Abstract

Objective:

Visual adaptation is a physiological and perceptual process by which the visual system adjusts to changes in the environment or visual stimuli. This process is fundamental to how we perceive the world around us and allows our visual system to efficiently encode and process visual information. The retina incorporates adaptation within its dozens of functionally distinct retinal ganglion cell types. Meanwhile, the field of retinal prostheses is increasing its understanding of electrical adaptation and cell-specific stimulation. However, very little is known about the interaction of visual and electrical stimulation on the adaptation of retinal ganglion cell types.

Methods/Approach:

Recording with a microelectrode array (MEA), we presented an ON and OFF full-field, visual stimulus to characterize various visual response parameters in healthy and degenerating rd10 mouse retinas. We then evaluated visual response changes before and after blocks of monophasic voltage-controlled electrical pulse stimulation.

Main results:

A history of electrical stimulation strengthened visual responses in WT retina, even when changes attributable to in vitro visual adaptation were considered. In rd10 retinas, electrical adaptation counteracted the baseline in vitro visual adaptation. In all cases, adaptation often affected the on and off visual response components differentially. Consequently, the ON/OFF classification of individual cells changed because of adaptation.

Significance:

Electrical stimulation-induced changes in the retina should be considered in the encoding of visual stimuli by retinal prosthetic devices. In vitro investigations for bionic vision should strive to probe electrical responsiveness after adaptation to ongoing electrical stimulation has achieved a steady state.

Introduction

The human visual system is a remarkable entity, capable of perceiving and adapting to a range of visual stimuli, from the dim glow of starlight to the brilliance of a sunny day (Webster, 2015). Over the past century, many luminaries like Ramón y Cajal, Hartline, Barlow, Kuffler, Lettvin, Hubel, and Wiesel have collectively contributed to today’s understanding of information processing via the visual system (Kuffler, 1953; Rodieck and Stone, 1965; Hubel and Wiesel, 1998; Hartline, 1938; Barlow, 1953; Lettvin et al., 1959). Central to the remarkable dynamism of the visual system is the retina’s capacity for adaptation—to recalibrate and fine-tune its responses to suit the ever-changing demands of the visual input. There is an ever-growing body of literature illuminating the various visual adaptations occurring in the retina (Baccus and Meister, 2002; Demb, 2002; Moore et al., 2011; Tikidji-Hamburyan et al., 2015; Tikidji-Hamburyan et al., 2017; Borghuis et al., 2018).

The development of retinal prosthetic devices offers a promising solution to the vision loss experienced by individuals with retinal degenerative diseases such as retinitis pigmentosa and age-related macular degeneration (Dagnelie, 2012; Ayton et al., 2020). However, the concept of retinal adaptation takes on profound significance in the context of retinal prosthetics where adaptation may be a barrier to ongoing perception (Fernandez, 2018). Many studies have also investigated electrical adaptation in the retina (Fried et al., 2006; Jensen and Rizzo, 2007; Ray, 2010; Freeman and Fried, 2011; Lorach et al., 2015; Im and Fried, 2016). Our own lab has previously demonstrated that voltage tuning curves displayed substantial hysteresis due to adaptation to electrical stimulation occurring over many seconds (Hosseinzadeh et al., 2018). Despite these efforts, little is known about the complex interaction between electrical and visual adaptation.

A recent study in mouse retina by Baden et al. (2016) has shown that, beyond the simple on, off, sustained, and transient response types, there are at least 30 different physiological types of retinal outputs. Other recent investigations into bionic vision as a cure for blindness have additionally shown that electrical input filters can vary according to retinal ganglion cell (RGC) type, suggesting the possibility for cell-specific stimulation (Sekhar et al., 2017; Ho et al., 2018). In more recent effort, we sought to compare the differences in preferred electrical input with this well-established catalog of dozens of RGC types, relying on their functional responses to light stimulation as a reference point (Shabani et al., 2021; Shabani et al., 2024). The present study provides insights into the adaptation of on and off responses that occur when retinas receive electrical stimulation. Understanding such adaptation is an essential aspect of maximizing the effectiveness of bionic vision devices and improving the perceptual experience of recipient patients.

