Visual feature tuning properties of short-latency stimulus-driven ocular position drift responses during gaze fixation

Ocular position drifts during gaze fixation are generally considered to be random walks. However, we recently identified a short-latency ocular position drift response, of approximately 1 min arc amplitude, that is triggered within <100 ms by visual onsets. This systematic eye movement response is feature-tuned and seems to be coordinated with a simultaneous resetting of the saccadic system by visual stimuli. However, much remains to be learned about the drift response, especially for designing better-informed neurophysiological experiments unraveling its mechanistic substrates. Here we systematically tested multiple new feature tuning properties of drift responses. Using highly precise eye tracking in three male rhesus macaque monkeys, we found that drift responses still occur for tiny foveal visual stimuli. Moreover, the responses exhibit size tuning, scaling their amplitude as a function of stimulus size, and they also possess a monotonically increasing contrast sensitivity curve. Importantly, short-latency drift responses still occur for small peripheral visual targets, which additionally introduce spatially-directed modulations in drift trajectories towards the appearing peripheral stimuli. Drift responses also remain predominantly upward even for stimuli exclusively located in the lower visual field, and even when starting gaze position is upward. When we checked the timing of drift responses, we found that it was better synchronized to stimulus-induced saccadic inhibition timing than to stimulus onset. These results, along with a suppression of drift response amplitudes by peri-stimulus saccades, suggest that drift responses reflect the rapid impacts of short-latency and feature-tuned visual neural activity on final oculomotor control circuitry in the brain. Significance During gaze fixation, the eye drifts slowly in between microsaccades. While eye position drifts are generally considered to be random eye movements, we recently found that they are modulated with very short latencies by some stimulus onsets. Here we characterized the feature-tuning properties of such stimulus-driven drift responses. Our results demonstrate that drift eye movements are not random, and that visual stimuli can impact them in a manner similar to how such stimuli impact microsaccades.


Introduc+on
The eye is never completely s?ll during gaze fixa?on (Barlow, 1952;Steinman et al., 1967;Steinman et al., 1973), resul?ng in subtle, but con?nuous, altera?ons of the re?nal image streams entering the visual system.Two primary components of fixa?onal eye movements are microsaccades and slow ocular posi?on driXs (Fig. 1A).While the neural control of microsaccades is rela?vely well established (Krauzlis et al., 2017;Hafed et al., 2021a), that of ocular posi?on driXs is less understood.Moreover, the ways with which external sensory transients interact with these two types of eye movements are not fully inves?gated.For microsaccades, visual transients in the environment rapidly reset the oculomotor rhythm, causing microsaccadic inhibi?on (Engbert and Kliegl, 2003;Hafed et al., 2021b;Buonocore and Hafed, 2023), and giving rise to important implica?ons on subsequent perceptual performance and visual neural sensi?vity (Hafed et al., 2015).Moreover, such inhibi?on is feature-tuned, altering its ?me course and strength as a func?on of the appearing visual pa8erns (Khademi et al., 2023).This likely reflects the tuning proper?es of visually-sensi?veneurons media?ng microsaccadic inhibi?on (Buonocore and Hafed, 2023).For driXs, we recently found that certain visual s?muli robustly trigger a short-latency change in driX sta?s?cs, which we refer to here as the driX response (Malevich et al., 2020).This response is characterized by a small predominantly upward displacement, superseding the ongoing driX direc?on, and being much slower than even the slowest microsaccades.For example, in Fig. 1B, aligning all eye posi?on epochs at the ?me of s?mulus onset reveals a predominantly rightward driX trajectory prior to s?mulus onset; this rightward driX was momentarily transformed into a predominantly upward driX pulse within less than 100 ms aXer s?mulus onset, with an even smaller downward component just prior to that (Fig. 1B,   C) (Malevich et al., 2020).Our previous work revealed that the driX response occurred when we presented rela?vely large s?muli (Malevich et al., 2020).We also found that this driX response, much like saccadic inhibi?on (Khademi et al., 2023), is feature-tuned.Specifically, it was stronger for low spa?al frequency pa8erns, as well as for certain gra?ng orienta?ons (Malevich et al., 2020).However, understanding the full mechanisms underlying the driX response requires much deeper characteriza?on of this response's func?onal proper?es.For example, might such a driX response s?ll occur for small visual s?muli, just like microsaccades can be affected by small eccentric targets (Hafed and Clark, 2002;Engbert and Kliegl, 2003)?And, would the predominantly upward nature of the driX response change if we only presented lower visual field s?muli rather than s?muli spanning both sides of the re?notopic horizon?(2) the eye driBs con-nuously with slow speeds in between saccades and microsaccades (black).(B) We recently found (Malevich et al., 2020) that large s-mulus onsets result in a short-latency change in ocular posi-on driB sta-s-cs, primarily marked by a small upward devia-on in eye posi-on (although an earlier, even smaller, downward movement component jumpstarts the whole response sequence).The figure shows average horizontal and ver-cal eye posi-ons (surrounded by SEM ranges; n = 882 trials) from an example condi-on and an example monkey (A) from Experiment 1 of the current study, replica-ng (Malevich et al., 2020).Posi-ve deflec-ons in each curve indicate rightward and upward eye posi-on devia-ons, respec-vely, and the data across trials were first aligned to eye posi-on at -me zero before averaging (Malevich et al., 2020) (Materials and Methods).As can be seen, the monkey exhibited rightward pre-s-mulus driBs; aBer s-mulus onset, there was a predominantly upward driB response, which was accompanied by a small leBward component to it.The upward driB response was also preceded by a much smaller and shorter-lived downward eye posi-on devia-on, although we primarily focus here on the overall upward nature of the whole response sequence.(C) Horizontal and ver-cal eye velocity curves (surrounded by SEM ranges) from the same trials as in B. The s-mulus-driven driB response was predominantly upward.Shaded regions on the x-axis indicate our measurement intervals of baseline (pres-mulus) and post-s-mulus eye veloci-es, for use in our summary sta-s-cs in the remainder of this ar-cle.
Here we answered these, and other, ques?ons, and we laid down a rich founda?on for tes?ng the neurophysiological underpinnings of not only the driX response, but also of the coordina?onbetween mul?ple types of fixa?onal and targe?ng eye movements with external sensory events.We first found that the driX response is size-tuned, and can s?ll happen for ?ny, foveal visual s?muli.We also characterized the contrast sensi?vity of the driX response, as well as its modula?on by small peripheral visual targets.Interes?ngly, and unlike our expecta?on(Malevich et al., 2020) that the driX response might reflect the preference of the superior colliculus (SC) for the upper visual field (Hafed and Chen, 2016;Fracasso et al., 2023), we found that the driX response is s?ll predominantly upward even for s?muli below the horizon.Finally, we characterized the temporal coordina?on between microsaccades and the driX response, as well as the altera?on of the driX response magnitude by the occurrence of peri-s?mulus microsaccades, mimicking the classic phenomenon of saccadic suppression (Zuber and Stark, 1966;Beeler, 1967;Hafed and Krauzlis, 2010;Idrees et al., 2020).Our results demonstrate that the "lens" through which the oculomotor system processes visual scenes may be similar for dicta?ng the visual feature tuning proper?es of both saccadic inhibi?on (Khademi et al., 2023) and driX responses, and that these two ubiquitous eye movement phenomena likely arise from a common underlying source.

Experimental animals and ethical approvals
We collected data from three adult, male rhesus macaque monkeys (macaca mula'a), referred to here as A, F, and M, respec?vely.The monkeys were aged 7-14 years, and they weighed 9.5-12.5 kg.All experiments were approved by ethics commi8ees at the regional governmental offices of the city of Tübingen.

Laboratory setup and animal procedures
Some experiments involved analysis of ocular posi?on driXs from our recent study, which only focused on saccades (Khademi et al., 2023).Other experiments were run specifically for the purposes of the current study, but in the same experimental setups as in (Khademi et al., 2023).The reader is referred to our recent publica?on for details on our laboratory equipment (Khademi et al., 2023).Briefly, we used precise eye tracking, using the scleral search coil technique (Robinson, 1963;Fuchs and Robinson, 1966;Judge et al., 1980), and a real-?me experimental control system based on PLDAPS (Eastman and Huk, 2012) and the Psychophysics Toolbox (Brainard, 1997;Pelli, 1997;Kleiner et al., 2007).The monkeys had their heads stabilized during the experiments, and they watched s?muli on a computercontrolled display in front of them.The display size was spanning approximately 31 deg horizontally and 23 deg ver?cally, and the experimental room was otherwise dark.

Experimental procedures
The experiments all involved gaze fixa?on, and we analyzed fixa?onal eye movements.The experimental procedures were described in detail recently (Khademi et al., 2023).In brief, the monkeys fixated a small, sta?onary fixa?on spot presented over a gray background (of luminance 26.11 or 36.5 cd/m 2 ).At a random ?me during fixa?on, a single-frame flash (~12 or ~7-8 ms) was presented.Across trials and experiments, the flash could have different feature proper?es (for example, full-screen flash or small, localized target, and so on).In what follows, we describe the experiment-specific details, explaining what image features the brief flashes had in the different experiments.Experiment 1: Size tuning This experiment was the same as that used recently (Khademi et al., 2023).In that study, we analyzed the saccades that took place around s?mulus onset.In the current study, we analyzed ocular posi?on driXs (in saccade-free epochs), as well as saccade-driX interac?ons, as we describe in more detail below.The s?mulus flash in this experiment consisted of a black circle of different radii across trials.The range of sizes tested included s?muli approximately as small as the fixa?on spot (0.09 deg radius), s?muli approximately as large as the en?re display (9.12 deg radius), and s?muli with sizes in between these two extremes.Moreover, the numbers of trials collected were the same as those reported in (Khademi et al., 2023).
For the numbers of trials that were analyzed, these depended on whether we picked driX response trials (saccade-free) or trials with peri-s?mulus microsaccades (see Data Analysis below for details).For example, as we describe in more detail below, for some analyses, we only considered trials in which there were no microsaccades in the interval from -100 ms to 200 ms rela?ve to s?mulus onset, and in some other analyses, we considered trials with microsaccades happening in the final 100 ms before s?mulus onset, and so on.That is why we document the specific numbers of trials included in the analyses of each figure shown in Results separately.Experiment 2: Contrast sensi9vity with full-screen s9muli This experiment was again the same as that used recently (Khademi et al., 2023).Briefly, the s?mulus onset could be a full-screen flash having one of five different Weber contrasts (5%, 10%, 20%, 40%, or 80%).Once again, we analyzed saccade-free driX response trials as well as trials having saccades within specific ?meintervals rela?ve to s?mulus onset (see Data Analysis below for more details).For each analysis, the numbers of trials included are documented individually in Results.DriX-only (saccade-free) trials were not analyzed previously in (Khademi et al., 2023).

