The sensorimotor contingency theory proclaims that perception consists in mastering the systematic relations between actions and their perceptual consequences (O'Regan and Noë, 2001). The prime example for sensorimotor contingencies are those movements we perform most often in real life, that is saccade eye movements and the associated motion stimulation they produce. Every time we perform a saccade the external world sweeps with high speed motion across the retina. Consequently, the visual system has to deal with two challenges: First, to keep in register where in external space each pre- and post-saccadic image originated from, and second, to suppress any motion sensation that would arise during the retinal displacement.

Learning sensorimotor contingencies requires a signal that indicates the initiation of a movement. In eye movement research, the concept of the efference copy arose, that is a signal which informs visual areas about the metrics of upcoming eye movements (Wurtz, 2018). To guarantee the spatial alignment across saccades between internal and external space, neurons in the intra-parietal cortex (Duhamel et al., 1992), in the frontal eye fields (Umeno and Goldberg, 1997), the superior colliculus (Walker et al., 1995) and in visual areas V2 and V3 (Nakamura and Colby, 2002) shift their receptive fields predictively, triggered by an efference copy signal (Sommer and Wurtz, 2006). The suppression of intra-saccadic motion perception might be driven by such an extra-retinal signal likewise (for a review, see Binda and Morrone, 2018).

As it has been shown that the visual system is able to detect high-speed motion (Burr et al., 1982), the question arises why we are not aware of the motion stimulation produced by a saccade. In the search for a mechanism that cancels trans-saccadic motion perception, researchers have found that sensitivity of visual contrast (Volkmann et al., 1978) and of motion (Burr et al., 1982; Shioiri and Cavanagh, 1989; Ilg and Hoffmann, 1993) drops down at the time of saccades. This process - called saccade omission - amounts to 0.5–1 log units, is strongest for low spatial frequencies (Volkmann, 1986; Burr et al., 1994) and is homogeneous across the visual field (Knöll et al., 2011). It starts ~50 ms before saccade initiation, peaks around saccade onset and fades away 50 ms later (Volkmann et al., 1978). It has been argued that the intra-saccadic decrease of visual sensitivity is the signature of a mechanism that shuts down the magnocelullar pathway at an early neural level to prevent the disturbing motion experience during saccades (Burr et al., 1994).

However, in contrast to an early shut-down resulting in motion blindness, Castet and Masson (2000) have demonstrated that motion detection during saccades is still possible if stimuli are optimized for high-speed motion detection. They used low-contrast gratings drifting at a constant high speed (360 dva/s) and found that the ability to perceive intra-saccadic motion depended on the difference between the peak eye velocity and grating speed. In their view, the magnocellular pathway is still active at the time of saccades and the peri-saccadic reduction of contrast sensitivity results from retinal factors (Castet et al., 2002). In a follow-up study the same authors could further show that this ability to perceive fast intra-saccadic motion declined when additional pre - and post-saccadic stimuli were visible. They suggested that these static images masked intra-saccadic motion thus reducing awareness.

An earlier report already showed that masking might cancel the retinal smear produced by the saccade (Campbell and Wurtz, 1978). Duyck et al. (2016) recently devised a method to test the perception of intra-saccadic smear objectively and thereby corroborated the idea that masking hides saccade-induced motion from perception. In that study smear-masking survived a dichoptic presentation thus suggesting a central origin. If masking and not saccade-related mechanisms explains saccadic omission, one should find masking-induced suppression even in the absence of saccades. Recently, Duyck et al. (2018) used simulated saccades and found motion perception to be much less salient in the presence of static objects.

While masking provides a very parsimonious account of how the retinal smear remains hidden from awareness, it is still an open question why we do not consciously perceive a motion transient from the pre-saccadic to the post saccadic image. It is long known that trans-saccadic displacements of a certain size go unnoticed by the observer (Bridgeman et al., 1975; Niemeier et al., 2003), a phenomenon termed saccadic suppression of displacement. The reason for the poor displacement detection is that the saccade covers the motion transient because the presence of motion transients leads to very high displacement sensitivity (Legge and Campbell, 1981).

Here, I offer a novel explanation of suppression which states that the brain stores and habituates to sensorimotor contingencies. This account does not require elimination of peri-saccadic perception via contrast reduction at an early processing level. I show that the saccadic suppression results from a habituation to intra-saccadic stimulation (see Figure 1). In this view, neurons in an extra-retinal storage mechanism will be informed by an efference copy about the initiation of a saccade, store saccade-induced visual stimulation and saturate their response to the visual information most dominant in the previous history of the last set of saccades that has been executed.

Figure 1

Download asset Open asset

This idea predicts that suppression magnitude can be modulated by the visual context in which saccades are performed. Performing saccades in a structured environment induces recurring intra-saccadic stimulation. In order to create an experimental environment in which intra-saccadic information could be systematically varied, I used a horizontal grating as a background on which horizontal saccades had to be performed. As the grating was oriented parallel to the saccade path it should have produced only negligible retinal motion information. Context-sensitivity would predict that performing many saccades on that background should lead to a minimal magnitude of saccade suppression compared to natural vision, where saccades are performed mostly across structured environments.

The intra-saccadic stimulation was displayed as soon as a saccade was detected. I used a grating displacement orthogonal to the rightward saccades, that is upward displacements of the horizontal grating, in order to disentangle the artificially produced shift (the grating displacement) from the retinal motion. Testing displacement discrimination parallel to the saccade path requires more sophisticated experimental designs (Castet and Masson, 2000) that would have led to an unjustifiable number of trials. If the sensorimotor system takes into account this exposure, intra-saccadic displacement detection should drastically decline. Consistent with context-sensitivity, saccade suppression magnitude was strong when observers performed many trials in which stimulation was presented contingent on saccade execution. However, saccade suppression was absent if the background grating remained stationary and observers were able to perceive intra-saccadic motion as well as during fixation.

I asked participants to perform horizontal, rightward saccades in a dark environment. Stimuli were presented on a computer screen that was covered by a semi-transparent foil in order to dampen visual stimulation as much as possible. The background consisted of a sinusoidal horizontal grating shown with a spatial frequency of 0.05 c/dva (see Figure 2A). At the start of each experimental session (except baseline), participants performed 105 saccades over the grating. In these context trials - depending on the session - the grating was displaced either during or after the saccade or it remained stationary.

Figure 2

Download asset Open asset

Shifting the grating during saccade execution allowed to systematically manipulate intra-saccadic stimulation. Figure 2B shows the time-course of events for sessions in which the grating was shifted upward as soon as the eye-tracker detected saccade initiation. After a fixation period of 1000–1500 ms the saccade target was presented only briefly and participants started their saccades after the disappearance of the fixation point. Since the experiment was conducted in a dark room and the monitor screen was covered by a semitransparent foil, no visual information except the shifting grating was present during saccade execution. I checked the timing of the grating displacement relative to saccade performance in an offline analysis. Depending on sessions, displacements were presented well within (see Figure 2C) or outside (see Figure 2D) the period of saccade execution. Figure 2E summarizes all five conditions tested in the experiment. Baseline sessions contained no context but only test trials. The remaining four sessions that did contain context trials, comprised a condition with no grating displacement and conditions with a displacement of the grating 37.89 ms (SEM 5.94 ms), 98.07 ms (SEM 8.59 ms) or 186.99 ms (SEM 7.92 ms) after saccade initiation.

After the presentation of the context trials, I tested displacement discrimination in a 2-alternative forced-choice task for a grating that was shifted either upwards or downwards (Figure 2F). The time-course of events was identical to that of the context trials except that the shift of the grating was applied at various times around saccade execution. I analyzed displacement discrimination performance relative to saccade onset in bins of 30 ms. Figure 2G shows displacement discrimination results of all participants from sessions in which the grating was shifted upward 35 ms after saccade onset in the context trials. Data were taken from a bin long before saccade initiation (<−50 ms to saccade onset, shown in gray) and from a bin close to saccade onset (−25 ms <saccade onset < 25 ms, shown in purple). Data were fitted by cumulative gaussian functions and performance determined by estimating the JND of the psychometric function.

Psychometric functions for grating displacements presented during the saccade were far shallower indicating higher thresholds or stronger suppression. Thus, suppression magnitude was strong when in the context trials the grating shifted upwards during saccade performance. Data shown in Figure 2H derive from sessions in which the grating was not displaced in the context trials. For all observers, JNDs of the psychometric functions from intra-saccadic bins were steeper than those measured after context trials in which the grating was displaced.

Figure 3A shows the full time course of displacement thresholds around the time of saccade execution for all observers. Data derive from sessions in which the grating was shifted 35 ms after saccade initiation. For all observers, intra-saccadic discrimination performance declined after they had performed saccades with intra-saccadic stimulation in the context trials. Figure 3B shows displacement thresholds from sessions where the grating was not shifted in the context trials. In order to quantify intra-saccade displacement discrimination, ie. suppression magnitude, I chose the peak thresholds that were contained in a bin ±50 ms around saccade initiation. For the quantification of displacement discrimination during fixation, I calculated the average over all bins lying at least 100 ms before and 100 ms after saccade onset.