In this study, we used a full-field flashing visual stimulus to characterize mouse retinal ganglion cell visual responses. We tested the hypothesis that ongoing electrical stimulation alters visual responses in healthy and degenerating rd10 mouse retinas. We further characterized the nature of these alterations and compared such adaptation to a pair of control conditions. We used an internal control condition to test whether the observed adaptation is specific to nearby, electrically responsive RGCs. We also tested whether the observed adaptation was specific to electrical stimulation or merely a byproduct of experimental and light probe conditions. The results enhance our understanding of adaptive mechanisms and electrical stimulation of the retina.

Materials and methods

A detailed treatment of our experimental methods can be found in Rathbun et al. (2024).

Animals

The animals were housed under standard 12 h light/12 h dark cycles with free and ample access to food and water. Adult wild-type C57Bl/6 J (Jackson Laboratory, Bar Harbor, ME, USA) and rd10 (on a C57Bl/6 J background; Jackson Laboratory) strains were used, with age ranging from postnatal day 28 to 37 for both strains. The rd10 strain is a well-established model for retinal degeneration in which the retina is unhealthy, but not yet blind at the ages examined. For each strain, three male and two female mice were used. For the external control condition, age-matched mice were used for both strains. For each strain of the control mice, two male mice were used. All procedures were approved by the Tübingen University committee on animal protection (Einrichtung für Tierschutz, Tierärztlichen Dienst und Labortierkunde directed by Dr. Franz Iglauer) and performed by the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in ophthalmic and visual research.

Retinal preparation

For dissecting the retina, the mice were anesthetized by CO2 inhalation. Following CO2 inhalation, the mice were checked for absence of withdrawal reflex by pinching the between-toe tissue and then euthanized by cervical dislocation. Under normal room lighting, the eyes were removed to carbogenated (95% O2 and 5% CO2) artificial cerebrospinal fluid (ACSF) solution containing the following (in mM): 125 NaCl, 3.5 KCl, 2 CaCl2, 1 MgCl2, 1.25 NaH2PO4, 25 NaHCO3, and 25 Glucose, pH 7.4.

The cornea, lens, vitreous, and ora serrata were removed, and the retina was detached from the pigment epithelium, and the optic nerve was cut at the base of the retina. Special care was taken to remove all traces of vitreous material from the inner surface of the retina to optimize contact between the nerve fiber layer and recording electrodes. The retinas were maintained in carbogenated ACSF until needed. For recording, a retinal half was mounted ganglion cell layer down on a planar microelectrode array (MEA). Two small paint brushes were used to orient and flatten the retinal half without risking damage to the MEA. A dialysis membrane (Cellu Sep, Membrane Filtration Products Inc., Seguin, Texas, USA) mounted on a custom Teflon ring was lowered onto the retina to press it into closer contact with the MEA (Meister et al., 1994). After securing the MEA under the preamplifier, the retina was continuously superfused with carbogenated ACSF (~6 mL/min) maintained at 33o C using both a heating plate and a heated perfusion cannula (HE-Inv-8 & PH01; Multi Channel Systems, Reutlingen, Germany) and stabilized for > 30 min prior to recording.

Microelectrode Array (MEA) and data acquisition

For recording the spiking responses from the RGCs, a planar MEA containing 59 circular titanium nitride electrodes (diameter: 30 μm, interelectrode spacing: 200 μm; Multi Channel Systems, Reutlingen, Germany) arrayed in a rectilinear grid. Electrode impedance was 200–250 kΩ at 1 kHz measured using a nanoZ impedance meter (Plexon Inc., TX, USA). The MEA60 system (MCS, Reutlingen, Germany) acquired data at 50 kHz/channel with a 1 Hz–3 kHz filter and 1,100 × amplification. Defective electrodes were grounded, and electrical stimulation waveforms assigned via MEA_Select software. The raw data were recorded with MC_Rack at a rate of 50 kHz/channel with a filter bandwidth ranging from 1 Hz–3 kHz and amplification gain of 1,100.

