An asymmetric outer retinal response to drifting sawtooth gratings

Abstract

Electroretinogram (ERG) studies have demonstrated that the retinal response to temporally modulated fast-ON and fast-OFF sawtooth flicker is asymmetric. The response to spatiotemporal sawtooth stimuli has not yet been investigated. Perceptually, such drifting gratings or diamond plaids shaded in a sawtooth pattern appear brighter when movement produces fast-OFF relative to fast-ON luminance profiles. The neural origins of this illusion remain unclear (although a retinal basis has been suggested). Thus we presented toad eyecups with sequential epochs of sawtooth, sine-wave, and square-wave gratings drifting horizontally across the retina at temporal frequencies of 2.5–20 Hz. All ERGs revealed a sustained direct-current (DC) transtissue potential during drift and a peak at drift offset. The amplitudes of both phenomena increased with temporal frequency. Consistent with the human perceptual experience of sawtooth gratings, the sustained DC potential effect was greater for fast-OFF cf. fast-ON sawtooth. Modeling suggested that the dependence of temporal luminance contrast on stimulus device frame rate contributed to the temporal frequency effects but could not explain the divergence in response amplitudes for the two sawtooth profiles. The difference between fast-ON and fast-OFF sawtooth profiles also remained following pharmacological suppression of postreceptoral activity with tetrodotoxin (TTX), 2-amino-4-phosphonobutric acid (APB), and 2,3 cis-piperidine dicarboxylic acid (PDA). Our results indicate that the DC potential difference originates from asymmetries in the photoreceptoral response to fast-ON and fast-OFF sawtooth profiles, thus pointing to an outer retinal origin for the motion-induced drifting sawtooth brightness illusion.

electrophysiological studies of retinal ON-OFF processing have demonstrated that the response to temporally modulated sawtooth flicker is asymmetric (Alexander et al. 2001; Barnes et al. 2002; Dryja et al. 2005; Khan et al. 2005; Kremers 2013; Kremers et al. 1993; Pangeni and Kremers 2013; Pangeni et al. 2012; Rodrigues et al. 2010; Vukmanic et al. 2014). It is unknown whether such asymmetries also occur for more complex spatiotemporal sawtooth stimuli (i.e., moving sawtooth grating or diamond patterns). Indeed, relatively little is known about electroretinographic responses to moving stimuli in any species. The motion electroretinograms (ERGs) recorded across studies to date have identified some similar waveforms, most notably a positivity following motion onset that increases in amplitude with temporal frequency (Bach and Hoffmann 2000; Dodt and Kuba 1995; Korth 1987; Korth et al. 2000). These studies, however, have primarily used short motion durations, square-wave stimuli, and alternating current (AC)-coupled recording systems (that do not allow study of sustained effects). Moreover, although the motion ERG has been compared to the pattern offset-onset response (Korth et al. 2000), no attempts have been made to dissect pharmacologically the level at which the response is generated.

In addition to extending the ON-OFF pathway and motion processing literature, ERG recordings of the retinal response to drifting sawtooth gratings may clarify the cellular origins of the perceptual effects that these stimuli elicit in humans. Such moving patterns generating fast-OFF profiles appear brighter than those moving in the opposite direction generating fast-ON profiles (despite no change in mean luminance). This illusion was explored by Cavanagh and Anstis (1986) using shaded grating patterns drifting leftward or rightward on the horizontal axis and has also been demonstrated for drifting shaded diamonds (Watanabe et al. 1995), drifting shaded boxes (Ashida and Scott-Samuel 2014), and spatially uniform sawtooth flicker (Cavanagh and Anstis 1986; Wu et al. 1996). An example of the illusion is provided in Supplemental Video S1 (available in the data supplement online at the Journal of Neurophysiology Web site).

