells 55,56 . Therefore, mixing inputs from several parallel signaling streams seems to be a general feature for most RGCs in the mammalian retina. Furthermore, it appears that the summation of inputs with slightly different kinetics (delay, decay) and sensitivity plays a crucial role in determining the ultimate kinetics work has to be designed to address this question. In general, we conclude that RGCs collect a cohort of excitatory and inhibitory signals for summation and use them to sculpt their own output signal with kinetics that suits their function prior to sending information to the brain. This output depending on the type of summated signals and their relative timing may result in a variety of RGC response kinetics on a rather wide transient/intermediary/sustained range.
According to this hypothesis, RGCs summate excitatory signals from multiple bipolar cell subtypes and gap junction coupled cell neighbors (amacrine and/or ganglion cells) and filter some of these signals out (via amacrine cell-mediated inhibition) to sway the dominance of input components to adapt to a specific visual function. This is in line with previous reports indicating that different RGC response properties fully emerge only after additional processing by currents in the bipolar cell axon terminal 51 , at synapses with amacrine cells 52 , and by RGCs.
It is clear that the response transience of these RGCs is not simply inherited from presynaptic bipolar cells but rather transformed to suit the specific RGC visual function.

The Visual Function of RGC Transience
We reported previously that RGC light response delays are subtype-specific and they are precisely fine-tuned by inner retinal microcircuits to achieve a better RGC performance 3 69,70,71,72 and, also others with sustained responses that perceive luminosity contrast 73 , color contrast 74 or object orientation 75 . While the first cohort of RGCs require a quick inactivation and corresponding decay of spiking frequency (transient response) in order to quickly recover and get ready for following changes in the visual scene, sustained RGCs allow for the summation of inputs over an extended time frame in order to get more sensitized for minuscule differences of light levels (e.g. grayscale or color) within their receptive fields. One interesting aspect of this hypothesis arises when stimulus-dependent changes in RGC response transience is taken into account. We showed one such incidence in this study when response transience values changed as a response to the modulation of stimulus intensity.
Does that mean that RGCs perform better in certain stimulating conditions than in others? We think that the answer to this question is 'yes'. We experience it daily that our vision is rather limited during the night and this limitation involves the reduction of contrast sensitivity in both the spatial and temporal domains of our vision. This latter phenomenon is expressed by the Ferry-Porter Law stating that the critical fusion frequency is proportional to the logarithm of the flickering stimulus luminance 76 . Therefore, the precise adjustment of RGC response temporal features including transience appears to be critical for our visual perception. Supported by our present investigation, most response transience altering mechanisms appear to be performed by the inner retinal microcircuits.

Animals and preparation
Adult (P20<) C57BL6 mice were used in this study. After overnight dark adaptation, animals were put under deep anesthesia using Forane (4%, 0.2ml/l) and terminated through cervical dislocation. Dissection and experimentation were carried out in mammalian Ringer's solution All efforts were made to minimize pain and discomfort. We also state that this manuscript is reported in accordance with ARRIVE guidelines.

Extracellular electrophysiology
Single-cell extracellular recordings were obtained from RGCs using tungsten microelectrodes were recorded. Data were filtered at 5 kHz with a Bessel filter and were sampled at 10 kHz.

Light Stimulation
A uniform full-field stimulus was used to evoke light responses (0.5s illumination every 2s). In experiments utilizing different stimulus intensity of light stimuli values were given in terms of the rate of photoisomerization that occurs in each rod in every second (Rh*/rod/s); we calculated with an average rod density of 437,000 rods/ mm2 [8] and quantum efficiency of 0.67 [9]. The intensity of the light stimuli varied from 1 to 6000 Rh*/rod/sec.

Data analysis
Spike sorting was carried out using Spike2 (CED, Cambridge, UK) and Offline Sorter (Plexon Instruments, Dallas, TX, USA). PSTHs measuring transiency were generated in NeuroExplorer (Plexon Instruments, Dallas, TX, USA). Gaussian smoothing (filter size: 3) was applied to all datasets. All transiency values were calculated using the PSTHτ method 20 , where PSTHt measures the time required for spiking frequency to decrease to 1/e of the peak firing amplitude. SPSS (v19, IBM, Armonk, NY, USA) and OriginPro (OriginLab Corp., Northampton, MA, USA) were used for statistical analysis. All responses included in this study were analyzed manually.