Electrical stimulation

In our previous study a detailed description of the electrical stimulation procedure is provided (Material and Methods, Stimulation; Jalligampala et al., 2017). Briefly, the stimulus pulses were generated using a stimulus generator (STG 2008, Multi Channel Systems, Reutlingen, Germany) and were delivered epiretinally via an interior electrode selected near robust neural signals. The stimulus (Figure 1a, left panel) consisted of monophasic rectangular voltage pulses with amplitudes (0.3, 0.5, 1.0, 1.5, 2.0, 2.5 V), included both cathodic (-V) and anodic (+V) stimuli, and durations (60, 100, 200, 300, 500, 1,000, 2,000, 3,000, 5,000 μs). While 0.1 V was also presented, the stimulator was later found to be unable to deliver the waveform faithfully. Accordingly, this stimulus was excluded from analysis. Durations were randomized in 5 sequential sets per voltage, with 5 s intervals between pulses for RGC recovery. Polarity was randomly chosen at the start of each block. Spontaneous activity was recorded for ~30 s before and after each block.

Experimental design. (a) Electrical stimuli were presented over a range of voltage/duration combinations. Constant-voltage stimulus blocks consisted of 5 repetitions of 9 durations. Within each repetition, the durations were randomized, 5 s separated pulses. Voltage blocks were presented from lowest to highest amplitude. Subsequent voltage blocks were separated by 150 s or greater. A block of full-field flash stimulus (20 repetitions of 2 s on and 2 s off) was interleaved between voltage blocks throughout the experiment. Voltage-duration combinations that exceeded safety limits were omitted (grey triangle). (b) The different test (ELECT-RESP) and control conditions (internal ELECT-CTL and external VIS-ONLY-CTL). Red denotes stimulating electrode. Open grey box indicates electrodes on which included cells were recorded. Grey box with cross indicates that cells recorded on these 9 electrodes were excluded for the ELECT-CTL condition. (c) Flash response characterization according to Carcieri et al., 2003. (top) Spike rasters of a full-field flash stimulus (20 trials, 2 s ON 2 s OFF). (middle) Peristimulus time histogram (PSTH) derived from the rasters. (bottom) Visual stimulus time course. A1 and A2 are relative (to baseline) response amplitude for on and off. Tp1 and Tp2 are time to peak (latency) for on and off. D1 and D2 are durations for on and off. (d) Visual response changes were evaluated by comparing First (red) to Second (blue) and First to Last (green) using the average PSTH binned at 2 ms intervals for all responses. Gaussian smoothing filter σ = 4 ms.

Visual stimulation

Visual stimuli were projected from below through the transparent MEA using a DLP projector (K10; Acer Inc., San Jose, California, USA). Full-field flashes (~3 × 4 mm) cycled 2 s ON (40 klx) and 2 s OFF (20 lx) for 20 repetitions per block (mean illuminance 20 klx, 99.9% Michelson contrast; Figure 1a). The six visual stimulus blocks were interleaved before, after, and within an electrical stimulation experiment that spanned ~80 min of recording time, including the First and Last flash blocks.

Test and control conditionsTest condition (ELECT-RESP)

RGCs were considered electrically responsive if they met two criteria: (1) ≥ 3 of 96 responses exceeded 2 SD above spontaneous firing rate; (2) firing rate ≥8.89 Hz (≥4 spikes in five 90-ms windows). Only cells recorded on the 8 electrodes surrounding the stimulating electrode (in red, Figure 1b, left panel) were included.

