* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Alpha-beta and Gamma Rhythms Subserve Feedback and
Artificial general intelligence wikipedia , lookup
Neuropsychology wikipedia , lookup
Metastability in the brain wikipedia , lookup
Emotional lateralization wikipedia , lookup
Cognitive neuroscience wikipedia , lookup
Aging brain wikipedia , lookup
Evolution of human intelligence wikipedia , lookup
History of neuroimaging wikipedia , lookup
Synaptic gating wikipedia , lookup
Visual selective attention in dementia wikipedia , lookup
Lateralization of brain function wikipedia , lookup
Dual consciousness wikipedia , lookup
Cognitive neuroscience of music wikipedia , lookup
Cortical cooling wikipedia , lookup
Visual memory wikipedia , lookup
Visual servoing wikipedia , lookup
Time perception wikipedia , lookup
Neuroplasticity wikipedia , lookup
Visual extinction wikipedia , lookup
Nervous system network models wikipedia , lookup
Human brain wikipedia , lookup
Neuroeconomics wikipedia , lookup
Neural correlates of consciousness wikipedia , lookup
Artificial intelligence for video surveillance wikipedia , lookup
Feature detection (nervous system) wikipedia , lookup
C1 and P1 (neuroscience) wikipedia , lookup
Cerebral cortex wikipedia , lookup
This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Georgios Michalareas1, Julien Vezoli1, Stan van Pelt1,2, Jan-Mathijs Schoffelen2,3, Henry Kennedy4,5 and Pascal Fries1,2 1 Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, Deutschordenstraße 46, 60528 Frankfurt, Germany. 2 Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Kapittelweg 29, 6525 EN Nijmegen, Netherlands. 3 Max Planck Institute for Psycholinguistics, Wundtlaan 1, 6525 XD Nijmegen, Netherlands. 4 Stem Cell and Brain Research Institute, INSERM U846, 18 avenue Doyen Lépine, 69675 Bron, France. 5 Université de Lyon, 37 rue du Repos, 69361 Lyon, France. Correspondence should be addressed to G.M. ([email protected]) and P.F. ([email protected]) Highlights: • • • • Gamma mediates forward and alpha-beta feedback influences among human visual areas Human inter-areal directed influences correlate with macaque laminar connectivity Rhythmic inter-areal influences establish a hierarchy of 26 human visual areas Alpha-beta influences differentially affect ventral- and dorsal-stream visual areas In Brief: Michalareas et al. show that in human visual cortex influences along feedforward projections predominate in the gamma band, whereas influences along feedback projections predominate in the alpha-beta band. These influences constrain a functional hierarchy in agreement with macaque anatomical hierarchy. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 1 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Summary Primate visual cortex is hierarchically organized. Bottom-up and top-down influences are exerted through distinct frequency channels, as was recently revealed in macaques by correlating inter-areal influences with laminar anatomical projection patterns. Because this anatomical data cannot be obtained in human subjects, we selected seven homologous macaque and human visual areas, and correlated the macaque laminar projection patterns to human inter-areal directed influences as measured with magnetoencephalography. We show that influences along feedforward projections predominate in the gamma band, whereas influences along feedback projections predominate in the alpha-beta band. Rhythmic inter-areal influences constrain a functional hierarchy of the seven homologous human visual areas that is in close agreement with the respective macaque anatomical hierarchy. Rhythmic influences allow an extension of the hierarchy to 26 human visual areas including uniquely human brain areas. Hierarchical levels of ventral and dorsal stream visual areas are differentially affected by inter-areal influences in the alpha-beta band. Introduction Non-human primate visual cortical areas are organized in a hierarchy with characteristic laminar patterns of feedforward and feedback projections (Barone et al., 2000; Felleman and Essen, 1991; Markov et al., 2014). Feedforward projections typically target layer 4. They originate predominantly from superficial layers, and this predominance increases with the number of hierarchical levels bridged by the projection. By contrast, feedback projections typically target layers 1 and 6. They originate predominantly from infragranular layers, and this predominance increases with the number of hierarchical levels bridged by the projection. In agreement with this, contextual feedback signals activate predominantly superficial layers in human V1 (Muckli et al., 2015; Olman et al., 2012). Cortical layers differ not only in their projection patterns, but also with regard to local rhythmic synchronization. Synchronization in the gamma-frequency band is strongest in superficial layers, whereas synchronization in the alpha-beta frequency band is strongest in infragranular layers (Buffalo et al., 2011; Xing et al., 2012). These results taken together with the laminar projection patterns suggest that gamma might subserve feedforward and alpha-beta feedback signaling (Fries, 2015; Lee et al., 2013). This is supported by patterns of inter-laminar delay and causality (Bollimunta et al., 2008; Livingstone, 1996; Plomp et al., 2014). Further, electrical stimulation of V1 induces enhanced gamma-band activity in V4, whereas area V4 stimulation induces enhanced alpha-betaband activity in V1 (van Kerkoerle et al., 2014). When directed inter-areal influences across 28 area pairs are assessed through large-scale highdensity electrocorticography in rhesus macaques and correlated with anatomical projection patterns, gamma is found systematically stronger in the feedforward direction and beta in the feedback direction (Bastos et al., 2015). The high consistency of this effect made it possible to construct a hierarchy of visual areas based on directed influences alone. These findings raise the intriguing possibility that a similar analysis in human cortex would reveal the functional hierarchy of visual areas. In human subjects, anatomical tract tracing methods, requiring active axonal transport, are neither possible in the living nor in the post-mortem brain. Further, tractographic analysis of diffusion MRI does not indicate the directionality of pathways and therefore cannot explore the hierarchical organization of inter-areal pathways. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 2 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. In the present study, we have first selected seven human areas showing strong homology to macaque areas. This enabled us to infer the structural hierarchy in the human cortex and to examine its functional hierarchy using magnetoencephalography. This shows that causal interactions along the feedforward and feedback pathways are exerted in distinct frequency bands. Based on these functional markers, we were able to derive the full hierarchy of 26 human visual areas based on the asymmetry of their feedforward and feedback interactions. Results Visually induced responses Forty-three human subjects were instructed to attentively monitor a visual stimulus for unpredictable changes, in order to engage both bottom-up and top-down influences. This paradigm reliably induces gamma-band activity in visual cortex (Hoogenboom et al., 2006). While subjects fixated centrally, a large circular sine-wave grating contracting towards the fixation point was presented. The stimulus could change speed at any time between 0.75 and 3 s after onset, which had to be reported within 0.5 s in order to obtain positive feedback (Figure 1A). Visual stimulation and task performance induced grand-average power reductions in the alpha and beta bands and enhancements in the gamma band in many sensors (Figure 1B), consistent with previous reports (Hoogenboom et al., 2006; van Pelt et al., 2012). Gamma power enhancements were strongest over occipital and parietal MEG sensors (Figure 1C). The sensors marked in Figure 1C show stimulusinduced gamma power enhancements that start around 0.1 s after stimulus onset and are sustained for the duration of visual stimulation and task performance (Figure 1D). Alpha and beta power shows sustained reductions at a slightly longer latency. Between 0.05 and 0.35 s, there is an additional theta (3-8 Hz) power enhancement, most likely reflecting the event-related field (ERF) (Jutai et al., 1984). To avoid the ERF and the effect of its non-stationarity on Granger Causality (GC) estimation (Wang et al., 2008), further analyses use the data from 0.365 s after stimulus onset up to the stimulus change, segmented into non-overlapping 0.365 s epochs (see Experimental Procedures). Induced gamma power change estimated at the level of the cortical sheet revealed an occipital peak extending into parietal and temporal regions (Figure 2A). In order to investigate GC among human visual areas and to compare it to macaque laminar connectivity, we selected seven visual areas, for which anatomical retrograde tracing data were available, and for which there is evidence of interspecies homology (see Experimental procedures). These areas and their homologies are illustrated in Figure 2B. Areas V1 and V2 are substantially larger than any one of the other areas (Figure 2B) and exhibit the largest gamma power enhancements (Figure 2A). To render V1 and V2 more comparable in size to the other areas, and to focus on their visually activated subregions, we used the vertices inside V1 and V2 that were within 2 cm from the local gamma maxima. These vertices and the vertices of the remaining areas are shown in Figure 2C. Granger Causality – Field spread effect MEG sensors pick up mixtures of signals from multiple sources in the brain. This mixing can be partly deciphered by projecting MEG data into source space. However, the closer two estimated sources are, the larger the remaining mixing, an effect referred to as “field spread” (Schoffelen and Gross, 2009). Field spread leads to artifactual correlation or coherence for source signals from proximate locations. Because field spread decays with distance between sources, the dominant spatial pattern of coherence to a given source (black parcel in Figure 3) is a smooth decay with distance from that source (Figure 3A, 3B). By contrast, the spatial pattern of GC exerted by that source does not show 3 Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. this smooth decay, but rather shows spatial peaks that have their maximum not at the source or in the immediate vicinity of the source (Figure 3C, 3D). This is likely due to the fact that field spread is instantaneous. GC eliminates instantaneous correlations, because they are not indicative of a causal influence. Granger Causality in feedforward and feedback directions GC was computed between all possible pairs of the seven selected visual areas. In order to estimate bias due to the GC metric used and the finite sample size, data epochs were randomized at each source location and GC computation was repeated. This bias estimation also entails an estimate of residual stimulus-locked components remaining after exclusion of the first 0.365 s after stimulus onset. The two hemispheres were analyzed independently. We averaged all inter-areal GC spectra and compared them to the average bias estimate in order to obtain a first estimate of overall inter-areal GC, in particular for the main spectral peaks, (Figure 4A). Average inter-areal GC exceeded the bias estimate across the spectrum and showed two clear peaks, one in the alpha-beta range, peaking at 11 Hz, and one in the gamma band, peaking around 60 Hz. Previous studies have shown clear differences between alpha and beta rhythms (Bressler and Richter, 2014; Gregoriou et al., 2015; Haegens et al., 2014; Wang, 2010), however, the present analysis revealed one peak spanning across alpha and beta frequency ranges (similar to (Buffalo et al., 2011)), which we therefore refer to as the alpha-beta band. Peak frequencies were essentially identical in the two hemispheres. We used macaque laminar connectivity data that quantifies the degree to which a given inter-areal projection has a feedforward or feedback character in order to investigate, whether GC differed between the feedforward and feedback directions. Retrograde tracer injection into a target area leads to labeling of projection neurons in the source areas. The more pronounced the feedforward (feedback) character of the anatomical projection, the higher (lower) is the proportion of supragranular projection neurons (Barone et al., 2000; Markov et al., 2014). This anatomical signature of the feedforward or feedback character of a projection is captured by the SLN index, i.e. the number of supragranular labeled neurons divided by the sum of supragranular and infragranular labeled neurons. The SLN value for all projections among the 7 selected macaque