Experiment 3: Upper and lower visual field s9muli
This experiment was collected specifically for this study (as well as related ongoing neurophysiological experiments).The general trial sequence was the same as that in the above two experiments.Specifically, the monkeys fixated a central spot.AXer a random ?me, one of five different events took place, depending on the trial type.The first trial type was just a sham condi?on: no s?mulus display update occurred at all, but we just used the sham event in the data file to study baseline driX trajectories and compare them to trajectories with a real s?mulus.The second trial type had the s?mulus being a 1 deg x 1 deg black square that was flashed for a single display frame.The loca?on of the flash was somewhere in the periphery rela?ve to the central fixa?on spot (approximately 3.5-11 deg), but this loca?on was constant within a given session.This loca?on was typically dictated by the loca?ons of recep?ve fields of neurons that we were recording simultaneously for other purposes, since this task was typically run while we recorded SC and/or primary visual cortex ac?vity.The third trial type was a 100% black full-screen flash (again with a dura?on of a single frame).Here, the s?mulus was basically similar to the s?muli used in Experiment 2 above.And, finally, the fourth and fiXh trial types were half-screen flashes.Specifically, we split the screen in half along the ver?cal dimension.In one condi?on, the flash was only in the upper half of the screen (above the midline defined by the ver?cal posi?on of the fixa?on spot), and in another condi?on, the flash was only in the lower half of the screen.We typically ran this task in daily blocks of approximately 100-500 trials per session, and we collected a total of 7524, 7521, and 7495 trials in monkeys A, F, and M, respec?vely.This resulted in 72-1208 trials per condi?on per animal for the saccade-free driX response analyses (like in Fig. 1B, C).Experiment 4: Small, localized s9muli across different visual field direc9ons Because the loca?ons of the small s?muli used in Experiment 3 were dictated by other experimental constraints (such as recep?ve field loca?ons), we ran an addi?onal experiment in which we sampled eccentric loca?ons more evenly.Specifically, the experiment consisted of the transient flash being a 1 deg x 1 deg black square at a 7.9 deg eccentricity from the display center.The square could appear in one of 8 equally spaced direc?ons, thus covering both right and leX as well as up and down visual field loca?ons.The flash loca?on was randomly interleaved across trials.We typically ran this task in daily blocks of 310-900 trials per session, and we collected a total of 5961, 4357, and 6048 trials in monkeys A, F, and M, respec?vely.This resulted in 65-383 analyzed trials per loca?on per animal for the basic saccade-free driX response analyses.We typically pooled mul?ple loca?ons for a given analysis, as we describe below, in order to increase sta?s?cal confidence in the results.Once again, all numbers of trials are documented in appropriate sec?ons of Results.Experiment 5: Gaze posi9on This task was the same as that in Experiment 2 above, with only one difference.Across sessions, the fixa?on spot could be at 4 deg to the right, leX, up, and down rela?ve to the display center.This task, therefore, allowed us to test whether the driX response (Fig. 1B, C) was substan?ally different if the star?ng gaze posi?on of the eye was different.We ran 4 sessions of this task in monkey A, collec?ng a total of 2206 trials.This resulted in 500-602 analyzed trials per eye posi?on for the basic saccade-free driX response analyses.

Data analysis
All saccades were analyzed as described recently (Khademi et al., 2023).Briefly, we detected saccades of all sizes using our established methods (Chen and Hafed, 2013;Bellet et al., 2019), and we included all detected saccades that took place around s?mulus onset.This allowed us to es?mate saccadic inhibi?on latency using the L50 parameter (Reingold andStampe, 2002, 2004;Rolfs et al., 2008;Khademi et al., 2023).Simply put, this parameter describes when the saccade rate curve drops by 50% of the dynamic range between pres?mulus (baseline) saccade rate and the minimum saccade rate during saccadic inhibi?on.The reader is referred to our detailed descrip?on of this parameter in (Khademi et al., 2023).We es?mated saccade rate using the method described in (Khademi et al., 2023): briefly, we calculated saccade onset likelihood within 50 ms moving windows that were stepped in ?me by 1 ms steps, and we did this on a per-trial basis; across-trial average rates were then obtained in order to calculate L50 from the global saccade rate.While we acknowledge that there might be other means to es?mate the latency of saccadic inhibi?on (Bompas et al., 2023), we used L50 because of its consistent use in other studies (Reingold andStampe, 2002, 2004;Rolfs et al., 2008;Khademi et al., 2023), and also because it does a good job in capturing the drop in saccade likelihood across condi?ons (see, for example, Fig. 7 later in Results).To visualize driX responses, we averaged the horizontal and ver?cal eye posi?on traces of a given animal and condi?on across trial repe??ons.Before such averaging, we realigned each trace to the posi?on of the eye at the ?me of s?mulus onset (Malevich et al., 2020).This allowed us to isolate visualiza?on of the driX sta?s?cs despite varia?ons in absolute eye posi?on at the ?me of s?mulus onset, due to con?nuous fixa?onal eye movements.We also visualized driX responses by plodng ver?cal eye velocity traces (e.g.Fig. 1C).We obtained these traces using a smooth differen?a?ng filter (Chen and Hafed, 2013;Malevich et al., 2020) applied to ver?cal eye posi?on on a trial-by-trial basis.We then averaged the individual trial velocity traces.For all analyses characterizing the driX response, we only picked trials without any saccades in the interval from -100 ms to 200 ms rela?ve to s?mulus onset.This was done for two reasons: to avoid masking the slow driX responses by large velocity pulses associated with saccades, and to avoid poten?al peri-saccadic modula?ons in the driX response strength.In some analyses, we specifically wanted to study such peri-saccadic modula?ons, as well as driX-saccade interac?ons in general.In that case, we replaced all velocity samples that were part of a saccade with not-a-number (NaN) labels before averaging the eye velocity traces across trials.For summary sta?s?cs, we es?mated the size of the driX response by calcula?ng average ver?cal eye velocity in a post-s?mulusresponse interval (70-150 ms; second gray interval on the x-axis in Fig. 1C) and subtrac?ng from it the baseline ver?cal eye velocity in a pres?mulus interval (first gray interval on the x-axis in Fig. 1C).We did this on a trial-by-trial basis, and we then averaged the difference measures across trials for popula?onsta?s?cs.Note that this velocity difference measure could quan?ta?vely be nega?ve,especially in the cases with weak or non-existent driX responses (Malevich et al., 2020).Note also that we picked the post-s?mulusresponse interval (70-150 ms) by inspec?ng driX responses across many different trials, condi?ons, and animals.While this interval was fixed for all analyses, it was long enough to avoid biasing our results in the cases in which the driX response was rendered a bit earlier or a bit later by specific visual feature dimensions.For analyzing the impacts of peri-s?mulus saccades on the driX response, we calculated the response strength measure just described above but now only for trials in which saccade onsets occurred within a specific ?mewindow rela?ve to s?mulus onset.This ?me window was defined by the purposes of the specific analysis (see Results).Finally, for analyzing effects of localized flash loca?ons on driX responses, we some?mes also measured eye posi?on rather than eye velocity.In this case, we grouped trials according to whether a flash was in the right or leX visual field (independent of its ver?cal posi?on), and we took the difference in eye posi?on (aXer aligning all traces at ?me zero like above) between the two groups of trials in a given post-s?mulusinterval.Similarly, we also grouped trials according to whether a flash was in the upper or lower visual field (independent of its horizontal posi?on), and we took the difference in eye posi?on between the two groups of trials (again, aXer all traces were aligned at the ?me of s?mulus onset, like described above).Using eye posi?on instead of eye velocity in these par?cular analyses allowed us to directly test whether there were spa?ally-directed modula?ons in driX sta?s?cs that were caused by eccentric s?mulus onsets (see Results), similar to how eccentric s?mulus onsets can bias microsaccade direc?ons (Hafed and Clark, 2002;Engbert and Kliegl, 2003).