Figure 3

Download asset Open asset

I analyzed saccade parameters in order to estimate a putative influence of saccade amplitudes on suppression strength (Volkmann et al., 1981; Stevenson et al., 1986) in my data. Figure 3D,E shows horizontal and vertical average saccade amplitudes for all five sessions. In all sessions, saccades show the typical undershoot of the required amplitude (20°). Similarly, no statistical differences were found between average peak velocities of saccades (see Figure 3C). It has been shown that background displacement can provoke saccade adaptation (Deubel, 1991; Robinson et al., 2000; Ditterich et al., 2000). The recurring upward displacement of the background might have induced adaptation of the vertical saccade component. I subtracted the last ten vertical saccade amplitudes of the context trials by the first ten to check for an adaptation of saccadic gain. Figure 3F shows saccade gain change for all four sessions that contained context trials. I calculated one-tailed paired t-tests separately for all four sessions. There was no statistical evidence for a significant gain change in any of the sessions (context 35 ms: t(3) = 1.18, p=0.32, context 98 ms: t(3) = −0.60, p=0.59, context 187 ms: t(3) = −0.66, p=0.55, context no motion: t(3) = 0.21, p=0.85).

Figure 3G shows displacement discrimination averaged across all four participants from the peri-saccade period (shown in red) and the fixation period (shown in gray) in baseline sessions where no context trials were applied. Observers demonstrated poor peri-saccadic displacement discrimination as is expected if suppression is not absent. Gratings had to be shifted upwards on average ~120 deg (phase shift) such that observers were able to tell the correct direction of the shift. A paired one-tailed T-test confirmed that displacement discrimination thresholds were significantly higher when the grating was shifted peri-saccadically than before saccade initiation (t(3) = −9.38, p=0.0026). The effect size of this difference was 1.7 (Cohen’s d) which is considered a very large effect. Figure 3H shows average displacement discrimination from sessions that included context trials. Intra-saccadic displacement discrimination thresholds were virtually identical to those of the baseline sessions if in the preceding context trials the grating was displaced during saccade execution, that is 35 ms after saccade onset. However, thresholds decreased massively after participants performed context trials in which the displacement occurred after the saccade had landed or in which there was no displacement at all. In these sessions, peri-saccade displacement discrimination was almost as good as during fixation. This finding provides clear evidence for the idea that the sensorimotor system habituates to peri-saccadic stimulation. In sessions where the grating was shifted outside the period of saccade execution, the system was relieved of suppression since no intra-saccadic stimulation occurred for more than 100 trials. For statistical analysis, I calculated the difference between baseline and adaptation sessions within each participant and averaged across all post-saccadic/no motion difference values (i.e. the 96, 187 ms and the No motion context data) within each participant. A paired one-tailed T-test confirmed that intra-saccadic stimulation in the context trials induced stronger suppression than post-saccadic/no displacement stimulation (t(3) = 3.68, p=0.017).

In order to better understand the effect of the intra-saccadic stimulation, I analyzed the biases of the displacement discrimination. Figure 3I shows biases for fixation (gray) and peri-saccade discrimination form the baseline sessions. A paired one-tailed T-test did not reveal a significant differences (t(3) = −1.63, p=0.20). Similarly in the discrimination data that followed context trials, no significant difference in biases between conditions was found (see Figure 3J). Thus, there was no evidence for a motion after-effect in the current data that would have manifested in changed bias, that is a shift of the psychometric functions.

Next, I sought to find out whether the habituation effect is direction-specific. To this end, I modified the display of the test trials to contain two horizontal gratings (see Figure 4A).

Figure 4

Download asset Open asset

The experiment contained context and test trials. In the test trials, participants were required to perform a saccade to the remembered position of the saccade target as in the first experiment. Then, they had to judge which of the two gratings was displaced. The displacement direction in the test trials was always upwards and applied either during saccade or after saccade execution. The displacement direction in the context trials was, in separate sessions, either upwards or downwards with the same displacement size of 57° phase shift as in the other experiments. A clear elevation of discrimination thresholds was observed only if the displacement in the context trials occurred during the saccade and had the same direction (upwards) as in the test trials (see Figure 4B). A non-parametric repeated measures ANOVA with the factors ‘displacement direction’ (upward/downward) and ‘displacement time’ (during/after saccade) revealed a significant main effect ‘displacement direction’ (F(1,4) = 15.65, p=0.017), confirming that thresholds were elevated selectively when the displacement direction in the context and the test trials matched. The significant interaction effect (F(1,4) = 8.18, p=0.046) revealed that the selectivity was observed only when the displacement occurred during but not after saccade execution. The factor ‘displacement time’ was not significant (F(1,4) = 6.668, p=0.061).

Displacing a stimulus during execution of a saccade produces a different retinal motion vector as a displacement during ocular fixation. The retinal motion vector is the sum of the physical displacement vector (i.e. upward) and the inverse saccade vector (i.e. leftward for rightward saccades). Even though the screen borders were covered by a foil and the room was dark, the vertical edges of the horizontal grating might have contributed to the experience of a horizontal motion component. In order to compare habituation for intra-saccadic and post-saccadic presentations with a matched retinal motion vector, I conducted a new experiment. In this Experiment 3, I added vertical bars to the horizontal grating to imitate leftward motion of the vertical grating edges. The spatial frequency of the vertical bars and the horizontal grating was kept identical. The procedure was identical to Experiment 2 except that only in the test trials the displacement now contained a horizontal component in addition to the vertical. The amount of the horizontal component was calculated to be identical to the saccade-induced horizontal shift of the grating displacement measured in Experiment 2 (see Materials and methods section for details). I tested sessions with an upward displacement in the context trials (see Figure 5B) and a downward displacement (see Figure 5C). In the test trials the physical displacement was either upward or downward when presented in the intra-saccadic period or it was upward + leftward or downward + leftward when presented in the post-saccadic period. As in Experiment 2, habituation was much stronger in the intra-saccadic period and selective for the direction of the displacement.

Figure 5

Download asset Open asset

A non-parametric repeated measures ANOVA was calculated separately for data deriving from sessions with upward displacements in the context trials (see Figure 5B) and from those including downward displacements (see Figure 5C). For contexts with upward displacements, a significant main effect ‘displacement time’ (F(1,4) = 65.556, p<0.001) and a significant interaction effect (F(1,4) = 19.883, p=0.011) was revealed, confirming that the selective decrease of discrimination occurred during but not after saccade performance. The factor ‘displacement direction’ was not significant (F(1,4) = 5.00, p=0.089). For contexts with downward displacements, a significant main effect ‘displacement time’ (F(1,4) = 16.337, p=0.016) and a significant interaction effect (F(1,4) = 25.600, p=0.007) was revealed, confirming that the selective decrease of discrimination occurred during but not after saccade performance. The factor ‘displacement direction’ was not significant (F(1,4) = 4.986, p=0.089).

The results of this study clearly demonstrate that saccade suppression magnitude is strongly modulated by the recent context of intra-saccadic stimulation. After hundreds of saccades without intra-saccadic stimulation, observers were able to detect intra-saccadic displacements almost as good as during fixation. However, if the displacement was presented systematically within the period of saccade execution, thresholds increased drastically. The visual system habituates context-dependent to saccade-contingent visual information. This results provides evidence for a possible mechanism of peri-saccadic suppression. Instead of a general shut-down of vision, neurons might saturate to the information most prevalent in the previous saccadic context and thereby cancel out disturbing intra-saccadic stimulation. Importantly, any visual adaptation induced was contingent on the execution of a saccade. No habituation effect was found when the stimulation in the context trials was presented after a saccade had finished. The context dependent habituation therefore requires an extra-retinal signal that informs visual areas about the upcoming saccade. Such a context-dependence would be consistent with a recent Kalman filter model of saccade suppression in which higher variance in sensory feedback yields a reduced sensory weight, thereby decreasing sensitivity (Crevecoeur and Kording, 2017). In the model, the strength of saccade suppression is dynamically linked to the predicted reliability of the intra-saccadic sensory signal.