Psychophysically, the strength of the drifting sawtooth grating illusion has been shown to increase with temporal frequency up to 3.75 Hz, the highest speed tested to date (Cavanagh and Anstis 1986). Related brightness perception measures for full-field sawtooth flicker (Wu et al. 1996) and moving sawtooth boxes (Ashida and Scott-Samuel 2014) suggest that the drifting sawtooth illusion may also occur at moderate temporal frequencies (i.e., up to 16 Hz). Although the neural basis of this perceptual effect remains unclear, Cavanagh and Anstis' (1986) psychophysical experiments led them to suggest that the illusion may result from saturation of transient responses to the fast phase of the sawtooth (although they did not specify the level of the visual system where they expected this to occur).

In the present study, we used direct-current (DC) ERG and pharmacological dissection to explore the retinal response to drifting sawtooth gratings. In doing so, we aimed to extend the temporal sawtooth flicker literature into the spatiotemporal domain and to elucidate the neural origins of Cavanagh and Anstis' (1986) drifting sawtooth illusion. Toad eyecups, for which robust retinal responses have been well-studied (Gallemore et al. 1997), were chosen as the model system as comparatively long recording times were required to facilitate pharmacological dissection of functional responses using a within-subjects design. In accordance with the temporal sawtooth flicker literature and the effects of spatiotemporal sawtooth on brightness perception in humans (Cavanagh and Anstis 1986), we expected that the DC ERG response amplitudes for fast-OFF and fast-ON sawtooth profiles would be significantly different and sensitive to the temporal frequency of stimulation.

METHODS

Subjects

Forty-six male cane toads (Rhinella marina) were obtained from a local supplier (Peter Douch, Mareeba, Queensland, Australia) and housed under a 12:12-h light-dark cycle before experimentation. Toads were anesthetized in a bath containing 0.5% MS-222 (Tricaine methanesulfonate; pH was adjusted to neutral with NaHCO₃) before pithing and tissue preparation. All procedures were conducted in strict accordance with the approved Swinburne University of Technology Animal Ethics Committee (AEC) protocols and adhere to the National Health and Medical Research Council (NHMRC) Code for the Care and Use of Animals for Scientific Purposes (2013).

Setup

Toads were removed from housing during the light cycle and remained under normal room lighting to facilitate light adaptation during the 30-min anesthesia and dissection protocol. Eyes were enucleated immediately following anesthesia, and the cornea and lens were removed. The remaining eyecup was supported by a custom-made wax mold in a bath of amphibian Ringer solution at room temperature (22–24°C). The Ringer solution, containing (in mM) 82.5 NaCl, 2 KCl, 27.5 NaHCO₃, 1 MgCl₂, 1.8 CaCl₂, and 10 glucose (Miller and Steinberg 1977), was oxygenated with 95% O₂-5% CO₂, resulting in a pH of ∼7.5.

Transtissue potential was recorded using Ag/AgCl electrodes with agar bridges (3% agar in 1 M KCl) on a Warner Dual Channel Epithelial Voltage Clamp (EC-825A). Analog voltage output (10× voltage amplification) was digitized at 1 kHz, notch-filtered (50 Hz), and low-pass filtered (20 Hz) using a PowerLab data acquisition system (ADInstruments) with LabChart software. Trigger outputs from the video stimulator marking the onset of stimulation blocks were recorded together with the electrophysiological recordings. Stimuli were presented using a DATAPixx video stimulator (VPixx Technologies) and a custom-made projection system consisting of a backlit liquid-crystal display (LCD) projector screen (8 × 5 cm, frame rate 60 Hz, native resolution 320 × 240) and Minolta camera lens (f/1.4, FL = 50 mm). The LCD screen, which was removed from an HDMI Portable Mini LED Projector (UC28 model), was driven by video graphics array (VGA) output from the DATAPixx and backlit with a DC light source. The screen was positioned at the top of a 50-cm length of polyvinyl chloride (PVC) tube that kept it at the hyperfocal distance from the lens. Lens focus was adjusted before start of each experiment such that when the eyecup was placed in the bath, the plane of focus fell at the level of the central retina.