Experimental design and sta9s9cal analyses
We always replicated all of our results in three monkeys (except for Experiment 5; see jus?fica?on below).Moreover, within each animal, we typically had hundreds to thousands of trial repe??ons per condi?on (see, for example, Fig. 1).This increased our confidence in our popula?onmeasures.Our choice of trial numbers to collect was guided by calcula?ng power es?mates before and during the experimental phases of the study.We also randomly interleaved all condi?ons in a given experiment, except when we were constrained by the experimental setup.For example, in Experiment 3, the loca?on of the small, localized flashes was constant within a given session, and this was dictated by other factors external to the study (like recep?ve field loca?ons).However, given the reflexive nature of our driX responses (see Results and Discussion), this should not have affected our interpreta?ons in any substan?almanner.More importantly, we also designed Experiment 4 with randomly interleaved target loca?onsexactly to compensate for the non-random nature of localized flash loca?ons in Experiment 3.For Experiment 5, we only ran it in one monkey.However, the results were virtually iden?cal, in a qualita?ve sense, to everything else that we had tested with the other two animals in other experiments.As a result, we decided that our conclusions from this experiment were already convincing.Similarly, we blocked gaze posi?on in this experiment, meaning that we tested each gaze posi?on condi?on in a block of con?guous trials (as opposed to randomly changing gaze posi?on from trial to trial).Again, this provided a stronger support for our conclusions that the driX response remains to be predominantly upward independent of gaze posi?on (see Results).All sta?s?cal tests and outcomes, as well as trial repe??on counts, are detailed in Results.We also performed sta?s?cal tests for each animal separately.

Results
We recently found that ocular posi?on driXs can be quite sensi?ve to visual s?mulus onsets, exhibi?ng short-latency, brief responses (Fig. 1) (Malevich et al., 2020).Here, we performed extensive addi?onal experiments characterizing the feature tuning proper?es of such s?mulus-driven driX responses.We used three rhesus macaque monkeys as our experimental subjects, and we did so for at least four reasons.First, we employed highly precise eye tracking in these animals, using the scleral search coil technique (Robinson, 1963;Fuchs and Robinson, 1966;Judge et al., 1980), to increase our confidence in the measurements.Commercial video-based eye trackers commonly used with human subjects would make measuring these ?ny driX responses very challenging (Wya8, 2010;Kimmel et al., 2012;Chen and Hafed, 2013;Choe et al., 2016;Malevich et al., 2020).Second, we could collect several experimental sessions per animal per condi?on, resul?ng in many trial repe??ons and sta?s?cally robust results across all of our experimental condi?ons (Materials and Methods).Third, these animals were already used in our characteriza?on of the closely related phenomenon of saccadic inhibi?on (Khademi et al., 2023), and we oXen used the very same data for characterizing driX responses here.Fourth, and most importantly, these animals are part of the ongoing efforts in our laboratory to explore the neurophysiological underpinnings of driX responses, which we hope to document in the near future.

The dri9 response exhibits size tuning
In our first experiment, we asked whether the ocular posi?on driX response is parametrically tuned to the size of the appearing visual s?mulus.In our ini?al characteriza?on of the driX response (Malevich et al., 2020), we mostly used large visual s?muli (full or half of our experimental s?mulus displays).This raises the ques?on of how small the visual target needs to be for the driX response to disappear.We instructed our monkeys to maintain fixa?on on a central fixa?on spot, and we presented a brief flash of a black circle centered on the fixa?on spot (Materials and Methods).The flash could be approximately as small as the fixa?on spot or as large as the en?re display, with intermediate radii in between, and we analyzed data from the same experiments in which we recently characterized saccadic inhibi?on as a func?on of s?mulus size (Khademi et al., 2023).The difference in the current study is that we specifically focused here on trials in which there were no microsaccades occurring within the interval between -100 ms and 200 ms from s?mulus onset (Materials and Methods; also see later for our separate analyses inves?ga?ng interac?ons between microsaccades and the driX response).The smallest foveal visual s?mulus could s?ll evoke a clear driX response.Figure 2A, B (yellow) shows average horizontal (Fig. 2A) and ver?cal (Fig. 2B) eye posi?on from monkey A when the smallest visual flash occurred.In each panel, we always aligned all eye posi?on traces across trials to the eye posi?on at ?me zero (s?mulus onset), in order to isolate the impact of the s?mulus event on driX sta?s?cs (despite variable eye posi?ons during gaze fixa?on; Materials and Methods) (Malevich et al., 2020).As can be seen, this monkey had a systema?c rightward driX trajectory before s?mulus onset (Fig. 2A, yellow); that is, the horizontal eye posi?on curve in Fig. 2A was steadily shiXing upward in the plot (meaning a rightward displacement) during the pre-s?mulus interval; the ver?cal eye posi?on curve in Fig. 2B was more-or-less steady.AXer s?mulus onset, Fig. 2B shows that there was s?ll a small upward driX response that occurred (not unlike that seen in Fig. 1B, C), despite the vanishingly small s?mulus size rela?ve to the size of the fixa?on spot.Such a small upward driX response was also clearly visible in monkey F (Fig. 2D, E, yellow curves), even though this monkey had a different pre-s?mulus driX trajectory (which was now predominantly leXward and downward).In monkey M, the smallest visual s?mulus barely modified the ongoing driX sta?s?cs (Fig. 2G, H, yellow curves), but this monkey also had the fastest pres?mulus driX speeds from among all three animals (compare the rates of change in eye posi?ons during the pre-s?mulus epochs across all panels).This faster baseline driX speed might have masked any poten?alimpacts of the smallest s?mulus size on driX eye movements in this monkey.Nonetheless, and as we describe next, driX responses were s?ll clearly visible in this animal for the slightly larger s?mulus radii of only 0.18 or 0.36 deg.Thus, in all three animals, even the smallest, foveal s?muli could s?ll evoke a reliable, predominantly upward, driX response.The driX response not only occurred for small, foveal s?muli, but its magnitude also systema?cally depended on s?mulus size.Specifically, the remaining curves of Fig. 2A, B, D,   E, G, H show eye posi?on traces from three addi?onals?mulus sizes that we used in our experiments, covering s?mulus radii larger than approximately 1 deg.In all cases, the driX response was rendered larger with larger s?muli.When we now considered all of our tested s?mulus sizes, we found that in both monkeys A and F, s?mulus sizes beyond a radius of about 1-2 deg systema?cally, and monotonically, increased the amplitude of the driX response.In monkey M, this monotonic rela?onship was evident even from the very smallest s?mulus sizes that we tested, well below 1 deg in radius.This la8er observa?on can be be8er appreciated from Fig. 2C, F, I, summarizing the rela?onship between driX response magnitude and s?mulus size.In these panels, and for each animal, we measured the driX response magnitude like we did in our earlier study (Malevich et al., 2020).Specifically, we took the difference in ver?cal eye velocity between two measurement intervals, a s?mulus response epoch and a pre-s?mulus baseline epoch (gray shaded regions in Fig. 1C; Materials and Methods).As can be seen from Fig. 2C, F, I, there was clear size tuning of the driX response magnitude in each animal: monkeys A and F showed a plateau (and even decreasing rela?onship in monkey A) up to about 1-2 deg, followed by a rise for larger s?muli; monkey M (generally having significantly faster baseline driX speeds) exhibited a monotonic increase with s?mulus size, even for s?muli smaller than 1 deg in radius.(n = 223, 219, 235, 266, 308, 339, 350, and 399 trials from the smallest to the largest s-mulus size).Note how this monkey also showed small transient oscilla-ons in both horizontal and ver-cal eye posi-ons at the very ini-al phases of the driB response.(G-I) Similar results for monkey M (n = 327, 369, 397, 423, 456, 420, 416, and 405 trials from the smallest to the largest s-mulus size).In all monkeys, the driB response was size-dependent, and it increased monotonically with sizes beyond 1-2 deg.We confirmed the above interpreta?onssta?s?cally.We performed, within each animal's data, a 1-way ANOVA rela?ng driX response magnitude to s?mulus size.In all three monkeys, there was a significant main effect of s?mulus size [p<0.0001for monkeys A, F, and M; F(7,6856) = 63.23 , F(7,2331) = 57.78, and F(7,3205) = 50.71for monkeys A, F, and M, respec?vely].Therefore, besides s?ll occurring for ?ny foveal s?muli, the driX response also clearly exhibits size tuning, which we will later link to the size tuning of saccadic inhibi?on that we recently characterized in the same experiments (Khademi et al., 2023).
It is also interes?ng to note that in all three animals, larger s?mulus sizes also increased the likelihood of observing a small transient modula?on of eye posi?on right at the very beginning of the overall driX response.For example, for the largest flashes, all three monkeys exhibited a small, but short-lived, downward change in eye posi?on before the upward driX pulse (Fig. 2B, E, H, largest s?mulus size), and this is similar to the downward transient that is evident in Fig. 1B.We frequently observed this small transient in our earlier study as well (Malevich et al., 2020).Monkey F addi?onally showed transient small oscilla?ons in eye posi?on at the beginning of the driX response for different sizes.The larger s?muli in the current experiment addi?onally increased the likelihood that the upward driX response had a horizontal component to it.For example, monkey A's upward driX response for large s?muli was accompanied by a slight leXward trajectory (Fig. 2A), and monkey M's upward driX response for large s?muli was accompanied by a rightward trajectory (Fig. 2G).Once again, we observed such horizontal devia?ons accompanying the upward driX response in our earlier experiments as well (Malevich et al., 2020).Therefore, our results so far demonstrate that the s?mulus-driven ocular posi?on driX response (Malevich et al., 2020) can s?ll happen for ?ny foveal visual transients, and that this driX response also exhibits size tuning (Fig. 2).As we will show below in more detail, it is interes?ng to note how this size tuning might relate to the size tuning of saccadic inhibi?on (Khademi et al., 2023).