However, from the present data it cannot be determined how stimulus-specific trans-saccadic habituation is. Although I found direction-selectivity, it cannot be ruled out that habituation would also occur for changes of form or any visual stimulus occurring transiently during the saccade. One plausible candidate for the habituation to trans-sacadic displacements might be saturation of motion-sensitive neurons. Saturation of neuronal responses after short-term motion adaptation has been reported to occur in area MT (Priebe et al., 2002). The idea of the extra-retinal storage (see Figure 1) does not exclude however, that habituation occurs for any stimulus feature that reoccurs during saccade performance. This question should be pursued in future research by testing whether habituation is selective for trans-saccadic motion stimuli or if it generalizes across other visual features. The experimental task that I applied can in principle be solved without relying on motion information. The direction of a displacement can be detected by two ways: By noticing the motion transient between the first and the second image or if that is not possible by comparing feature locations of the first and the second image. Displacement detection is strong during ocular fixation (Legge and Campbell, 1981) but weak during saccades (Bridgeman et al., 1975; Deubel et al., 1996; Niemeier et al., 2003). In order to compare the results of the current study to the phenomenon of suppression of displacement, one should look at studies that displaced the saccade target constantly in the same direction before testing displacement detection. This was done when the effect of saccade adaptation on displacement detection was tested (Collins et al., 2009). In this study, no effect of saccade adaptation trials on the displacement thresholds was observed (see their Figure 3C). However what they did report was a shift of the bias in displacement detection after saccade adaptation. In my study however, the bias of displacement discrimination was statistically indistinguishable before and after the context trials. In the absence of a conscious displacement detection, the oculomotor system might be aware of the shift and - in the case of systematic displacements - adapt to it. Some studies have shown that saccades not only adapt to intra-saccadic jumps of the saccade target but also to displacements of the background (Deubel, 1991; Robinson et al., 2000; Ditterich et al., 2000). I compared the first and the last 10 saccade amplitudes of the context trials and did not find significant changes that would indicate adaptation. The most likely reason for the absence of adaptation is that the previous studies used background images containing focal elements that allow estimating the distance to the eye position. An increase of visual sensitivity starting around 50 ms before saccade onset can be observed at the focus of attention, that is the saccade target (Rolfs and Carrasco, 2012). However, I used a low-frequency sinusoidal grating without clear edge information. In the experiments of the present study, it cannot be decided to what extent participants employed a comparison between the pre-and post-saccadic image and how much they used the motion transient to judge the displacement. In real life we constantly perform saccades of various sizes and directions, producing different retinal motion stimulations. The direction-selectivity that I found suggests that different saccade vectors are connected to habituation of different displacement directions. If there would a general insensitivity to displacements, direction-selectivity could not have been observed.

It has been suggested that suppression of trans-saccadic displacements is determined by peri-saccadic suppression of contrast sensitivity. In this view, an active mechanism transiently shuts down the magnocellular pathway and thereby prevents an intra-saccadic motion sensation (Burr et al., 1994). Others have demonstrated that some intra-saccadic motion perception is possible if stimuli are made optimal to compensate for the retinal displacement (Castet and Masson, 2000). Testing whether suppression of visual sensitivity is context-sensitive as well could answer the question of a link between peri-saccadic contrast sensitivity and displacement discrimination directly. Under the assumption of such a linkage, performing a number of saccades on a background with high contrast (even without motion information) should increase peri-saccadic displacement thresholds. However, the direction selectivity that I found indicates already that the habituation cannot be explained by the decrease in contrast sensitivity which would affect all displacement directions likewise.

Peri-saccadic increase in contrast thresholds might underlie a similar habituation mechanism. It has been repeatedly reported that contrast suppression (Brooks and Fuchs, 1975; Brooks et al., 1980; Mitrani et al., 1971; Richards, 1969) and the neural modulations in area MT (Price et al., 2005) are absent when tested under dark conditions. Consistently, intra-saccadic motion becomes visible if the light is switched on only during execution of the saccade (Campbell and Wurtz, 1978). However, suppression remains strong when saccades are made under ocular diffuser or in a Ganzfeld (Riggs and Manning, 1982). Since the peri-saccadic stimulation is identical in the dark and the Ganzfeld and only the illumination differs, the differences between suppression under both conditions cannot be attributed to backward masking. The visual system can habituate to intra-saccadic motion only under light conditions, thus probably explaining the absence of suppression in the dark. If suppression of visual contrast depends on context, performing a number of saccades in the Ganzfeld should release habituation and thereby increase intra-saccadic contrast sensitivity.

In conclusion, the data of the present study indicate that saccade suppression is context dependent. The experiments suggest a simple mechanism that explains how the disturbing intra-saccadic experience is prohibited by saccade-contingent neuronal habituation.

All measured data have been deposited in OSF at https://doi.org/10.17605/OSF.IO/TZA7F.

1. 1. Zimmermann E

   (2019) Open Science Framework

   Saccade suppression depends on context.

   https://doi.org/10.17605/OSF.IO/TZA7F

1. 1. Binda P
   2. Morrone MC

   (2018) Vision during saccadic eye movements

   Annual Review of Vision Science 4:193–213.

   https://doi.org/10.1146/annurev-vision-091517-034317

   • PubMed
   • Google Scholar
2. 1. Bridgeman B
   2. Hendry D
   3. Stark L

   (1975) Failure to detect displacement of the visual world during saccadic eye movements

   Vision Research 15:719–722.

   https://doi.org/10.1016/0042-6989(75)90290-4

   • PubMed
   • Google Scholar
3. 1. Brooks BA
   2. Impelman DM
   3. Lum JT

   (1980) Influence of background luminance on visual sensitivity during saccadic eye movements

   Experimental Brain Research 40:322–329.

   https://doi.org/10.1007/BF00237797

   • Google Scholar
4. 1. Brooks BA
   2. Fuchs AF

   (1975) Influence of stimulus parameters on visual sensitivity during saccadic eye movement

   Vision Research 15:1389–1398.

   https://doi.org/10.1016/0042-6989(75)90196-0

   • PubMed
   • Google Scholar
5. 1. Burr DC
   2. Holt J
   3. Johnstone JR
   4. Ross J

   (1982) Selective depression of motion sensitivity during saccades

   The Journal of Physiology 333:1–15.

   https://doi.org/10.1113/jphysiol.1982.sp014434

   • Google Scholar
6. 1. Burr DC
   2. Morrone MC
   3. Ross J

   (1994) Selective suppression of the magnocellular visual pathway during saccadic eye movements

   Nature 371:511–513.

   https://doi.org/10.1038/371511a0

   • Google Scholar
7. 1. Campbell FW
   2. Wurtz RH

   (1978) Saccadic omission: why we do not see a grey-out during a saccadic eye movement

   Vision Research 18:1297–1303.

   https://doi.org/10.1016/0042-6989(78)90219-5

   • PubMed
   • Google Scholar
8. 1. Castet E
   2. Jeanjean S
   3. Masson GS

   (2002) Motion perception of saccade-induced retinal translation

   PNAS 99:15159–15163.

   https://doi.org/10.1073/pnas.232377199

   • Google Scholar
9. 1. Castet E
   2. Masson GS

   (2000) Motion perception during saccadic eye movements

   Nature Neuroscience 3:177–183.

   https://doi.org/10.1038/72124

   • PubMed
   • Google Scholar
10. 1. Collins T
    2. Rolfs M
    3. Deubel H
    4. Cavanagh P

    (2009) Post-saccadic location judgments reveal remapping of saccade targets to non-foveal locations

    Journal of Vision 9:29.

    https://doi.org/10.1167/9.5.29

    • PubMed
    • Google Scholar
11. 1. Crevecoeur F
    2. Kording KP

    (2017) Saccadic suppression as a perceptual consequence of efficient sensorimotor estimation

    eLife 6:e25073.

    https://doi.org/10.7554/eLife.25073

    • PubMed
    • Google Scholar
12. 1. Deubel H

    (1991)

    Presbyopia Research. Perspectives in Vision Research

    Adaptive Control of Saccade Metrics, Presbyopia Research. Perspectives in Vision Research, Boston, Springer, 10.1007/978-1-4757-2131-7.

    • Google Scholar
13. 1. Deubel H
    2. Schneider WX
    3. Bridgeman B

    (1996) Postsaccadic target blanking prevents saccadic suppression of image displacement

    Vision Research 36:985–996.

    https://doi.org/10.1016/0042-6989(95)00203-0

    • PubMed
    • Google Scholar
14. 1. Ditterich J
    2. Eggert T
    3. Straube A

    (2000) The role of the attention focus in the visual information processing underlying saccadic adaptation

    Vision Research 40:1125–1134.

    https://doi.org/10.1016/S0042-6989(00)00018-3

    • Google Scholar
15. 1. Duhamel JR
    2. Colby CL
    3. Goldberg ME

    (1992) The updating of the representation of visual space in parietal cortex by intended eye movements

    Science 255:90–92.

    https://doi.org/10.1126/science.1553535

    • PubMed
    • Google Scholar
16. 1. Duyck M
    2. Collins T
    3. Wexler M

    (2016) Masking the saccadic smear

    Journal of Vision 16:1–13.

    https://doi.org/10.1167/16.10.1

    • PubMed
    • Google Scholar
17. 1. Duyck M
    2. Wexler M
    3. Castet E
    4. Collins T

    (2018) Motion masking by stationary objects: a study of simulated saccades

    I-Perception 9:1–11.

    https://doi.org/10.1177/2041669518773111

    • Google Scholar
18. 1. Ilg UJ
    2. Hoffmann KP

    (1993) Motion perception during saccades

    Vision Research 33:211–220.

    https://doi.org/10.1016/0042-6989(93)90159-T

    • PubMed
    • Google Scholar
19. 1. Knöll J
    2. Binda P
    3. Morrone MC
    4. Bremmer F

    (2011) Spatiotemporal profile of peri-saccadic contrast sensitivity

    Journal of Vision 11:15.

    https://doi.org/10.1167/11.14.15

    • PubMed
    • Google Scholar
20. 1. Legge GE
    2. Campbell FW

    (1981) Displacement detection in human vision

    Vision Research 21:205–213.

    https://doi.org/10.1016/0042-6989(81)90114-0

    • PubMed
    • Google Scholar
21. 1. Luke SG

    (2017) Evaluating significance in linear mixed-effects models in R

    Behavior Research Methods 49:1494–1502.

    https://doi.org/10.3758/s13428-016-0809-y

    • PubMed
    • Google Scholar
22. 1. Mitrani L
    2. Yakimoff N
    3. Mateeff S

    (1971) Is saccadic suppression really saccadic?

