Historically, vision research has focused on the cortical response to boundaries and edges (e.g. Albrecht and Hamilton, 1982; Hubel and Wiesel, 1968). Perception, however, requires mechanisms by which areas enclosed by boundaries are ‘filled-in’. As surface perception requires spreading or integration of information over a large visual field, these mechanisms have been hypothesized to be localized in high-level visual areas (e.g. Dennett, 1991; Gregory, 1972; von der Heydt et al., 2003), and a role of feedback projections in perceptual filling-in has been suggested (Devinck and Knoblauch, 2019). Several studies have indeed claimed that early visual cortex does not contribute to the processing of surfaces (Cornelissen et al., 2006; Friedman et al., 2003; Perna et al., 2005). Nevertheless, a large number of human fMRI studies (Hsieh and Tse, 2010; Kok and de Lange, 2014; Mendola et al., 1999; Pereverzeva and Murray, 2008; Sasaki and Watanabe, 2004) as well as cat (Rossi et al., 1996; Rossi and Paradiso, 1999) and monkey electrophysiological recording studies (De Weerd et al., 1995; and in Komatsu et al., 2000; Lamme, 1995; Lamme, 1999; reviewed in Lamme and Roelfsema, 2000; Lu and Roe, 2007; Roe et al., 2005; Zipser et al., 1996) have demonstrated retinotopic signals in response to the perception of surface brightness, colour, and texture. These surface-related neural signals in early visual cortex have raised the question to what extent they reflect feedback. As feedback projections target predominantly superficial and deep layers in early visual cortex (Anderson and Martin, 2009; Rockland and Pandya, 1979; Rockland and Virga, 1989), this leads to a clear prediction for activity distributions across cortical depth induced by feedback. In the domain of surface perception, only a handful of neurophysiological studies in animals have successfully tested layer-specific distributions of activity during feedback. Using texture-defined surfaces, two neurophysiological studies in monkeys (Self et al., 2013; van Kerkoerle et al., 2014) revealed complex temporal patterns engaging both deep and superficial layers. A single human fMRI study using a static surface induced in a Kanizsa display (Kok et al., 2016) reported cortical deep layer activity compatible with a role of feedback in surface perception. These experiments align with anatomical data indicating that feedback projections can target both superficial and deep layers. Recent optogenetics studies in mice have moreover confirmed that the correlates of feedback in V1 causally depend on activity in high-level visual cortex (Schnabel et al., 2018).

Surface perception is thought to interact tightly with mechanisms of contour reconstruction. A number of computational models of surface perception (Grossberg, 1987a; Grossberg, 1987b; see also Keil et al., 2005) have proposed that diffusion-like spreading in a surface feature system is contained within proper retinotopic bounds by local inhibition delivered by boundary representations. Neurophysiological observations of contour-related responses in V2 (von der Heydt et al., 1984) and in V1 (Grosof et al., 1993) and surface related responses in V1, V2 and V3 (De Weerd et al., 1995; Huang and Paradiso, 2008) have emphasized the role of early visual areas in this interaction between surface and contour processing.

Separating responses to edges from responses to the interior of a surface is of utmost importance, as contour responses themselves involve feedback (Lee and Nguyen, 2001; Wokke et al., 2013), and may show a depth distribution of activity in early visual cortex similar to that elicited by responses to surfaces. In the only depth-specific human fMRI study on surface perception to date, Kok et al., 2016 presented participants with Kanizsa stimuli containing illusory surfaces and contours. The illusory stimuli caused a response at deep cortical depths in V1, suggesting feedback originating from higher cortical areas. However, due to stimulus design and choice of the region-of-interest (ROI), the feedback related signal could be due to the illusory contour or to the illusory surface, because the ROI could have captured activity related to both.

Research using visual illusions to study the neural correlates of surface perception has predominantly used static surfaces with induced percepts of brightness, colour, or texture, while these features were physically absent in these surfaces. We are aware of only one previous study that measured responses to induced motion of a uniform surface, that is without local changes in retinotopic input (Akin et al., 2014). In fMRI studies focusing on motion interpolation, feedback-related responses in V1 were most likely driven by contours rather than surfaces (Meng et al., 2005; Muckli et al., 2005; Seghier et al., 2000), and other motion-related V1 responses may have been driven by local elements in a non-uniform surface (Muckli et al., 2002).

By contrast, here we used a stimulus (adapted from Akin et al., 2014, see Table 1 for a detailed comparison of stimulus parameters) that consisted of a centrally fixated, luminance-defined disk, of which a sector was removed. The removed sector was limited to the right hemifield, and rotated clockwise and anticlockwise within the right hemifield, thereby inducing a motion percept of the disk. In the left hemifield, the entire half of the disk was static, remained physically identical, and did not contain local elements inducing the movement percept. Two control conditions that eliminated the illusory motion kept the half of the disk in the left hemifield identical as well. That is, the three stimuli differed in global and local perceptual quality, while being physically identical in the left half of the visual field.

These stimuli, hence, provide several advantages: First, because the motion percept is induced without relying on local elements, an fMRI correlate of surface motion cannot be reduced to merely a modified processing of local elements. Second, because the retinal image of illusory and control stimuli was identical in the left hemifield, and because transcallosal connections are restricted to the vertical meridian in primate early visual cortex (Clarke and Miklossy, 1990; Essen and Zeki, 1978; Glickstein and Whitteridge, 1976; Wong-Riley, 1974), any difference between stimulus conditions can be attributed unambiguously to top-down feedback effects. Third, the stimulus was large enough so that contributions to the fMRI signal from the surface were separable from contributions from the contour, enabling any feedback signal to be attributed solely to the surface.

Furthermore, we used ultra-high field (UHF) 7T fMRI to test whether the attribution of motion to a locally static, luminance-defined surface leads to a cortical depth-resolved pattern of activity consistent with feedback processing in early visual cortex. While the tools to perform layer-specific recordings have been available in invasive neurophysiology in animals for decades, the analysis of depth-specific activity in humans has only recently become within reach thanks to UHF fMRI and advances in data analysis (Guidi et al., 2016; Huber et al., 2015; Koopmans et al., 2010; Koopmans et al., 2011; Lawrence et al., 2019; Marquardt et al., 2018; Olman et al., 2012; Polimeni et al., 2010; Ress et al., 2007). Our analysis included not only V1 (as in Kok et al., 2016), but was extended to V2 and V3.

Notably, in the non-depth resolved fMRI study that inspired our stimulus design, a smaller BOLD response was reported to the grey figure region than to the textured background, which may have reflected a stronger sensory response driven by the textured than by the homogeneously grey surface. Irrespective of whether the BOLD response to the grey figure was negative or positive, we hypothesized that the illusory perception of surface motion would be associated with enhanced activity in superficial and/or deep layers compared to control conditions, in accordance with a contribution of feedback in early visual cortex.

In accordance with a previous report using a similar stimulus (see our Figure 1 and Akin et al., 2014), but contrary to what could be the generally expected positive response to a luminance increase, we observed widespread negative signal change in the retinotopic representation of our stimuli in early visual cortex of the right hemisphere. This is illustrated here for the experimental condition inducing the illusory motion percept (Figure 2, see also Figure 1—figure supplement 1). A band of positive activation with transient peaks at the beginning and end of the stimulation was observed at the cortical representation of the stimulus edge (Figure 2E,F). The pattern of negative responses to the surface interior and positive activation at the stimulus edge was similar across stimulus conditions (Figure 3; and Figure 5—figure supplement 1). Control experiments supported the idea that the negative sign of the response was related to the much stronger response to the texture in the background than to the homogenous grey in the figure (see Figure 7—figure supplement 1 and Figure 7—figure supplement 2).

Figure 1 with 2 supplements see all

Download asset Open asset

Figure 2 with 1 supplement see all

Download asset Open asset

Figure 3

Download asset Open asset

In the cortical representation of the surface (Figure 2C), we found increased activity due to the illusory percept of motion in the experimental condition, compared to the control conditions where this percept was absent. Using a mixed-effects model comparison, we found differential activity among the experimental and control conditions with a magnitude that differed among brain areas V1, V2, and V3, as confirmed by a significant ROI (V1, V2, V3) by condition (motion induction, static control, dynamic control) interaction (likelihood ratio (df): 39.6 (4), p<0.0001). Moreover, cortical depth profiles of the activity increase were significantly different between brain areas (likelihood ratio (df) of model comparison with/without cortical depth by ROI interaction: 30.2 (2), p<0.0001).

Figure 4—figure supplement 1A plots the contrasts between experimental stimulus and static control, between experimental stimulus and dynamic control, and among the two control conditions, as a function of depth in areas V1, V2 and V3, after spatial deconvolution. The amount of signal change between the control conditions appeared to differ between areas V2 and V3. To investigate this apparent difference between the control conditions in V2 and V3, we ran a mixed-effects model comparison, but restricted to V2 and V3 and the two control conditions. We found a significant ROI by condition interaction (likelihood ratio (df): 16.7 (1), p<0.0001). In other words, we found a statistically significant difference in the pattern of stimulus-induced activation caused by the control conditions in V2 and V3. Moreover, we found a ROI by depth interaction, showing that the depth profiles differed between ROIs (likelihood ratio (df): 3.9 (1), p=0.0491).

Another relevant outcome of the anatomically restricted analysis was that in V3, the dynamic and stationary controls were not equivalent (mixed effects model comparison, limited to V3 and the two control conditions, testing for an effect of ‘condition’; likelihood ratio (df): 35.8 (1), p = <0.0001). As the physical motion in the right hemisphere in the dynamic control condition is better matched to the physical motion of the Pac-Man contours in the experimental stimulus, we considered the dynamic control to be superior over the static control condition. Accordingly, Figure 4 shows the cortical depth profile of the signal gain corresponding to the induced motion effect for the cortical representation of the stimulus centre, using the difference between motion induction and dynamic control condition. The peak of the apparent motion effect was located at ~25% in V1,~50% in V2, and ~40% in V3, relative to the pial surface (where 100% cortical depth corresponds to the white/grey matter boundary).

Figure 4 with 3 supplements see all

Download asset Open asset

The ratio of superficial peak positions in the cortical depth profiles of the condition contrast ‘motion induction stimulus’ vs. ‘dynamic control stimulus’ was compared with a chi-squared test. The null hypothesis of no difference in the ratio of superficial peak positions between areas V1, V2, and V3 was rejected (chi-squared (df) = 6.82 (2), p=0.033). As a more specific follow-up, we tested whether the ratio of superficial peaks differed between V1 versus V2 and V3 together. Again, the null hypothesis of no difference was rejected (chi-squared (df) = 6.43 (1), p=0.011). Thus, the ratio of superficial peak positions (in the single subject cortical depth profiles) is significantly higher in striate than in extrastriate cortex (i.e. the peak is closer to the cortical surface in the striate cortex).

For the cortical representation of the stimulus edge, the stimulus conditions also caused differential activation among visual areas (likelihood ratio (df) of linear mixed effects model comparison with/without ROI by condition interaction: 22.8 (4), p<0.0001). However, there was no evidence for differences between stimulus conditions in the cortical depth profiles at the stimulus edge (likelihood ratio (df) of model comparison with/without cortical depth by condition interaction: 1.6 (2), p=0.46); see Figure 4—figure supplement 2 for cortical depth profiles of apparent motion effect at stimulus edge). This is likely due to the strong feedforward drive due to local contrast at the figure’s edge, which may engage neurons about equally across cortical depth.

Temporal response pattern

In areas V1, V2, and V3, the central region of interest for all conditions exhibited a sustained negative response, whereas the edge region responded with a transient positive signal change at stimulus onset and offset (Figure 5). Separately for the sustained and transient responses, we determined response onset time as the first time point at which the signal was significantly different from zero (one-sample t-test, p<0.05, Bonferroni corrected). Interestingly, this revealed that the onset of the transient response at the cortical representation of the stimulus edge preceded the onset of the sustained response in the surface representation by one MRI acquisition time point (i.e. ~2 s; Figure 5). The pattern of positive transient and negative sustained responses at the stimulus edge and centre, respectively, was consistent across areas, conditions (Figure 5—figure supplement 1) and subjects (Figure 5—figure supplement 2). An additional control experiment was performed to investigate whether the temporal dynamics of the responses were similar for a longer stimulus duration (Figure 5—figure supplement 3). The results indicate that this was indeed the case, and that the negative response to the centre of the PacMan surface was sustained over long stimulus durations (25 s, compared to ~10 s in the main experiment). Separate cortical depth profiles of the early and late response phases show no evidence for temporal differences in the laminar activation profile (Figure 5—figure supplement 4).

Figure 5 with 5 supplements see all

Download asset Open asset

Similar to the main experiment (Figure 5), a trend towards an onset time difference was also observed in the control experiment (Figure 6). In this case, the onset difference is found between stimulus conditions, at the same retinotopic location. However, this onset time difference is not statistically significant, probably due to the small sample size of the control experiment (n = 2).