Experimental Protocols

Data were collected across three experiments. The main experiment used a within-subjects design to measure the DC ERG response to drifting sawtooth, square-wave, and sine-wave gratings before and after drug delivery. Two additional experiments were conducted to test whether the response to drifting sawtooth diamonds was similar to that for drifting sawtooth gratings and to measure toad ERGs for temporally varied sawtooth flicker. The protocols for these experiments are outlined in Fig. 1.

Fig. 1.

Methods summary for drifting grating and diamond and temporal sawtooth flicker experiments.

Drifting grating experiment.

Sawtooth, sine-wave, and square-wave gratings (cropped examples in Fig. 2 of results) with an equal number of dark and light bars were created using VPixx (v.3.03). When projected onto the eyecup, the gratings covered a 5- × 4-mm patch (subtending 4.6 × 5.7° at the lens) with a spatial frequency of 1.53 cycles per degree and a mean luminance of 22 cd/m². The stimulus was surrounded by a rectangular black mask with a luminance of 0.41 cd/m². During a presentation sequence, stimuli drifted leftward for 2 s, remained stationary for 3 s, drifted rightward for 2 s, and then remained stationary for 3 s before the sequence was repeated. This horizontal drift was lined up with the long direction of the rectangular patch. Leftward and rightward drift of sawtooth gratings produced local fast-OFF and fast-ON temporal luminance profiles, respectively, with no change in net luminance of the stimulus. Preliminary experiments confirmed that flipping the luminance profile of the sawtooth stimulus sequence also reversed the amplitude of the ERG response (when recording sequentially from the same eyecups). This reversed stimulus was not included in subsequent experiments due to time constraints (recovery time was required following the flip). Sine- and square-wave stimuli produced a symmetric profile during leftward and rightward drift. Preliminary experiments confirmed that flipping the luminance profile had no effect on ERG responses to these symmetric stimuli.

Fig. 2.

Mean DC ERG responses (± SE) to directionally moving sawtooth gratings (A), sawtooth diamonds (B), square-wave gratings (C), and sine-wave gratings (D) drifting across the retina at 5 Hz. Each stimulus (cropped examples shown in insets) drifted leftward and rightward (resulting in fast-ON and fast-OFF temporal profiles for the sawtooth stimuli). Shading indicates drift offset and the beginning of the 3-s stationary period. Note that although the DC potential shift and drift offset peak were not strong at 5 Hz for sine-wave and fast-ON gratings, these waveforms become larger at higher temporal frequencies (see Fig. 3).

The presentation sequence described above was used to test the effects of grating profile and temporal frequency on ERG response amplitudes in 20 eyecups (see Fig. 1. for schematic flowchart of experimental protocol). Preliminary data collection indicated that experiments should be limited to 2 h per eyecup to ensure stable and healthy recordings. To keep within this time frame, sawtooth, sine-wave, and square-wave grating sequences were presented 7 times to each eyecup. Each presentation sequence was repeated across 5 drift frequencies (2.5, 5, 7.5, 15, and 20 Hz) with order counterbalanced between eyecups. This series of presentations formed a “run” lasting ∼20 min. Spatially uniform temporal square-wave flash ERGs were recorded at the beginning and end of each run to assess the ongoing integrity of the eyecup preparation.

Each eyecup was presented with three stimulus runs. Run 1 was recorded with Ringer superfusate for all eyecups (n = 20). Subsequently, eyecups were divided into three groups, and drugs were added to the bath to suppress cellular activity as follows. Postreceptoral ON pathway activity was suppressed by specific inhibition of the metabotropic glutamate receptor of ON-bipolar cells (mGluR6) using 1 mM 2-amino-4-phosphonobutric acid (APB) in Ringer (Jardon et al. 1989; Slaughter and Miller 1981). Postreceptoral OFF pathway activity was suppressed by inhibition of ionotropic glutamate receptors found on hyperpolarizing second-order (OFF-bipolar and horizontal) cells using 1 mM 2,3 cis-piperidine dicarboxylic acid (PDA) in Ringer (Jardon et al. 1989). Spiking amacrine and ganglion cell activity was suppressed using 2 or 10 μM tetrodotoxin (TTX) in Ringer (Awatramani et al. 2001). Drug dosage choices were based on the published amphibian studies referenced above. See Fig. 1 for eyecup numbers and the sequence of pharmacological suppression used in each condition.