    Vision Research 11:1157–1161.

    https://doi.org/10.1016/0042-6989(71)90119-2

    • PubMed
    • Google Scholar
23. 1. Nakamura K
    2. Colby CL

    (2002) Updating of the visual representation in monkey striate and extrastriate cortex during saccades

    PNAS 99:4026–4031.

    https://doi.org/10.1073/pnas.052379899

    • Google Scholar
24. 1. Niemeier M
    2. Crawford JD
    3. Tweed DB

    (2003) Optimal transsaccadic integration explains distorted spatial perception

    Nature 422:76–80.

    https://doi.org/10.1038/nature01439

    • PubMed
    • Google Scholar
25. 1. O'Regan JK
    2. Noë A

    (2001) A sensorimotor account of vision and visual consciousness

    Behavioral and Brain Sciences 24:939–973.

    https://doi.org/10.1017/S0140525X01000115

    • PubMed
    • Google Scholar
26. 1. Price NS
    2. Ibbotson MR
    3. Ono S
    4. Mustari MJ

    (2005) Rapid processing of retinal slip during saccades in macaque area MT

    Journal of Neurophysiology 94:235–246.

    https://doi.org/10.1152/jn.00041.2005

    • PubMed
    • Google Scholar
27. 1. Priebe NJ
    2. Churchland MM
    3. Lisberger SG

    (2002) Constraints on the source of short-term motion adaptation in macaque area MT. I. the role of input and intrinsic mechanisms

    Journal of Neurophysiology 88:354–369.

    https://doi.org/10.1152/jn.00852.2001

    • PubMed
    • Google Scholar
28. 1. Richards W

    (1969) Saccadic suppression

    Journal of the Optical Society of America 59:617–624.

    https://doi.org/10.1364/JOSA.59.000617

    • PubMed
    • Google Scholar
29. 1. Riggs LA
    2. Manning KA

    (1982)

    Saccadic suppression under conditions of whiteout

    Investigative Ophthalmology & Visual Science 23:138–143.

    • PubMed
    • Google Scholar
30. 1. Robinson F
    2. Noto C
    3. Watanabe S

    (2000) Effect of visual background on saccade adaptation in monkeys

    Vision Research 40:2359–2367.

    https://doi.org/10.1016/S0042-6989(00)00079-1

    • PubMed
    • Google Scholar
31. 1. Rolfs M
    2. Carrasco M

    (2012) Rapid simultaneous enhancement of visual sensitivity and perceived contrast during saccade preparation

    Journal of Neuroscience 32:13744–13752.

    https://doi.org/10.1523/JNEUROSCI.2676-12.2012

    • Google Scholar
32. 1. Shioiri S
    2. Cavanagh P

    (1989) Saccadic suppression of low-level motion

    Vision Research 29:915–928.

    https://doi.org/10.1016/0042-6989(89)90106-5

    • Google Scholar
33. 1. Sommer MA
    2. Wurtz RH

    (2006) Influence of the thalamus on spatial visual processing in frontal cortex

    Nature 444:374–377.

    https://doi.org/10.1038/nature05279

    • PubMed
    • Google Scholar
34. 1. Stevenson SB
    2. Volkmann FC
    3. Kelly JP
    4. Riggs LA

    (1986) Dependence of visual suppression on the amplitudes of saccades and blinks

    Vision Research 26:1815–1824.

    https://doi.org/10.1016/0042-6989(86)90133-1

    • Google Scholar
35. 1. Umeno MM
    2. Goldberg ME

    (1997) Spatial processing in the monkey frontal eye field. I. predictive visual responses

    Journal of Neurophysiology 78:1373–1383.

    https://doi.org/10.1152/jn.1997.78.3.1373

    • PubMed
    • Google Scholar
36. 1. Volkmann FC
    2. Riggs LA
    3. White KD
    4. Moore RK

    (1978) Contrast sensitivity during saccadic eye movements

    Vision Research 18:1193–1199.

    https://doi.org/10.1016/0042-6989(78)90104-9

    • Google Scholar
37. 1. Volkmann FC
    2. Riggs LA
    3. Ellicott AG
    4. Moore RK

    (1981)

    The dependence of saccadic suppression on the amplitude of eye movement

    Investigative Ophthalmology & Visual Science 20:52.

    • Google Scholar
38. 1. Volkmann FC

    (1986) Human visual suppression

    Vision Research 26:1401–1416.

    https://doi.org/10.1016/0042-6989(86)90164-1

    • PubMed
    • Google Scholar
39. 1. Walker MF
    2. Fitzgibbon EJ
    3. Goldberg ME

    (1995) Neurons in the monkey superior colliculus predict the visual result of impending saccadic eye movements

    Journal of Neurophysiology 73:1988–2003.

    https://doi.org/10.1152/jn.1995.73.5.1988

    • Google Scholar
40. Conference

    1. Wobbrock J
    2. Findlater L
    3. Gergle D
    4. Higgins J

    (2011) The aligned rank transform for nonparametric factorial analyses using only ANOVA procedures

    Proceedings of the ACM Conference on Human Factors in Computing Systems (CHI). pp. 143–146.

    https://doi.org/10.1145/1978942.1978963

    • Google Scholar
41. 1. Wurtz RH

    (2018) Corollary discharge contributions to perceptual continuity across saccades

    Annual Review of Vision Science 4:215–237.

    https://doi.org/10.1146/annurev-vision-102016-061207

    • PubMed
    • Google Scholar

Thank you for submitting your article "Saccade suppression of motion depends on context" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Floris de Lange as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Martin Rolfs (Reviewer #2); Therese Collins (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

In this study, the author investigated the impact of recent visual experience on suppression of intra-saccadic visual stimulation. This is a timely study and the idea is intriguing, the experiments are well designed, and the results seem robust. But some major and other theoretical concerns should be addressed. The reviewers suggest an experiment that should be conducted which would address some major concerns. In addition, many methodological and data analysis issues should be stated and/or clarified. Lastly, a suggested additional analysis would strengthen the manuscript.

Essential revisions:

1) It is not clear that the author studies motion perception at all. Instead, the author uses a displacement task, which seems to rely on position rather than motion information. Indeed, the author might not be studying the perception and suppression of motion (as studied by Burr, Castet, and others), but of displacement (i.e., Bridgeman, Hendry, and Stark, 1975; Niemeier, Crawford, and Tweed, 2003). Both processes have been associated with rather different theoretical implications, and with an emphasis on different aspects of visual stability. Indeed, it seems that the study says more about the visual system's ability to disregard displacements of the retinal image across the saccade rather than about the perceptual omission of intra-saccadic motion. In my view, this does not reduce the novelty of the approach, but leads to a rather different interpretation of the meaning of these results with respect to understanding visual stability.

2) There are insufficient links to previous work about motion perception during saccades. In the Abstract, the author states that "…the mechanism that inhibits intra-saccadic motion detection is unknown". The author cites only Castet and Masson, 2000, but further work from Eric Castet and colleagues should be consulted: Castet, Jeanan and Masson, 2002; Duyck, Wexler, Castet and Collins, 2018, as well as Duyck, Collins and Wexler, 2016. These studies show that visual masking is a powerful determinant of saccadic suppression (as did Campbell and Wurtz, 1978). Thus, although some aspects of saccadic suppression of motion do remain to be fully described and uncovered, it is a misrepresentation of the literature to state that they are unknown.

Indeed, there is a dominant view/theory in the field: that the mechanism of reduced motion sensitivity is the peri-saccadic suppression of the magnocellular pathway, instantiated via efference-copy signals at the earliest stages of visual processing (Diamond et al., 2000; Thier et al; Burr et al., 1994; Burr and Ross, 1982; reviewed in Binda and Morrone, 2018). It seems that the present manuscript provides an alternative account to this dominant view and that should be emphasized here.

3) The article would also benefit from a clearer explanation of the difference between what Burr et al., 1994, showed, which is that there is a small decrease in contrast sensitivity (which they hypothesized might contribute to decreased motion sensitivity during saccades, but which they did not test); and what Castet and colleagues showed, which is that suppression of motion during saccades results not from this decrease of contrast sensitivity but from masking of static images before and after saccades.

4) Specify what other areas in addition to intra-parietal cortex shift their receptive fields predictively. There should be a more explicit link between the sentence of shifting receptive fields (Introduction) and the following sentence dealing with suppression of intra-saccadic motion perception. Additionally, it would be good to mention if some of the areas in which receptive fields shift predictively are likely to play a role in suppression of intra-saccadic motion perception.