Internal control (ELECT-CTL)

This control included electrically nonresponsive cells >300 μm from the stimulating electrode, recorded in the same tissue (Figure 1b, middle panel). A caveat to this control condition is that electrical responses have been shown to extend at least as far as 800 μm from the stimulating electrode (Eickenscheidt et al., 2012; Ryu et al., 2009; Stett et al., 2007; Jalligampala et al., 2017; Wilke et al., 2011).

External control (VIS-ONLY-CTL)

This control incorporated cells exposed only to visual stimulation at the same timing as the original protocol. Stimulating visually at the same time points provided a control for any changes occurring during the in vitro recording due to all factors excluding electrical stimulation (Figure 1b, right panel).

Data processing and inclusion criteria

Spike sorting software was used to process raw data (Offline Sorter, Plexon Inc., TX, USA). Raw traces were first filtered using a low-cut, 12-point Bessel filter at 51 Hz to exclude line noise, and thresholded at 4 standard deviations below the mean of the amplitude histogram. Traces were sorted into noise and unit-specific clusters with an automated routine (Standard Expectation Maximization), manually inspected for accuracy, and units were included only if they exhibited: (1) clear lock-out in ISI histogram/autocorrelogram, (2) no peaks in cross-correlograms indicative of split units, (3) distinct principal component separation of biphasic waveforms, and (4) stable waveform shape and firing rate throughout the experiment. The sorted spike timestamps were collected with NeuroExplorer (Plexon Inc., TX, USA) and exported to MATLAB for further analysis. Total RGC counts were 2078 WT cells (16 retinal halves) and 1880 rd10 cells (9 halves) for test/internal controls; 366 WT cells (3 halves) and 573 rd10 cells (3 halves) for external controls (Figure 2).

RGC counts for WT and rd10 for test and control conditions. “distances” are recording distances relative to the stimulating electrode. (For the definition of electrically “responsive” see Methods: Test and Control Conditions, and Jalligampala et al., 2017, Methods–Data Analysis).

Data analysis

Determining the visual response parameters: For each visual block (20 repetitions), responses were quantified following Carcieri et al. (2003) (Figure 1c). A peristimulus time histogram (PSTH) was constructed by aligning and averaging the 20 responses (4 s each, 10 ms bins), then smoothing with a Gaussian filter (σ = 50 ms). Baseline firing rate and SD were calculated from the last 250 ms of each ON and OFF phase. For each phase, time to peak (Tp1, Tp2), and baseline-subtracted amplitude (A1, A2) were measured. Response duration (D1, D2) was defined as the time for the response to decay from the peak (A) to A/e (e = Euler’s number). Responses were excluded if amplitude did not exceed baseline + 2 SD or if peaks were likely caused by carryover between phases (latency < 100 ms). Response polarity was quantified using the ON/OFF index: (A1 − A2) / (A1 + A2), classifying cells as off (−1 to −0.5), on–off (−0.5 to 0.5), or on (0.5 to 1).

Visual response changes

To test whether electrical stimulation alters visual responses, we compared visual response parameters before and after stimulation for the test and both control conditions (Figure 1d). The visual block recorded before any electrical stimulation was defined as the “First” block, and the block recorded after completion of the full protocol (~80 min) as the “Last” block. Because many cells showed changes after only low-voltage stimulation (0.3 and 0.1 V; ~20 min; Figure 1d), the block recorded after this initial period was defined as the “Second” block and used to assess early effects. Although 0.1 V stimuli were delivered, these data were excluded from analysis due to unreliable waveform and charge delivery (Jalligampala et al., 2017).

Statistics

To test the hypothesis that visual response parameter medians were unchanged between First and Last as well as between First and Second blocks, Wilcoxon’s ranksum test was used (MATLAB; The Mathworks, Natick, MA) at a significance of 0.05. To examine whether significant response parameter changes differed between test and each control condition, Wilcoxon’s ranksum test, significance of 0.05, was used to test the hypothesis that the two change distributions have equal medians.