visual areas was calculated (Markov et al., 2014). If, between area A and area B, the SLN for the A-to-B projection was higher than the SLN for the B-to-A projection, then A-to-B was considered a feedforward and B-to-A a feedback projection. GC values were averaged independently for both directions in each area pair (Figure 4B, 4C for left and right hemisphere, respectively). The same was done with the surrogate data in order to estimate bias. GC significantly exceeded the bias estimate across the entire spectrum. GC in the alpha-beta band was stronger in feedback than in feedforward directions. In contrast, GC in the gamma band was substantially stronger in feedforward than in feedback directions. The statistical comparison revealed significant differences exclusively in these two frequency bands. Results are very similar in both hemispheres. The alpha-beta band cluster (P<0.05) ranged from 7 to 17 Hz in both hemispheres. The gamma-band cluster (P<0.05) ranged from 42 to 93 Hz in the left hemisphere and from 41 to 84 Hz in the right hemisphere. Figure 4D shows the GC in feedforward and feedback directions for all area pairs. GC peaks in the gamma and alpha-beta bands for all area pairs of both hemispheres. For all area pairs, GC exceeded the bias estimate for frequencies up to the upper end of the gamma band. When GC is compared Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 4 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. between feedforward and feedback directions, most area pairs confirm the above pattern. The results for the alpha-beta band are as follows: In the left hemisphere, alpha-beta GC was stronger in the feedback than in the feedforward direction in 13 out of 21 area pairs (61.91%), weaker in 1 area pair (MT-TEO; 4.76%) and not significantly different in another 7 (33.33%), i.e. it was more frequently significantly stronger in the feedback than in the feedforward direction (p=0.0018, sign test here and for the following tests). In the right hemisphere, alpha-beta GC was stronger in the feedback than in the feedforward direction in 13 out of 21 area pairs (61.91%), weaker in 2 area pairs (MT-TEO and MT-DP; 9.52%) and not significantly different in another 6 (28.57%) (p=0.0074). The results for the gamma band are as follows: In the left hemisphere, gamma GC was stronger in the feedforward than in the feedback direction in 17 out of 21 area pairs (80.95%), weaker in 1 area pair (DP-7A; 4.76%) and not significantly different in 3 other pairs (14.29%), i.e. it was more frequently significantly stronger in the feedforward than in the feedback direction (p=0.00015). In the right hemisphere, gamma GC was stronger in the feedforward than in the feedback direction in 16 out of 21 area pairs (76.19%), weaker in 1 area pair (DP-7A; 4.76%) and not significantly different in 4 (19.05%) (p=0.00028). The peak in the GC spectra that we have identified as the alpha-beta peak spans the classical alphaand beta-frequency bands. Alpha- and beta-band rhythms vary in frequency across subjects and can be influenced by stimulus and task conditions (Haegens et al., 2014). We calculated the average inter-areal GC among the seven visual areas separately per subject. When Gaussians were fitted to these averages in the 4-20 Hz range, most alpha-beta peaks were well approximated by a single Gaussian and the resulting peak frequencies extended from 7 to 19 Hz (Figure S1A, B; mean±SD = 11.02±2.45 Hz). Likewise, when Gaussians were fitted to the GC averages in the 30-100 Hz range, the resulting gamma peak frequencies extended from 52 to 69 Hz (Figure S1C, D; mean±SD = 59.16±4.16 Hz). Similar cross-subject variability has been described in previous studies using the same stimulus and task as used here and has been related to genetic factors (Hoogenboom et al., 2006; van Pelt et al., 2012). To account for this inter-subject variability in peak frequencies, we repeated the above GC analyses after aligning alpha-beta and gamma peak frequencies, respectively, across subjects (Figure 5). After alignment, GC values tended to increase, and the above differences between GC in the feedforward and feedback direction were confirmed. Significant differences remained identical except in a few cases, the most notable being V1-V2 in the right hemisphere, changing from an insignificant difference to a stronger alpha-beta GC in the feedforward than in the feedback direction. Because the peak-aligned analysis accounts for inter-subject variability, it was used for all further analyses (results without peak alignment are provided as supplemental information and are qualitatively identical). Correlation between human inter-areal Granger Causality and macaque anatomical projections So far, we have used the SLN metric to decide for each area pair, which one of the two reciprocal projections is feedforward and which feedback. As explained above, the SLN of an anatomical projection is the number of supragranular neurons normalized by the number of supragranular and infragranular neurons that give rise to the projection. Thus, the SLN quantifies the feedforward or feedback character of a projection in a graded metric. We next wanted to correlate this graded anatomical metric with a similarly graded functional metric, which captures the above mentioned GC Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 5 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. asymmetries. Therefore, we defined as in (Bastos et al., 2015) the Directed influence Asymmetry Index (DAI) between areas A and B as DAI(A-to-B) = (GC(A-to-B) – GC(B-to-A)) / (GC(A-to-B) + GC(B-to-A)) The numerator captures the predominant or net direction of GC, while the denominator provides a normalization factor such that the asymmetry in causal interactions can be compared between different area pairs and between different frequencies, for which raw GC values can differ substantially. Note that DAI is defined per frequency of the GC spectrum. To test whether the GC asymmetry is systematically related to the feedforward-feedback character of the anatomical projections, we calculated the Spearman correlation coefficient between DAI values (from human MEG data) and SLN values (from macaque retrograde tracing data) across all visual area pairs. The seven visual areas under consideration give rise to 42 pairs, of which 38 had anatomical data that reached threshold (i.e. more than 10 labeled neurons). The DAI-SLN correlation was determined per GC