5) The author should provide a clearer description of the model that he has in mind to account for reduced visibility of motion/displacement during saccades. In the Introduction, he states: "In this view, motion-sensitive neurons saturate their response to the visual information most dominant in the previous history of intra-saccadic stimulation." The idea expressed here is extended in the Discussion, but it needs further explanation. The author suggests that the system stores information (e.g. due to habituation) about motion that was recently experienced, but how could that motion be exclusive to saccades? Are the motion- (or displacement-) sensitive neurons that undergo habituation tuned to extremely high speeds? Or is the habituation linked to an efference-copy signal, forming a sensorimotor contingency? Perhaps visualize this with a schematic diagram.

6) The author states that saccade suppression starts 50 ms before saccade onset (Introduction). Interestingly, during this timing presaccade attention enhances perception (e.g. Kowler et al., 1995; Deubel and Schneider, 1996; Rolfs and Carrasco, 2012; Li, Barbot and Carrasco, 2016; Szinte, Jonikaitis, Rangelov and Deubel, 2018). What are your thoughts regarding whether and how these two phenomena may interact?

7) The author suggests a saccade-contingent habituation to intra-saccadic motion. It would be good to specify what 'previous history' means here (Introduction). Also, how would previous history translate to real life in which we constantly make saccades in all directions?

Methods and Analysis

8) The Materials and methods section does not contain any information on data pre-processing and data analysis, and many important aspects of the experimental methods are not specified.

9) Why was the design of Experiment 2 unbalanced? In other words, why weren't there test trials with downward motion? The result would be more convincing with a balanced design showing that motion detection thresholds were elevated in downward motion test trials as well. This is particularly important given known differences between upwards and downwards motion perception (e.g., due to gravity). If robust, the results reported here should be valid for both motion directions.

10) It is not always obvious what one is looking at in the figures, and understanding requires flipping back and forth between the Materials and methods and Results. It would be more conducive to comprehension if some details were made available at the appropriate parts of the Results section. For example, the psychometric functions in Figure 1G-H: aside from the fact that the figure is too small for comfortable reading, it seems that there must be a mistake in the x-axis label. It currently reads "time rel. to saccade onset", but should probably be "motion direction in test trials".

Another example, in Figure 2F-H: it is unclear whether the x-axis refers to time relative to saccade onset in context or test trials. Unfortunately, the text and figure captions do not make it any clearer (especially since the text says that motion discrimination performance in the test trials was analyzed in bins of 30 ms, but the x-axis is not in bins of 30 ms).

11) The timing of the stimuli with respect to the saccade is very critical in these experiments and more information needs to be provided. The author suggests that stimuli targeted one of 3 different times relative to saccade onset. Did the author confirm accurate timing of the displacements offline? How did the detected onset of a saccade online compare to the saccade onset detected offline? The author should provide measurements of their stimulus timing (using a photodiode or similar) relative to this ground truth.

Schweitzer and Rolfs (2019; https://www.biorxiv.org/content/10.1101/693309v1) have recently provided a detailed description of the delays contributing here, many of which are often unknown and need to be measured in a particular setup.

12) The author states, "In my experiments the adapting stimulus was a single-frame displacement only, whose effectivity might have accumulated over the bunch of trials.” To assess whether this was the case, the author should analyze separately the first vs. the second halves of the experiments and compare them.

Additional experiment

13) Across the two experiments, the author tests the specificity of a reduction in saccadic suppression to presentations occurring (1) during the saccade and (2) in directions that are congruent with those experienced during the context trials. However, given that the vertical displacement occurs during the saccade and given that the grating has vertical boundaries at the edges of the screen (they must be there, however minimal they may be), the vector of motion on the retina is changed due to the movement of the retina itself. Thus, the stimulus' motion across the retina has a vertical (due to stimulus motion on the screen) as well as a horizontal component (in the direction opposite the saccade). As a consequence, the context trials in which the motion/displacement is shown during the saccade should habituate the system to this combined retinal motion direction. If, as the author argues, the habituation is direction-specific, the consequence of this habituation (higher suppression) would thus only show up during test trials in which the motion/displacement is shown during the saccade, as the motion in these trials is again of the same direction. It would not be expected to show up in conditions (or to a reduced extent, compatible with the data) in which the context trials showed vertical motion while the eyes were stationary, as motion/displacement during the test trials (with intra-saccadic tests) would no longer be congruent with that experienced during the context trials.

An experiment to test this would use two different sets of context-trials in which the retinal direction of the stimulus is matched between intra-saccadic and post-saccadic context conditions. Alternatively, the context trials could be kept the way they were done in Experiment 2, but now the test trials in both intra-saccadic and post-saccadic context conditions could have one of two retinal directions: the direction imposed on the retina during saccades (vertical + horizontal) or the direction imposed during fixation (purely vertical). These experiments would reveal if the habituation is indeed specific to saccades or if the specificity to saccades found in the first two experiments is merely a consequence of differences in retinal motion trajectories. Moreover, the experiment should be run with a balanced design; i.e., habituation to upwards and downwards motion should be compared.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for re-submitting your article "Saccade suppression of motion depends on context" for consideration by eLife. Your article has been re-reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Floris de Lange as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Martin Rolfs (Reviewer #2); Therese Collins (Reviewer #3).

Our decision has been reached after consultation among the reviewers and the action editor. The reviewers appreciate that you have added new experiments to address the concerns raised in the previous round. They think that the findings from Experiment 3 are convincing, but that the findings from the Supplementary experiment are problematic for your proposal that this study deals with motion discrimination rather than displacement discrimination. Based on these discussions and the individual reviews below, we are inviting you to resubmit your manuscript, we are proposing a way forward for you to consider.

Experiment 3 seems convincing because it tries to match saccade-induced and screen motion, uses a balanced design (both upwards and downwards context motions are tested), and provides converging evidence for the directional tuning of the contextual habituation effect. Thus, the results of Experiment 2 are unlikely to be a consequence of differences in retinal motion vectors between different context conditions.

In the supplementary experiment a full-field random pattern mask flashed during saccade execution diminishes performance (comparison of pre- and post-saccadic target locations). The author concludes that because the mask abolished the motion transient, it is the motion transient that drives performance.

However, there is a serious logical flaw in this experiment. The results seem to be inconsistent with the author's theoretical proposal; i.e., that saccadic omission results from habituation to the motion transients that accompany every saccade (and therefore there is no need for special trans-saccadic mechanisms as claimed by many in the literature). In support of this claim, Experiment 1 shows that exposing observers to artificial intra-saccadic motion degrades motion discrimination performance (relative to performance without exposure to artificial intra-saccadic motion). The supplementary experiment is essentially the same as Experiment 1, except that now the upwards motion is masked by a full-field random pattern (although it still remains to be shown whether such a pattern actually does mask motion during normal viewing, i.e. fixation). If habituation to intra-saccadic motion explains poor motion discrimination, then elimination of intra-saccadic motion should release discrimination performance. Performance should therefore improve or, at least, remain comparable to the no-exposure condition (because there is, of course, still the usual saccade-induced motion transient in the no-exposure condition). Instead, results show the reverse – performance gets worse – which cannot easily be reconciled with the theoretical claim the author is making. Even if we consider that the mask itself generates motion on the retina during the saccade, the vector of that motion is horizontal and should not –according to the author's proposal – affect vertical motion detection.

Indeed, in the rebuttal, the author argues that his results do speak to motion perception because detecting the displacement change across saccades requires visibility of the motion transient. In support of his argument, he cites Shiori and Cavanagh, 1989, who showed that sensitivity to the motion transient was abolished during saccades. But this doesn't make much sense: if the motion transient is not perceived during the saccade, how can it be used to drive displacement reports? He does not cite (but could have) experiments performed by Deubel and Schneider in which they showed that performance increases dramatically when there is a blank (that diminishes the motion transient) between pre- and post-saccadic targets. So it seems that there is evidence that diminishing the motion transient may actually help trans-saccadic location performance.

Previous work from the author and Sabine Born showed that it is possible to mimic poor performance in a fixation task in which a mask occurs between the two targets. However, being able to degrade performance in two different ways (either by presenting targets before and after a saccade, or by presenting them before and after a mask) does not mean that the same processes are involved.

An alternative interpretation of the results is that the intra-saccadic mask might induce uncertainty about the initial position of the grating, making the displacement harder to judge (rather than rendering the motion invisible); for instance, the masking pattern may have itself induced transients on the retina that caused the system to down-regulate any input during subsequent saccades. This would argue that the author's protocol does not study motion perception (as expressed in the title of the manuscript), but displacement detection.

To justify the conclusion that is the motion transient that drives performance, two further pieces of information would be needed. (1) Does such a full-field mask abolish a motion transient of an embedded target (in a fixation condition)? (2) Why is the invisibility of the motion transient during a saccade (as shown by Shiori and Cavanagh) not sufficient? The author would have to provide evidence that the manipulation does indeed reduce motion perception while leaving the ability to judge displacements intact.