Figure 6

Download asset Open asset

Background dependence of the negative response

A control experiment was conducted to investigate the effect of the background and of the stimulus shape on the processing of a surface stimulus. The results revealed that the directionality and temporal course of the response is heavily affected by the type of background, but not by the shape of the stimulus. A negative surface response was only observed when the stimuli were presented on a texture background, irrespective of the stimulus shape (Figure 7B D). When presented on a homogenous background, as luminance stimuli are usually presented, the interior of the surface and its edges evoked a positive response (Figure 7A C). An additional control experiment showed that the full-screen texture pattern evokes a positive response when contrasted against a uniform rest condition (Figure 7—figure supplement 1). Thus, it is safe to assume that the negative surface response was caused by the texture background.

Figure 7 with 3 supplements see all

Download asset Open asset

The temporal dynamics of the response in the texture background condition (Figure 6, green and blue lines) closely resembled the results from the main experiment (Figure 5); showing a transient positive response at the edges and a sustained, delayed, negative response at the surface interior. In contrast, the response to both the interior of the surface and to its edges was positive and sustained in case of a uniform background (Figure 6, red and orange lines). Interestingly, these results imply that the temporal shape of the edge response changed as a function of the background condition; in other words, whether the edge response is sustained or transient depends on whether the stimuli are presented on a texture pattern or on a uniform background.

Because the delay in the negative response is observed under all three stimulus conditions in the main experiment (Figure 5, Figure 5—figure supplement 1), the delay is unlikely to be related to the (apparent) motion of the stimuli. The delay may be due to a slow decrease in activation, following an elevated baseline caused by the texture background. However, we have also observed delayed negative BOLD responses in V1 and V2 under very different stimulus conditions (without texture background), even in the absence of any change in local retinotopic input (see Marquardt et al., 2018, p. 176, Figure 5.7 E and F, and p. 177, Figure 5.8, https://doi.org/10.26481/dis.20190829im). Moreover, we have at present no explanation for the temporal dynamics of the response at the stimulus edge (Figure 6B). Thus, it is possible that varying physiological contributions to the negative BOLD response contribute to the observed temporal dynamics, and more research will be needed to clarify the origin of these delays.

We have studied neural correlates of perceived surface motion induced in a locally static grey surface on a dark, textured background (Figure 1). The motion percept was caused by local edge movement in the contralateral hemifield and spread over the entire surface in the ipsilateral hemifield. We report three main findings: First, the induced percept of surface motion was associated with an fMRI signal increase in the representation of the surface in areas V1, V2 and V3 (Figure 4). As the enhanced signal was measured far away from the location where the perceived motion was induced, this signal likely derives from feedback. In addition, the differences in the cortical depth distribution of motion-percept related signal gain among visual areas also supported a feedback origin. Second, we found that the response to the edge preceded the response to the surface by approximately 2 s (Figure 5). Third, we observed a negative BOLD signal in the figure representation (Figure 3), which depended on the presence of a textured background and was eliminated when the background texture was removed (Figure 7). Hence, the signal gain due to the motion percept represented an increase in signal from a negative BOLD signal in the control condition to a less negative BOLD signal in the illusory movement-condition.

Top-down feedback

The main and control stimuli were ‘physically’ identical in the left visual field, while the global perceptual quality of the stimulus depended on the right half of the stimuli (Figure 1; videos of the stimuli are available online: https://doi.org/10.5281/zenodo.2583017). This stimulus design offers three advantages: First, the surface itself was homogenously grey and did not contain local moving elements, thereby avoiding the interpretative question whether enhanced fMRI activity during surface perception reflects enhanced processing of local elements or an integrated surface motion percept. Second, any changes in activity correlating with a perceptual change from static to moving in the left hemifield were induced by the right hemifield. Anatomical investigations have shown that direct, transcallosal, interhemispheric connections are restricted to the proximity of the vertical meridian in primate early visual cortex (Clarke and Miklossy, 1990; Essen and Zeki, 1978; Glickstein and Whitteridge, 1976; Wong-Riley, 1974). This, combined with the fact that the surface motion percept in the left hemifield was induced in the absence of physical changes to the left-hemifield stimulus, renders top-down feedback from higher areas, rather than within-area horizontal interactions, the most plausible source of the motion percept and associated depth distributions of activity. Third, the cortical region that retinotopically represents the physically constant left side of the stimulus and the one which induces the motion percept (i.e. the ‘mouth’ of the Pac-Man) were far apart. Thus, it is very unlikely that imprecisions in the retinotopic maps could confound our results. By the same token, the size of our stimulus enabled us to separate responses to the surface from responses to the contours.

Although there was considerable variability among subjects, overall, the cortical depth profiles of the enhanced response due to the illusory motion effect in V1, V2, and V3 suggests that top-down signals may have re-entered at superficial layers in V1, where most of the signal gain due to motion perception was concentrated (Figure 4). Re-entrant connections via superficial V1 have been reported in neurophysiological (McManus et al., 2011), anatomical (Martinez-Conde et al., 1999), and high-field fMRI studies (Lawrence et al., 2019; Muckli et al., 2015). This re-entrant information may have propagated to V2 and V3 via feedforward pathways, in line with anatomical evidence that the strongest forward projections from V1 to V2 originate in superficial V1 layers 3B and 4B, and arrive across the full extent of layer four in V2 (Douglas and Martin, 2004; Felleman and Van Essen, 1991). Furthermore, forward projections originating in superficial V1 layers and superficial V2 layers also target layer four in V3 (Rockland and Pandya, 1979; Van Essen et al., 1986). This pattern of forward projections may explain the activity peak at intermediate depths of areas V2 and V3 (Figure 8A). Therefore, although our data do not permit a direct test of the directionality and precise temporal dynamics of information flow, re-entrant feedback at the level of V1 is a plausible interpretation of the present results.

Figure 8

Download asset Open asset

An additional contribution to the depth-pattern of activity observed in extrastriate areas may have originated from the pulvinar, the LGN, and possibly other subcortical structures (Standage and Benevento, 1983; Trojanowski and Jacobson, 1977). The middle layers of extrastriate cortex are the target of projections from the pulvinar (Benevento and Rezak, 1975; Figure 8B; Benevento and Rezak, 1976; Ogren and Hendrickson, 1977; Rezak and Benevento, 1979), a structure that is sometimes referred to as a ‘higher-order relay’ because of its role in cortico-cortical interaction (Sherman and Guillery, 2002). The pulvinar has been shown to regulate cortico-cortical communication in the visual system based on attentional demands (Saalmann et al., 2012). Experiments in humans (Villeneuve et al., 2005; Villeneuve et al., 2012) and cats Merabet et al., 1998 have demonstrated a role of the pulvinar in higher-order motion processing (i.e. coherent motion of entire objects, as opposed to local motion). In line with this, Shimono et al., 2012 have found evidence for an involvement of the pulvinar in the interhemispheric integration of motion information (2012). Moreover, an involvement of the LGN in the perception of illusory motion has been observed by Akin et al., 2014, using an experimental design very similar to ours. As the V1 cortical depth profile we observed suggests similar levels of activity in mid-level to superficial layers (Figure 4), it is possible that a feedback signal assigning motion to the grey surface re-entered the LGN, was fed-forward to V1, and from V1 to V2 and V3. In summary, both cortical and subcortical sources of re-entrant feedback in lower-level visual areas may have contributed to the observed depth-resolved responses (see Figure 8).

The increase BOLD contribution associated with the illusory percept of surface motion is in line with other fMRI studies for a range of surface illusions (Hsieh and Tse, 2010; Kok and de Lange, 2014; Mendola et al., 1999; Pereverzeva and Murray, 2008; Sasaki and Watanabe, 2004). Compared to Kok et al., 2016, who reported a fMRI response enhancement limited to the deepest cortical layers during the percept of an illusory Kansiza triangle, the signal gain we found was focused on superficial to middle layer compartments. Our results resemble somewhat more the superficial activity reported in Muckli et al., 2015 in response to the completion of occluded visual scenes, and in Lawrence et al., 2019 associated with feature-based attention. These differences in activity depth profiles could reflect fundamental differences in feedback mechanisms engaged in the stimulus paradigms in the different studies, which is a possibility that should be investigated further. Irrespective of the differences in observed activity profiles over depth, they all support re-entrant feedback signals, which is in line with mounting evidence that, even for the simplest displays, feedback from the highest level of the visual system plays a role (Lamme and Roelfsema, 2000; McManus et al., 2011; Roelfsema et al., 2002; Schnabel et al., 2018).

The amplitude of the apparent motion effect (Figure 4) is small, compared with the overall response strength (e.g. Figure 5). This result is consistent with previous studies, finding small effect sizes of top-down effects relative to bottom-up effects. For example, Kok et al., 2016 reported a top-down modulation of about 0.12% (their Figure 2A). In our data, the apparent motion effect is strongest at mid-cortical depths in V2, where it reaches an amplitude of slightly more than 0.3%. In previous studies, top-down effects were observed in experiments that contrasted stimulus conditions evoking a positive response, whereas in our case the stimulus induced responses were negative compared to the baseline. We see no principled objection against the interpretability of a condition contrast among control and experimental responses that both yield BOLD responses smaller than the BOLD baseline response.

The fMRI dataset, experimental stimuli, and analysis code are publicly available. The fMRI dataset is available on Zenodo (https://doi.org/10.5281/zenodo.3366301). The software used for the presentation of retinotopic mapping stimuli, and for the corresponding analysis, is available on github (https://github.com/ingo-m/pyprf). Example videos of the main experimental stimuli are available on Zenodo (https://doi.org/10.5281/zenodo.2583017). If you would like to reproduce the experimental stimuli, the respective PsychoPy code can be found on github (https://github.com/ingo-m/PacMan/tree/master/stimuli/experiment). The respective repository also contains the analysis code and a brief description how to reproduce the analysis (https://github.com/ingo-m/PacMan; copy archived at https://github.com/elifesciences-publications/PacMan). High-level visualisations (e.g. cortical depth profiles & signal timecourses) and group-level statistical tests are implemented in a separate repository (https://github.com/ingo-m/py_depthsampling/tree/PacMan; copy archived at https://github.com/elifesciences-publications/py_depthsampling).

1. 1. Marquardt I
   2. De Weerd P
   3. Schneider M
   4. Gulban OF
   5. Ivanov D
   6. Uludağ K

   (2019) Zenodo

   Dataset: Feedback contribution to surface motion perception in the human early visual cortex.

   https://doi.org/10.5281/zenodo.3366301

1. Book

   1. Ahrens J

   (2005) ParaView: An End-User Tool for Large-Data Visualization

   In: Hansen C. D, Johnson C. R, editors. The Visualization Handbook. Elsevier Butterworth–Heinemann. pp. 717–731.

   https://doi.org/10.1016/B978-012387582-2/50038-1

   • Google Scholar
2. 1. Akin B
   2. Ozdem C
   3. Eroglu S
   4. Keskin DT
   5. Fang F
   6. Doerschner K
   7. Kersten D
   8. Boyaci H

   (2014) Attention modulates neuronal correlates of interhemispheric integration and global motion perception

   Journal of Vision 14:30.

   https://doi.org/10.1167/14.12.30

   • PubMed
   • Google Scholar
3. 1. Albrecht DG
   2. Geisler WS
   3. Frazor RA
   4. Crane AM

   (2002) Visual cortex neurons of monkeys and cats: temporal dynamics of the contrast response function

   Journal of Neurophysiology 88:888–913.

   https://doi.org/10.1152/jn.2002.88.2.888

   • PubMed
   • Google Scholar
4. 1. Albrecht DG
   2. Hamilton DB

   (1982) Striate cortex of monkey and cat: contrast response function

   Journal of Neurophysiology 48:217–237.

   https://doi.org/10.1152/jn.1982.48.1.217

   • PubMed
   • Google Scholar
5. 1. Anderson JC
   2. Martin KA

   (2009) The synaptic connections between cortical Areas V1 and V2 in macaque monkey

   Journal of Neuroscience 29:11283–11293.

   https://doi.org/10.1523/JNEUROSCI.5757-08.2009

   • PubMed
   • Google Scholar
6. 1. Andersson JL
   2. Skare S
   3. Ashburner J

   (2003) How to correct susceptibility distortions in spin-echo echo-planar images: application to diffusion tensor imaging

   NeuroImage 20:870–888.

   https://doi.org/10.1016/S1053-8119(03)00336-7

   • PubMed
   • Google Scholar
7. Book

   1. Ayachit U

   (2015)

   Updated for ParaView version 4.3

   In: Avila L, editors. The ParaView Guide. Kitware. pp. 1–276.