Data availability
All raw data of this manuscript as well as detailed protocols will be promptly available upon request.   PSTHτ values of this RGC clearly changed during this experiment as the stimulus intensity was gradually increased. Similar to the cell in panel a this OFF RGC showed a very sensitive but rather delated response peak (red arrow) and a faster but less sensitive (light blue arrow) peak. The two signal components differed in their delays and sensitivities and a slight alteration in PSTHτ values occurred as a result of the summation of components (mostly in mesopic conditions -middle panels). While the distinction of response components can clearly be differentiated for the ON RGC in a, this OFF cell (and most examined RGCs) showed a less obvious and less separable summation of incoming signals. c and d. Diagrams show that, similar to cells shown in panels a and b (values of these cells appear in black and red in the diagrams), most recorded RGCs displayed stimulus strength driven changes of PSTHτ values (grey curves). e. Diagram shows minimum/maximum PSTHτ value pairs for the recorded cells during the course of the stimulus intensity recording paradigm. The examined RGCs showed ~18-73% PSTHτ changes during the course of this experiment.     . These two inputs have dissimilar delays (due to differential bipolar cell signaling and/or different location of synapses over the RGC dendritic arbor) and therefore the summation of the responses results in an intermediate or sustained RGC spiking response. b. This RGC receives excitatory inputs from two sources, from a transient bipolar cell (light blue EPSC) and from a gap junction coupled amacrine cell (purple depolarization). If the dynamics of these two inputs differ their summation will induce intermediate and/or sustained RGC spiking. c. This RGC receives excitation from a bipolar cell (light blue EPSC) and delayed inhibition (red IPSC) from an amacrine cell resulting in a transient RGC response. d. An RGC that receives excitation from a bipolar cell (light blue EPSC) and inhibition (red IPSC) from an amacrine cell. In this scenario the two inputs have about the same delays therefore the excitation will be truncated and the RGC output is an intermediate/sustained spiking.    recorded as a response to stimuli of various strength (intensity values are re ected in the right top corner of each -panel). PSTHτ values of this RGC changed non-monotonously during this experiment when the stimulus intensity was gradually increased (see also panel c). PSTHs clearly show a peak of very sensitive signal component (red arrow) evoked by weak, scotopic stimuli. This sensitive response component appears relative delayed when it is compared to the less sensitive but brisk signal component (light blue arrow). These two signal components differ in their delays but appear similar in response decay, therefore PSTHτ values are shifted towards the sustained range when the two signals are summated (mesopic conditions -2nd, 3rd and 4th panels), whereas remain transient when only one signal is present (scotopic condition -1st panel) or dominates over the other component (photopic conditions -5th. 6th and 7th panels; see also panel c). b. Representative OFF RGC light-evoked rasters (left column) and PSTHs (right column) recorded as a response to varying intensity stimuli (intensity values are re ected in the right top corner of each -panel). PSTHτ values of this RGC clearly changed during this experiment as the stimulus intensity was gradually increased. Similar to the cell in panel a this OFF RGC showed a very sensitive but rather delated response peak (red arrow) and a faster but less sensitive (light blue arrow) peak. The two signal components differed in their delays and sensitivities and a slight alteration in PSTHτ values occurred as a result of the summation of components (mostly in mesopic conditions -middle panels). While the distinction of response components can clearly be differentiated for the ON RGC in a, this OFF cell (and most examined RGCs) showed a less obvious and less separable summation of incoming signals. c and d. Diagrams show that, similar to cells shown in panels a and b (values of these cells appear in black and red in the diagrams), most recorded RGCs displayed stimulus strength driven changes of PSTHτ values (grey curves). e. Diagram shows minimum/maximum PSTHτ value pairs for the recorded cells during the course of the stimulus intensity recording paradigm. The examined RGCs showed ~18-73% PSTHτ changes during the course of this experiment.     Summary Drawing of Potential Signal Summation Mechanisms that Affecting RGC Response Transience. a. Two bipolar cells of different subtypes provide transient inputs to the same RGC (light blue EPSC curves). These two inputs have dissimilar delays (due to differential bipolar cell signaling and/or different location of synapses over the RGC dendritic arbor) and therefore the summation of the responses results in an intermediate or sustained RGC spiking response. b. This RGC receives excitatory inputs from two sources, from a transient bipolar cell (light blue EPSC) and from a gap junction coupled amacrine cell (purple depolarization). If the dynamics of these two inputs differ their summation will induce intermediate and/or sustained RGC spiking. c. This RGC receives excitation from a bipolar cell (light blue EPSC) and delayed inhibition (red IPSC) from an amacrine cell resulting in a transient RGC response. d. An RGC that receives excitation from a bipolar cell (light blue EPSC) and inhibition (red IPSC) from an amacrine cell. In this scenario the two inputs have about the same delays therefore the excitation will be truncated and the RGC output is an intermediate/sustained spiking.