In sum, the only condition in which performance increased after context trials was the context in which the author minimized changes on the retina during the saccade (no displacement context trials). It seems, therefore, that any visual changes during saccades result in insensitivity during subsequent saccades. The plasticity in intrasaccadic omission remains an interesting finding, but it deviates quite strongly from the claim in the title/Abstract/conclusion that this paper is about saccade suppression of motion. Thus, we invite you to provide an alternative interpretation to saccadic suppression of motion for your current findings. Note that this would entail a re-writing of the entire manuscript, theory included. Please note that in eLife it is not customary to have many review rounds; thus, if you decide to take this route please try to make a comprehensive and clear a case as possible.

Analysis

In several of the new results, the author conducted multiple t-tests rather than an ANOVA, which is not good statistical practice. Take as an example: in Experiment 2, the author tests for direction-specificity of the contextual habituation with four conditions resulting from the crossing of two factors, Direction of Motion in Test Trials (Upward/Downward) and Time of shift (intra-sacc: 35 ms/post-sacc: 98 ms). In the results he only presents t-test on the Direction Factor, independently for each time of shift. Because the t-test is significant in intra-sacc, and not in post-sacc, the author concludes that there is direction-specific intra-saccadic habituation. Of course, this conclusion only holds if there is a statistically reliable interaction between the two factors. Factorial designs should be analyzed with ANOVAs throughout, instead of t-tests. To respond to the sample size critique, the author could use a linear mixed model with p-values estimated by the methods explained by Luke (2017, Behav Research Methods), which apparently is less sensitive to sample size than ANOVAs and t-tests.

One issue that further complicates the interpretation of the results of the supplementary experiment is that it is difficult to have an idea of power. How big of a difference in slope can 4 subjects reveal (given the variability seen in the other experiments, and especially when several cannot be fit by a cumulative Gaussian)?

Essential revisions:1) It is not clear that the author studies motion perception at all. Instead, the author uses a displacement task, which seems to rely on position rather than motion information. Indeed, the author might not be studying the perception and suppression of motion (as studied by Burr, Castet, and others), but of displacement (i.e., Bridgeman, Hendry, and Stark, 1975; Niemeier, Crawford, and Tweed, 2003). Both processes have been associated with rather different theoretical implications, and with an emphasis on different aspects of visual stability. Indeed, it seems that the study says more about the visual system's ability to disregard displacements of the retinal image across the saccade rather than about the perceptual omission of intra-saccadic motion. In my view, this does not reduce the novelty of the approach, but leads to a rather different interpretation of the meaning of these results with respect to understanding visual stability.

I agree that this is an important issue. Since I used a displacement, a link to the SSD phenomenon seems likely. However, there are arguments that speak against such a connection. For this reason, I avoided this topic in my first version of the manuscript. However, now I have added a large part in the Discussion section.

The question here is – as the reviewer point out – whether observers in the displacement task rely on position rather than motion. In the SSD literature the poor trans-saccadic displacement detection is usually explained by a saccadic suppression of the motion transient. Consequently, if observers can rely only on position their discrimination performance drops down. The reviewers suggest that through the context trials in which there is no motion observers improve their ability to rely on position, i.e. to compare pre and post-saccadic locations.

In order to test the ability of observers to compare pre-and post-saccadic location, I conducted an extra experiment. I modified my Experiment 1 such that the displacement in the test trials contained a whole-field random pattern mask that was presented for 24 ms exactly at the time when the displacement occurred. The motion transient of the displacement was therefore covered by the mask. The mask was presented short enough to be only on screen during saccade execution. If observers would be able to compare pre-and post-saccadic locations in the absence of the motion transient, they should be able to solve this task. However, for most of the data, a psychometric function could not even be estimated because the discrimination performance was very poor. Therefore, I conclude that the motion transient is necessary for trans-saccadic discrimination performance and that my context trials modulate the visibility of the motion transient.

I now write:

“In principle, the experimental task that I applied can be solved without motion information. […] Together, these arguments let it seem very unlikely that observers compared pre- and post-saccadic locations of image parts.”

2) There are insufficient links to previous work about motion perception during saccades. In the Abstract, the author states that "…the mechanism that inhibits intra-saccadic motion detection is unknown". The author cites only Castet and Masson, 2000, but further work from Eric Castet and colleagues should be consulted: Castet, Jeanan and Masson, 2002; Duyck, Wexler, Castet and Collins, 2018, as well as Duyck, Collins and Wexler, 2016. These studies show that visual masking is a powerful determinant of saccadic suppression (as did Campbell and Wurtz, 1978). Thus, although some aspects of saccadic suppression of motion do remain to be fully described and uncovered, it is a misrepresentation of the literature to state that they are unknown.

I have now re-written large parts of the Introduction:

“As it has been shown that the visual system is able to detect high-speed motion (Burr et al., 1982) the question arises why we are not aware of the motion stimulation produced by a saccade. […] Recently, Duyck, Wexler, Castet and Collins, 2018, used simulated saccades and found motion perception to be much less salient in the presence of static objects.”

Indeed, there is a dominant view/theory in the field: that the mechanism of reduced motion sensitivity is the peri-saccadic suppression of the magnocellular pathway, instantiated via efference-copy signals at the earliest stages of visual processing (Diamond et al., 2000; Thier et al.; Burr et al., 1994; Burr and Ross, 1982; reviewed in Binda and Morrone, 2018). It seems that the present manuscript provides an alternative account to this dominant view and that should be emphasized here.

I thank the reviewers for giving me the chance to point out the novelty of my account. I now write:

“Here, I offer a novel explanation of suppression which states that the brain stores and habituates to sensorimotor contingencies. […] In this view, motion-sensitive neurons store saccade-induced visual stimulation and saturate their response to the visual information most dominant in the previous history of the last bunch of saccades that has been executed.”

3) The article would also benefit from a clearer explanation of the difference between what Burr et al., 1994, showed, which is that there is a small decrease in contrast sensitivity (which they hypothesized might contribute to decreased motion sensitivity during saccades, but which they did not test); and what Castet and colleagues showed, which is that suppression of motion during saccades results not from this decrease of contrast sensitivity but from masking of static images before and after saccades.

I now describe the differences between the approaches by Burr et al., 1994, and Castet more clearly in the Introduction. Please see my reply to the reviewers’ point 2.

Additionally, I now write in the Discussion:

“It is unclear if suppression of motion transients is determined by peri-saccadic suppression of contrast sensitivity. […] However, the motion direction selectivity that I found indicates already that the habituation cannot be explained by the decrease in contrast sensitivity which would affect all motion directions likewise.”

Please note that already in the first version of the manuscript I stated in the last sentence of this paragraph (last sentence of the above quoted paragraph) that the decrease in contrast sensitivity does not explain the unawareness of intra-saccadic motion. I hope the new version now is more explicit on this point.

4) Specify what other areas in addition to intra-parietal cortex shift their receptive fields predictively. There should be a more explicit link between the sentence of shifting receptive fields (Introduction) and the following sentence dealing with suppression of intra-saccadic motion perception. Additionally, it would be good to mention if some of the areas in which receptive fields shift predictively are likely to play a role in suppression of intra-saccadic motion perception.

This section was meant to introduce the concept of efference copy which is primarily discussed in the remapping literature. I did not mean to say that remapping might somehow be related to suppression. I now rewrote that section and I also specify the other areas in which receptive field shifts have been measured. This section now reads:

“Learning sensorimotor contingencies requires a signal that indicates the initiation of a movement. […] The suppression of intra-saccadic motion perception might be driven by such an extra-retinal signal likewise (for a review, see Binda and Morrone, 2018).”

5) The author should provide a clearer description of the model that he has in mind to account for reduced visibility of motion/displacement during saccades. In the Introduction, he states: "In this view, motion-sensitive neurons saturate their response to the visual information most dominant in the previous history of intra-saccadic stimulation." The idea expressed here is extended in the Discussion, but it needs further explanation. The author suggests that the system stores information (e.g. due to habituation) about motion that was recently experienced, but how could that motion be exclusive to saccades? Are the motion- (or displacement-) sensitive neurons that undergo habituation tuned to extremely high speeds? Or is the habituation linked to an efference-copy signal, forming a sensorimotor contingency? Perhaps visualize this with a schematic diagram.

Yes indeed, I propose that storage of the past intra-saccadic motion stimulation is saccade-contingent. Indeed, this storage is linked to an efference-copy signal and thereby exclusive to saccades. I have now created a schematic diagram (see Figure 1) explaining this idea, as the reviewer suggested.

6) The author states that saccade suppression starts 50 ms before saccade onset (Introduction). Interestingly, during this timing presaccade attention enhances perception (e.g. Kowler et al., 1995; Deubel and Schneider, 1996; Rolfs and Carrasco, 2012; Li, Barbot and Carrasco, 2016; Szinte, Jonikaitis, Rangelov and Deubel, 2018). What are your thoughts regarding whether and how these two phenomena may interact?