   • Google Scholar
8. 1. Bazin PL
   2. Cuzzocreo JL
   3. Yassa MA
   4. Gandler W
   5. McAuliffe MJ
   6. Bassett SS
   7. Pham DL

   (2007) Volumetric neuroimage analysis extensions for the MIPAV software package

   Journal of Neuroscience Methods 165:111–121.

   https://doi.org/10.1016/j.jneumeth.2007.05.024

   • PubMed
   • Google Scholar
9. 1. Behnel S
   2. Bradshaw R
   3. Citro C
   4. Dalcin L
   5. Seljebotn DS
   6. Smith K

   (2011) Cython: the best of both worlds

   Computing in Science & Engineering 13:31–39.

   https://doi.org/10.1109/MCSE.2010.118

   • Google Scholar
10. 1. Benevento LA
    2. Rezak M

    (1975) Extrageniculate projections to layers VI and I of striate cortex (area 17) in the rhesus monkey (Macaca mulatta)

    Brain Research 96:51–55.

    https://doi.org/10.1016/0006-8993(75)90569-7

    • PubMed
    • Google Scholar
11. 1. Benevento LA
    2. Rezak M

    (1976) The cortical projections of the inferior pulvinar and adjacent lateral pulvinar in the rhesus monkey (Macaca mulatta): an autoradiographic study

    Brain Research 108:1–24.

    https://doi.org/10.1016/0006-8993(76)90160-8

    • PubMed
    • Google Scholar
12. 1. Boorman L
    2. Kennerley AJ
    3. Johnston D
    4. Jones M
    5. Zheng Y
    6. Redgrave P
    7. Berwick J

    (2010) Negative blood oxygen level dependence in the rat: a model for investigating the role of suppression in neurovascular coupling

    Journal of Neuroscience 30:4285–4294.

    https://doi.org/10.1523/JNEUROSCI.6063-09.2010

    • PubMed
    • Google Scholar
13. 1. Bressler DW
    2. Fortenbaugh FC
    3. Robertson LC
    4. Silver MA

    (2013) Visual spatial attention enhances the amplitude of positive and negative fMRI responses to visual stimulation in an eccentricity-dependent manner

    Vision Research 85:104–112.

    https://doi.org/10.1016/j.visres.2013.03.009

    • PubMed
    • Google Scholar
14. 1. Briggs F
    2. Usrey WM

    (2007) A fast, reciprocal pathway between the lateral geniculate nucleus and visual cortex in the macaque monkey

    Journal of Neuroscience 27:5431–5436.

    https://doi.org/10.1523/JNEUROSCI.1035-07.2007

    • PubMed
    • Google Scholar
15. 1. Bullier J
    2. Henry GH

    (1980) Ordinal position and afferent input of neurons in monkey striate cortex

    The Journal of Comparative Neurology 193:913–935.

    https://doi.org/10.1002/cne.901930407

    • PubMed
    • Google Scholar
16. 1. Clarke S
    2. Miklossy J

    (1990) Occipital cortex in man: organization of callosal connections, related myelo- and Cytoarchitecture, and putative boundaries of functional visual Areas

    The Journal of Comparative Neurology 298:188–214.

    https://doi.org/10.1002/cne.902980205

    • PubMed
    • Google Scholar
17. 1. Cornelissen FW
    2. Wade AR
    3. Vladusich T
    4. Dougherty RF
    5. Wandell BA

    (2006) No functional magnetic resonance imaging evidence for brightness and color filling-in in early human visual cortex

    Journal of Neuroscience 26:3634–3641.

    https://doi.org/10.1523/JNEUROSCI.4382-05.2006

    • PubMed
    • Google Scholar
18. 1. De Weerd P
    2. Gattass R
    3. Desimone R
    4. Ungerleider LG

    (1995) Responses of cells in monkey visual cortex during perceptual filling-in of an artificial scotoma

    Nature 377:731–734.

    https://doi.org/10.1038/377731a0

    • PubMed
    • Google Scholar
19. Book

    1. Dennett DC

    (1991)

    Consciousness Explained 

    Little, Brown and Co.

    • Google Scholar
20. 1. Devinck F
    2. Knoblauch K

    (2019) Central mechanisms of perceptual filling-in

    Current Opinion in Behavioral Sciences 30:135–140.

    https://doi.org/10.1016/j.cobeha.2019.08.003

    • Google Scholar
21. 1. Devor A
    2. Tian P
    3. Nishimura N
    4. Teng IC
    5. Hillman EM
    6. Narayanan SN
    7. Ulbert I
    8. Boas DA
    9. Kleinfeld D
    10. Dale AM

    (2007) Suppressed neuronal activity and concurrent arteriolar vasoconstriction may explain negative blood oxygenation level-dependent signal

    Journal of Neuroscience 27:4452–4459.

    https://doi.org/10.1523/JNEUROSCI.0134-07.2007

    • PubMed
    • Google Scholar
22. 1. Douglas RJ
    2. Martin KA

    (2004) Neuronal circuits of the neocortex

    Annual Review of Neuroscience 27:419–451.

    https://doi.org/10.1146/annurev.neuro.27.070203.144152

    • PubMed
    • Google Scholar
23. 1. Dumoulin SO
    2. Wandell BA

    (2008) Population receptive field estimates in human visual cortex

    NeuroImage 39:647–660.

    https://doi.org/10.1016/j.neuroimage.2007.09.034

    • PubMed
    • Google Scholar
24. 1. Essen DC
    2. Zeki SM

    (1978) The topographic organization of rhesus monkey prestriate cortex

    The Journal of Physiology 277:193–226.

    https://doi.org/10.1113/jphysiol.1978.sp012269

    • PubMed
    • Google Scholar
25. 1. Feinberg DA
    2. Moeller S
    3. Smith SM
    4. Auerbach E
    5. Ramanna S
    6. Gunther M
    7. Glasser MF
    8. Miller KL
    9. Ugurbil K
    10. Yacoub E

    (2010) Multiplexed Echo planar imaging for sub-second whole brain FMRI and fast diffusion imaging

    PLOS ONE 5:e15710.

    https://doi.org/10.1371/journal.pone.0015710

    • PubMed
    • Google Scholar
26. 1. Felleman DJ
    2. Van Essen DC

    (1991) Distributed hierarchical processing in the primate cerebral cortex

    Cerebral Cortex 1:1–47.

    https://doi.org/10.1093/cercor/1.1.1

    • PubMed
    • Google Scholar
27. 1. Friedman HS
    2. Zhou H
    3. von der Heydt R

    (2003) The coding of uniform colour figures in monkey visual cortex

    The Journal of Physiology 548:593–613.

    https://doi.org/10.1113/jphysiol.2002.033555

    • PubMed
    • Google Scholar
28. 1. Friston KJ
    2. Williams S
    3. Howard R
    4. Frackowiak RS
    5. Turner R

    (1996) Movement-related effects in fMRI time-series

    Magnetic Resonance in Medicine 35:346–355.

    https://doi.org/10.1002/mrm.1910350312

    • PubMed
    • Google Scholar
29. 1. Gattass R
    2. Gross CG
    3. Sandell JH

    (1981) Visual topography of V2 in the macaque

    The Journal of Comparative Neurology 201:519–539.

    https://doi.org/10.1002/cne.902010405

    • PubMed
    • Google Scholar
30. 1. Glickstein M
    2. Whitteridge D

    (1976) Degeneration of layer III pyramidal cells in area 18 following destruction of callosal input

    Brain Research 104:148–151.

    https://doi.org/10.1016/0006-8993(76)90655-7

    • PubMed
    • Google Scholar
31. 1. Goense JBM
    2. Logothetis NK

    (2008) Neurophysiology of the BOLD fMRI signal in awake monkeys

    Current Biology 18:631–640.

    https://doi.org/10.1016/j.cub.2008.03.054

    • Google Scholar
32. 1. Gregory RL

    (1972) Cognitive contours

    Nature 238:51–52.

    https://doi.org/10.1038/238051a0

    • PubMed
    • Google Scholar
33. 1. Greve DN
    2. Fischl B

    (2009) Accurate and robust brain image alignment using boundary-based registration

    NeuroImage 48:63–72.

    https://doi.org/10.1016/j.neuroimage.2009.06.060

    • PubMed
    • Google Scholar
34. 1. Grosof DH
    2. Shapley RM
    3. Hawken MJ

    (1993) Macaque VI neurons can signal ‘illusory’ contours

    Nature 365:550–552.

    https://doi.org/10.1038/365550a0

    • Google Scholar
35. 1. Grossberg S

    (1987a) Cortical dynamics of three-dimensional form, color, and brightness perception: I. monocular theory

    Perception & Psychophysics 41:87–116.

    https://doi.org/10.3758/BF03204874

    • PubMed
    • Google Scholar
36. 1. Grossberg S

    (1987b) Cortical dynamics of three-dimensional form, color, and brightness perception: ii. binocular theory

    Perception & Psychophysics 41:117–158.

    https://doi.org/10.3758/BF03204875

    • PubMed
    • Google Scholar
37. 1. Grossberg S
    2. Hong S

    (2006) A neural model of surface perception: lightness, anchoring, and filling-in

    Spatial Vision 19:263–321.

    https://doi.org/10.1163/156856806776923399

    • PubMed
    • Google Scholar
38. 1. Guidi M
    2. Huber L
    3. Lampe L
    4. Gauthier CJ
    5. Möller HE

    (2016) Lamina-dependent calibrated BOLD response in human primary motor cortex

    NeuroImage 141:250–261.

    https://doi.org/10.1016/j.neuroimage.2016.06.030

    • PubMed
    • Google Scholar
39. 1. Gulban OF
    2. Schneider M
    3. Marquardt I
    4. Haast RAM
    5. De Martino F

    (2018) A scalable method to improve gray matter segmentation at Ultra high field MRI

    PLOS ONE 13:e0198335.

    https://doi.org/10.1371/journal.pone.0198335

    • PubMed
    • Google Scholar
40. 1. Halchenko YO
    2. Hanke M

    (2012) Open is not enough. let's Take the Next Step: An Integrated, Community-Driven Computing Platform for Neuroscience

    Frontiers in Neuroinformatics 6:22.

    https://doi.org/10.3389/fninf.2012.00022

    • PubMed
    • Google Scholar
41. Preprint

    1. Havlicek M
    2. Uludag K

    (2019) A dynamical model of the laminar BOLD response

    bioRxiv.

    https://doi.org/10.1101/609099

    • Google Scholar
42. 1. Horiguchi H
    2. Nakadomari S
    3. Misaki M
    4. Wandell BA

    (2009) Two temporal channels in human V1 identified using fMRI

    NeuroImage 47:273–280.

    https://doi.org/10.1016/j.neuroimage.2009.03.078

    • PubMed
    • Google Scholar
43. 1. Houzel JC
    2. Milleret C

    (1999) Visual inter-hemispheric processing: constraints and potentialities set by axonal morphology

    Journal of Physiology-Paris 93:271–284.

    https://doi.org/10.1016/S0928-4257(00)80056-X

    • PubMed
    • Google Scholar
44. 1. Hsieh PJ
    2. Tse PU

    (2010) "Brain-reading" of perceived colors reveals a feature mixing mechanism underlying perceptual filling-in in cortical area V1

    Human Brain Mapping 31:1395–1407.

    https://doi.org/10.1002/hbm.20946

    • PubMed
    • Google Scholar
45. 1. Huang X
    2. Paradiso MA

    (2008) V1 response timing and surface filling-in

    Journal of Neurophysiology 100:539–547.

    https://doi.org/10.1152/jn.00997.2007

    • PubMed
    • Google Scholar
46. 1. Hubel DH
    2. Wiesel TN

    (1968) Receptive fields and functional architecture of monkey striate cortex

    The Journal of Physiology 195:215–243.

    https://doi.org/10.1113/jphysiol.1968.sp008455

    • PubMed
    • Google Scholar
47. 1. Huber L
    2. Goense J
    3. Kennerley AJ
    4. Trampel R
    5. Guidi M
    6. Reimer E
    7. Ivanov D
    8. Neef N
    9. Gauthier CJ
    10. Turner R
    11. Möller HE

    (2015) Cortical lamina-dependent blood volume changes in human brain at 7 T

    NeuroImage 107:23–33.

    https://doi.org/10.1016/j.neuroimage.2014.11.046

    • PubMed
    • Google Scholar
48. 1. Hunter JD

    (2007) Matplotlib: a 2D graphics environment

    Computing in Science & Engineering 9:90–95.

    https://doi.org/10.1109/MCSE.2007.55

    • Google Scholar
49. 1. Jenkinson M
    2. Bannister P
    3. Brady M
    4. Smith S

    (2002) Improved optimization for the robust and accurate linear registration and motion correction of brain images

    NeuroImage 17:825–841.

    https://doi.org/10.1006/nimg.2002.1132

    • PubMed
    • Google Scholar
50. 1. Jenkinson M
    2. Smith S

    (2001) A global optimisation method for robust affine registration of brain images

    Medical Image Analysis 5:143–156.

    https://doi.org/10.1016/S1361-8415(01)00036-6

    • Google Scholar
51. Software

    1. Kaczmarzyk J
    2. Goncalves M
    3. Halchenko Y
    4. Mitchell R
    5. Nielson D
    6. Jarecka D

    (2017) Kaczmarj/Neurodocker, version 0.3.2

    Zenodo.