Drifting diamond experiment.

The sawtooth perceptual illusion in humans occurs for both drifting gratings and drifting diamond plaids (Cavanagh and Anstis 1986; Watanabe et al. 1995). To test whether the DC ERG response to sawtooth gratings and diamond plaids is also similar, ERGs were recorded from 11 additional eyecups presented with shaded diamonds drifting at 5 Hz (see Fig. 1 for an outline of procedures and Fig. 2 for a cropped example of the diamond plaid stimulus). These recordings were conducted with Ringer superfusate only.

Temporal sawtooth flicker experiment.

Although the DC ERG response to drifting sawtooth gratings has not previously been examined in any species, many studies have investigated the retinal response to temporal sawtooth flashes (as noted in the Introduction and discussion). To provide a basis for comparison with these studies, ERG responses to 1-Hz spatially uniform fast-ON and fast-OFF sawtooth flicker were recorded from an additional 15 eyecups. As with the drifting grating experiment, these data were collected across multiple runs to allow pharmacological dissection (see Fig. 1 for details).

Data Analysis

Electrophysiological data were analyzed with LabVIEW (v.11.01; National Instruments). The preprocessing stage involved removing high-amplitude spiking interference and applying a 10-Hz digital low-pass filter. Low-frequency drift was observed across experimental runs [presumably resulting from changes in electrode junction potentials (Barry and Diamond 1970), slight variations in bath temperature, and declining tissue integrity over time]. To control for this slope in the signal that was not due to experimental conditions, the fourth-order polynomial regression fit line (chosen for goodness of fit) was subtracted from the entire 20-min trace. Starting from the onset of drift, 5-s epochs were extracted for each leftward and rightward drift sequence (2 s drift, 3 s stationary). Data for each condition were averaged relative to a baseline taken from the 10-ms stationary presentation preceding each drift onset trigger. Averaged epochs for each eyecup were downsampled to 100 Hz. As there were no significant differences in electrophysiological responses from leftward/rightward motions of symmetric sine- and square-wave gratings, these data were averaged for each eyecup.

RESULTS

Drifting Gratings and Diamonds

ERG responses to 5-Hz drift are illustrated in Fig. 2. In general, these grand mean ERGs revealed a sustained positive shift in transtissue DC potential during drift and a peak at drift offset. For both gratings and diamonds, the DC potential shift was larger for the fast-OFF sawtooth (perceptual brightening) than for the fast-ON sawtooth (perceptual darkening). Oscillations at the fundamental frequency of local flicker were apparent in the responses to gratings but not diamond plaids. Although both of the projected patterns were designed to be equated for light and dark regions, there can be no guarantee that all sections of the retina were uniformly responsive. The effects of a small imbalance in effective light and dark regions would be greater in gratings cf. diamonds due to their simpler spatiotemporal luminance pattern and to the fact that the orientation of the grating pattern was parallel to the edge of the stimulus. Two parameters were extracted for further analyses: the DC potential shift was defined as the average potential from 1 to 2 s postdrift onset and the drift offset peak as the trough-to-peak amplitude of the response at the cessation of stimulus motion.

The responses displayed in Fig. 3 illustrate mean DC shift (Fig. 3A) and drift offset peak potentials (Fig. 3B) for the grating profiles at each temporal frequency. As shown in Fig. 3A, the difference in DC potential shift between the two sawtooth modulations increased with temporal frequency up to 20 Hz. Over all temporal frequencies, the square-wave DC amplitude was consistently greater than that of the sine-wave profile and for frequencies >5 Hz was roughly parallel (Fig. 3A). However, the sawtooth profiles showed lesser slope with temporal frequency compared with either square- or sine-wave stimulation.

Fig. 3.