This is a challenging question as attention shifts (increase of contrast sensitivity (Rolfs and Carrasco, 2012)) and saccade suppression (decrease of contrast sensitivity) lead to opposite effects. Both processes can co-exist because the benefits of attention shifts are at a focal region around the attended target which is usually the saccade target (Kowler et al., 1995; Deubel and Schneider, 1996) or under special conditions an increased region around the saccade target (Szinte, Puntiroli and Deubel, 2019). Suppression is mostly studied with gratings or bars that fill a large region of the visual field that is not the focus of attention. Interestingly, focal light points indeed are not suppressed (Brook and Fuchs, 1975). One could go further and argue that the insensitivity of contrast for the unattended regions is a consequence of the enhancement at the attended region. However, there is so far no evidence for a causal link between both processes. I now write in the Introduction:

“An increase of visual sensitivity starting around 50 ms before saccade onset can be observed at the focus of attention, i.e. the saccade target (Rolfs and Carrasco, 2012).”

7) The author suggests a saccade-contingent habituation to intra-saccadic motion. It would be good to specify what 'previous history' means here (Introduction). Also, how would previous history translate to real life in which we constantly make saccades in all directions?

I now write “in the previous history of the last bunch of saccades that has been executed.”. In the Discussion, I now suggest that in real life different saccade directions must involve different habituations. This is suggested by the direction-selectivity of the habituation as I now write:

“In real life we constantly perform saccades of various sizes and directions, producing different motion stimulations. The direction-selectivity that I found suggests that different saccade vectors are connected to habitation of different motion directions. If there would a general insensitivity to motion, direction-selectivity could not have been observed.”

Methods and Analysis8) The Materials and methods section does not contain any information on data pre-processing and data analysis, and many important aspects of the experimental methods are not specified.

I have now added the missing information.

9) Why was the design of Experiment 2 unbalanced? In other words, why weren't there test trials with downward motion? The result would be more convincing with a balanced design showing that motion detection thresholds were elevated in downward motion test trials as well. This is particularly important given known differences between upwards and downwards motion perception (e.g., due to gravity). If robust, the results reported here should be valid for both motion directions.

I agree and have now included a balanced design in my new Experiment 3.

10) It is not always obvious what one is looking at in the figures, and understanding requires flipping back and forth between the Materials and methods and Results. It would be more conducive to comprehension if some details were made available at the appropriate parts of the Results section. For example, the psychometric functions in Figure 1G-H: aside from the fact that the figure is too small for comfortable reading, it seems that there must be a mistake in the x-axis label. It currently reads "time rel. to saccade onset", but should probably be "motion direction in test trials".Another example, in Figure 2F-H: it is unclear whether the x-axis refers to time relative to saccade onset in context or test trials. Unfortunately, the text and figure captions do not make it any clearer (especially since the text says that motion discrimination performance in the test trials was analyzed in bins of 30 ms, but the x-axis is not in bins of 30 ms).

Thanks. I have corrected these errors. I have also put more details into the description of the results and I hope that it is more comprehensive now.

11) The timing of the stimuli with respect to the saccade is very critical in these experiments and more information needs to be provided. The author suggests that stimuli targeted one of 3 different times relative to saccade onset. Did the author confirm accurate timing of the displacements offline? How did the detected onset of a saccade online compare to the saccade onset detected offline? The author should provide measurements of their stimulus timing (using a photodiode or similar) relative to this ground truth.Schweitzer and Rolfs (2019; https://www.biorxiv.org/content/10.1101/693309v1) have recently provided a detailed description of the delays contributing here, many of which are often unknown and need to be measured in a particular setup.

Yes, this is an important point. I have carefully measured the display lag, i.e. the duration between the command to draw the stimulus is sent from the stimulus program and the actual appearance of the stimulus on the screen. This was done the following way as described also in the new section “Timing of stimulus presentation”:

“Timing of stimulus presentation

In order to test the exact time of stimulus presentation on the screen, I used a photodiode that was connected to an electronic circuit and sent a TTL impulse to the parallel port of the eyelink PC for every luminance change on the screen. […] With the CRT monitor that I used this display lag was measured to be on average 4.35 ms (S.E.M. 0.29). All data in the analysis were shifted in time by 4.35 ms.”

12) The author states, "In my experiments the adapting stimulus was a single-frame displacement only, whose effectivity might have accumulated over the bunch of trials.” To assess whether this was the case, the author should analyze separately the first vs. the second halves of the experiments and compare them.

As the reviewer suggested, I split my data into two halves and looked at the suppression strength separately. I found that already in the first half, suppression was reduced in the sessions without motion in the context trials. The strength of this effect was nearly identical to the result obtained when analyzing the total data set. What I meant to say with the sentence above is that it is likely that the habituation grows over several experiences of intra-saccadic displacements. It will surely not just happen with a single trial only. As for saccade adaptation, it is a separate question how many trials are needed to induce the effect (in my case: the habituation). This should be tested by varying the amount t of context trials in the beginning of a test session. For now, I have dropped the sentence quoted by the reviewer.

Additional experiment13) Across the two experiments, the author tests the specificity of a reduction in saccadic suppression to presentations occurring (1) during the saccade and (2) in directions that are congruent with those experienced during the context trials. However, given that the vertical displacement occurs during the saccade and given that the grating has vertical boundaries at the edges of the screen (they must be there, however minimal they may be), the vector of motion on the retina is changed due to the movement of the retina itself. Thus, the stimulus' motion across the retina has a vertical (due to stimulus motion on the screen) as well as a horizontal component (in the direction opposite the saccade). As a consequence, the context trials in which the motion/displacement is shown during the saccade should habituate the system to this combined retinal motion direction. If, as the author argues, the habituation is direction-specific, the consequence of this habituation (higher suppression) would thus only show up during test trials in which the motion/displacement is shown during the saccade, as the motion in these trials is again of the same direction. It would not be expected to show up in conditions (or to a reduced extent, compatible with the data) in which the context trials showed vertical motion while the eyes were stationary, as motion/displacement during the test trials (with intra-saccadic tests) would no longer be congruent with that experienced during the context trials.An experiment to test this would use two different sets of context-trials in which the retinal direction of the stimulus is matched between intra-saccadic and post-saccadic context conditions. Alternatively, the context trials could be kept the way they were done in Experiment 2, but now the test trials in both intra-saccadic and post-saccadic context conditions could have one of two retinal directions: the direction imposed on the retina during saccades (vertical + horizontal) or the direction imposed during fixation (purely vertical). These experiments would reveal if the habituation is indeed specific to saccades or if the specificity to saccades found in the first two experiments is merely a consequence of differences in retinal motion trajectories. Moreover, the experiment should be run with a balanced design; i.e., habituation to upwards and downwards motion should be compared.

I thank the reviewer for giving me the chance to provide a clear answer on that question. I have carried out as the reviewers suggested. I describe the details of this experiment in the corresponding Materials and methods and Results sections.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The reviewers appreciate that you have added new experiments to address the concerns raised in the previous round. They think that the findings from Experiment 3 are convincing, but that the findings from the Supplementary experiment are problematic for your proposal that this study deals with motion discrimination rather than displacement discrimination. Based on these discussions and the individual reviews below, we are inviting you to resubmit your manuscript, we are proposing a way forward for you to consider.

I agree with the reviewers that the data from the supplementary experiment alone do not provide sufficient evidence to decide whether this study deals with motion suppression. I now discuss this explicitly in the manuscript. Furthermore, I have made changes in the Abstract and Discussion accordingly. I have restructured the Discussion section and added a paragraph stating that the current data do not specify which visual features trigger habituation. I mention saturation of motion-sensitive neurons as one possible mechanism followed by a paragraph outlining the idea of displacement detection via pre-and post saccadic image comparisons. I now write:

“However, from the present data it cannot be determined how stimulus-specific transsaccadic habituation is. […] The experimental task that I applied can in principle be solved without motion information.”

and:

“In the experiments of the present study, it cannot be decided to what extent participants employed a comparison between the pre-and post-saccadic image and how much they used the motion transient to judge the displacement.”

Experiment 3 seems convincing because it tries to match saccade-induced and screen motion, uses a balanced design (both upwards and downwards context motions are tested), and provides converging evidence for the directional tuning of the contextual habituation effect. Thus, the results of Experiment 2 are unlikely to be a consequence of differences in retinal motion vectors between different context conditions.

I agree and stick to the initial statements regarding the directional tuning.