frequency and per subject. Figure 6 shows the mean and SEM across the 43 subjects. The results for the left (Figure 6A) and right (Figure 6B) hemisphere were qualitatively identical. Also, the results remained qualitatively unchanged when the analysis was repeated without alignment to the subject-wise alpha-beta- and gamma-band peak frequencies (Figure S2). Feedforward (feedback) anatomical projections predominantly originate in supragranular (infragranular) layers and are therefore characterized by a large (small) SLN (supragranular labeled neuron proportion), typically above (below) 0.5. Correspondingly, when the correlation is calculated between SLN values and DAI values, a positive DAI-SLN correlation for a given frequency range indicates that this frequency is systematically stronger (weaker) in the direction of feedforward (feedback) projections. Thus, the significant positive DAI-SLN correlation for the gamma band indicates that the net influence in this frequency is feedforward (left (right) hemisphere: mean=0.465 (0.467), p=0.0019 (0.0019)). The significant negative correlation for the alpha-beta band indicates that the net influence is feedback (left (right) hemisphere: mean=-0.226 (-0.178), p=0.0159 (0.0139)). A Granger-Causality-based hierarchy correlates with the anatomical hierarchy SLN values have been used to establish an anatomical hierarchy of primate visual areas (Markov et al., 2014). In this hierarchy, the feedforward-feedback character of most inter-areal projections is in conformity with a global hierarchy, in which each area occupies a particular hierarchical level. For example, if the A-to-B projection is feedforward and the B-to-C projection is feedforward, then the Ato-C projection is typically strongly feedforward. Note there is no a-priori reason that prevents the Ato-C projection from being feedback, because neurons in A projecting to B and C are typically almost completely non-overlapping (Salin and Bullier, 1995), and neurons are exclusively either feedback or feedforward (Markov et al., 2014). The fact that inter-areal projections tend to confirm to a global hierarchy is an empirical finding in the large set of available SLN data (Markov et al., 2014). Given that the macaque SLN is correlated to the human DAI, we asked whether the human DAI also establishes a hierarchy of human visual areas, similar to the DAI-based hierarchy of macaque visual areas established previously (Bastos et al., 2015). Within each subject, we considered each area in turn as a seed area. The DAI values of this seed area relative to all areas were ranked (with the DAI of the seed to itself being zero, according to the DAI definition). Average ranking across all seven seed areas was established for each subject. Figure 7 shows average rankings and their SEMs across the 43 subjects. Figure 7A shows the rankings derived from the DAI in the gamma-frequency band. The human GC-based ranking is plotted against the macaque anatomy-based ranking (Markov et al., 6 Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. 2014) and the two hierarchies are significantly correlated (left hemisphere: R=0.962, p=0.0005; right hemisphere: R=0.945, p=0.0013; Pearson correlations in this and the following tests). Note, an essentially perfect correlation existed for all areas, except TEO, which ranked slightly higher in the human GC-based than the macaque anatomy-based hierarchy. Note also that hierarchical levels span almost the entire possible range of 7 levels, suggesting slightly less than 7 levels as a first estimate of a lower bound for the number of hierarchical levels present in the data (see below for a higher estimate, when more areas are included). The same procedure was repeated with the DAI in the alpha-beta frequency band (after inverting its sign because of its negative correlation to SLN). The resulting relation to anatomy was less pronounced, because early visual areas V1 and V2 were placed higher and the ventral-stream areas V4 and TEO were placed lower than in the anatomical hierarchy (Figure 7B) (left hemisphere: R=0.75, p=0.032; right hemisphere: R=0.812, p=0.0267). A hierarchy based on average DAI values from both the gamma and the alpha-beta band (after inverting the latter) shows excellent correspondence between functional and anatomical hierarchies (Figure 7C) (left hemisphere: R=0.901, p=0.0057; right hemisphere: R=0.923, p=0.003). In comparison to the pure gamma-based hierarchy, the ventralstream areas V4 and TEO are moved lower. This might be due to strong feedback influences exerted in the alpha-beta band (see Discussion). Functional hierarchies built separately from the GC values in the two hemispheres were qualitatively identical (compare the two columns of Figure 7). Further, the results remained qualitatively unchanged when the analysis was repeated without alignment to the subject-wise peak frequencies in the alpha-beta and gamma band (Figure S3). When the analysis was repeated for areas randomly defined as contiguous patches of dipoles (with sizes similar to the actual areas), this produced correlations distributed around zero. All observed correlations exceeded the 97.5th percentile of these randomization distributions, i.e. they were significantly different in a two-sided test. As an independent test for the presence of functional hierarchies, we computed repeated measures one-way ANOVAs across subjects with the factor AREA and the dependent variable HIERARCHICAL LEVEL. The observed F-values were compared to the distribution of F-values from 1000 trial shuffles. All hierarchies were highly significant (p-values of 0 to 0.003). Granger Causality establishes a functional hierarchy across 26 human visual areas The analysis so far was limited to 7 areas, for which the homology between human and macaque is well pronounced and for which quantitative retrograde tracing data were available. Given that the anatomical hierarchy was strongly correlated with a GC-based hierarchy, we next investigated whether a hierarchy can also be built for a larger set of human visual areas: V1, V2, V3, V3A, V3B, V3C, V3D, V4, V4t, LO1, LO2, PITd, PITv, MT, VO1, MST, FST, ER, V7, IPS1, IPS2, IPS3, IPS4, 7A, FEF, 46 (see Experimental procedures for description). For these 26 areas, the functional hierarchical level was derived based on the DAI metric, in the same fashion as for the 7 areas. Figure 8A shows the 26 areas according to their functional hierarchical level. If the hierarchical