Effects of grating profile and temporal frequency (2.5, 5, 7.5, 15, and 20 Hz) on the sustained potential increase during drift and the peak amplitude at drift offset. A: mean ERG amplitudes (± SE) for the 1–2 s epoch following drift onset for all grating profiles across temporal frequencies (epoch time frame indicated relative to an example response at 5 Hz in inset figure). B: mean trough-peak amplitude (± SE) for the drift offset peak (example offset peak indicated in 5-Hz inset figure).

A 2 (sawtooth profile) × 5 (frequency) within-groups ANOVA confirmed that across temporal frequency, the sustained DC potential during drift was greater for the fast-OFF than fast-ON sawtooth [F(1, 17) = 34.29, P <0.001, partial η² = 0.67]. The sawtooth profile by frequency interaction shows that the effect of temporal frequency on the DC potential was significantly different for the fast-OFF and fast-ON profiles [F(4,68) = 21.32, P <0.001, partial η² = 0.56]. Although we were specifically interested in the DC potential shift during drift (corresponding to the time frame when illusory brightness has been demonstrated in humans), we observed a similar pattern of results for the drift offset peak (Fig. 3B). As with the DC potential, both the main effect of temporal frequency [F(1, 17) = 45.11, P < 0.001, partial η² = 0.73] and the sawtooth profile by frequency interaction [F(4, 68) = 14.59, P < 0.001, partial η² = 0.46] were significant.

It should be noted that the frame-based display used here affected the stimulus in a temporally dependent manner. As the temporal frequency increases, there are progressively fewer luminance steps covered across one cycle of a moving grating. Thus at the 60-Hz frame rate used here and drift frequencies of 7.5, 15, and 20 Hz, a single photoreceptor will receive one cycle of the stimulus in eight, four, and three steps of luminance, respectively. For the sawtooth stimuli, this has the unusual effect that the proportion of increments and decrements occurring on each frame across the retinal photoreceptors becomes more equal as the drift frequency increases (i.e., as the number of frames per cycle of the pattern decreases).

The effects of temporal frequency of stimulation on the different stimulus waveforms were modeled using LabVIEW software. Sample waves (square, sine, and sawtooth) were defined over 144 sample points for a single cycle. Whereas the linear estimates of brightening and darkening are 0 for all waves (corresponding to the 0 sum of deviations away from the mean over a full cycle), a nonlinear estimate based on the root mean square (RMS) of the changes for each frame of stimulation showed generally increasing values with increasing temporal frequency over the range of 2–20 Hz (Fig. 4). There is a strong qualitative similarity between the empirical data (Fig. 3) and the model (Fig. 4), particularly with respect to the lower slope with temporal frequency for sawtooth compared with square or sine wave stimulation. The square-wave showed a greater RMS effect than either the sine or the sawtooth waves, whereas the RMS effects of the sine and sawtooth waves cross over at ∼10 Hz (at approximately the same point as the sine and the average of the fast-ON and fast-OFF sawtooth waves in the empirical data). The outstanding difference, not predicted in some form by the model, is the difference between the two sawtooth profiles. Although there is a clear empirical difference, the model prediction is identical for the fast-ON and fast-OFF sawtooth RMS effects as a function of temporal frequency.

Fig. 4.

Modeling the effects of alteration in temporal contrast on the root-mean-square (RMS) temporal contrast per point. For square, sine, and sawtooth waveforms, each defined on 144 points per cycle, the effects of increasing temporal frequency of drift were modeled by estimating the RMS difference per stimulus frame over a spatial cycle of each wave. The 2 sawtooth polarities (fast-ON and fast-OFF) produced identical results (shown as sawtooth).