In the supplementary experiment a full-field random pattern mask flashed during saccade execution diminishes performance (comparison of pre- and post-saccadic target locations). The author concludes that because the mask abolished the motion transient, it is the motion transient that drives performance.However, there is a serious logical flaw in this experiment. The results seem to be inconsistent with the author's theoretical proposal; i.e., that saccadic omission results from habituation to the motion transients that accompany every saccade (and therefore there is no need for special trans-saccadic mechanisms as claimed by many in the literature). In support of this claim, Experiment 1 shows that exposing observers to artificial intra-saccadic motion degrades motion discrimination performance (relative to performance without exposure to artificial intra-saccadic motion). The supplementary experiment is essentially the same as Experiment 1, except that now the upwards motion is masked by a full-field random pattern (although it still remains to be shown whether such a pattern actually does mask motion during normal viewing, i.e. fixation). If habituation to intra-saccadic motion explains poor motion discrimination, then elimination of intra-saccadic motion should release discrimination performance. Performance should therefore improve or, at least, remain comparable to the no-exposure condition (because there is, of course, still the usual saccade-induced motion transient in the no-exposure condition). Instead, results show the reverse – performance gets worse – which cannot easily be reconciled with the theoretical claim the author is making. Even if we consider that the mask itself generates motion on the retina during the saccade, the vector of that motion is horizontal and should not – according to the author's proposal – affect vertical motion detection.

Please note that the mask was only displayed in test trials. Please see my description in the previous version of the manuscript:

“The presentation of the grating, fixation point and saccade target in the context trials was identical to Experiment 1. However, in contrast to Experiment 1 the mask presentation and the grating displacement in the test trials was triggered by saccade detection and occurred 35 ms after saccade onset in every trial (see Figure 3—figure supplement 1A,B).”

If the mask would have been presented in the context trials, then indeed, there would be the logical flaw that the reviewers pointed out. The idea for this experiment was to find out how well subjects can detect displacements if the motion transients is covered by the mask. The reviewers suggest that the efficiency of the mask has to be demonstrated in a fixation experiment. I agree. When setting up the experiment, I saw that – even in fixation – discriminating the displacement was very hard. I took this to be line with earlier work by Sabine Born and me. But surely, this should be tested for the exact stimulus that I used in the present saccade experiment.

It is therefore doubtful how much the supplementary experiment adds with regard to the main finding of the current manuscript (the plasticity). I am convinced the logic of the experiment can be defended, but again more experiments would be necessary. I therefore have decided to remove the supplementary experiment.

Indeed, in the rebuttal, the author argues that his results do speak to motion perception because detecting the displacement change across saccades requires visibility of the motion transient. In support of his argument, he cites Shiori and Cavanagh, 1989, who showed that sensitivity to the motion transient was abolished during saccades. But this doesn't make much sense: if the motion transient is not perceived during the saccade, how can it be used to drive displacement reports?

I think the reason for the confusion my statements induced might be the usage of terms like “absence of motion transients” or “insensitivity to displacements”. Saccadic suppression is never a complete insensitivity to contrast or motion transients. In my own data presented in this manuscript, psychometric functions could be estimated, clearly suggesting that subjects did perceive the displacement to some degree.

He does not cite (but could have) experiments performed by Deubel and Schneider in which they showed that performance increases dramatically when there is a blank (that diminishes the motion transient) between pre- and post-saccadic targets. So it seems that there is evidence that diminishing the motion transient may actually help trans-saccadic location performance.

I agree that this is a classic study that should be cited and I thank the reviewers for pointing this out. However, I disagree with the reviewer’s interpretation of this study. In the Deubel and Schneider’s “no blank" condition, the motion transient fell into the saccadic period and was thereby diminished. Consequently, subjects showed poor discrimination performance. When there was a blank, the target reappeared after the saccadic period, thus making the motion transient visible again. Under this condition, subjects saw the displacements well.

Previous work from the author and Sabine Born showed that it is possible to mimic poor performance in a fixation task in which a mask occurs between the two targets. However, being able to degrade performance in two different ways (either by presenting targets before and after a saccade, or by presenting them before and after a mask) does not mean that the same processes are involved.

The argument of our articles on that topic was that by using the mask we degraded performance in a way similar enough to trans-saccadic stimulation to conclude that the same process is involved. It is true that a full field mask is not the best match for what is on the retina during the saccade but it might be close enough. This result shows that the loss of the motion transient decreases discrimination of displacements. But I see the reviewer’s point, one could mimic the saccadic disturbance better with a mirror that moves with saccadic speed. This condition could also be used to test the alternative interpretation that the reviewers mention in the following point below. As I have decided to remove the supplementary experiment, this experiment has to be carried out in future work.

An alternative interpretation of the results is that the intra-saccadic mask might induce uncertainty about the initial position of the grating, making the displacement harder to judge (rather than rendering the motion invisible); for instance, the masking pattern may have itself induced transients on the retina that caused the system to down-regulate any input during subsequent saccades. This would argue that the author's protocol does not study motion perception (as expressed in the title of the manuscript), but displacement detection.

This alternative interpretation might be compatible with the model by Crevecoeur and Körding, 2017 which states that suppression scales with the unreliability of trans-saccadic vision. As I have dropped the supplementary experiment, this issue is not discussed further in the new version of the manuscript.

To justify the conclusion that is the motion transient that drives performance, two further pieces of information would be needed. (1) Does such a full-field mask abolish a motion transient of an embedded target (in a fixation condition)?

I agree that such a control is necessary. Please also see my reply above.

2) Why is the invisibility of the motion transient during a saccade (as shown by Shiori and Cavanagh) not sufficient?

In most tests saccadic suppression is not complete. We are not functionally blind during saccades as often claimed. For motion stimuli, it has been shown that the trans-saccadic visibility depends on the stimulus characteristics (Castet and Masson, 2000). The same holds true for saccadic suppression of contrast: point-like stimuli are detected less than background changes (Brook and Fuchs, 1975). In the study of Shiori and Cavanagh performance was at chance level. However, their stimulus (random dot motion) was much different than my grating displacement. As a matter of fact, there was no complete suppression of the grating displacement in the current study, otherwise I would not have been able to estimate a psychometric function.

The author would have to provide evidence that the manipulation does indeed reduce motion perception while leaving the ability to judge displacements intact.

Indeed, the ability to judge displacements by comparing pre-and post-saccadic images should be left intact after reductions of motion perception. Performance could be compared to trials in which isoluminant stimuli are used as they are not “seen” by the motion detectors. I will pursue this issue in a follow-up study.

In sum, the only condition in which performance increased after context trials was the context in which the author minimized changes on the retina during the saccade (no displacement context trials). It seems, therefore, that any visual changes during saccades result in insensitivity during subsequent saccades. The plasticity in intrasaccadic omission remains an interesting finding, but it deviates quite strongly from the claim in the title/Abstract/conclusion that this paper is about saccade suppression of motion. Thus, we invite you to provide an alternative interpretation to saccadic suppression of motion for your current findings. Note that this would entail a re-writing of the entire manuscript, theory included. Please note that in eLife it is not customary to have many review rounds; thus, if you decide to take this route please try to make a comprehensive and clear a case as possible.

I have followed this advice and removed all instances of “motion suppression” from the title, Abstract and the main text. As the supplementary experiment is judged inconclusive the statement “that any visual changes during saccades result in insensitivity” is not really shown by the present data. I keep the term “saccadic suppression” instead of “saccadic omission”

AnalysisIn several of the new results, the author conducted multiple t-tests rather than an ANOVA, which is not good statistical practice. Take as an example: in Experiment 2, the author tests for direction-specificity of the contextual habituation with four conditions resulting from the crossing of two factors, Direction of Motion in Test Trials (Upward/Downward) and Time of shift (intra-sacc: 35 ms/post-sacc: 98 ms). In the results he only presents t-test on the Direction Factor, independently for each time of shift. Because the t-test is significant in intra-sacc, and not in post-sacc, the author concludes that there is direction-specific intra-saccadic habituation. Of course, this conclusion only holds if there is a statistically reliable interaction between the two factors. Factorial designs should be analyzed with ANOVAs throughout, instead of t-tests. To respond to the sample size critique, the author could use a linear mixed model with p-values estimated by the methods explained by Luke (2017, Behav Research Methods), which apparently is less sensitive to sample size than ANOVAs and t-tests.

Many thanks for pointing this article out! I have now calculated non-parametric ANOVAs and estimated the p-values with the Kenward-Roger approximations. I describe the analysis in the section on data analysis:

“For statistical analysis of Experiments 2 and 3, a non-parametric repeated measures ANOVA was calculated, using the Aligned Rank Transform (Wobbrock et al., 2011). Significance was determined by applying the Kenward-Roger approximation to estimate p-values, a procedure that has been shown to produce acceptable Type 1 error rates even for smaller samples (Luke, 2017).”

One issue that further complicates the interpretation of the results of the supplementary experiment is that it is difficult to have an idea of power. How big of a difference in slope can 4 subjects reveal (given the variability seen in the other experiments, and especially when several cannot be fit by a cumulative Gaussian)?

I agree. For this reason and the other argument listed above, I decided to drop that experiment from the manuscript.

Author details

1. Eckart Zimmermann

   Institute for Experimental Psychology, Heinrich Heine University Düsseldorf, Düsseldorf, Germany

   Contribution

   Conceptualization, Software, Formal analysis, Funding acquisition, Validation, Investigation, Visualization, Methodology

   For correspondence

   eckart.zimmermann@hhu.de

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1964-2711

• 1,620

  Page views

• 205

  Downloads

• 8

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.