Results

Here, our goal is to gain a deeper understanding of how ongoing electrical stimulation affects the visual response parameters of different RGC types. In a previous study from our group (Jalligampala et al., 2017), we established an experimental and analysis framework, by which one could identify the optimal stimulus that will activate a majority of RGCs indirectly via network stimulation from the epiretinal side. This stimulus was optimal for “blind” experiments where the specific response properties for each cell were unknown. During the entire duration of the experiment, six visual stimulus blocks of full-field “flash” stimulus were applied to monitor the stability of RGC responses. These visual blocks were interleaved before, after, and between electrical stimulation blocks spanning ~80 min of the entire recording time (Figure 1a). Apart from monitoring the stability of the RGC responses, the visual stimulus provided us with an opportunity to classify the cells into different physiological cell types based on their response to the visual stimulus. Surprisingly, we found evidence that visual responses might change over the course of the experiment. In that previous study, we pooled all the RGCs and did not distinguish them into different physiological cell types. However, in this study, we classified RGCs into different response types to better understand how various visual response parameters change during ongoing electrical stimulation. For this study, the primary data set came from the previous study (Jalligampala et al., 2017) and was used to test our hypothesis that ongoing electrical stimulation alters visual responses in healthy and degenerating mouse retinas. A preliminary version of this study was reported using only that original data (Jalligampala et al., 2015). The additional VIS-ONLY-CTL dataset was collected to account for any non-electrical effects intrinsic to the experimental design.

In an effort to subdivide the RGC population into visual response categories, we examined visual response parameters in the context of a previous study (Carcieri et al., 2003). However, we failed to find strong agreement between our data and the statistically determined response distribution boundaries that delineated types in that study (see Supplementary Figures S1, S2). Nevertheless, due to the predominance of the on, off, and on- off types in the established literature (Morgan and Wong, 2007), we divided our data according to the approximate ON/OFF index boundaries of Carcieri et al. (2003) (see Methods).

Diversity in alteration of visual responses to electrical stimulation

To evaluate how electrical stimulation alters the visual response parameters, we plotted the rastergram and peristimulus time histograms (PSTHs) of the cell’s response to full-field flash stimulus (Figure 3). These example cells were both electrically and visually responsive. The rastergram shows the visual response to all six flash blocks presented before, after, and between the electrical stimulation blocks. The PSTH (above) represents the average response (20 trials) of the “First block,” i.e., before electrical stimulation. The PSTH (below) shows the average response (20 trials) of the “Last block,” i.e. after the entire electrical stimulation protocol was over. We observed diversity in the alteration of the visual responses. For our study, our definition of neural adaptation was a change in the cell’s firing rate. Therefore, to evaluate the effect of electrical stimulation, we observed how the cell’s firing rate changed following a period of electrical stimulation. Apart from the changes in firing rate, shown in all the example cells (Figures 3a–d), we also observed changes in other visual response parameters (latency, duration, and ON/OFF index). For some cells, we observed that the latency of the responses became shorter following electrical stimulation (Figure 3d, on response). For some cells, we observed a change in the transiency of the responses, e.g., some cells which were transient before electrical stimulation became sustained following electrical stimulation (Figures 3c,d). Interestingly, we observed for some cells, a change in ON/OFF index, i.e., before electrical stimulation the cell responded to a single phase of the flash stimulus, however after electrical stimulation the cell responded to both phases of the flash stimulus (Figure 3d, on to on–off). Similar heterogeneous changes in response properties were also observed in the VIS-ONLY-CTL condition, indicating that visual stimulation alone can produce varied adaptation effects even in the absence of electrical stimulation.

Diversity of visual response changes. (a–d) Rastergram depicts responses for all six visual blocks. Response differences between blocks 1 (First), 2 (Second), and 6 (Last) were examined. (Top) Average PSTH for First block. (Bottom) Average PSTH for Last block.