    https://doi.org/10.5281/zenodo.1058998
52. 1. Keil MS
    2. Cristóbal G
    3. Hansen T
    4. Neumann H

    (2005) Recovering real-world images from single-scale boundaries with a novel filling-in architecture

    Neural Networks 18:1319–1331.

    https://doi.org/10.1016/j.neunet.2005.04.003

    • PubMed
    • Google Scholar
53. 1. Kok P
    2. Bains LJ
    3. van Mourik T
    4. Norris DG
    5. de Lange FP

    (2016) Selective activation of the deep layers of the human primary visual cortex by Top-Down feedback

    Current Biology 26:371–376.

    https://doi.org/10.1016/j.cub.2015.12.038

    • PubMed
    • Google Scholar
54. 1. Kok P
    2. de Lange FP

    (2014) Shape perception simultaneously up- and downregulates neural activity in the primary visual cortex

    Current Biology 24:1531–1535.

    https://doi.org/10.1016/j.cub.2014.05.042

    • PubMed
    • Google Scholar
55. 1. Komatsu H
    2. Kinoshita M
    3. Murakami I

    (2000) Neural responses in the retinotopic representation of the blind spot in the macaque V1 to stimuli for perceptual filling-in

    The Journal of Neuroscience 20:9310–9319.

    https://doi.org/10.1523/JNEUROSCI.20-24-09310.2000

    • PubMed
    • Google Scholar
56. 1. Koopmans PJ
    2. Barth M
    3. Norris DG

    (2010) Layer-specific BOLD activation in human V1

    Human Brain Mapping 31:1297–1304.

    https://doi.org/10.1002/hbm.20936

    • PubMed
    • Google Scholar
57. 1. Koopmans PJ
    2. Barth M
    3. Orzada S
    4. Norris DG

    (2011) Multi-echo fMRI of the cortical laminae in humans at 7 T

    NeuroImage 56:1276–1285.

    https://doi.org/10.1016/j.neuroimage.2011.02.042

    • PubMed
    • Google Scholar
58. 1. Kuhlmann L
    2. Vidyasagar TR

    (2011) A computational study of how orientation Bias in the lateral geniculate nucleus can give rise to orientation selectivity in primary visual cortex

    Frontiers in Systems Neuroscience 5:81.

    https://doi.org/10.3389/fnsys.2011.00081

    • PubMed
    • Google Scholar
59. 1. Lamme VA

    (1995) The neurophysiology of figure-ground segregation in primary visual cortex

    The Journal of Neuroscience 15:1605–1615.

    https://doi.org/10.1523/JNEUROSCI.15-02-01605.1995

    • PubMed
    • Google Scholar
60. 1. Lamme VAF

    (1999) Separate processing dynamics for texture elements, boundaries and surfaces in primary visual cortex of the macaque monkey

    Cerebral Cortex 9:406–413.

    https://doi.org/10.1093/cercor/9.4.406

    • Google Scholar
61. 1. Lamme VA
    2. Roelfsema PR

    (2000) The distinct modes of vision offered by feedforward and recurrent processing

    Trends in Neurosciences 23:571–579.

    https://doi.org/10.1016/S0166-2236(00)01657-X

    • PubMed
    • Google Scholar
62. 1. Lawrence SJD
    2. Norris DG
    3. de Lange FP

    (2019) Dissociable laminar profiles of concurrent bottom-up and top-down modulation in the human visual cortex

    eLife 8:e44422.

    https://doi.org/10.7554/eLife.44422

    • Google Scholar
63. 1. Lee TS
    2. Nguyen M

    (2001) Dynamics of subjective contour formation in the early visual cortex

    PNAS 98:1907–1911.

    https://doi.org/10.1073/pnas.98.4.1907

    • PubMed
    • Google Scholar
64. 1. Logothetis NK
    2. Pauls J
    3. Augath M
    4. Trinath T
    5. Oeltermann A

    (2001) Neurophysiological investigation of the basis of the fMRI signal

    Nature 412:150–157.

    https://doi.org/10.1038/35084005

    • PubMed
    • Google Scholar
65. 1. Lu HD
    2. Roe AW

    (2007) Optical imaging of contrast response in macaque monkey V1 and V2

    Cerebral Cortex  17:2675–2695.

    https://doi.org/10.1093/cercor/bhl177

    • Google Scholar
66. 1. Markuerkiaga I
    2. Barth M
    3. Norris DG

    (2016) A cortical vascular model for examining the specificity of the laminar BOLD signal

    NeuroImage 132:491–498.

    https://doi.org/10.1016/j.neuroimage.2016.02.073

    • PubMed
    • Google Scholar
67. 1. Marquardt I
    2. Schneider M
    3. Gulban OF
    4. Ivanov D
    5. Uludağ K

    (2018) Cortical depth profiles of luminance contrast responses in human V1 and V2 using 7 T fMRI

    Human Brain Mapping 39:2812–2827.

    https://doi.org/10.1002/hbm.24042

    • PubMed
    • Google Scholar
68. 1. Marques JP
    2. Kober T
    3. Krueger G
    4. van der Zwaag W
    5. Van de Moortele PF
    6. Gruetter R

    (2010) MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field

    NeuroImage 49:1271–1281.

    https://doi.org/10.1016/j.neuroimage.2009.10.002

    • PubMed
    • Google Scholar
69. 1. Martinez-Conde S
    2. Cudeiro J
    3. Grieve KL
    4. Rodriguez R
    5. Rivadulla C
    6. Acuña C

    (1999) Effects of feedback projections from area 18 layers 2/3 to area 17 layers 2/3 in the cat visual cortex

    Journal of Neurophysiology 82:2667–2675.

    https://doi.org/10.1152/jn.1999.82.5.2667

    • PubMed
    • Google Scholar
70. 1. McManus JN
    2. Li W
    3. Gilbert CD

    (2011) Adaptive shape processing in primary visual cortex

    PNAS 108:9739–9746.

    https://doi.org/10.1073/pnas.1105855108

    • PubMed
    • Google Scholar
71. 1. Mendola JD
    2. Dale AM
    3. Fischl B
    4. Liu AK
    5. Tootell RBH

    (1999) The representation of illusory and real contours in human cortical visual Areas revealed by functional magnetic resonance imaging

    The Journal of Neuroscience 19:8560–8572.

    https://doi.org/10.1523/JNEUROSCI.19-19-08560.1999

    • Google Scholar
72. 1. Meng M
    2. Remus DA
    3. Tong F

    (2005) Filling-in of visual phantoms in the human brain

    Nature Neuroscience 8:1248–1254.

    https://doi.org/10.1038/nn1518

    • PubMed
    • Google Scholar
73. 1. Merabet L
    2. Desautels A
    3. Minville K
    4. Casanova C

    (1998) Motion integration in a thalamic visual nucleus

    Nature 396:265–268.

    https://doi.org/10.1038/24382

    • PubMed
    • Google Scholar
74. 1. Millman KJ
    2. Aivazis M

    (2011) Python for scientists and engineers

    Computing in Science & Engineering 13:9–12.

    https://doi.org/10.1109/MCSE.2011.36

    • Google Scholar
75. 1. Moeller S
    2. Yacoub E
    3. Olman CA
    4. Auerbach E
    5. Strupp J
    6. Harel N
    7. Uğurbil K

    (2010) Multiband multislice GE-EPI at 7 tesla, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI

    Magnetic Resonance in Medicine 63:1144–1153.

    https://doi.org/10.1002/mrm.22361

    • PubMed
    • Google Scholar
76. 1. Muckli L
    2. Singer W
    3. Zanella FE
    4. Goebel R

    (2002) Integration of multiple motion vectors over space: an fMRI study of transparent motion perception

    NeuroImage 16:843–856.

    https://doi.org/10.1006/nimg.2002.1085

    • PubMed
    • Google Scholar
77. 1. Muckli L
    2. Kohler A
    3. Kriegeskorte N
    4. Singer W

    (2005) Primary visual cortex activity along the apparent-motion trace reflects illusory perception

    PLOS Biology 3:e265.

    https://doi.org/10.1371/journal.pbio.0030265

    • PubMed
    • Google Scholar
78. 1. Muckli L
    2. De Martino F
    3. Vizioli L
    4. Petro LS
    5. Smith FW
    6. Ugurbil K
    7. Goebel R
    8. Yacoub E

    (2015) Contextual feedback to superficial layers of V1

    Current Biology 25:2690–2695.

    https://doi.org/10.1016/j.cub.2015.08.057

    • PubMed
    • Google Scholar
79. 1. Müller NG
    2. Kleinschmidt A

    (2004) The attentional 'spotlight's' penumbra: center-surround modulation in striate cortex

    NeuroReport 15:977–980.

    https://doi.org/10.1097/00001756-200404290-00009

    • PubMed
    • Google Scholar
80. 1. Ogren MP
    2. Hendrickson AE

    (1977) The distribution of pulvinar terminals in visual Areas 17 and 18 of the monkey

    Brain Research 137:343–350.

    https://doi.org/10.1016/0006-8993(77)90344-4

    • PubMed
    • Google Scholar
81. 1. Oliphant TE

    (2007) Python for scientific computing

    Computing in Science & Engineering 9:10–20.

    https://doi.org/10.1109/MCSE.2007.58

    • Google Scholar
82. 1. Olman CA
    2. Harel N
    3. Feinberg DA
    4. He S
    5. Zhang P
    6. Ugurbil K
    7. Yacoub E

    (2012) Layer-specific fMRI reflects different neuronal computations at different depths in human V1

    PLOS ONE 7:e32536.

    https://doi.org/10.1371/journal.pone.0032536

    • PubMed
    • Google Scholar
83. 1. Paradiso MA
    2. Nakayama K

    (1991) Brightness perception and filling-in

    Vision Research 31:1221–1236.

    https://doi.org/10.1016/0042-6989(91)90047-9

    • PubMed
    • Google Scholar
84. 1. Pasley BN
    2. Inglis BA
    3. Freeman RD

    (2007) Analysis of oxygen metabolism implies a neural origin for the negative BOLD response in human visual cortex

    NeuroImage 36:269–276.

    https://doi.org/10.1016/j.neuroimage.2006.09.015

    • PubMed
    • Google Scholar
85. 1. Peirce JW

    (2007) PsychoPy--psychophysics software in python

    Journal of Neuroscience Methods 162:8–13.

    https://doi.org/10.1016/j.jneumeth.2006.11.017

    • PubMed
    • Google Scholar
86. 1. Peirce JW

    (2008) Generating stimuli for neuroscience using PsychoPy

    Frontiers in Neuroinformatics 2:2008.

    https://doi.org/10.3389/neuro.11.010.2008

    • Google Scholar
87. 1. Pereverzeva M
    2. Murray SO

    (2008) Neural activity in human V1 correlates with dynamic lightness induction

    Journal of Vision 8:8.

    https://doi.org/10.1167/8.15.8

    • PubMed
    • Google Scholar
88. 1. Perna A
    2. Tosetti M
    3. Montanaro D
    4. Morrone MC

    (2005) Neuronal mechanisms for illusory brightness perception in humans

    Neuron 47:645–651.

    https://doi.org/10.1016/j.neuron.2005.07.012

    • PubMed
    • Google Scholar
89. 1. Pessoa L
    2. Mingolla E
    3. Neumann H

    (1995) A contrast- and luminance-driven multiscale network model of brightness perception

    Vision Research 35:2201–2223.

    https://doi.org/10.1016/0042-6989(94)00313-0

    • PubMed
    • Google Scholar
90. Software

    1. Pinheiro J
    2. Bates D
    3. DebRoy S
    4. Sarkar D

    (2017) Nlme: linear and nonlinear mixed effects models, version 1.0

    cran.r-project.

    https://CRAN.R-project.org/package=nlme
91. 1. Polimeni JR
    2. Fischl B
    3. Greve DN
    4. Wald LL

    (2010) Laminar analysis of 7T BOLD using an imposed spatial activation pattern in human V1

    NeuroImage 52:1334–1346.

    https://doi.org/10.1016/j.neuroimage.2010.05.005

    • PubMed
    • Google Scholar
92. 1. Poltoratski S
    2. McCormack D
    3. Tong F

    (2016) Orientation-Tuned surround suppression in the human LGN

    Journal of Vision 16:875.

    https://doi.org/10.1167/16.12.875

    • Google Scholar
93. 1. Poser BA
    2. Koopmans PJ
    3. Witzel T
    4. Wald LL
    5. Barth M

    (2010) Three dimensional echo-planar imaging at 7 tesla

    NeuroImage 51:261–266.

    https://doi.org/10.1016/j.neuroimage.2010.01.108

    • PubMed
    • Google Scholar
94. Software

    1. R Development Core Team

    (2017) R: A Language and Environment for Statistical Computing, version 4.0.1

    R Foundation for Statistical Computing, Vienna, Austria.