The cellular basis of differences in response to the two sawtooth grating polarities was further investigated by addition of TTX, APB, and PDA to the bath superfusate. Flash ERGs recorded at the beginning and end of each stimulus block confirmed the expected elimination of the b-wave/ON-bipolar response by APB and a substantial but not complete suppression of the OFF-bipolar response with PDA (Hare and Ton 2002; Stockton and Slaughter 1989). Figure 5 illustrates the effects of pharmacological manipulation on the response to 15-Hz fast-ON and fast-OFF sawtooth gratings. As shown in this figure, TTX and PDA did not affect the DC shift during drift or peak at drift offset for either sawtooth polarity (P > 0.05 for all paired-samples t-tests). Application of APB, however, increased the DC shift [fast-ON: t(5) = 10.74, P < 0.001; fast-OFF: t(5) = 7.04, P = 0.001] and completely abolished the peak at drift offset for both fast-OFF [t(5) = 6.36, P = 0.001] and fast-ON [t(5) = 6.18, P = 0.002] gratings [this effect is evident in APB only (red trace) and APB+PDA (purple trace) responses]. Importantly, as illustrated in Fig. 5C, the DC shift amplitudes for fast-OFF and fast-ON gratings remained significantly different following APB delivery [t(5) = 4.01, P = 0.010].

Fig. 5.

Effects of pharmacological suppression on mean DC ERG responses (± SE) to directionally moving fast-OFF (A) and fast-ON (B) sawtooth gratings drifting at 15 Hz. C shows the difference wave (fast-OFF minus fast-ON) for each drug condition. Each set of 3 graphs (fast-OFF, fast-ON, and difference) explores responses collected from the same eyecups before and after drug delivery. For TTX eyecups, superfusate changes were as follows: run 1 = Ringer, run 2 = 2 μM TTX, and run 3 = 10 μM TTX. For APB eyecups: run 1 = Ringer, run 2 = 1 mM APB, and run 3 = 1 mM APB + 1 mM PDA. For PDA eyecups: run 1 = Ringer, run 2 = 1 mM PDA, and run 3 = 1 mM APB + 1 mM PDA.

This pattern of responses was consistent across all of the temporal frequencies tested. Figure 6 illustrates the mean DC shift (Fig. 6A) and drift offset peak (Fig. 6B) potentials for fast-OFF and fast-ON sawtooth grating profiles drifting at 2.5–20 Hz. Separate 2 (sawtooth profile) by 2 (drug delivery) by 5 (drift frequency) repeated-measures ANOVAs were performed to test the effects of TTX, PDA, and APB on the DC potential shift and the peak after drift offset. Neither TTX nor PDA had a significant main effect on the DC potential shift or peak at drift offset (P > 0.05). Furthermore, there were no significant interactions between drug delivery and sawtooth polarity on the DC shift or peak at drift offset for either PDA or TTX (P > 0.05). Application of APB enhanced the DC shift during drift [main effect: F(1, 5) = 33.85, P = 0.002, partial η² = 0.87] and abolished the peak at drift offset for both waveforms across temporal frequency [main effect: F(1, 5) = 23.05, P = 0.005, partial η² = 0.82]. The significant drug by sawtooth profile interaction indicates that APB has a greater effect on the DC potential shift for fast-ON than fast-OFF waveforms [F(1, 5) = 76.01, P < 0.001, partial η² = 0.94].

Fig. 6.

Effects of pharmacological suppression on the DC potential increase during drift (A) and the peak amplitude at drift offset (B) for sawtooth gratings across temporal frequency. Mean ERG amplitudes (± SE) for the 1- to 2-s epoch following onset of leftward drift (fast-OFF) and rightward drift (fast-ON) across temporal frequencies (2.5, 5, 7.5, 15, and 20) are shown in A. Mean trough-peak amplitude (± SE) for the drift offset peak are shown in B. The final line of graphs in both A and B shows the difference (fast-OFF minus fast-ON) for each drug condition. Each set of 6 graphs (fast-OFF, fast-ON, and difference in A and B) explores responses collected from the same set of eyecups before and after drug delivery. For TTX eyecups, superfusate changes were as follows: run 1 = Ringer, run 2 = 2 μM TTX, and run 3 = 10 μM TTX. For APB eyecups: run 1 = Ringer, run 2 = 1 mM APB, and run 3 = 1 mM APB + 1 mM PDA. For PDA eyecups: run 1 = Ringer, run 2 = 1 mM PDA, and run 3 = 1 mM APB + 1 mM PDA. For an example of epoch time frames relative to original waveforms, see Fig. 3.