As stated, our primary measure of visual response adaptation to electrical stimulation was the change in firing rate, quantified by on and off response amplitudes. We also analyzed changes in response latency and duration. These parameters are shown in notched box-whisker plots (median, 95% CI, quartiles, and outliers) to illustrate variability across conditions (Figures 4–6). We first compared responses between the first and last blocks. Because changes appeared as early as ~20 min (second block, after 0.1 and 0.3 V stimulation), we also compared the first and second blocks. Finally, to assess slower adaptation, we compared the second and last blocks (~60 min apart). All comparisons were made between the test condition (ELECT-RESP) and control conditions (ELECT-CTL and VIS-ONLY-CTL; see Methods). Statistical results are shown in Figure 7, cell numbers in Supplementary Figure S6, and outliers in Supplementary Figures S3–S5. Supplementary Figure S8 has been added to show the difference in response amplitude medians between ELECT-RESP and each of the control conditions. The main results are as follows. In all three conditions, WT cells showed increased firing rates to both light onset and offset, with increased response duration to light onset. Amplitude increased shortly after electrical stimulation, whereas duration increased significantly only later, and both effects were stronger in electrically responsive cells. In contrast, such adaptation was not consistently observed in rd10 retina. Other changes in latency and duration occurred but did not show a strong or consistent effect of stimulation. Notably, in rd10 retina under control conditions, the latency and duration of light-offset responses decreased over time, though not consistently relative to the stimulation condition. A detailed analysis follows.

Box-whisker plots for response amplitude. Box plots for on (solid line) and off (dashed) responses shown for first (red), second (blue), and last (green) flash blocks for WT (a,c,e) and rd10(b,d,f) retinas for the test (ELECT-RESP, (a,b) and control [ELECT-CTL, (c,d) and VIS-ONLY-CTL, (e,f) conditions]. Horizontal lines of the box plot demarcate 25, 50, and 75% quartiles. Notches are 95% confidence intervals for the median (50% quartile). Whiskers denote data range excluding outliers. Some outlier cutoffs are clipped to show detail. Refer to Figure 7 for pairwise statistical tests. Asterisks: significantly different distributions for FIRST-SECOND, FIRST-LAST, and SECOND-LAST comparisons—p-values in Figure 7.

Box-whisker plots for response latency. Box plots for on (solid line) and off (dashed) responses shown for first (red), second (blue), and last (green) flash blocks for WT (a,c,e) and rd10(b,d,f) retinas for the test [ELECT-RESP, (a,b) and control [ELECT-CTL, (c,d) and VIS-ONLY-CTL, (e,f) conditions]. Horizontal lines of the box plot demarcate 25, 50, and 75% quartiles. Notches are 95% confidence intervals for the median (50% quartile). Whiskers denote data range excluding outliers. Some outlier cutoffs are clipped to show detail. Refer to Figure 7 for pairwise statistical tests. Asterisks: significantly different distributions for FIRST-SECOND, FIRST-LAST, and SECOND-LAST comparisons—p-values in Figure 7.

Box-whisker plots for response duration. Box plots for on (solid line) and off (dashed) responses shown for first (red), second (blue), and last (green) flash blocks for WT (a,c,e) and rd10(b,d,f) retinas for the test [ELECT-RESP, (a,b) and control [ELECT-CTL, (c,d) and VIS-ONLY-CTL, (e,f) conditions]. Horizontal lines of the box plot demarcate 25, 50, and 75% quartiles. Notches are 95% confidence intervals for the median (50% quartile). Whiskers denote data range excluding outliers. Some outlier cutoffs are clipped to show detail. Refer to Figure 7 for pairwise statistical tests. Asterisks: significantly different distributions for FIRST-SECOND, FIRST-LAST, and SECOND-LAST comparisons—p-values in Figure 7.