The dri9 response is stronger for high contrast s=muli
We next turned our a8en?on to the contrast sensi?vity curve of the driX response.We had the three monkeys view brief, transient full-screen flashes while they fixated their gaze at the center of the display.Across trials, the flashes (which were all darker than the background) could have a different Weber contrast (Materials and Methods).In all three animals, the driX response magnitude monotonically increased with s?mulus contrast, increasing quasi-linearly as a func?on of log-contrast.These results can be seen in Fig. 3, which is organized similarly to Fig. 2. Specifically, Fig. 3A, B, D, E, G, H shows horizontal and ver?cal eye posi?on traces from all three monkeys for three example contrast levels.The lowest tested contrast (5%; yellow curves) s?ll showed a reliable driX response in all three monkeys.Moreover, the driX response magnitude increased with increasing contrast.To summarize these results, we again calculated the driX response size as described above (difference in ver?cal eye velocity between a response and a baseline epoch; Materials and Methods), and we plo8ed it as a func?on of s?mulus contrast for each animal.These plots are shown in Fig. 3C, F, I, and they demonstrate the contrast sensi?vity curve of the driX response.Sta?s?cally, there was a clear effect of contrast on driX response magnitude in each animal [p<0.0001across all animals; 1-way ANOVA on driX response magnitude as a func?on of contrast; F(4,3626) = 56.65,F(4, 959) = 46.71,and F(4, 2142) = 45.43 for monkey A, F, and M, respec?vely].Therefore, to the extent that s?mulus-driven neural responses somewhere in the visual/oculomotor system might mediate short-latency ocular posi?on driX responses (Malevich et al., 2020), these visual responses are expected to monotonically depend on s?mulus contrast.Given the short ?me interval between s?mulus onset and the actual eye movement modula?ons, we hypothesize (Buonocore and Hafed, 2023;Khademi et al., 2023) that these visual responses that are relevant for the driX response can be observed late in the oculomotor control circuitry, perhaps even in the brainstem pre-motor network.

The dri9 response is predominantly upward even for lower visual field s=muli
Speaking of oculomotor control circuitry, a candidate brain structure possessing shortlatency visual responses and having direct access to the oculomotor system is the SC, and it is also a structure that can contribute to smooth eye movements (Krauzlis et al., 1997;Basso et al., 2000;Krauzlis et al., 2000;Hafed et al., 2008;Hafed and Krauzlis, 2008).Because the SC has stronger visual sensi?vity for the upper visual field (Hafed and Chen, 2016;Fracasso et al., 2023), and seems to also magnify its representa?on for the upper visual field (Hafed and Chen, 2016), we hypothesized earlier that the predominantly upward nature of the driX response (for s?muli spanning both the upper and lower visual fields) might be mediated, at least par?ally, by SC visual ac?vity (Malevich et al., 2020).If so, then presen?ng s?muli exclusively in the lower visual field (below the line of sight) should make the driX response downward instead, since it now shiXs the balance of SC visual ac?vity in favor of the lower visual field.We, therefore, next tested how the driX response was affected by presen?ng a half-screen brief flash either only in the upper half of the en?re display or in the lower half (Materials and Methods).We also interleaved sham trials (without any flashes) as well as trials with small, localized flashes in the periphery (Materials and Methods).We note here that our earlier half-screen experiments (Malevich et al., 2020) involved splidng the screen area along the horizontal rather than ver?cal dimension (giving rise to either right or leX visual field s?mula?on rather than upper/lower visual field s?mula?on); thus, these experiments s?ll contained equal s?mulus energy in the upper and lower visual fields and could not conclusively test the original hypothesis about upper visual field SC preference.The driX response was s?ll predominantly upward even for lower visual field half-screen s?muli.Figure 4 shows the eye posi?on and velocity measures from this experiment in a manner similar to how we presented data in the earlier figures (Figs. 2, 3).The cri?cal comparison here is between the upper and lower visual field s?mulus condi?ons (red and purple colors in Fig. 4).In these condi?ons, the brief flash could consist of a black rectangle covering either exactly the top half or bo8om half of the display.In each monkey, the driX response was s?ll predominantly upward for lower visual field flashes (Fig. 4B, E, H), which is inconsistent with the hypothesis that SC visual responses dictate the upward direc?on of the driX response.Moreover, across the animals, there was no systema?crela?onship between the strength of the upward driX response and the visual field loca?on of the s?mulus.For example, in monkeys A and M, the overall driX response magnitude was similar for the upper and lower visual field s?muli (Fig. 4B for monkey A and Fig. 4H for monkey M).On the other hand, for monkey F, upper visual field s?muli did indeed cause a stronger upward component of the driX response than lower visual field s?muli (Fig. 4E).Sta?s?cal tests between the velocity difference measures of the two condi?onsconfirmed these observa?ons, as can be seen in Fig. 4C, F, I.In monkey A, there was no significant difference between upper and lower visual field flashes in Fig. 4C (p=0.26,t-test, t = -1241).For monkey F, the driX response magnitude was significantly stronger for the upper visual field s?muli (p=0.0079,t-test, t = 2.6642; Fig. 4F).And, for monkey M, there was again no reliable difference between the upper and lower visual field s?muli (p=0.77,t-test, t = -0.2818;Fig. 4I).The figure is otherwise organized as Fig. 2. Therefore, the driX response remains to be predominantly upward even with lower visual field s?muli, and the strength of this driX response may or may not reflect the presence of lower or upper visual field s?mulus energy (also see later for further tests of this idea with small, localized flashes).The other condi?ons shown in Fig. 4 were also informa?ve in the broader context of this study.For example, in all animals, the driX response was always the strongest for the largest s?mulus flashes (full-screen s?muli; blue colors in Fig. 4).This is consistent with our observa?ons in Fig. 2. Interes?ngly, in the present experiments, we also interleaved trials with a 1 deg x 1 deg localized s?mulus flash in the periphery rela?ve to the fixa?on spot loca?on (Materials and Methods; this is complementary to the small, foveal flashes of Fig. 2).Remarkably, there was s?ll a small upward driX response in this case (all yellow curves in Fig. 4).This prompted us to inves?gate the influences of small, localized eccentric (rather than foveal) flashes on ocular posi?on driXs in much more detail, as we describe next.
Small, localized s=muli addi=onally cause spa=ally-directed dri9 modula=ons Our results so far demonstrate that the upward driX response occurs under a large variety of s?mulus condi?ons, which hints that this driX response may be a reflexive movement of some kind.Indeed, the driX response remains predominantly upward even for lower visual field flashes (Fig. 4), and it also occurs for small foveal (Fig. 2) and eccentric (Fig. 4) targets.However, whether the driX response is a reflex or not, it is s?ll likely the outcome of readout of s?mulus-driven neural ac?vity in the oculomotor control network.For small, localized targets, such ac?vity can be highly spa?ally localized, especially in topographically organized structures like the SC.Might it then be the case that spa?ally localized visual bursts somewhere in the oculomotor system may play a modulatory role on ocular posi?on driXs during fixa?on?Indeed, we recently found that at the ?me of saccade readout, spa?ally localized SC spiking systema?cally altered saccade metrics and kinema?cs even when such spiking was not part of the movements' motor bursts (Buonocore et al., 2021), and the ques?on now becomes whether a similar effect can be seen in ocular posi?on driXs as well.In previous work with peripheral cueing, we uncovered evidence that peripheral s?mulus onsets can indeed give rise to spa?ally-directed driX trajectories (Tian et al., 2018), but our localized s?mulus experiments in the driX response study of (Malevich et al., 2020) did not exhaus?velystudy spa?ally-directed effects.Moreover, the s?mulus loca?ons for the localized targets in Fig. 4, and in (Tian et al., 2018), were not distributed enough to explore different spa?ally-directed modula?ons (Materials and Methods).Therefore, we explicitly ran an addi?onal experiment with localized s?mulus flashes, this ?mesystema?cally sampling different direc?ons rela?ve to the line of sight.The experiment consisted of the monkeys fixa?ng a central spot, and a brief black flash of 1 deg x 1 deg size occurred at an eccentricity of 7.9 deg.The flash could occur at one of eight equally spaced direc?ons rela?ve to the fixa?on spot (see inset schema?c in Fig. 5C).To robustly infer (from a sta?s?cal perspec?ve) poten?al spa?ally-directed driX modula?ons, we first grouped all target loca?onsalong the horizontal direc?on.That is, any localized flash that was in the right visual field was grouped into the rightward target group, and any localized flash that was in the leX visual field was grouped into the leXward target group (see the two different colors in the schema?cinset of Fig. 5C).We then analyzed the eye posi?ons of the three animals in the two groups of trials.We focused, here, on eye posi?ons rather than eye veloci?es (like we did in earlier analyses) because we wanted to directly assess the poten?al spa?al biasing that was caused by the s?mulus onsets.rightward or upward differences between the cyan and blue curves.Horizontal eye posi-on reflected the spa-al layout of the flashes, and this difference increased with -me.Ver-cal eye posi-on did not.(D-F) Similar observa-ons for monkey F (n = 349 and 398 trials for the right and leB s-mulus loca-ons, respec-vely).This monkey showed an even clearer driB response modula-on by s-mulus loca-on, also consistent with the same monkey's performance in earlier experiments (Tian et al., 2018).(G-I) Similar analyses for monkey M (n = 649 and 1091 trials for the right and leB s-mulus loca-ons, respec-vely).This monkey did not show horizontal modula-on of driBs by s-mulus loca-on, but this monkey also had significantly faster baseline driB speed than the other two monkeys.As with the other two monkeys, there was s-ll an upward s-mulus-triggered driB response component (H).P-values indicate results of t-tests comparing eye posi-ons within a given measurement interval.Horizontal eye posi?on driXs systema?cally reflected the peripheral hemifield loca?ons of the brief, localized flashes, confirming our earlier observa?ons that ocular posi?on driXs can be spa?ally-directed (Tian et al., 2018).For example, Fig. 5A shows the horizontal eye posi?on of monkey A for the two groups of s?mulus loca?ons (see inset schema?c in Fig. 5C).As in all of our other analyses, we aligned eye posi?ons at ?me zero to be8er appreciate the s?mulus-driven changes in driX sta?s?cs.Shortly aXer s?mulus onset, the monkey's horizontal eye posi?on deviated more rightward for the rightward flashes than for the leXward flashes, and the eye posi?on devia?on between the two s?mulus groups increased in size with ?me.This modula?on was riding on top of the upward driX response that we described above, as can also be seen from Fig. 5B.Here, the ver?cal eye posi?on of the same animal and in the same trials showed an upward driX pulse, which (unlike horizontal eye posi?on) was largely not differen?a?ng between s?mulus loca?ons (especially in the early phases of the response).Thus, small, localized eccentric targets along the horizontal direc?on were associated with both an upward driX pulse as well as horizontal modula?on of ocular posi?on driXs reflec?ng the horizontal loca?ons of the targets.We summarized these observa?ons by measuring the eye posi?on difference between the two curves of Fig. 5A or Fig. 5B at two different post-s?mulus ?mes (shaded gray bars near the x-axes in Fig. 5A, B).This difference was significant for horizontal eye posi?on but not for ver?cal eye posi?on, as can be seen from Fig. 5C.Moreover, the horizontal difference in eye posi?on was larger for the later ?me interval (Fig. 5C).These observa?ons were virtually iden?cal in monkey F (Fig. 5D-F), despite the monkey's different baseline (pre-s?mulus) driX trajectory.Thus, there can indeed be spa?ally-directed driX modula?ons in addi?on the upward driX pulse.For monkey M, there was no clear evidence of spa?ally-directed driX modula?ons in the horizontal direc?on, but this monkey did exhibit a clear upward driX pulse (Fig. 5G-I).As men?oned earlier, this monkey had the fastest baseline driX speeds from among the three animals, rendering a weak modula?on by spa?ally localized peripheral ac?vity harder to see.This is similar to our observa?ons of the size tuning experiments described above (Fig. 2).In all, the results of Fig. 5 confirm that ocular posi?on driXs are not always random or stochas?c (Kowler and Steinman, 1979b, a;Ahissar et al., 2016;Tian et al., 2018;Skinner et al., 2019;Bowers et al., 2021;Reiniger et al., 2021;Clark et al., 2022;Nghiem et al., 2022), and that these driXs can reliably reflect localized s?mulus loca?ons in addi?on to exhibi?ng a (poten?allyreflexive) upward driX pulse.Having said that, true dependence of ocular posi?on driXs on localized s?mulus loca?ons should include evidence of spa?ally-directed driX trajectories for the ver?cal dimension as well.Thus, we next regrouped our trials according to the ver?cal loca?ons of the localized flashes (see inset schema?c of Fig. 6C).In this case, all three monkeys showed evidence that ver?cal eye posi?on deviated more upward for upper visual field target loca?onsthan for lower visual field target loca?ons (Fig. 6); the effect was weakest in monkey A, but the trend was s?ll clearly there.Moreover, in all cases except for monkey M, horizontal eye posi?on devia?ons were similar to each other for the upper and lower visual field targets, exactly complementary to the results of Fig. 5. Thus, in Fig. 5, it was horizontal eye posi?on that was most affected by horizontal target loca?ons, and in Fig. 6, it was ver?cal eye posi?on instead that was most affected by ver?cal target loca?ons.Such a complementary nature of the results of Figs. 5, 6 is consistent with the interpreta?on that spa?ally-directed driX responses can indeed occur.Once again, these spa?ally-directed effects were occurring in addi?on to a global upward driX response, which was similar to what we saw in all of our earlier analyses with other types of s?muli.Therefore, ocular posi?on driXs exhibit a s?mulus-driven upward driX response for a large range of s?mulus types (including small foveal and peripheral targets; Figs.1-4), and they also undergo spa?ally-directed modula?ons by spa?ally localized flashes (Figs. 5, 6).These spa?ally-directed modula?ons likely reflect localized visual bursts in oculomotor control circuits, such as the SC, that have an impact on eye movement genera?on in the brain.It would be interes?ng in the future to understand why large (non-spa?ally-specificflashes) in the upper and lower visual field (Fig. 4) did not systema?callymodulate the driX response in the ver?cal eye posi?on direc?on across all three animals even though small targets did (compare the ver?cal eye posi?on results of Fig. 4 and Fig. 6).5, except that we now grouped the trials according to whether the localized s-mulus flashes were in the upper or lower visual field (see inset schema-c in C for the color codes).All monkeys showed a ver-cal driB response that was predominantly upward.On top of that, the s-mulus loca-ons now modulated the ver-cal component of eye posi-ons more than the horizontal component, again consistent with the idea that localized s-muli can s-ll have a modulatory effect on ocular posi-on driBs (compare the eye posi-on traces to those in Fig. 5).Also note that the ver-cal posi-on difference measurements in the later -me interval did not increase rela-ve to those in the earlier -me interval as in Fig. 5 for the case of horizontal posi-on difference (in monkeys A and F).This is likely because the spa-ally-driven modula-on in the ver-cal dimension was riding on a driB response that was already predominantly ver-cal in the current case.