    https://www.R-project.org
95. 1. Rao RP
    2. Ballard DH

    (1999) Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects

    Nature Neuroscience 2:79–87.

    https://doi.org/10.1038/4580

    • PubMed
    • Google Scholar
96. 1. Ress D
    2. Glover GH
    3. Liu J
    4. Wandell B

    (2007) Laminar profiles of functional activity in the human brain

    NeuroImage 34:74–84.

    https://doi.org/10.1016/j.neuroimage.2006.08.020

    • PubMed
    • Google Scholar
97. 1. Rezak M
    2. Benevento LA

    (1979) A comparison of the organization of the projections of the dorsal lateral geniculate nucleus, the inferior pulvinar and adjacent lateral pulvinar to primary visual cortex (area 17) in the macaque monkey

    Brain Research 167:19–40.

    https://doi.org/10.1016/0006-8993(79)90260-9

    • PubMed
    • Google Scholar
98. 1. Rockland KS
    2. Pandya DN

    (1979) Laminar origins and terminations of cortical connections of the occipital lobe in the rhesus monkey

    Brain Research 179:3–20.

    https://doi.org/10.1016/0006-8993(79)90485-2

    • PubMed
    • Google Scholar
99. 1. Rockland KS
    2. Virga A

    (1989) Terminal arbors of individual "feedback" axons projecting from area V2 to V1 in the macaque monkey: a study using immunohistochemistry of anterogradely transported Phaseolus vulgaris-leucoagglutinin

    The Journal of Comparative Neurology 285:54–72.

    https://doi.org/10.1002/cne.902850106

    • PubMed
    • Google Scholar
100. 1. Roe AW
     2. Lu HD
     3. Hung CP

     (2005) Cortical processing of a brightness illusion

     PNAS 102:3869–3874.

     https://doi.org/10.1073/pnas.0500097102

     • PubMed
     • Google Scholar
101. 1. Roelfsema PR
     2. Lamme VA
     3. Spekreijse H
     4. Bosch H

     (2002) Figure-ground segregation in a recurrent network architecture

     Journal of Cognitive Neuroscience 14:525–537.

     https://doi.org/10.1162/08989290260045756

     • PubMed
     • Google Scholar
102. 1. Rossi AF
     2. Rittenhouse CD
     3. Paradiso MA

     (1996) The representation of brightness in primary visual cortex

     Science 273:1104–1107.

     https://doi.org/10.1126/science.273.5278.1104

     • PubMed
     • Google Scholar
103. 1. Rossi AF
     2. Paradiso MA

     (1999) Neural correlates of perceived brightness in the retina, lateral geniculate nucleus, and striate cortex

     The Journal of Neuroscience 19:6145–6156.

     https://doi.org/10.1523/JNEUROSCI.19-14-06145.1999

     • Google Scholar
104. 1. Saalmann YB
     2. Pinsk MA
     3. Wang L
     4. Li X
     5. Kastner S

     (2012) The pulvinar regulates information transmission between cortical Areas based on attention demands

     Science 337:753–756.

     https://doi.org/10.1126/science.1223082

     • PubMed
     • Google Scholar
105. 1. Sasaki Y
     2. Watanabe T

     (2004) The primary visual cortex fills in color

     PNAS 101:18251–18256.

     https://doi.org/10.1073/pnas.0406293102

     • PubMed
     • Google Scholar
106. Preprint

     1. Schnabel UH
     2. Kirchberger L
     3. van Beest E
     4. Mukherjee S
     5. Barsegyan A
     6. Lorteije JAM
     7. van der Togt C
     8. Self MW
     9. Roelfsema PR

     (2018) Feedforward and feedback processing during figure-ground perception in mice

     bioRxiv.

     https://doi.org/10.1101/456459

     • Google Scholar
107. 1. Seghier M
     2. Dojat M
     3. Delon-Martin C
     4. Rubin C
     5. Warnking J
     6. Segebarth C
     7. Bullier J

     (2000) Moving illusory contours activate primary visual cortex: an fMRI study

     Cerebral Cortex 10:663–670.

     https://doi.org/10.1093/cercor/10.7.663

     • PubMed
     • Google Scholar
108. 1. Self MW
     2. van Kerkoerle T
     3. Supèr H
     4. Roelfsema PR

     (2013) Distinct roles of the cortical layers of area V1 in figure-ground segregation

     Current Biology 23:2121–2129.

     https://doi.org/10.1016/j.cub.2013.09.013

     • PubMed
     • Google Scholar
109. 1. Self MW
     2. Roelfsema PR

     (2015) Scene perception in early vision: figure-ground organization in the lateral geniculate nucleus

     PNAS 112:6784–6785.

     https://doi.org/10.1073/pnas.1507097112

     • PubMed
     • Google Scholar
110. 1. Setsompop K
     2. Gagoski BA
     3. Polimeni JR
     4. Witzel T
     5. Wedeen VJ
     6. Wald LL

     (2012) Blipped-controlled aliasing in parallel imaging for simultaneous multislice Echo planar imaging with reduced g-factor penalty

     Magnetic Resonance in Medicine 67:1210–1224.

     https://doi.org/10.1002/mrm.23097

     • PubMed
     • Google Scholar
111. 1. Sherman SM
     2. Guillery RW

     (2002) The role of the thalamus in the flow of information to the cortex

     Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 357:1695–1708.

     https://doi.org/10.1098/rstb.2002.1161

     • PubMed
     • Google Scholar
112. 1. Shimono M
     2. Mano H
     3. Niki K

     (2012) The brain structural hub of interhemispheric information integration for visual motion perception

     Cerebral Cortex 22:337–344.

     https://doi.org/10.1093/cercor/bhr108

     • PubMed
     • Google Scholar
113. 1. Shmuel A
     2. Yacoub E
     3. Pfeuffer J
     4. Van de Moortele PF
     5. Adriany G
     6. Hu X
     7. Ugurbil K

     (2002) Sustained negative BOLD, blood flow and oxygen consumption response and its coupling to the positive response in the human brain

     Neuron 36:1195–1210.

     https://doi.org/10.1016/S0896-6273(02)01061-9

     • PubMed
     • Google Scholar
114. 1. Shmuel A
     2. Augath M
     3. Oeltermann A
     4. Logothetis NK

     (2006) Negative functional MRI response correlates with decreases in neuronal activity in monkey visual area V1

     Nature Neuroscience 9:569–577.

     https://doi.org/10.1038/nn1675

     • PubMed
     • Google Scholar
115. 1. Silver MA
     2. Ress D
     3. Heeger DJ

     (2007) Neural correlates of sustained spatial attention in human early visual cortex

     Journal of Neurophysiology 97:229–237.

     https://doi.org/10.1152/jn.00677.2006

     • PubMed
     • Google Scholar
116. 1. Slotnick SD
     2. Schwarzbach J
     3. Yantis S

     (2003) Attentional inhibition of visual processing in human striate and extrastriate cortex

     NeuroImage 19:1602–1611.

     https://doi.org/10.1016/S1053-8119(03)00187-3

     • PubMed
     • Google Scholar
117. 1. Smith SM
     2. Jenkinson M
     3. Woolrich MW
     4. Beckmann CF
     5. Behrens TE
     6. Johansen-Berg H
     7. Bannister PR
     8. De Luca M
     9. Drobnjak I
     10. Flitney DE
     11. Niazy RK
     12. Saunders J
     13. Vickers J
     14. Zhang Y
     15. De Stefano N
     16. Brady JM
     17. Matthews PM

     (2004) Advances in functional and structural MR image analysis and implementation as FSL

     NeuroImage 23 Suppl 1:S208–S219.

     https://doi.org/10.1016/j.neuroimage.2004.07.051

     • PubMed
     • Google Scholar
118. 1. Somers DC
     2. Dale AM
     3. Seiffert AE
     4. Tootell RB

     (1999) Functional MRI reveals spatially specific attentional modulation in human primary visual cortex

     PNAS 96:1663–1668.

     https://doi.org/10.1073/pnas.96.4.1663

     • PubMed
     • Google Scholar
119. 1. Standage GP
     2. Benevento LA

     (1983) The organization of connections between the pulvinar and visual area MT in the macaque monkey

     Brain Research 262:288–294.

     https://doi.org/10.1016/0006-8993(83)91020-X

     • Google Scholar
120. 1. Tootell RB
     2. Hadjikhani N
     3. Hall EK
     4. Marrett S
     5. Vanduffel W
     6. Vaughan JT
     7. Dale AM

     (1998) The retinotopy of visual spatial attention

     Neuron 21:1409–1422.

     https://doi.org/10.1016/S0896-6273(00)80659-5

     • PubMed
     • Google Scholar
121. 1. Tosun D
     2. Rettmann ME
     3. Han X
     4. Tao X
     5. Xu C
     6. Resnick SM
     7. Pham DL
     8. Prince JL

     (2004) Cortical surface segmentation and mapping

     NeuroImage 23 Suppl 1:S108–S118.

     https://doi.org/10.1016/j.neuroimage.2004.07.042

     • PubMed
     • Google Scholar
122. 1. Trojanowski JQ
     2. Jacobson S

     (1977) The morphology and laminar distribution of cortico-pulvinar neurons in the rhesus monkey

     Experimental Brain Research 28-28:51–62.

     https://doi.org/10.1007/BF00237085

     • Google Scholar
123. 1. Uludağ K

     (2008) Transient and sustained BOLD responses to sustained visual stimulation

     Magnetic Resonance Imaging 26:863–869.

     https://doi.org/10.1016/j.mri.2008.01.049

     • PubMed
     • Google Scholar
124. 1. Uludağ K
     2. Blinder P

     (2018) Linking brain vascular physiology to hemodynamic response in ultra-high field MRI

     NeuroImage 168:279–295.

     https://doi.org/10.1016/j.neuroimage.2017.02.063

     • PubMed
     • Google Scholar
125. 1. van der Walt S
     2. Colbert SC
     3. Varoquaux G

     (2011) The NumPy array: a structure for efficient numerical computation

     Computing in Science & Engineering 13:22–30.

     https://doi.org/10.1109/MCSE.2011.37

     • Google Scholar
126. 1. Van Essen DC
     2. Newsome WT
     3. Maunsell JH
     4. Bixby JL

     (1986) The projections from striate cortex (V1) to Areas V2 and V3 in the macaque monkey: asymmetries, areal boundaries, and patchy connections

     The Journal of Comparative Neurology 244:451–480.

     https://doi.org/10.1002/cne.902440405

     • PubMed
     • Google Scholar
127. 1. van Kerkoerle T
     2. Self MW
     3. Dagnino B
     4. Gariel-Mathis MA
     5. Poort J
     6. van der Togt C
     7. Roelfsema PR

     (2014) Alpha and gamma oscillations characterize feedback and feedforward processing in monkey visual cortex

     PNAS 111:14332–14341.

     https://doi.org/10.1073/pnas.1402773111

     • PubMed
     • Google Scholar
128. 1. Villeneuve MY
     2. Kupers R
     3. Gjedde A
     4. Ptito M
     5. Casanova C

     (2005) Pattern-motion selectivity in the human pulvinar

     NeuroImage 28:474–480.

     https://doi.org/10.1016/j.neuroimage.2005.06.015

     • PubMed
     • Google Scholar
129. 1. Villeneuve MY
     2. Thompson B
     3. Hess RF
     4. Casanova C

     (2012) Pattern-motion selective responses in MT, MST and the pulvinar of humans

     European Journal of Neuroscience 36:2849–2858.

     https://doi.org/10.1111/j.1460-9568.2012.08205.x

     • PubMed
     • Google Scholar
130. 1. Viswanathan S
     2. Jayakumar J
     3. Vidyasagar TR

     (2015) Contrast invariance of orientation tuning in the lateral geniculate nucleus of the feline visual system

     European Journal of Neuroscience 42:2250–2257.

     https://doi.org/10.1111/ejn.12991

     • PubMed
     • Google Scholar
131. 1. Viswanathan A
     2. Freeman RD

     (2007) Neurometabolic coupling in cerebral cortex reflects synaptic more than spiking activity

     Nature Neuroscience 10:1308–1312.

     https://doi.org/10.1038/nn1977

     • PubMed
     • Google Scholar
132. 1. von der Heydt R
     2. Peterhans E
     3. Baumgartner G

     (1984) Illusory contours and cortical neuron responses

     Science 224:1260–1262.

     https://doi.org/10.1126/science.6539501

     • PubMed
     • Google Scholar
133. Book

     1. von der Heydt R
     2. Friedman HS
     3. Zhou H

     (2003) Searching for the neural mechanism of color filling-in

     In: Pessoa L, De Weerd P, editors. Filling-In: From Perceptual Completion to Cortical Reorganization. Oxford University Press. pp. 106–127.

     https://doi.org/10.1093/acprof:oso/9780195140132.001.0001

     • Google Scholar
134. 1. Waehnert MD
     2. Dinse J
     3. Weiss M
     4. Streicher MN
     5. Waehnert P
     6. Geyer S
     7. Turner R
     8. Bazin PL