ranking had not been consistent across seed areas within subjects, or across subjects, the different areas would rank roughly at the same hierarchical level. Similarly, if ranking were not consistent across subjects, the error bars would be large relative to the rank range. Thus, the fact that in Figure 8, the 26 areas span almost the entire dynamic range of 26 hierarchical levels, with relatively small error bars, suggests slightly less than 26 levels as a lower bound for the number of hierarchical levels present in the data. It also suggests that the organization of most area-pair-wise GC influences corresponds to a global hierarchy. In addition, the resulting hierarchy is in good agreement with the mosaic of evidence from Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 7 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. different sources including quantitative anatomy (Barone et al., 2000; Markov et al., 2014), general anatomical connectivity and position (Felleman and Essen, 1991) and cognitive architectures (Dehaene, 2005). The location of the frontal eye field (FEF) low in the hierarchy is in agreement with recent quantitative analysis of retrograde tracing (Barone et al., 2000; Markov et al., 2014). In terms of hierarchical discriminability of areas, it is clear that areas are more separable at the upper and lower ends of the hierarchy, than they are in the middle of the hierarchy. The same procedure repeated in the alpha-beta band led to several areas changing their hierarchical level. To illustrate this, Figure 8B retains the x-axis ordering established in Figure 8A on the basis of the gamma-band DAI, but reports on the y-axis the hierarchical level based on the alpha-beta band DAI (a copy of each area’s gamma-DAI based hierarchical level is shown as a gray dot). In general, ventral-stream areas (labeled green) V4, PITv, PITd and VO1 are moved to lower levels, whereas many dorsal-stream areas (labeled red) including V3C, V3D, MST, FST, IPS3, IPS4 and 7A are shifted to higher levels. Similarly V1 and V2 are moved to higher levels. These shifts could be due to relatively strong alpha-beta-band influences exerted by the dorsal-stream areas onto the ventral stream areas, and exempting the earliest areas V1 and V2 (see Discussion). Likewise, FEF is shifted to a higher level, suggesting that it exerts relatively strong alpha-beta feedback. The split between dorsal- and ventral-stream areas becomes even more apparent, when the functional hierarchy is based on the DAIs from the gamma and alpha-beta bands combined (Figure 8C). Compared to a pure gamma-based hierarchy (gray dots), it shows a consistent upward shift for most dorsal-stream areas (V3C, V3D, MST, FST, V7, IPS1 to 4) and a consistent downward shift for ventral-stream areas (V4, PITv, PITd, VO1). We repeated the ANOVA analysis for the 26 areas, and again all hierarchies were significant (p-values ranging from 0 to 0.016). Discussion We used MEG to study the spectral specificity of causal influences along inferred anatomical feedforward and feedback projections in the human visual cortex. Projections were defined as feedforward or feedback based on retrograde tracing data in homologous macaque visual areas. Frequency-resolved Granger causality between human visual cortical areas showed that causal influences along feedforward projections were predominant in the gamma band, while causal interactions along feedback projections were predominant in the alpha-beta band. Based on these results, we used all inter-areal GC values to build a functional hierarchy in human visual cortex, which was strongly correlated to the hierarchy derived from macaque retrograde tracing data. Finally, we extrapolated this to 26 human visual areas. The present analyses are based on MEG data from 43 human subjects, which allowed us to use statistics that are non-parametric counterparts of random-effect statistics. The previous analysis using ECoG in macaques was restricted to two animals, as is typical for non-human primate studies, and thereby to a fixed-effect analysis. Whereas a fixed-effect analysis allows inference only on the tested sample, our present analysis allows generalization to the population of human subjects. Fast versus slow oscillations for feedforward and feedback signaling Several studies have reported that feedforward and feedback communication use respectively higher and lower frequency ranges. Some studies compared task conditions and found that conditions expected to emphasize feedforward signaling lead to stronger synchronization in relatively higher 8 Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. frequency bands, whereas conditions expected to emphasize feedback signaling lead to stronger synchronization in relatively lower frequency bands (Arnal et al., 2011; Buschman and Miller, 2007; von Stein et al., 2000). Other studies compared anatomical structures and found that pathways or layers involved in feedforward signaling show stronger synchronization at higher frequencies, and pathways or layers involved in feedback signaling show stronger synchronization in lower frequencies. For example, feedforward communication from medial entorhinal cortex to hippocampal area CA1 uses higher gamma, whereas feedback communication from CA3 to CA1 uses lower gamma (Colgin et al., 2009; Schomburg et al., 2014). In visual cortex, superficial layers, sending feedforward projections, show predominantly gamma-band synchronization, whereas infragranular layers, sending feedback projections, show predominantly alpha-beta-band synchronization (Buffalo et al., 2011; Xing et al., 2012). Across cortical sublayers, gamma shows phase shifts consistent with feedforward signaling (Livingstone, 1996; van Kerkoerle et al., 2014), and alpha shows phase shifts or GC consistent with feedback signaling (Bollimunta et al., 2011; van Kerkoerle et al., 2014). Direct quantification of directed influences by means of GC showed that inter-areal feedforward influences are carried by gamma and feedback influences by alpha or beta rhythms. Between human auditory area A1 and auditory association cortex, feedforward signaling uses gamma, and feedback signaling delta/beta rhythms (Fontolan et al., 2014). Between macaque visual areas V1 and V4, feedforward signaling uses gamma and feedback signaling alpha or beta (Bosman et al., 2012; van Kerkoerle et al., 2014). Finally, a recent study investigated eight visual areas in macaque and found