The dri9 response is synchronized with saccadic inhibi=on
Our analyses so far focused on trials in which there were no saccades in the interval from -100 ms to 200 ms rela?ve to s?mulus onset.This was important to allow us to best observe the driX response, because saccades would cause much larger velocity pulses that would mask such a response (but see our later analyses in which we directly tackled the ques?on of saccade-driX interac?ons).We also know from our recent work (Malevich et al., 2020) that the driX response is complementary to saccade genera?on, in the sense that it occurs near the ?me of saccadic inhibi?on.Having said that, our current study afforded us a much be8er chance at exploring this complementary nature between saccade genera?on and the driX response in more detail.Specifically, we know from our most recent work that the ?me of saccadic inhibi?on in our size tuning and contrast sensi?vity experiments varied systema?cally as a func?on of s?mulus type (Khademi et al., 2023).If the driX response is indeed obligatorily synchronous to saccadic inhibi?on, then we should also see evidence that the ?ming of the driX response (not just its magnitude like in our earlier analyses above) should depend on the s?mulus feature.This would, in turn, imply that the driX response and saccadic inhibi?on may be generated by common neural circuitry.We explored this idea by plodng driX responses and saccades together in the same graphs, and we checked whether driX response ?ming co-varied with saccadic inhibi?on ?ming. Figure 7 illustrates this for the size tuning experiment.For each monkey, the individual rasters indicate individual saccade ?mes across trials, grouped by s?mulus size (different colors).These rasters were reproduced from our earlier study (Khademi et al., 2023), since we analyzed driX responses from the same set of experiments.Superimposed on the rasters, we addi?onally plo8ed average ver?cal eye posi?ons for each s?mulus size (similar to the example ver?cal eye posi?on plots in Fig. 2).Each eye posi?on curve was scaled to fit within the similar-colored group of saccade rasters, and posi?on scale bars for each curve are included (on the leX side of the curve) for reference.As can be seen, the driX response latency appeared synchronized with the latency of saccadic inhibi?on, as es?mated by the L50 parameter (dark green ver?cal lines; Materials and Methods).This parameter is rou?nely used to characterize the latency of saccadic inhibi?on (Reingold andStampe, 2002, 2004;Rolfs et al., 2008;Khademi et al., 2023), and Fig. 7 shows that when L50 was late, so was the onset of the driX response, and vice versa.experiment, we recently found that the -ming of saccadic inhibi-on depends on s-mulus size (Khademi et al., 2023).This is indicated here, for monkey A, by the raw saccade onset -mes (-ck marks) and a measure (ver-cal dark green lines marked with L50) of saccadic inhibi-on -ming (Materials and Methods) (Khademi et al., 2023).
Each row of -ck marks represents a single trial, and each -ck mark represents the onset -me of a saccade.The L50 line in each condi-on (dark green color) indicates our es-mate of the saccadic inhibi-on -ming (Khademi et al., 2023), and all trials of a given s-mulus size are grouped together according to the color legend.Within each group of trials, we also plohed the driB response (on trials without saccades; Materials and Methods) by showing ver-cal eye posi-on aligned on s-mulus onset (scale bars are shown on the leB of each curve).Despite the variable saccadic inhibi-on -ming, the driB response was synchronized with such -ming.That is, both the -ming of the driB response (on trials without saccades) and the -ming of saccadic inhibi-on (on trials with saccades) depended on the s-mulus proper-es (also see Figs. 8, 9).(B) Similar observa-ons from monkey F. (C) Similar observa-ons from monkey M. The saccade data in B were directly replohed from (Khademi et al., 2023) (CC-BY) since they came from the same experiments.Numbers of trials in the saccade data can be inferred from the rasters and from (Khademi et al., 2023); numbers of trials in the smooth driB data were reported in Fig. 2. We next checked this synchrony idea further by asking whether our driX response curves across s?mulus sizes were be8er aligned to s?mulus onset or to the onset of saccadic inhibi?on.For each animal, we plo8ed in Fig. 8 the average ver?cal eye posi?on traces for all s?mulus sizes (the curves were displaced ver?cally from each other for easier viewing).In the top row of the figure (Fig. 8A, C, E), the traces were aligned to s?mulus onset like in our earlier analyses, and the small ver?cal ?ck marks indicate the ?me of saccadic inhibi?on (L50) as we recently calculated it (Khademi et al., 2023).In the bo8om row (Fig. 8B, D, F), the same traces were now aligned to the ?me of L50, with the small ver?cal ?ck marks now indica?ngs?mulus onset ?me.In all three monkeys, the driX response curves were be8er synchronized with L50 than with s?mulus onset.That is, the curves across the different s?mulus sizes were less ji8ered in ?me rela?ve to each other when they were referenced to L50 than to s?mulus onset ?me.Thus, there seems to be an obligatory ?ming rela?onship between saccadic inhibi?on and driX response latency.(Khademi et al., 2023).Consistent with Fig. 7, saccadic inhibi-on -me varied with s-mulus size (Khademi et al., 2023), and the driB response followed this rela-onship.(B) This is beher seen when aligning the driB response curves of A to the -me of L50 rather than to the -me of s-mulus onset.Here, all the curves were beher aligned in -me.The ver-cal -ck marks now indicate s-mulus onset -me.(C, D) Similar results for monkey F. (E, F) Similar results for monkey M. In all cases, the driB response was rela-vely well synchronized with the -ming of saccadic inhibi-on, poten-ally sugges-ng a common mechanism underlying both phenomena.The numbers of trials underlying each curve were reported in Fig. 2. Such an obligatory rela?onship also held in our contrast sensi?vity experiment.In this experiment, lower contrasts were generally associated with later saccadic inhibi?on (Khademi et al., 2023).As Fig. 9 shows, such contrasts were also associated with later driX responses, and across s?mulus contrasts, the ?ming of the driX responses appeared to be be8er temporally aligned to the ?ming of saccadic inhibi?on across s?mulus features (Fig. 9B, D, F).Therefore, across mul?ple tasks associated with mul?ple different ?mes of saccadic inhibi?on (Khademi et al., 2023), we found that the driX response was synchronized with the reflexive interrup?on of saccade genera?on rhythms caused by visual onsets in the environment.saccadic inhibi-on -me depended on s-mulus property (Khademi et al., 2023), and once again, the driB response was synchronized with the -ming of saccadic inhibi-on.The figure is otherwise formahed iden-cally to Fig. 8, and the numbers of trials underlying each curve were reported in Fig. 3.