     (2014) Anatomically motivated modeling of cortical laminae

     NeuroImage 93 Pt 2:210–220.

     https://doi.org/10.1016/j.neuroimage.2013.03.078

     • PubMed
     • Google Scholar
135. 1. Wokke ME
     2. Vandenbroucke AR
     3. Scholte HS
     4. Lamme VA

     (2013) Confuse your illusion: feedback to early visual cortex contributes to perceptual completion

     Psychological Science 24:63–71.

     https://doi.org/10.1177/0956797612449175

     • PubMed
     • Google Scholar
136. 1. Wong-Riley MT

     (1974) Demonstration of geniculocortical and callosal projection neurons in the squirrel monkey by means of retrograde axonal transport of horseradish peroxidase

     Brain Research 79:267–272.

     https://doi.org/10.1016/0006-8993(74)90415-6

     • PubMed
     • Google Scholar
137. 1. Yushkevich PA
     2. Piven J
     3. Hazlett HC
     4. Smith RG
     5. Ho S
     6. Gee JC
     7. Gerig G

     (2006) User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability

     NeuroImage 31:1116–1128.

     https://doi.org/10.1016/j.neuroimage.2006.01.015

     • PubMed
     • Google Scholar
138. 1. Zhang Y
     2. Brady M
     3. Smith S

     (2001) Segmentation of brain MR images through a hidden markov random field model and the expectation-maximization algorithm

     IEEE Transactions on Medical Imaging 20:45–57.

     https://doi.org/10.1109/42.906424

     • PubMed
     • Google Scholar
139. 1. Zhao F
     2. Wang P
     3. Kim SG

     (2004) Cortical depth-dependent gradient-echo and spin-echo BOLD fMRI at 9.4T

     Magnetic Resonance in Medicine 51:518–524.

     https://doi.org/10.1002/mrm.10720

     • PubMed
     • Google Scholar
140. 1. Zipser K
     2. Lamme VAF
     3. Schiller PH

     (1996) Contextual modulation in primary visual cortex

     The Journal of Neuroscience 16:7376–7389.

     https://doi.org/10.1523/JNEUROSCI.16-22-07376.1996

     • Google Scholar

Acceptance summary:

This paper showcases the use of ultra-high field (7 Tesla) fMRI for disentangling the feedforward and feedback interactions across the cortical hierarchy underlying perception. The authors perform a detailed quantification of the laminar-specific patters of neural activity in human early visual cortex (areas V1-V3) during motion-based surface segmentation. This allows them to identify a signal component associated with illusory motion perception that is consistent with feedback from a higher-tier cortical area.

Decision letter after peer review:

Thank you for submitting your article "Feedback contribution to surface motion perception in the human early visual cortex" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Tobias Donner as the Reviewing Editor and Christian Büchel as the Senior Editor.

The following individuals involved in review of your submission have agreed to reveal their identity: Tomas Knapen (Reviewer #1); Lars Muckli (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:

This study investigates the role of feedback in surface motion perception. The authors use high-field (7T) fMRI to distinguish which layers of the early visual cortex are activated by global motion perception of a surface. This study provides new insights into neuronal processes underlying perception of surface motion – an important and ubiquitous perceptual phenomenon. The perception of surfaces is a construction taking into consideration nearby contextual information, in this instance, cues that give rise to perception of surface motion. The main novel contribution lies in the fact that the authors separate center from surface responses, which has not been done in humans before. Reviewers applaud the authors for approaching important aspects of high-resolution fMRI of visual cortex (anatomical segmentation, unwarping, registration, pRF-based retinotopy) with great care and state-of-the-art methodology. However, reviewers also raised a number of substantive concerns, which would need to be addressed in revision.

Essential revisions:

1) Address the negative overall BOLD response.

The responses in the surface region are dominated by a large negative BOLD response. This negative BOLD effect seems to dwarf the cortical depth-dependent results by a factor of around 20. Reviewers wondered whether a less negative BOLD response for the dynamic Pacman condition could be the result of 1) an increase in activity, or 2) a decrease in negative evoked activity.

In general, the physiological basis of negative BOLD is not well understood. Is it possible that activation in the right visual field could cause negative BOLD in the right V1, not because it induces neural activity in right V1 but for haemodynamic reasons (e.g. “blood stealing”) that are not well understood? In this case, given that right visual field stimulation is different in the three different conditions, could there be different amplitudes of negative BOLD in right V1 due to this? In other words: can we assume that any BOLD effect in right V1 is not influenced by neural activity in left V1? This could present a problem in the interpretation of the current results, which would need to be addressed. Also, reviewers thought that the presence of the big negative BOLD response creates a discrepancy in the paper, which focusses on the depth-dependent results.

Reviewers felt the study would benefit substantially from testing whether the depth-dependent results hold even when the sign of the figure response is changed. This could be achieved by manipulating the background and inverting the sign of the figure-related BOLD response – the feedback-related findings should be the same in the different background conditions.

After some discussion, reviewers and Reviewing Editor agreed that you could approach this general issue in one of two ways:

i) You run an additional experiment that repeats the figure segmentation experiment without the textured background, so as to avoid the large negative BOLD response.

ii) You discuss this issue more prominently in the paper, and acknowledge the resulting limitations in terms of interpretation in the Discussion.

While option (i) would be the preferred course of action (allowing for a conclusive answer to this point), we (and reviewers) leave it to you to decide which of those options you choose.

2) Present laminar results in more detail; in particular, show individual subject data.

The depth-dependent results are presented relatively minimalistically and the analysis is done in summary form using linear mixed models. Two specific suggestions for improvement:

a) It seems that he peak of the depth-level BOLD response is shifted towards CSF for V1 relative to V2 and V3. Can the authors also make this point by an analysis that stays a little closer to the data, for example (and this is just an example) by fitting a quadratic function to the depth-level bold responses and showing that the peak indeed shifts upward for V1?

b) Layer-specific fMRI effects are still a challenging field. To improve the interpretation it would be good to show individual results – alignment of the depth layers to the activity – and main effects of profile across subjects (as supplementary figure).

3) Discuss relationship to (discrepancies with) animal physiology.

Similar questions have been addressed with electrophysiological recordings in animals, particularly by Pieter Roelfsema's lab. In the current study, you report that the laminar profile of surface segmentation differs across the different regions of the early visual cortex, with the effect peaking in superficial layers in V1, and in middle layers in V2 and V3. You propose a neural mechanism for these profiles, with top-down feedback targeting the superficial layers of V1, which subsequently target the middle layers of V2 and (directly or indirectly) V3.

These findings differ from effects of surface segmentation using electrophysiology in animals, which report both superficial and deep layers of V1 being involved. Please discuss possible sources of this apparent discrepancy.

4) Address the role of attention.

A similar interpretational problem as with the negative BOLD issue (point 1) arises with the role of attention. Attention has been suggested to have a similar laminar profile (Lawrence et al., 2019). Also note that Akin et al. suggest that the effect reported here is dependent on attention. It seems possible that the current signals do not at all reflect the neural correlate of surface segmentation, but rather only the secondary effect of top-down attention directed at a (perceptually moving) segmented surface? If the reported signal is an exclusively attentional one, rather than the perceptual signal that captured attention, this would change the interpretation quite profoundly.

As with point 1, the most direct way of addressing of this issue would be to run an additional experiment that controls attention more effectively than the current fixation task – either using a more challenging fixation task and showing psychophysically that this puts the stimuli of interest in the "near-absence" of attention (see work by Jochen Braun, Christof Koch, David Heeger, Sang-Hun Lee, and others); or manipulate attentional load parametrically and show that the current signals are (largely) independent of this. the alternative would be to acknowledge this issue and the resulting limitations in terms of interpretability. Again, we leave the decision about the course of action to you; but please note that points 1) and 2) could be addressed in a single experiment.

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

Thank you for submitting your article "Feedback contribution to surface motion perception in the human early visual cortex" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by Tobias Donner as the Reviewing Editor and Christian Büchel as the Senior Editor. The following reviewer has agreed to reveal their identity: Tomas Knapen (Reviewer #1).

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

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

This is an overall well-conducted high-resolution fMRI study on the cortical mechanisms of visual surface perception. The authors have thoroughly revised their work based on the reviewer comments. Reviewers and Editors feel that most of the comments were addressed adequately, with some additional control data and additional discussion of the interpretation of the effects. This has improved the manuscript. Yet, there do remain some lingering issues, which we feel would need to be addressed more explicitly through textual revision before this paper can be accepted for publication at eLife. We feel the authors should invest more into incorporating the reviewer comments, rather than simply defending their original conclusions.

Revisions for this paper:

1) Relationship with attention.

The authors have opted not to run an additional control experiment, but rather address the potential role of attention in their findings in Discussion. That is fine in principle, but we do find the new paragraph to be unsatisfying and unclear. Perhaps there may be some confusion about the nature of the concern. We realize that attention may play an intricate (perhaps necessary) role in the cortical computation underlying surface segmentation. But that is not the issue here: the issue is whether the reported effects are (i) due to this computation per se (whether or not some form of attention is involved), or (ii) due to a purely secondary effect of attention being drawn to a segmented surface.

This matters for how the results are conceptualized and understood by the broad readership: The first would constitute a specific neural correlate of surface segmentation per se, whereas the second would be a non-specific signature of any form of object perception (compared to no object).

We acknowledge that arbitrating between these two scenarios is generally a hard problem in visual neuroscience; yet, there are a few compelling demonstrations in the literature for doing so. We feel that, without using similar manipulations here, there is just now way of knowing, which of these two scenarios accounts for the current results. Future experiments will be needed to pinpoint this.

These points, and in particular the two possible scenarios, should be made explicit in the new Discussion paragraph.

2) Negative BOLD response due to stimulus with textured background.

We remain puzzled by the negative BOLD. Intriguingly, the onset of a whole-field textured background (Figure 6—figure supplement 3) elicits an fMRI response that is comparable in positive amplitude to the negative effects within the stimulus region (Figure 4 and 7). This might indicate that in the main experiment, the textured background causes ongoing positive responses throughout the run, presumably maintained by ongoing micro saccades and fixational instability. Then, the presentation onset of a stimulus has a two-fold effect; it decreases the ongoing response to this background (because it disappears) while also playing the role of an activating stimulus very locally (dark patches in the texture are brighter, whereas light patches are darker, stimulating both on and off visual channels temporarily). In this scenario, the negative response shown in Figure 4 should be seen as a combination of a positive and a negative component, that summed together produce the appearance of a delayed negative response.

If the negative BOLD response is indeed a composite of a positive and negative BOLD response, it would be interesting to see how the laminar effects can perhaps decompose this composite. A laminar analysis conducted on the first and later parts of the response separately may be highly insightful here. We understand that SNR may be insufficient to conclusively accommodate this analysis.

This idea (possibly others) deserves unpacking, over and above the brief hint towards this possibility the authors mention in subsection “Background dependence of the negative response”. The fact that the authors have seen similar responses in other contexts doesn't necessarily speak to mechanism in this case.

3) Relationship/discrepancy with previous animal neurophysiology studies:

The authors state in rebuttal: "Finally, electrophysiological experiments typically measure responses over a time window of only a few hundred milliseconds after stimulus onset (300 ms in case of Self et al., 2013). However, the initial and the sustained response to a surface stimulus can have different laminar profiles in V1 (Maier et al., 2011)."

If anything, this makes the discrepancy more puzzling: sustained responses, which the authors imply are more strongly reflected in the BOLD response, are predominant in the deep, not the middle and superficial layers (Maier et al., 2011; Self et al., 2013). This part of the Discussion needs revision.

4) Possibility of visual cortex effects being the result of top-down feedback to the LGN being fed forward to V1.

The authors seem to simply dismiss this possibility, which we find puzzling; a previous study (Akin et al) showed that the experimental modulation used here affects LGN, and depth dependent profiles as reported here in Figure 3, i.e. mostly middle and superficial activations, seems to be very plausibly in line with the effect arising from the LGN, especially when taking the deep-to-superficial draining vein effects into account. Of course, the authors performed a deconvolution to address drain, but we cannot assume that this got rid of any draining vein effects completely. Again, this alternative explanation is worth discussing, rather than simply dismissing it.

Essential revisions:

1) Address the negative overall BOLD response.

The responses in the surface region are dominated by a large negative BOLD response. This negative BOLD effect seems to dwarf the cortical depth-dependent results by a factor of around 20. Reviewers wondered whether a less negative BOLD response for the dynamic Pacman condition could be the result of 1) an increase in activity, or 2) a decrease in negative evoked activity.

In general, the physiological basis of negative BOLD is not well understood. Is it possible that activation in the right visual field could cause negative BOLD in the right V1, not because it induces neural activity in right V1 but for haemodynamic reasons (e.g. “blood stealing”) that are not well understood? In this case, given that right visual field stimulation is different in the three different conditions, could there be different amplitudes of negative BOLD in right V1 due to this? In other words: can we assume that any BOLD effect in right V1 is not influenced by neural activity in left V1? This could present a problem in the interpretation of the current results, which would need to be addressed. Also, reviewers thought that the presence of the big negative BOLD response creates a discrepancy in the paper, which focusses on the depth-dependent results.