Similar drug effects occurred for square-wave and sine-wave gratings (data not shown). It should also be noted that there was some variation in baseline ERG response amplitudes across the three drug groups (i.e., mean Ringer traces in TTX, APB, and PDA graphs of Figs. 5–6). These small variations in the quality of response obtained from each set of eyecups are not expected to affect the results given that all statistical analyses used a within-subjects design.

Temporal Sawtooth Flicker

Figure 7 illustrates ERG responses to 1-Hz temporal sawtooth flicker before and after addition of TTX, APB, and PDA to the bath. Although TTX can alter the timing and magnitude of the square-wave flash ERG b-wave (Dong and Hare 2000; Hare and Ton 2002), it caused no obvious changes to the sawtooth flicker waveforms recorded here. By comparison, APB abolished the ON response and enhanced the OFF response to both fast-ON and fast-OFF sawtooth. PDA effects were weaker, primarily resulting in a suppression of the OFF response to both sawtooth profiles.

Fig. 7.

Mean DC ERG responses to 1-Hz spatially uniform fast-ON and fast-OFF sawtooth flicker in Ringer and following pharmacological suppression with TTX, APB, PDA, or APB+PDA. As with the drifting grating data, these full-field flicker responses were recorded over successive runs. Run 1 was recorded with Ringer. Run 2 was recorded following pharmacological suppression with 2 μM TTX, 1 mM APB, or 1 mM PDA. For eyecups with APB or PDA in run 2, a 3rd run was recorded following suppression of all postreceptoral activity with 1 mM APB + 1 mM PDA.

DISCUSSION

This study demonstrated that the DC ERG response to fast-ON and fast-OFF drifting sawtooth gratings is asymmetric. At temporal frequencies >5 Hz, all grating and diamond profiles elicited a sustained positive DC potential shift during drift and a peak at drift offset. These waveforms were also evident at 2.5 and 5 Hz for all stimuli except sine-wave and fast-ON gratings, which showed weak responses at low temporal frequencies. Consistent with the temporal sawtooth flicker literature (Alexander et al. 2001; Barnes et al. 2002; Dryja et al. 2005; Khan et al. 2005; Kremers 2013; Kremers et al. 1993; Pangeni and Kremers 2013; Pangeni et al. 2012; Rodrigues et al. 2010; Vukmanic et al. 2014) and brightness perception measures in humans (Cavanagh and Anstis 1986; Watanabe et al. 1995), the DC potential shift for spatiotemporal fast-OFF sawtooth was larger than that for fast-ON sawtooth across all of the conditions tested (shaded gratings drifting at 2.5–20 Hz and shaded diamonds drifting at 5 Hz).

Modeling suggested that differences in root-mean-square (RMS) temporal luminance contrast resulting from the stimulus device frame rate may have contributed to ERG response patterns for drifting stimuli, particularly when considering the effects of temporal frequency. This modeling shows that as temporal frequency increases, so does the RMS temporal contrast and in a way that depends on the particular grating, sine versus square versus sawtooth. These effects are quite general for all frame-based stimuli involving moving spatiotemporal patterns, including those used in previous studies of the drifting sawtooth illusion (Ashida and Scott-Samuel 2014; Cavanagh and Anstis 1986). Importantly, RMS temporal luminance contrast was identical for the two directions of sawtooth drift at any given temporal frequency. Thus the DC ERG responses recorded for these stimuli reflect a physiological asymmetry in the encoding of fast-ON and fast-OFF spatiotemporal sawtooth by toad retina.

Pharmacological dissection of ERG responses to drifting sawtooth gratings suggested an ON-bipolar source for the drift offset peak. The DC potential during drift appears to have more complex origins. The amplitude of this potential was not altered by TTX, which inhibits sodium-based action potentials of spiking amacrine and ganglion cells (Narahashi et al. 1964). It was also unaltered by PDA, an antagonist of ionotropic glutamate receptors on OFF-bipolar and horizontal cells and kainite receptors in the inner retina. In contrast, mGluR6 agonist APB increased the DC shift, indicating that the ON-bipolars make an antagonistic contribution to this waveform. This increase in the DC potential following APB occurred for both fast-ON and fast-OFF profiles. Thus, if we compare fast-ON and fast-OFF responses following APB or APB+PDA delivery, they remain significantly different. This suggests that, although the ON-bipolars contribute to the DC potential, the asymmetry originates in the outer retina.