The dri9 response occurs with different star=ng eye posi=ons
In addi?on to ini?ally men?oning the poten?al rela?onship between the driX response and saccadic inhibi?on, we also suggested in our earlier work that the driX response occurs independently of star?ng eye posi?on (Malevich et al., 2020).However, in that study, we only used the natural variability of eye posi?ons during fixa?on to test whether the driX response s?ll occurred when the eye was momentarily fixa?ng below or above some central value (such as the median eye posi?on across trials).This leX open the ques?on of whether the driX response might depend on significantly larger eye posi?on devia?ons from the primary posi?on.To answer this, we performed a new version of our contrast sensi?vity experiment, in which we now explicitly required gaze fixa?on away from the display center.Specifically, in each block of trials, we placed the fixa?on spot at 4 deg eccentricity from the center of the display, either to the right of it, to the leX of it, above it, or below it (Fig. 10A).In all cases, the driX response s?ll occurred, and it was largely independent of the star?ng eye posi?on. Figure 10B shows ver?cal eye posi?on traces for the highest contrast s?mulus from each gaze posi?on condi?on.Of course, and as with all of our earlier analyses, we aligned all traces to the eye posi?on at s?mulus onset, and that is why all curves are aligned to zero eye posi?on on the y-axis despite the different star?ng gaze posi?on condi?ons.As can be seen, the upward driX response always happened, irrespec?ve of star?ng eye posi?on.Interes?ngly, the pre-s?mulus driX trajectory did depend on gaze posi?on.For example, when gaze was up (purple curve), pre-s?mulus driX in ver?cal eye posi?on was downward, and when gaze was down (blue curve), pre-s?mulus driX in ver?cal eye posi?on was upward.Nonetheless, and as just stated, there was s?ll an upward driX response in both cases.and leB gaze fixa-on condi-ons, respec-vely).The upward driB response always occurred, even when the eye was gazing down.Note that the pre-s-mulus driB direc-on showed some dependence on gaze posi-on.For example, downward gaze posi-on was associated with more upward pre-s-mulus eye posi-on driB, whereas upward gaze posi-on was associated with more downward pre-s-mulus eye posi-on driB (compare the blue and purple curves).However, in both cases, the s-mulus-driven response was s-ll upward.(C) Our measure of the driB response magnitude as a func-on of s-mulus contrast and fixa-on gaze posi-on.The driB response was stronger with higher contrasts.However, there was no systema-c dependence on gaze posi-on -a two-way ANOVA revealed a significant main effect of s-mulus contrast [F(4,1184) = 16.42;p<0.0001] but not star-ng eye posi-on [F(3,1184) = 1.36; p = 0.25].This extends our earlier findings with much smaller star-ng gaze posi-on devia-ons (Malevich et al., 2020).The numbers of trials per condi-on were as follows: 63, 66, 44, and 49 for up, down, leB, and right, respec-vely (5% contrast); 60, 73, 45, and 59 for up, down, leB, and right, respec-vely (10% contrast); 61, 71, 52, and 60 for up, down, leB, and right, respec-vely (20% contrast); 61, 78, 55, and 49 for up, down, leB, and right, respec-vely (40% contrast); 62, 71, 58, and 55 for up, down, leB, and right, respec-vely (80% contrast).Across all s?mulus contrasts, we replicated the contrast sensi?vity curve of Fig. 3 for each gaze posi?on condi?on (Fig. 10C).Indeed, there was no effect of gaze posi?on on driX response magnitude, but there was a clear effect of s?mulus contrast; sta?s?cal results are presented in the legend of Fig. 10.Therefore, even with substan?al devia?ons of gaze posi?ons, the driX response s?ll occurs, and it is s?ll predominantly upward.Moreover, pres?mulus driX trajectories can depend on gaze posi?on, likely reflec?ng a pulling force (whether biomechanical or neural) to return the eye back to the primary posi?on.Nonetheless, rela?ve to these changed baseline driX sta?s?cs, the driX response magnitudes are more-or-less constant (Fig. 10C).
The dri9 response magnitude is affected by the occurrence of peri-s=mulus saccades Finally, and s?ll on the general theme of interac?ons with saccades (Figs.7-9) and gaze posi?ons (Fig. 10), we next explored modula?ons in the driX response magnitude by the occurrence peri-s?mulus saccades.In our earlier work (Malevich et al., 2020), a coarse analysis suggested minimal (or even poten?allyno) interac?on with peri-s?mulus saccades.However, due to data sparsity, the analysis that we conducted at the ?me was not specific enough in its ?me course resolu?on.For example, rather than tes?ng trials with saccade onsets occurring within only a constrained ?me interval (as we would typically do for studying transient modula?ons by saccades), we tested trials with "saccades up to" some par?cular ?me point.Such an analysis might have excessively blurred transient changes in driX response magnitude caused by the occurrence of peri-s?mulus saccades (indeed, perisaccadic effects can be very transient in nature).With our current experiments, we had an opportunity to explore such transient changes in more detail.Indeed, because suppression of both visual sensi?vity and percep?on by peri-s?mulus saccades is jumpstarted already in the re?na (Idrees et al., 2020;Idrees et al., 2022), it would be remarkable if the driX response magnitude was completely unaffected by saccades.This would suggest that whatever visual response is media?ng the driX response would be immune to peri-saccadic suppression.This ques?on, therefore, warranted more detailed analysis in the current study.
Here, we binned our data for inves?ga?ons of poten?al "saccadic suppression" as we usually do for analyzing visual neural sensi?vity (Hafed and Krauzlis, 2010;Chen and Hafed, 2017;Fracasso et al., 2023) or percep?on (Idrees et al., 2020;Baumann et al., 2021).For example, for a given s?mulus condi?on, we took all trials in which there was a saccade onset occurring within the interval between -100 ms and 0 ms rela?ve to s?mulus onset (green shaded region in Fig. 11A).These trials would be expected to exhibit suppressed visual sensi?vity if saccadic suppression does take place.We also took trials in which there was a saccade onset 175-275 ms aXer s?mulus onset (yellow shaded region in Fig. 11A).These trials, instead, would be expected to not experience saccadic suppression (since the saccades occurred far away in ?me from s?mulus onset).Finally, we took trials in which there were no saccades at all in the interval from -100 ms to 200 ms rela?ve to s?mulus onset (shaded gray region in Fig. 11A), and these trials cons?tuted our "standard" driX response trials (like in our other analyses above).The driX response magnitude was suppressed by the presence in peri-s?mulus saccades.In Fig. 11B, for an example monkey and condi?on, we compared the standard driX response (gray curve in both panels A and B of Fig. 11) to the response when the s?mulus occurred right aXer microsaccades during pre-s?mulusfixa?on (green).As can be seen, the upward s?mulus-evoked velocity pulse was smaller in peak amplitude when the microsaccades occurred than when they did not occur.On the other hand, for microsaccades distant in ?me from s?mulus onset (yellow in Fig. 11), the driX response was recovered (Fig. 11C).Thus, for a brief moment in ?me when s?mulus onset occurred near saccade onset, the subsequent s?mulus-driven driX response was systema?cally suppressed.This is qualita?vely very similar to the classic phenomenon of saccadic suppression.to analyze no-saccade driB responses.Note that we did not sample all peri-s-mulus saccade -mes with high resolu-on; this was done to increase robustness of our observa-ons, especially given how noisy velocity measures can be with small numbers of trials.Nonetheless, we had sufficient data to check whether s-mulus onsets immediately aBer nearby saccades (shaded green interval) had altered driB responses.(B) For such trials (green), the driB response magnitude was suppressed.Error bars denote SEM (n = 168 and 879 for the green and gray curves, respec-vely), and no eye velocity data are shown in the green curve in the interval from -100 to 0 ms because saccades were occurring.As with the case of saccadic suppression (Hafed and Krauzlis, 2010;Chen and Hafed, 2017), the driB response was suppressed, sugges-ng that it might depend on circuits in which visual responses experience saccadic suppression; note that this observa-on was also categorically different from postsaccadic enhancement (Chen and Hafed, 2013).(C) For trials with a saccade occurring 175 to 275 ms aBer s-mulus onset (well away from s-mulus onset), the driB response was recovered.Error bars again denote SEM (n = 171 and 879 for the colored and gray curves, respec-vely).Also see Fig. 12 for summary data of suppression and recovery across other condi-ons and tasks.This observa?on was consistent across all monkeys and in all condi?ons that we checked.For example, for each s?mulus condi?on in both the contrast sensi?vity (5 s?mulus condi?ons) and size tuning (8 s?mulus condi?ons) tasks, we measured the driX response magnitude (as we did earlier; Figs.2-4, 10) and plo8ed it as a func?on of which ?me window of Fig. 11A the par?cular trials came from.For trials with saccades -100-0 ms from s?mulus onset, the driX response magnitude was always smaller than the driX response magnitude in the absence of peri-s?mulus saccades (Fig. 12; compare the response in the peri-s?mulus?me bin centered on -50 ms to the corresponding baseline response and its associated horizontal dashed line).Moreover, for trials with saccades 175-275 ms from s?mulus onset, the driX response magnitude was recovered and much closer to the standard driX response magnitude in the absence of peri-s?mulus saccades (Fig. 12; compare the response in the later ?me bin to that in the associated horizontal dashed line).We also confirmed these observa?ons sta?s?cally.For example, a two-way ANOVA in the contrast sensi?vity task revealed a main effect of both s?mulus contrast [p<0.0001 in monkeys A, F, and M] and saccade ?me rela?ve to s?mulus onset [p<0.0001 in monkeys A, F, and M].There was also a significant interac?on between saccade ?me and s?mulus contrast in monkey A [F(4,1343) = 3.76; p = 0.0048] but not in either monkey F [F(4,1744) = 0.54; p = 0.70] or monkey M [F(4, 1162) = 0.89; p = 0.47].Similarly, a two-way ANOVA in the size tuning task revealed a main effect of both s?mulus radius [p<0.0001 in monkeys A, F, and M] and saccade ?me [p<0.0001 in monkeys A, F, and M] in all three monkeys.However, once again there were no consistent interac?on effects.Monkey A showed no significant interac?on between s?mulus radius and saccade ?me [F(7,2633) = 1.38; p =0.21], monkey F showed a significant interac?on [F(7, 4118)= 5.17; p < 0.0001], and monkey M showed no significant interac?on [F(7,1542) = 1.7; p = 0.11].In each curve with connec-ng lines between the data points, the x-axis shows the center of the -me bin in which saccades occurred rela-ve to s-mulus onset (see Fig. 11A), and the y-axis shows our measure of the driB response strength (Materials and Methods).The floa-ng data points (and associated horizontal dashed lines) in each plot show the no-saccade driB response strength for a given condi-on (e.g.gray curves in Fig. 11).
Each color shows one tested contrast, and error bars denote SEM.As can be seen, the driB response magnitude was suppressed for saccades occurring near s-mulus onset and recovered for farther saccades (n >= 97, 149, 105 trials in monkeys A, F, and M, respec-vely, across all condi-ons of the experiment).(D-F) Similar results for the size tuning experiment (n >= 112, 182, or 50 trials across all condi-ons in monkeys A, F, and M, respec-vely).Therefore, evoked visual responses media?ng the driX response are likely suppressed by the presence of peri-s?mulus saccades, much like visual responses in some oculomotor areas including the SC (Hafed and Krauzlis, 2010;Chen and Hafed, 2017;Fracasso et al., 2023).Of course, we are not sugges?ngat all that SC responses mediate the driX response, especially given the results of Fig. 4. Rather, our results mean, instead, that other visual responses impac?ng the oculomotor system must exhibit saccadic suppression, and it would be interes?ng to iden?fy in the near future which of these visual responses mediate the driX response.