Reviewers felt the study would benefit substantially from testing whether the depth-dependent results hold even when the sign of the figure response is changed. This could be achieved by manipulating the background and inverting the sign of the figure-related BOLD response – the feedback-related findings should be the same in the different background conditions.

After some discussion, reviewers and Reviewing Editor agreed that you could approach this general issue in one of two ways:

i) You run an additional experiment that repeats the figure segmentation experiment without the textured background, so as to avoid the large negative BOLD response.

ii) You discuss this issue more prominently in the paper, and acknowledge the resulting limitations in terms of interpretation in the Discussion.

While option (i) would would be the preferred course of action (allowing for a conclusive answer to this point), we (and reviewers) leave it to you to decide which of those options you choose.

We thank the reviewers for their thorough assessment of our manuscript. We agree that follow-up experiments investigating the negative response (and also the temporal dynamics of the response, especially at the stimulus edge) would be very interesting, but we think that the current data by itself provides important novel insights to justify publication, and discuss the insights and limitations of this study in detail. Moreover, we present new data from limited additional control experiments.

Your remark comprises several important considerations that we would like to address in turn:

1) The relatively low effect size of the top-down motion effect, relative to the amplitude of the negative BOLD response

2) The physiological basis of the negative BOLD response

3) A possible interhemispheric interaction

1) The small effect of the top down effect

The amplitude of the negative response (caused by the texture background) is indeed large, compared to the smaller top-down, apparent-motion effect. However, such a discrepancy in the magnitude of (presumably) bottom-up and top-down effects is not uncommon. We have added the following paragraph to the Discussion section to address this issue. (See also following point, related to the interpretation of negative response.)

“The amplitude of the apparent motion effect (Figure 3) is small, compared with the overall response strength (e.g. Figure 4). […] We see no principled objection against the interpretability of a condition contrast among control and experimental responses that both yield BOLD responses smaller than the BOLD baseline response.”

2) Negative BOLD

To better understand the negative response to the figure in our experiments, we conducted a few limited post-hoc experiments in single participants. First, we contrasted the texture against an equiluminant, dark background in a single participant. This experiment showed that the texture yields a strong, positive response. The methods for this experiment are described in “Control experiment”, results presented in paragraph one of the Results and subsection “ Background dependence of the negative response”, and discussed in sub-section “Negative BOLD response”.

It should also be mentioned that the reduced size of the stimuli in the figures, gives a somewhat misleading idea on the visibility of the texture. Figure 1—figure supplement 1, if rendered such that the radius of the disk is 3.75cm gives, an impression of what it would have looked like in the scanner, when viewed from ~57cm. The clear granularity of the texture with peak-to-peak Michelson contrast of 92% is in line with a significant texture-driven response, which could be the basis of the negative BOLD responses in our experiments.

At the end of first paragraph of Materials and methods section, we have added a sentence to make sure the readers have a correct impression of the stimulus and its background: “Note that in our figures, the texture backgrounds are proportionally reduced with stimulus size and do not convey a good impression of the granularity and contrast of the texture (e.g. compare Figure 1—figure supplement 1 and Figure 1—figure supplement 2).”

In a further experiment (see new Figure 6—figure supplement 3), we presented one participant with the following conditions: Baseline condition (homogenous dark); Condition A grey square on black homogenous background; Condition B grey square on black textured background; Condition C whole field texture background. The fixation spot was placed at the centre of the screen, the square was 6.95 degree x 6.95 degree in size, with its centre at a 3.82 deg eccentricity. Analysis was done in a ROI corresponding to the centre of the square representation in V1.

As in the previous control experiment, contrasting full-field texture (C) with equiluminous homogenous black (Baseline) gave a strong positive BOLD response in the square ROI, similar to the negative BOLD in the conditions of our main experiment. There was also a limited positive BOLD response when contrasting the grey square on texture (B) to the grey square on homogenous black (A), suggesting some leakage of the texture-driven response into the square representation. In addition, subtracting activity for the square on texture (B) from full-field texture (C) yielded a negative BOLD response, again similar to what was observed in our main experiment. Taken together, these results suggest that for the stimuli we have chosen, a straightforward explanation of the negative BOLD inside the figure is the relative lack of activation driven by a homogenous stimulus compared to a texture surface.

The BOLD response in the figure hence is not due to vascular steal: the current consensus is that vascular steal does not occur in healthy subjects and is rather a sign of pathology, and is caused by a large, decrease in neural activation (e.g. see Shmuel et al., 2002; Pasley et al., 2007; Shmuel et al., 2006; Boorman et al., 2010; Devor et al., 2007). Instead the results support the interpretation that the negative BOLD was due to the relative lack of stimulation in the grey figure region compared to the textured background.

The following sentence was added to the first paragraph of the Results section, to make clear to the reader early on that the texture can explain the negative response in the main experiment:

“Control experiments supported the idea that the negative sign of the response was related to the much stronger response to the texture in the background than to the homogenous grey in the figure (see Figure 6—figure supplement 2 and Figure 6—figure supplement 3).”

Moreover, the following paragraph was added to the Discussion subsection “Negative BOLD response”:

“A control experiment revealed that the negative response was only observed when the experimental stimuli were presented on a texture background (Figure 6 and 7). […] Instead, the results of our control experiments support the interpretation that the negative BOLD was due to the relative lack of stimulation in the grey figure region compared to the textured background.”

3) Differences in right-field stimulation affecting interhemispheric balance and contributing to the correlate of the illusion

Our design used a stimulus manipulation in the right hemifield, in order to induce an illusion in the left hemifield. This raises the possibility of interhemispheric interactions that may cause a stronger negative BOLD signal in the illusory condition compared to the control condition. In principle, a combination of transcallosal, interhemispheric connections (Clarke and Miklossy, 1990; Essen and Zeki, 1978; Glickstein and Whitteridge, 1976; Houzel and Milleret, 1999; Van Essen et al., 1982; Wong-Riley, 1974) as well as feedback of lateral interactions taking place in higher-order visual cortex could affect the interhemispheric balance of activity in V1, V2 and V3. These higher level interactions may also include attentional imbalances. Several studies have demonstrated that spatial attention causes both a positive response in the cortical representation of attended locations, and a negative response at unattended locations (Bressler et al., 2013; Müller and Kleinschmidt, 2004; Silver et al., 2007; Slotnick et al., 2003; Somers et al., 1999; Tootell et al., 1998).

As we indicated above, we propose that the primary cause of the negative BOLD in the grey surface is primarily related to the absence of texture inside the figure compared to the texture baseline. Nevertheless, we verified whether an imbalance between the constant area of the stimulus in one hemifield (where the illusion is probed), and the part of the stimulus in the other hemifield (where the illusion is either generated or prevented in the control conditions) could explain the pattern of results we obtained.

Our data directly speak against this possibility. The right field stimulation in the motion illusion stimulus (Figure 1—figure supplement 2A) and the dynamic control stimulus (Figure 1—figure supplement 2C) were designed to be similarly high, whereas the right field stimulation in the static control condition (Figure 1—figure supplement 2B) was designed to be low. If an activity imbalance between hemifields had been the primary cause of our pattern of results, we would have expected that the dynamic control condition and the illusion condition would both have induced a similar decrease in activity in the left hemifield representation compared to the static control condition. In addition, in the left hemifield’s representation, an increased (less negative) BOLD would have been observed for the dynamic than for the static control condition. Instead, we find that in V1 and V2, the static and dynamic control conditions yield the same activity levels in the left hemifield representation, and that in the left hemifield representation the motion illusion condition gives the highest (least negative) response, i.e., the illusory motion condition yields activity that exceeds (i.e. is less negative than) the activity in both control conditions (see Figure 3—figure supplement 1). This shows that the dynamic stimulation in the right hemifield in itself (and a potential associated imbalance in attention) is unlikely to be responsible for the observed pattern of results in V1 and V2.

Note that in V3, we found that the response in the representation of the left part of the stimulus (where the illusion is probed) is smaller for the dynamic control condition than for the static control condition, which suggests that V3, in contrast to V1 and V2, is sensitive to differences in activity between hemifields. Accordingly, when contrasting the motion illusion condition against the dynamic control condition, the positive difference is smaller than when contrasting the motion illusion condition against the static control condition. This shows that, in V3, the former rather than the latter statistical contrast provides the best estimation of the motion illusion, and that overall the dynamic control condition was superior over the static control condition.

We have included the above four paragraphs integrally in a new sub-section in the Discussion (“Hemispheric imbalances in stimulation”). Also note that we have backed up the last paragraph, which is based on a difference in activity levels in the left-hemisphere part of the stimulus between static and dynamic control conditions, by additional analysis that is included in the Results section. We slightly changed the flow on these two pages to be able to include the results of these extra analyses.

Related to the apparent difference between the control conditions in V2 and V3, we ran a mixed-effects model comparison, but restricted to V2 and V3 and the two control conditions. We found a significant ROI by condition interaction (likelihood ratio (df): 16.7 (1), p < 0.0001). In other words, we found a statistically significant difference in the pattern of stimulus-induced activation caused by the control conditions in V2 and V3. Moreover, we found a ROI by depth interaction showing that the depth profiles differed between ROIs (likelihood ratio (df): 3.9 (1), p = 0.0491).

Another relevant outcome of the anatomically restricted analysis was that in V3, the dynamic and stationary controls were not equivalent (mixed effects model comparison, limited to V3 and the two control conditions, testing for an effect of “condition”; likelihood ratio (df): 35.8 (1), p = <0.0001). We included this result as well as it further supports the fact that the depth resolved activity profiles were different among visual areas.

2) Present laminar results in more detail; in particular, show individual subject data.

The depth-dependent results are presented relatively minimalistically and the analysis is done in summary form using linear mixed models. Two specific suggestions for improvement:

a) It seems that he peak of the depth-level BOLD response is shifted towards CSF for V1 relative to V2 and V3. Can the authors also make this point by an analysis that stays a little closer to the data, for example (and this is just an example) by fitting a quadratic function to the depth-level bold responses and showing that the peak indeed shifts upward for V1?

Thank you for pointing out this issue, we agree that a more direct test of the distribution of peak positions is required. First, we include a new Figure with single subject cortical depth profiles (Figure 3—figure supplement 2). Second, we provide an additional statistical test. As you point out, the question we would like to address is whether the distribution of peak positions is shifted towards superficial layers in V1, relative to V2 and V3. We propose to address this question by a chi-square test on the distribution of superficial peaks, as follows (see new paragraphs in Materials and methods and Results sections respectively):

Materials and methods:

“We investigated the shape of cortical depth profiles in more detail by comparing the distribution of peak positions in superficial layers between cortical areas. […] The ratio of superficial peaks (in the single subject profiles) was compared between areas with a chi-squared test.”

Results

“The ratio of superficial peak positions in the cortical depth profiles of the condition contrast “motion induction stimulus” vs. “dynamic control stimulus” was compared with a chi-squared test. […] Thus, the ratio of superficial peak positions (in the single subject cortical depth profiles) is significantly higher in striate than in extrastriate cortex.”

b) Layer-specific fMRI effects are still a challenging field. To improve the interpretation it would be good to show individual results – alignment of the depth layers to the activity – and main effects of profile across subjects (as supplementary figure).

We agree that it is good practice to be as transparent as possible, and particularly with a relatively novel technique as layer-specific fMRI. The updated Figure 3—figure supplement 2 presents single-subject cortical depth profiles. The single-subject profiles are noisy, in line with the relatively low SNR of the depth-specific fMRI signal, as can also be seen in previous publications (see e.g. Muckli et al., 2015; Kok et al., 2016. Figure S3).

3) Discuss relationship to (discrepancies with) animal physiology.

Similar questions have been addressed with electrophysiological recordings in animals, particularly by Pieter Roelfsema's lab. In the current study, you report that the laminar profile of surface segmentation differs across the different regions of the early visual cortex, with the effect peaking in superficial layers in V1, and in middle layers in V2 and V3. You propose a neural mechanism for these profiles, with top-down feedback targeting the superficial layers of V1, which subsequently target the middle layers of V2 and (directly or indirectly) V3.

These findings differ from effects of surface segmentation using electrophysiology in animals, which report both superficial and deep layers of V1 being involved. Please discuss possible sources of this apparent discrepancy.

We agree that this aspect deserves more attention, and have expanded the Discussion section accordingly, as follows:

“Relationship to electrophysiological findings

Self et al. (Self et al., 2013) studied the laminar profile of figure-ground segregation in monkey V1. They observed neuronal activity related to feedforward, horizontal, and feedback mechanisms, that are thought to reflect processing of stimulus texture, borders, and figure-ground segregation, respectively. […] Results based on experimental designs similar to the one used in the present study have, to the best of our knowledge, not been reported.”