For the plots in Figures 4–6, the nonparametric Wilcoxon’s ranksum test (MATLAB, p < 0.05) was used to determine significant changes. For each condition (responsive and nearby ELECT-RESP, nonresponsive and distant ELECT-CTL, and the visual-only control VIS-ONLY-CTL) and each mouse strain (WT and rd10) we tested whether each visual response parameter differed between First vs. Second, First vs. Last, and Second vs. Last flash stimulus blocks. For each of these tests, the quartiles of the First, Second, and Last response parameter distributions are provided along with the p-value of the test. For testing whether the parameter changes were significantly different between test and control conditions, p-values are presented before the middle and right data blocks. Green boxes identify significant increase. Red boxes identify significant decreases. Bold p-values highlight significant changes. Red p-values indicate significance between test and control conditions.

WT retinas

First vs. Last Block–Amplitude: For both ON and OFF responses, relative amplitude increased significantly from the First to the Last block across all three conditions (Figures 4a,c,e). However, this increase was significantly greater in the test condition (ELECT-RESP) than in the controls (ELECT-CTL, VIS-ONLY-CTL; Figure 7), indicating that electrical stimulation enhances the amplitude increase beyond visual stimulation alone. Latency: ON response latency showed no significant change across conditions. OFF response latency decreased significantly in all three conditions, with a slightly greater decrease in ELECT-RESP compared to VIS-ONLY-CTL, but no difference between ELECT-RESP and ELECT-CTL (Figures 5a,c,e). Duration: ON response duration increased significantly in all three conditions, with no significant differences between test and controls (Figures 6a,c,e). OFF response duration increased significantly only in the control conditions, with no significant effect of electrical stimulation.

First vs. Second Block–Amplitude: For both ON and OFF responses, there was a significant increase in the relative amplitude from the First to the Second block for all three conditions. When comparing the magnitude of pre- and post-stimulation changes between the test and control conditions, we found the increase in ON amplitudes to be significantly greater for ELECT-RESP. However, the increase in OFF response amplitudes was only significantly greater for ELECT-RESP in comparison to the ELECT-CTL, but not VIS-ONLY-CTL (Figures 4a,c,e). Latency: The latency of the ON responses significantly decreased from the First to the Second block only for ELECT-CTL, and this change was not significantly different when compared to ELECT-RESP (Figures 5a,c,e). For the OFF response latency, there was a significant decrease in latency from the First to the Second block in all conditions. On comparing the test and control conditions, this decrease was significantly greater for ELECT-RESP compared to VIS-ONLY-CTL, but not ELECT-CTL. Duration: There was a significant increase in duration of the ON responses from the First to the Second block only for ELECT-CTL. For the ELECT-RESP, there was a significant decrease in duration of the OFF responses from the First to the Second block, and this change was significant compared to the unchanged OFF duration of ELECT-CTL and the increased OFF duration of VIS-ONLY-CTL (Figures 6a,c,e).

Second vs. Last Block - Amplitude: The only amplitude change from the Second to the Last block was for the on response amplitude of VIS-ONLY-CTL. However, comparing this to ELECT-RESP, there was no statistical significance (Figures 4a,c,e). Latency: Only the ELECT-CTL condition exhibited latency changes from Second to Last block. While the on response latency increased, the off latency decreased (Figures 5a,c,e). Duration: There was a significant increase in duration of the on responses for all three conditions. Furthermore, comparing the test to the control conditions the magnitude of increase in on response duration was significantly greater for the ELECT-RESP (Figures 6a,c,e). For all three conditions, there was no significant change in off duration from the Second to Last block.

rd10 retinas

First vs. Last Block–Amplitude: The only significant amplitude change from First to Last block was an increase in on and off response amplitudes for VIS-ONLY-CTL (Figures 4b,d,f). This increase was only significantly different from ELECT-RESP for the off response. Latency: There was a significant increase in on response latency from the First to Last block for ELECT-RESP. However, this was not a significant change compared to the control conditions. For both control conditions, there was a significant decrease in off response latency from the First to the Last block, and this change was also significant relative to the test condition (Figures 5b,d,f). Duration: The only response duration changes from the First to the