Discussion
Ocular posi?on driX eye movements have interested and intrigued neuroscien?sts for many decades (Ratliff and Riggs, 1950;Barlow, 1952;Nachmias, 1959Nachmias, , 1961;;Kowler and Steinman, 1979a, b).The interac?ons between these eye movements and exogenous sensory events have, however, garnered significantly less a8en?on.We recently observed a robust s?mulusdriven ocular posi?on driX response for some visual s?muli (Malevich et al., 2020), and our goal in the present study was to inves?gate its func?onal proper?es much more deeply.Such inves?ga?on provides an important founda?on for pinpoin?ng the neurophysiological mechanisms giving rise to this driX response, which is itself an important endeavor given how li8le knowledge we currently have about the neural control of ocular posi?on driXs in general.
Our inves?ga?on revealed several interes?ng proper?es of the driX response, most notable of which is its robustness even for small foveal and peripheral visual s?muli.There was always a subtle, predominantly upward devia?on in ocular posi?on driX trajectories with such s?muli.Given that this devia?onalters the spa?o-temporal pa8erns of images impinging on the re?na (Kuang et al., 2012;Rucci and Victor, 2015;Ahissar et al., 2016), this suggests that visual onsets in a variety of neuroscien?fic and cogni?ve experiments can have sensory representa?onal changes embedded within them, which are directly mediated by s?mulus-driven ocular posi?on driXs (in addi?on to whatever other experimental variables that were being considered by the experimenters).This idea has an interes?ng parallel in the field of microsaccades; in that related field, it has been suggested that these ?ny eye movements can have a significant impact on interpre?ngvarious perceptual and cogni?ve phenomena (Hafed, 2013;Chen et al., 2015;Hafed et al., 2015;Tian et al., 2016).The ubiquitous nature of the upward velocity pulse that we observed under a variety of condi?ons might suggest that it is a reflexive eye movement.However, it seems to be too small to be related to a poten?al dorsal light reflex in lower animals (Brodsky, 1999), and it is also binocular (Malevich et al., 2020) and occurring under binocular visual s?mula?on condi?ons.The driX response is also not a general gaze posi?on response to darkness (Malevich et al., 2020).Nonetheless, in the same general theme of linking ancient reflexes to effects in primate vision (Brodsky, 1999), the driX response might help us to learn about lowlevel, evolu?onarily old components of the oculomotor control network, which are s?ll present and ac?ve in the primate brain.In fact, given the discrepancy between the results of Fig. 4 and our original hypothesis about the SC media?ng the driX response (Malevich et al., 2020), we now seriously ponder the possibility that visual responses downstream of the SC might be more important for observing this response.This might explain why the driX response happens so ubiquitously across many different s?mulus types, since visual responses downstream of the SC are bound to influence eye movements, if ever so subtly (by mere proximity to the final oculomotor muscle drive).Having said that, the driX response as we defined it in the introduc?on(Fig. 1) is not the only ocular posi?on driX phenomenon that takes place aXer the onset of small, localized visual s?muli.Indeed, our results from Figs. 4-6 clearly show that there can be spa?allydirected driX modula?ons reflec?ng the loca?on of a peripheral visual s?mulus.This is consistent with our earlier observa?ons about ocular posi?on driXs in peripheral Posner-like cueing tasks (Tian et al., 2018).An important implica?on of this is that ocular posi?on driXs are not en?rely random movements, consistent with other evidence (Murphy et al., 1975;Kowler and Steinman, 1979b, a;Ahissar et al., 2016;Tian et al., 2018;Skinner et al., 2019;Bowers et al., 2021;Reiniger et al., 2021;Clark et al., 2022;Nghiem et al., 2022).This evidence again has parallels in the field of microsaccades, which were thought to be random un?l two decades ago (Hafed and Clark, 2002;Engbert and Kliegl, 2003).Mechanis?cally,spa?ally-directed driX modula?ons can emerge from readout of topographically organized visual-motor maps, like in the SC (Robinson, 1972;O8es et al., 1986;Chen et al., 2019).For example, we recently found that at the ?me of saccade triggering, even spontaneous spiking in movement-unrelated loca?ons of the SC map can be instantaneously readout by the oculomotor system to modify the flight trajectory of saccades (Buonocore et al., 2021).In a similar light, spa?al readout of the en?re landscape of SC ac?vity can dictate the smooth posi?on devia?ons during gaze fixa?on, and such landscape will have clear spa?al biases when some SC neurons discharge visual bursts aXer localized, peripheral s?mulus onsets.The spa?ally-directed driX effects that we observed would then reflect these biases.Such a mechanism would be consistent with how the SC contributes to the much faster (rela?ve to the driX response) smooth pursuit eye movements in general, like when tracking an invisible moving goal that is being represented in a spa?ally broad manner across the SC map (Hafed and Krauzlis, 2008).Such a mechanism would also be consistent with the idea that the upward driX pulse that accompanies spa?ally-directed driX modula?ons can be mediated by some other circuit opera?ons (poten?ally even downstream of the SC).Returning to the more reflex-like, predominantly upward driX response (Fig. 1), as we said, it is likely dissociated from SC ac?vity because it remains predominantly upward even when SC neurons represen?ng the lower visual field are expected to be burs?ngaXer s?mulus onset (Fig. 4).This idea can and should be explicitly tested by recording SC ac?vity from the same task of Fig. 4. We also think that other evidence in our data could point to a dissocia?on of the driX response from SC ac?vity.Specifically, we oXen observed a transient eye posi?on modula?on right before the upward velocity pulse, a clear example of which is seen in Fig. 1B, C. Such a transient modula?on jumpstarts the whole driX response sequence, and it seems to also be feature-tuned.That is, it was modulated in strength and ?ming as a func?on of some s?mulus proper?es, like size and contrast (Figs. 2,3).This could suggest that visual bursts media?ng the driX response (wherever they may actually be in the end) could ini?ally cause such transients, and that the subsequent upward driX pulse could reflect various ?meconstants of the oculomotor control network and oculomotor plant (Robinson, 1964).For example, using a systems control perspec?ve, imagine a nega?ve feedback control loop driving an eye plant, and now drive the whole circuit with a temporal impulse func?on.Part of the resul?ng response would reflect the ?me constants of not only the control loop but also the eye plant.If that is the case, then future experiments need to understand why driving the oculomotor control network with a temporal impulse func?on (a brief visual burst) would eventually lead to a predominantly upward eye movement, as opposed to downward or horizontal or in some random direc?on, aXer the ini?al transient modula?on.
Regardless of the mechanism, all of the above evidence suggests that the driX response falls in a class of eye movement phenomena that may be evoked directly by visual bursts in the oculomotor system, as we recently discussed (Buonocore and Hafed, 2023;Khademi et al., 2023).These phenomena also include express saccades (Fischer and Boch, 1983;Edelman and Keller, 1996;Marino et al., 2015;Hall and Colby, 2016) and saccadic inhibi?on (Reingold and Stampe, 1999, 2002, 2004;Edelman and Xu, 2009;Khademi et al., 2023).In fact, we think that saccadic inhibi?on and the driX response are likely mediated by the same structures , further emphasizing the idea that the driX response might be reflexive.If so, one might make some neurophysiological predic?ons here.Specifically, if the hypothesis (Hafed et al., 2021b;Buonocore and Hafed, 2023) holds that omnipause neurons in the brainstem have visual pa8ern responses explaining the feature tuning proper?es of saccadic inhibi?on, and if driX responses are also triggered by these neural bursts, then one predic?on is that visual bursts in these omnipause neurons might act as the "temporal impulse func?on" that jumpstarts the driX response, which we alluded to above.If so, this would implicate omnipause neurons in more than just the interrup?on of saccades (Keller and Edelman, 1994;Kaneko, 1996;Keller et al., 1996;Gandhi and Keller, 1999), and the next ques?onwill be why brief burst impulses in omnipause neuron ac?vity could cause a small, but smooth, eye posi?on devia?ons (in addi?on to inhibi?ng saccade genera?on).Finally, regardless of whether these ideas are experimentally validated or not, it is also important to consider our observa?onthat the driX response was suppressed by the occurrence of peri-s?mulus saccades (Figs. 11,12).Some smooth eye movement phenomena are actually enhanced when s?muli occur right aXer microsaccades (Chen and Hafed, 2013), but these phenomena typically involve ocular following of moving s?muli (Chen and Hafed, 2013).In our case, the driX response was not to follow a moving target or pa8ern.Its suppression, thus, predicts that visual bursts media?ng the driX response (wherever they may be) must be suppressed by peri-s?mulus saccades.It would be interes?ng to also test for this idea neurophysiologically.