4) Address the role of attention.

A similar interpretational problem as with the negative BOLD issue (point 1) arises with the role of attention. Attention has been suggested to have a similar laminar profile (Lawrence et al., 2019). Also note that Akin et al. suggest that the effect reported here is dependent on attention. It seems possible that the current signals do not at all reflect the neural correlate of surface segmentation, but rather only the secondary effect of top-down attention directed at a (perceptually moving) segmented surface? If the reported signal is an exclusively attentional one, rather than the perceptual signal that captured attention, this would change the interpretation quite profoundly.

As with point 1, the most direct way of addressing of this issue would be to run an additional experiment that controls attention more effectively than the current fixation task – either using a more challenging fixation task and showing psychophysically that this puts the stimuli of interest in the "near-absence" of attention (see work by Jochen Braun, Christof Koch, David Heeger, Sang-Hun Lee, and others); or manipulate attentional load parametrically and show that the current signals are (largely) independent of this. the alternative would be to acknowledge this issue and the resulting limitations in terms of interpretability. Again, we leave the decision about the course of action to you; but please note that points 1) and 2) could be addressed in a single experiment.

The overall magnitude of the negative BOLD is more likely explained by the absence of texture in the figure region. The small differences in activity (less negative BOLD for induced motion compared to control conditions) reflect the differences among the experimental and control stimuli. With respect to attention, we have excluded one major factor related to potential imbalances in stimulation and attention to the left and right halves of the stimulus display (see point 1). In addition, a comparison with results in a recent paper by Lawrence et al., 2018, suggests that the depth distribution of activity in our data did not reflect a mere attentional effect. In Lawrence et al., 2018, attention to a stationary grid resulted in a bias towards superficial layers that was similar in V1, V2 and V3. By contrast, in our study, motion induction inside a surface yielded more superficial activity in V1, but deeper activity in V2 and V3. So, whereas attention likely contributed to both our own data as well as those of Lawrence et al., 2018, it is plausible that the different results between the two studies are caused by the different manners in which the target of attention were defined in the two studies, namely a luminance defined grid in one case, and an induced motion feature in the other case.

Nevertheless, we acknowledge that the question what the precise balance is of attentional contributions versus motion induction is interesting. We have considered the reviewer’s suggestions, and reasoned as follows: If we impose in a few conditions an increase in the strictness of the fixation task, and we observed that the magnitude of the increased signal in the right hemisphere stimulus representation (compared to control conditions) decreases, this can represent both a decrease in attention or attentional capture, and a decrease of the assignment of motion to the surface. So, a parametric decrease in the signal with increasing strictness of the fixation task is not easily interpretable. Alternatively, it could be the case that a differential signal (between experimental and control stimuli) survives, irrespective of the attentional manipulation. This finding in itself is also not interpretable. To show that this signal is specifically related to the motion assignment, and not to other aspects that segregate the figure from the background, one would also have to do a parametric variation of the motion strength and at least one other physical stimulus variable (e.g., the contrast difference between surface and background). In addition, a precise interpretation of the fMRI data would require parallel psychophysical experiments that would have to calibrate double-task paradigms in order to separately titrate effects of attention (to fixation) and motion effects (induced inside the figure surface). In sum, we believe we have plausible arguments against more general exclusively attentional explanations and suggest that the detailed experiments to try to isolate contributions of the motion percept per se from attentional contributions fall outside the scope of the present work.

In the light of the above considerations, we have added a paragraph in the Discussion in which we offer the following interpretation of our data (as well as its limitations):

“Attention

Neurophysiological data suggest that the perceptual aspect of a surface is not filled in in the absence of attention (e.g., Poort et al., 2012) and that neural correlates of surface perception are not observed during anesthesia (Lamme et al., 1998). […] However, separating the contributions of attention on the one hand and motion analysis in visual scenes on the other hand to depth-resolved feedback-related fMRI signals is challenging and requires a further series of fMRI and psychophysical experiments outside the scope of the present work.”

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

Revisions for this paper:

1) Relationship with attention.

The authors have opted not to run an additional control experiment, but rather address the potential role of attention in their findings in Discussion. That is fine in principle, but we do find the new paragraph to be unsatisfying and unclear. Perhaps there may be some confusion about the nature of the concern. We realize that attention may play an intricate (perhaps necessary) role in the cortical computation underlying surface segmentation. But that is not the issue here: the issue is whether the reported effects are (i) due to this computation per se (whether or not some form of attention is involved), or (ii) due to a purely secondary effect of attention being drawn to a segmented surface.

This matters for how the results are conceptualized and understood by the broad readership: The first would constitute a specific neural correlate of surface segmentation per se, whereas the second would be a non-specific signature of any form of object perception (compared to no object).

We acknowledge that arbitrating between these two scenarios is generally a hard problem in visual neuroscience; yet, there are a few compelling demonstrations in the literature for doing so. We feel that, without using similar manipulations here, there is just now way of knowing, which of these two scenarios accounts for the current results. Future experiments will be needed to pinpoint this.

These points, and in particular the two possible scenarios, should be made explicit in the new Discussion paragraph.

We acknowledge that an interpretation of the top-down effect as a secondary effect of attention being drawn to the surface cannot be excluded, and have adjusted the respective paragraph in the Discussion section to clarify this:

“Attention

Although our data do not support that an imbalance in stimulation or attention drove the observed pattern of results, the fixation task we used likely permitted at least a minimal level of attention to the visual field as a whole. […] We cannot fully exclude that our observations reflect a general attentional contribution rather than computations specifically related to the perception of surface motion. These are exciting open questions, and it is encouraging that solving these questions is now within reach of high-field fMRI.”

2) Negative BOLD response due to stimulus with textured background.

We remain puzzled by the negative BOLD. Intriguingly, the onset of a whole-field textured background (Figure 6—figure supplement 3) elicits an fMRI response that is comparable in positive amplitude to the negative effects within the stimulus region (Figure 4 and 7). This might indicate that in the main experiment, the textured background causes ongoing positive responses throughout the run, presumably maintained by ongoing micro saccades and fixational instability. Then, the presentation onset of a stimulus has a two-fold effect; it decreases the ongoing response to this background (because it disappears) while also playing the role of an activating stimulus very locally (dark patches in the texture are brighter, whereas light patches are darker, stimulating both on and off visual channels temporarily). In this scenario, the negative response shown in Figure 4 should be seen as a combination of a positive and a negative component, that summed together produce the appearance of a delayed negative response.

If the negative BOLD response is indeed a composite of a positive and negative BOLD response, it would be interesting to see how the laminar effects can perhaps decompose this composite. A laminar analysis conducted on the first and later parts of the response separately may be highly insightful here. We understand that SNR may be insufficient to conclusively accommodate this analysis.

This idea (possibly others) deserves unpacking, over and above the brief hint towards this possibility the authors mention in subsection “Background dependence of the negative response”. The fact that the authors have seen similar responses in other contexts doesn't necessarily speak to mechanism in this case.

To shed more light on the issue of the negative BOLD response, we included two additional supplementary figures, presenting a simulation on a possible explanation for the delayed negative response, and a laminar analysis regarding the early and late phase of the negative BOLD response. We reference the new supplementary figure at the appropriate location in the Discussion:

“[…] The results lend further support to the hypothesis that the negative signal in the main experiment was driven by a relative lack of activation in response to a uniform stimulus compared to a texture surface. A simulation lends further support to this interpretation (Figure 4—figure supplement 4). Thus, we suggest that the negative surface response resulted primarily from an elevated baseline activity due to the texture background.”

The new simulation suggests that an elevated baseline (caused, for example, by ongoing activation due to the texture background and saccadic eye movements) and a small positive response (caused by the grey stimulus surface) could indeed account for the observed delayed negative response. The signal amplitudes chosen for the simulation are based on empirical data from the additional control experiments, and are therefore plausible.

Regarding the cortical depth profiles of the early and late response phases, we analyzed the early and late phase of the responses, as the reviewer suggested and found no evidence for a qualitatively different laminar profile. A reference to this new figure was added:

“Separate cortical depth profiles of the early and late response phases show no evidence for temporal differences in the laminar activation profile (Figure 4—figure supplement 5).”

3) Relationship/discrepancy with previous animal neurophysiology studies:

The authors state in rebuttal: "Finally, electrophysiological experiments typically measure responses over a time window of only a few hundred milliseconds after stimulus onset (300 ms in case of Self et al., 2013). However, the initial and the sustained response to a surface stimulus can have different laminar profiles in V1 (Maier et al., 2011)."

If anything, this makes the discrepancy more puzzling: sustained responses, which the authors imply are more strongly reflected in the BOLD response, are predominant in the deep, not the middle and superficial layers (Maier et al., 2011; Self et al., 2013). This part of the Discussion needs revision.

We have revised the respective section, and have removed the unclear argument. Several possible explanations for the discrepancy with electrophysiological studies remain, but we can obviously only speculate on this:

“Relationship to electrophysiological findings

Self et al. (Self et al., 2013) studied the laminar profile of figure-ground segregation in monkey V1. They observed neuronal activity related to feedforward, horizontal, and feedback mechanisms, that are thought to reflect processing of stimulus texture, borders, and figure-ground segregation, respectively. […] Furthermore, beyond the differences in signal measured, stimulus, and experimental design, it is difficult to use the few 100ms typically measured post-stimulus onset in neurophysiological experiments as a predictor for fMRI activity measured 10s and more after stimulus onset.”

4) Possibility of visual cortex effects being the result of top-down feedback to the LGN being fed forward to V1.

The authors seem to simply dismiss this possibility, which we find puzzling; a previous study (Akin et al) showed that the experimental modulation used here affects LGN, and depth dependent profiles as reported here in Figure 3, i.e. mostly middle and superficial activations, seems to be very plausibly in line with the effect arising from the LGN, especially when taking the deep-to-superficial draining vein effects into account. Of course, the authors performed a deconvolution to address drain, but we cannot assume that this got rid of any draining vein effects completely. Again, this alternative explanation is worth discussing, rather than simply dismissing it.

We agree that the LGN might have played a role, and have adjusted the Discussion section accordingly:

“An additional contribution to the depth-pattern of activity observed in extrastriate areas may have originated from the pulvinar, the LGN, and possibly other subcortical structures (Standage and Benevento, 1983; Trojanowski and Jacobson, 1977). […] Moreover, an involvement of the LGN in the perception of illusory motion has been observed by Akin et al., 2014, using an experimental design very similar to ours. As the V1 cortical depth profile we observed suggests similar levels of activity in mid-level to superficial layers (Figure 3), it is possible that a feedback signal assigning motion to the grey surface re-entered the LGN, was fed-forward to V1, and from V1 to V2 and V3. In summary, both cortical and subcortical sources of re-entrant feedback in lower-level visual areas may have contributed to the observed depth-resolved responses (see Figure 8).”

Author details

1. Ingo Marquardt

   Department of Cognitive Neuroscience, Maastricht Brain Imaging Centre (MBIC) Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Contributed equally with

   Peter De Weerd

   For correspondence

   ingo.marquardt@posteo.de

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-5178-9951

2. Peter De Weerd

   1. Department of Cognitive Neuroscience, Maastricht Brain Imaging Centre (MBIC) Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands
   2. Maastricht Center of Systems Biology (MACSBIO), Faculty of Science & Engineering, Maastricht University, Maastricht, Netherlands

   Contribution

   Conceptualization, Formal analysis, Supervision, Methodology, Writing - original draft, Writing - review and editing

   Contributed equally with

   Ingo Marquardt

   For correspondence

   p.deweerd@maastrichtuniversity.nl

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2252-5548

3. Marian Schneider

   Department of Cognitive Neuroscience, Maastricht Brain Imaging Centre (MBIC) Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands

   Contribution

   Conceptualization, Software, Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3192-5316

4. Omer Faruk Gulban

   Department of Cognitive Neuroscience, Maastricht Brain Imaging Centre (MBIC) Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands

   Contribution

   Conceptualization, Software, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7761-3727

5. Dimo Ivanov

   Department of Cognitive Neuroscience, Maastricht Brain Imaging Centre (MBIC) Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands

   Contribution

   Investigation

   Competing interests

   No competing interests declared

6. Yawen Wang

   Department of Cognitive Neuroscience, Maastricht Brain Imaging Centre (MBIC) Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, Netherlands

   Contribution

   Formal analysis, Investigation, Visualization

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0002-0768

7. Kâmil Uludağ

   1. Center for Neuroscience Imaging Research, Institute for Basic Science and Department of Biomedical Engineering, N Center, Sungkyunkwan University, Jangan-gu, Republic of Korea
   2. Techna Institute and Koerner Scientist in MR Imaging, University Health Network, Toronto, Canada

   Contribution

   Conceptualization, Supervision, Funding acquisition, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   Kamil.Uludag@rmp.uhn.ca

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2813-5930

• 1,255

  Page views

• 184

  Downloads

• 4

  Citations

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