To our knowledge, this is the first demonstration of a larger outer retinal response to fast-OFF versus fast-ON spatiotemporal sawtooth. This finding builds on an extensive cross-species literature demonstrating that the retinal response to temporal fast-OFF and fast-ON sawtooth is asymmetric (Alexander et al. 2001; Barnes et al. 2002; Dryja et al. 2005; Khan et al. 2005; Kremers 2013; Kremers et al. 1993; Pangeni and Kremers 2013; Pangeni et al. 2012; Rodrigues et al. 2010; Vukmanic et al. 2014). Within this temporal domain, the toad ERG response appears comparable to that of humans and monkeys, where the fast phase of fast-ON flicker generally elicits slightly larger ON-bipolar responses relative to the combined photoreceptor and OFF-bipolar response elicited by fast-OFF flicker (Alexander et al. 2001; Barnes et al. 2002; Khan et al. 2005; Kremers 2013; Pangeni and Kremers 2013; Vukmanic et al. 2014). Moreover, like in monkey (Khan et al. 2005), our pharmacological dissection of the toad temporal sawtooth ERG (Fig. 7) revealed push-pull interactions between ON- and OFF-bipolar cell contributions to both fast-ON and fast-OFF waveforms. In this respect, our pharmacological results for temporal and spatiotemporal sawtooth profiles are somewhat similar, as APB increased the DC potential shift in response to drifting fast-ON and fast-OFF gratings (Figs. 5 and 6A) and substantially increased the amplitude of OFF responses to both fast-ON and fast-OFF temporal flicker (Fig. 7).

It should be noted that this study is also the first, to our knowledge, to dissect the DC motion ERG pharmacologically. The only previous studies to investigate ERG responses to sustained motion (i.e., >500 ms) were conducted using square-wave stimuli in humans (Dodt and Kuba 1995; Korth et al. 2000). These studies identified an initial fast positive waveform [not seen in toad, where temporal resolution is slow (Donner et al. 1990)] followed by a sustained potential with similarities to the DC shift identified in the present study. As noted above, our pharmacological findings suggest that this sustained waveform primarily reflects photoreceptoral and ON-bipolar activity. Also similar to the present study, Korth et al. (2000) identified a peak following motion offset in human motion ERGs that appeared related to the pattern onset response. Our pharmacological results suggest that this peak could be generated by the ON-bipolars rather than cells in the proximal retina.

In summary, this study aimed to characterize the retinal response to spatiotemporal sawtooth stimuli and to clarify the neural basis of the perceptual effects that such stimuli elicit in humans. We demonstrated pharmacologically that the response to drifting fast-OFF and fast-ON sawtooth gratings is asymmetric at the photoreceptoral level in toad. Further comparison of our findings with published AC motion ERG and sawtooth flicker studies demonstrated that the toad response to both temporal and spatiotemporal stimuli shares many similarities with that of humans and monkeys. Together, these findings point to an outer retinal origin for the perceptual brightness illusion generated by drifting sawtooth stimuli.

GRANTS

This research was supported by Australia Research Council Grant DP110103784 to D. P. Crewther and S. G. Crewther.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

N.R., L.H., S.G.C., and D.P.C. conception and design of research; N.R., L.H., and J.J. performed experiments; N.R., L.H., and D.P.C. analyzed data; N.R., L.H., J.J., S.G.C., and D.P.C. interpreted results of experiments; N.R., L.H., and D.P.C. prepared figures; N.R. and L.H. drafted manuscript; N.R., L.H., J.J., S.G.C., and D.P.C. edited and revised manuscript; N.R., L.H., J.J., S.G.C., and D.P.C. approved final version of manuscript.