Figure 1 S
Figure 1 S*mulus-driven ocular posi*on dri6 responses.(A) Accurate gaze fixa-on is characterized by two prominent features: (1) microsaccades occur occasionally to re-align the line of sight (red); and (2) the eye driBs

Figure 2
Figure 2 Size tuning of ocular posi*on dri6 responses.(A) Average horizontal eye posi-on from monkey A for four example s-mulus sizes (0.09 deg, 1.14 deg, 4.56 deg, and 9.12 deg).Error bars denote SEM, and the numbers of trials were 827, 804, 927, and 882 for the four s-mulus sizes, respec-vely.Upward deflec-ons in the plot denote rightward eye posi-on deflec-ons.(B) Average ver-cal eye posi-on from the same trials as in A; error bars again denote SEM, and upward deflec-ons denote upward eye posi-on deflec-ons.A clear dependence of the ocular posi-on driB response on s-mulus size can be seen.Note also how the smallest tested s-mulus (0.09 deg) s-ll caused a ver-cal driB response, but its ini-al smaller downward component was missing.(C) Our measure of the driB response magnitude (average baseline-corrected ver-cal eye velocity in the interval 70-150 ms aBer s-mulus onset; Fig. 1C; Materials and Methods) for all tested s-mulus sizes in monkey A(n = 827, 729,   872, 868, 804, 885, 927, and 882  trials from the smallest to the largest s-mulus size).Error bars denote SEM.(D-F) Similar results for monkey F(n = 223, 219, 235, 266, 308, 339, 350, and 399  trials from the smallest to the

Figure 3
Figure 3 Contrast sensi*vity of ocular posi*on dri6 responses.(A) Average horizontal eye posi-on from monkey A for three example s-mulus contrasts (5%, 20%, and 80%).Error bars denote SEM, and the numbers of trials were 689, 739, and 750 for the three contrasts, respec-vely.(B) Average ver-cal eye posi-on from the same trials as in A (error bars again denote SEM).A clear dependence of the ocular posi-on driB response on contrast can be seen.(C) Our measure of the driB response magnitude for all tested s-mulus contrasts in monkey A (n = 689, 699, 739, 754, and 750 trials from the lowest to the highest contrast).Error bars denote SEM.(D-F) Similar results for monkey F (n = 135, 165, 179, 223, and 262 trials from the lowest to the highest contrast).(G-I) Similar results for monkey M (n = 384, 412, 443, 433, and 475 trials from the lowest to the highest contrast).The figure is otherwise organized as Fig. 2.

Figure 4
Figure 4 Predominantly upward ocular posi*on dri6 responses even with lower visual field s*muli.(A) Average horizontal eye posi-on from monkey A in the visual field experiment.Gray indicates sham s-mulus onsets (n = 899 trials), yellow indicates a small localized flash eccentric from the fixa-on spot (Materials and Methods) (n = 833 trials), red indicates a s-mulus onset in the lower half of the display (n = 890 trials), purple indicates a s-mulus onset in the upper half of the display (n = 848 trials), and blue indicates a full-screen flash (n = 474 trials).Error bars denote SEM.(B) Average ver-cal eye posi-on from the same trials (error bars again denote SEM).The driB response was predominantly upward even for lower visual field s-mulus onsets (red).Note, however, how the ini-al downward component of the global driB response was weaker for the upper visual field s-mulus onsets.(C) Our measure of the driB response magnitude for all condi-ons.Sham and localized s-mulus onsets had weak driB responses (also see Figs. 5, 6); upper and lower visual field s-mulus onsets had generally similar driB response magnitudes (and were both globally upward); and full-screen s-muli had stronger driB response magnitudes (consistent with the size tuning effects of Fig. 2).(D-F) Similar results for monkey F (n = 401, 341, 372, 415, and 72 trials for the shown condi-ons: sham, localized, lower visual field, upper visual field, and full-screen flashes, respec-vely).(G-I) Similar results for monkey M (n = 835, 439, 1208, 1143, and 553 trials).

Figure 5
Figure 5 Spa*ally-directed dri6 modula*ons with localized s*muli along the horizontal direc*on.(A) Average horizontal eye posi-on from monkey A when localized flashes (1 x 1 deg squares; 7.9 deg eccentricity) appeared in the right (cyan) or leB (blue) visual field (see inset schema-c in C).Error bars denote SEM (n = 1120 and 999 trials for right and leB s-mulus loca-ons, respec-vely).DriB trajectory was affected by s-mulus loca-on, and the effect increased with -me.The two gray bars near the x-axis indicate measurement intervals for comparing eye posi-ons between the two groups of flash loca-ons.(B) Ver-cal eye posi-on from the same trials as in A. There was a general upward driB component, which was similar for rightward or leBward flashes.(C) We measured the difference between the cyan and blue curves in A, B for the two measurement intervals.Posi-ve values mean

Figure 6
Figure 6 Spa*ally-directed dri6 modula*ons with localized s*muli along the ver*cal direc*on.This figure is organized exactly like Fig. 5, except that we now grouped the trials according to whether the localized s-mulus (A-C) n = 1100 and 1006 trials for upper and lower visual field s-mulus loca-ons, respec-vely.(D-F) n = 303 and 312 trials for upper and lower visual field s-mulus loca-ons, respec-vely.(G-I) n = 984 and 881 trials for upper and lower visual field s-mulus loca-ons, respec-vely.

Figure 7
Figure 7 Coincidence between dri6 response onset and saccadic inhibi*on *ming.(A) In our size tuning

Figure 8
Figure 8 Alignment of the dri6 response onset to saccadic inhibi*on *ming.(A) Average ver-cal eye posi-on in each condi-on of the size tuning experiment from monkey A. Each curve was slightly offset ver-cally from the others for easier viewing.The ver-cal -ck mark in each curve indicates the -me of saccadic inhibi-on for the condi-on, as es-mated by the parameter L50 (Materials and Methods) (Khademi et al., 2023).Consistent with

Figure 9
Figure 9 Alignment of the dri6 response onset to saccadic inhibi*on *ming in another task.This figure is organized similarly to Fig. 8, but now showing results from the contrast sensi-vity experiment.Once again,

Figure 10
Figure 10 Independence of the dri6 response from star*ng eye posi*on.(A) We performed the contrast sensi-vity experiment, but now requiring gaze fixa-on at 4 deg eccentricity from the center of the display (either to the right, leB, up, or down from display center).(B) Average ver-cal eye posi-on from the four condi-ons with the highest contrast s-mulus (error bars denote SEM, and n = 62, 71, 58, and 55 trials for the up, down, right,

Figure 11
Figure 11 Saccadic suppression of dri6 responses.(A) Example saccade raster plot and driB response (shown by ver-cal eye velocity) from one monkey (A) and one condi-on (9.12 deg radius in the size tuning experiment).The shaded colored bars indicate how we picked trials to check for an interac-on between peri-s-mulus saccades and driB responses.For each such bar, we picked only trials from the same condi-on having saccade onsets occurring within the bar's -me window.The shaded gray bar, on the other hand, indicates our standard approach

Figure 12
Figure 12 Suppression of the dri6 response strength by the occurrence of peri-s*mulus saccades.(A-C) Summary plots of saccadic suppression of the driB response strength for each monkey in the contrast sensi-vity experiment.In each curve with connec-ng lines between the data points, the x-axis shows the center of the -me