that across the numerous inter-areal projections, feedforward connectivity was systematically related to GC in the theta and gamma bands, and feedback connectivity systematically to GC in the beta band (Bastos et al., 2015). These earlier results from macaques agree fully with the present results for humans in the gamma band, but they agree only partly in the other frequency bands. Our present analysis in humans suggests that top-down influences are exerted by rhythms spanning alpha and beta frequency ranges, even though previous studies clearly suggest separate alpha and beta rhythms (Gregoriou et al., 2015; Wang, 2010; Womelsdorf et al., 2014). It is important to note that our analysis merely reveals whether directed influences in a frequency band relate to feedforward or feedback projections. Our results suggest that both alpha and beta subserve top-down signaling. Yet, this does not contradict differential mechanisms and sub-functions of top-down influences (Fries, 2015). Alpha might mediate top-down influences suppressing irrelevant background (van Kerkoerle et al., 2014), whereas beta might mediate top-down influences, that facilitate the bottom-up communication of attended stimuli (Bastos et al., 2015; Lee et al., 2013). Our present analysis did not reveal a convincing peak of GC influences in the theta-frequency range, even though theta-band influences have been found in the earlier study with macaques (Bastos et al., 2015). The reason for this difference is currently unclear, and we can only speculate that it might be due to differences between ECoG and MEG, or between the stimulus and/or task conditions. Compared to infragranular connectivity, supragranular connectivity shows fine-grained spatial structure and interconnects some brain regions, like e.g. frontal cortex and ventral visual areas (Markov et al., 2014), possibly allowing frontal areas to change their functional hierarchical level dynamically (Bastos et al., 2015). In agreement with this, gamma-band GC influences between macaque FEF and V4 predominate in the FEF-to-V4 direction after an attentional cue, but subsequently predominate in the V4-to-FEF direction (Gregoriou et al., 2009). Between human frontal cortex and high-level ventral visual cortex, gamma-band coherence shows a 20 ms phase lead of frontal cortex (Baldauf and Desimone, 2014). Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 9 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. The present results, indicating a functional hierarchy of the human visual cortex, are relevant to future investigation of predictive coding in human cortex. Feedforward and feedback influences have been hypothesized to subserve the signaling of prediction errors and predictions, respectively. These concepts are central to the theory of predictive coding, stating in essence that higher areas constantly derive predictions based on incoming evidence and prior experience, and feed those predictions back to lower areas, where they are subtracted from sensory evidence, such that only prediction errors need to be forwarded to higher areas (Bastos et al., 2012). Crucially, in this framework, the updating of predictions entails the accumulation of prediction errors over time, so that predictions change more slowly than prediction errors. Therefore, if predictions and prediction errors are mediated through inter-areal rhythmic synchronization, the communication of prediction errors requires higher frequencies than the communication of predictions. This is in agreement with several recent studies (Bastos et al., 2015; Fontolan et al., 2014; van Kerkoerle et al., 2014) and our finding that feedforward signaling is dominated by gamma and feedback signaling is dominated by alpha-beta rhythms. Ventral stream areas While the functional human hierarchy derived from the DAI was largely in agreement with the macaque anatomical hierarchy, some important deviations were observed, especially when the hierarchy was computed from the feedback causal asymmetries in the alpha-beta band. In this case, the ventral stream areas V4 and TEO were much lower in the functional hierarchy than in the anatomical one. Feedback causal influences were strongest onto area V4 and progressively decreased towards the primary visual cortex. A similarly decreasing pattern has been observed for the effect of attention on firing rates in areas V1, V2 and V4 (Buffalo et al., 2010). In this work, a reverse order of attentional effects was shown, with a stronger and earlier enhancement in firing rates in V4, and progressively smaller and later enhancement in V2 and in V1. This strong feedback influence on V4 was also prominent in the hierarchy derived from the combined feedforward and feedback causal asymmetries as manifested in the unexpectedly low hierarchical position of V4 (at a similar level to V2). These results suggest that feedback influences from higher ventral and dorsal areas primarily address mid-level ventral stream areas such as V4 and TEO, and to a lesser degree affect the earlier visual areas V2 and V1. It has already been suggested that early ventral stream areas and especially area V4 is well positioned for integrating top-down influences with bottom-up information (Roe et al., 2012), mainly due to strong corticocortical (Cloutman, 2013; Markov et al., 2014) and corticothalamocortical (Van Essen, 2005) anatomical connections with temporal, prefrontal and parietal regions. Also, details of the task may differentially impact the patterns of alpha-beta band top-down influences. Our task required subjects to detect a speed change in a drifting grating and may therefore predominantly engage the dorsal stream. It is conceivable that this contributed to strong alpha-beta band influences from dorsal to ventral stream areas, which in turn places dorsal stream areas higher and ventral stream areas lower in the functional hierarchy. Future studies might systematically investigate changes in the functional hierarchy as a function of task requirements. Functional hierarchy of 26 visual areas Using the asymmetry in causal interactions, we estimated the functional hierarchy for 26 areas of the human visual system. For several of the 26 areas, it is not possible to establish a clear homology to Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 10 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. areas in the non-human primate brain. Yet, for the areas, for which a homology can be hypothesized, there was a good agreement between functional and anatomical hierarchical level (Felleman and Essen, 1991; Markov et al., 2014). Furthermore, for areas defined only in humans, the functional hierarchical level agreed with expectations derived from overall position in the brain and functional characteristics. The strong effect of feedback causal influences on mid-level ventral stream areas was observed also in this hierarchy. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 11 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Experimental procedures Experimental paradigm The experimental paradigm (Figure 1A), was adopted from (van Pelt et al., 2012) and (Hoogenboom et al., 2006) and is known to induce strong gamma band oscillations in occipital areas. Each trial was initiated with a fixation point (Gaussian of diameter 0.5 deg) at the center of the screen for 1.5 s. The fixation period was followed by the presentation of a circular sine-wave grating, centered at the fixation point and contracting towards its center (spatial frequency: 3 cycles/deg, velocity: 0.66 deg/s, contrast: 100%). The screen background was black. At a random moment between 0.75 and 3 s after stimulus onset, velocity increased (step to 1 deg/s), which was reported by the subjects with their right index finger. After the response followed a rest period of 1 s during which feedback was given. A total of 180 trials were presented to each subject. Visual presentation was performed on a translucent screen, situated 0.9 m in front of subjects’ eyes by using an LCD projector with a 60 Hz update rate. We quantified the MEG signal components locked to the LCD update for sensors over visual cortex (Figure 1C), and found them to account for a very small part of the total power at 60 Hz (<1.94% of 60 Hz power during baseline, <1.42% of 60 Hz power during stimulation). Note that data were acquired in Europe, i.e. the power line frequency was 50 Hz. Data Acquisition MEG: Recordings were performed with a whole-head MEG system (CTF Systems Inc.) comprising 275 axial gradiometers. MRI: Structural MR images were acquired using an Avanto 1.5 T whole body MRI scanner (Siemens, Germany). Further details can be found in the corresponding section of Supplemental Experimental Procedures. Subjects Datasets of 43 subjects were selected from a pool of 160 subjects, which had been recorded for genotyping of parameters derived from visually induced gamma-band activity. Further details on selection criteria can be found in the corresponding section of Supplemental Experimental Procedures. All subjects gave written informed consent according to guidelines of the local ethics committee (Commissie Mensgebonden Onderzoek Regio Arnhem-Nijmegen, The Netherlands). Data preprocessing and spectral estimation All MEG data analysis was performed in MATLAB, partly using the FieldTrip toolbox (Oostenveld et al., 2011). Data segments which contained artifacts originating from jumps in the SQUIDs (superconducting quantum interference devices), muscle activity and eyes-movements were identified and removed using semi-automated FieldTrip routines. Power line artifacts were estimated and removed using a discrete Fourier Transform at 50, 100 and 150 Hz with a bin width of ±0.25 Hz. We used data from 0.365 s after stimulus onset until the visual stimulus changed speed, thereby excluding stimulus-onset or stimulus–change related non-stationarities. These data were divided into nonoverlapping 0.365 s epochs. This fixed epoch length provided a good compromise between optimizing the use of the variable-length trials and maximizing the spectral resolution for the nonparametric spectral factorization employed in the calculation of Granger causality. Epochs taken from early or late in the trial (median split) did not differ in their GC and were therefore pooled. Spectral estimation was first performed at the sensor level. Per sensor, each data epoch was demeaned, Hann-tapered, zero-padded to 1 s length and Fourier transformed to give the spectrum between 1 and 140 Hz (Percival and Walden, 2000). Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 12 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Human Brain Model For source analysis, structural MR images were acquired for each individual subject, and the cortical mantle was extracted with the FreeSurfer longitudinal stream pipeline (Reuter et al., 2012). Cortical parcellation and area definition used a recent visuotopic atlas provided by (Abdollahi et al., 2014), the FreeSurfer software suite (http://FreeSurfer.net/) (Desikan et al., 2006; Destrieux et al., 2010; Fischl et al., 2004) and the Caret software suite (http://brainvis.wustl.edu/) (Glasser and Van Essen, 2011; Van Essen et al., 2012). Further details on the construction of the source space with the investigated visual areas can be found in the corresponding section of Supplemental Experimental Procedures. Selection of homologous brain areas between human and macaque We aimed at relating functional connectivity derived from human MEG recordings to anatomical connectivity derived from macaque retrograde tracing experiments. Therefore, the availability of retrograde tract tracing data dictated the selection of visual areas. From the set of areas in the occipital, parietal and temporal cortical regions with such tracing data, a subset was selected for which evidence of homology between human and macaque brains exists in the literature. These areas were V1, V2, MT, V4, DP, TEO and 7A and are shown in Figure 2B. The numbers of dipoles per area are listed in Supplemental Table 1. More details about the selected areas and their homology between species can be found in the corresponding section of Supplemental Experimental Procedures. Parcels representing areas V1 and V2 are quite extensive in comparison to the other parcels and contain a large number of voxels that are far from the main locus of activation, as defined by visually induced gamma power enhancement (Figure 2). In order to minimize the effect of the resulting extensive smoothing, a more spatially confined representation for these areas has been employed. In this representation, V1 and V2 parcels consist of the vertices within 2 cm from the local gamma power maxima within the original representation of these areas. Inverse solution Subject-wise structural MRI scans were used to derive individual single-shell volume conductor models (Nolte, 2003). The inverse solution was then performed using Linearly Constrained Minimum Variance (LCMV) adaptive spatial filtering, often referred to as “Beamforming”. When determining the LCMV solution, we used a regularization parameter of 20% of the covariance matrix. For further details on the inverse solution please see the corresponding section in Supplemental Experimental Procedures. Granger Causality (GC) In order to study the strength and directionality of causal influences between the investigated brain areas, we employed GC (Granger, 1969), a statistical measure that quantifies the extent to which one time series can predict another. The principle idea behind GC is that if the addition of the history of signal A improves the prediction of signal B, as compared to the prediction of signal B based on its own history alone, then signal A is said to ‘”Granger-cause” signal B. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 13 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. This measure of predictive causality can be estimated either in the time domain or in the frequency domain (Geweke, 1982). In the latter case GC provides a spectrum of causal interactions between the two signals as a function of frequency. This is of great utility for the analysis of MEG data, because oscillations at distinct frequencies can be simultaneously realized in different brain areas and likely subserve inter-areal communication (Fries, 2005; Fries, 2015). GC can be calculated using parametric methods, based on multivariate autoregressive models. This approach requires the selection of model parameters and can fail to capture complex spectral features (Mitra and Pesaran, 1999). Therefore, we used a non-parametric approach, which estimates GC directly from a factorization of the cross-spectral density (CSD) matrix (Dhamala et al., 2008). For the present study, the CSD of two brain sources is simply the scalar product between their complex and complex-conjugate spectral coefficient sequences. These coefficients are computed by multiplying the sensor-level spectral decomposition with the spatial filter for each location. The visuotopic source space used in this work consists of the set of points distributed within the brain areas of interest, referred to as “parcels”. To calculate GC between two parcels, we first averaged the CSD between all inter-parcel dipole pairs. Within parcels, the power spectral density was computed by averaging all the within-parcel power-spectral density values. Retrograde tracing database Description of the anatomical dataset acquisition and analysis has been reported in (Markov et al., 2014). Please see also Supplemental Experimental Procedures for additional information. Selection of 26 visual areas from the human brain For this analysis, we extended the initial set of 7 visual areas, for which anatomical connectivity information was available, to an extended set of 26 areas involved in visual processing. The additional areas were V3, V3A, V3B, V3C, V3D, V4t, LO1, LO2, VO1, MST, FST from the atlas of (Abdollahi et al., 2014), and areas IPS1, IPS2, IPS3, IPS4, 46, ER from the visuotopic atlas of Caret (Van Essen et al., 2012). Area PIT from the 7-area analysis was split into its two constituent areas PITv, PITd from the atlas of (Abdollahi et al., 2014). Frontal eye fields (FEF) were manually selected. For further details on the 26 selected visual areas, see Supplemental Experimental Procedures. Statistical testing Figures 4 and 5 use randomization-based non-parametric statistical tests with cluster-based multiple comparison correction (Maris and Oostenveld, 2007). Each subject provided one feedforward and one feedback GC spectrum, which were compared across subjects via t-tests. The t-values of spectrally adjacent, significant t-tests were summed to generate the observed (first-level, clusterbased) test metric (not used for inference). We then performed 1000 randomizations. In each randomization, feedforward and feedback labels were randomly exchanged per subject, and the test metric was computed as for the non-randomized data. Per randomization, the largest test metric entered into the randomization distribution. The observed test metrics were compared against the randomization distribution. Comparisons between feedforward or feedback GC versus the bias estimates proceeded accordingly. Similarly, the correlation coefficients between DAI and SLN values (computed as explained in the Results section) were tested against zero (Figure 6) and against bias estimates (Figure S2), i.e. each subject provided one correlation spectrum and either one null spectrum or one bias-estimate spectrum. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 14 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Figure legends Figure 1. Experimental paradigm and visually induced changes in MEG signal power in frequency, space and time (A) Experimental paradigm. The fixation period was followed by the presentation of a circular sinewave grating, centered at the fixation point and contracting towards its center. At a random moment between 0.75 and 3 s after stimulus onset, velocity increased, which was reported by the subjects with their right index finger. Thicker red arrows indicate the speed change. (B) Spectrum of power change during stimulation versus baseline (-0.5 to -0.2 sec). Each line represents the average across subjects for one MEG sensor. (C) Sensor-level topography of power change from (B) in the frequency range of 40 to 75 Hz. (D) Average power change as a function of time and frequency for selected sensors over occipito-parietal areas, shown with * in (C). Color bar applies to (C) and (D). Figure 2. Gamma power distribution on the cortical sheet and human-macaque homology definition (A) Gamma-power change distribution on the cortical sheet as derived from inverse solution. (B) Selected human-macaque homologous visual areas. Displayed on flat cortical maps. The small black spheres on the macaque flat cortex indicate the injection sites of retrograde tracer published in (Markov et al., 2014). (C) Vertices representing the human homologous areas from (B) in the MEG inverse solution on the very inflated Conte69 template brain. Figure 3. Granger Causality limits the effect of field spread For this investigation, each hemisphere was randomly segmented into 400 parcels according to the Fast Marching algorithm (Sethian, 1999). There were on average 10 vertices per parcel. The parcel closest to the peak of the average gamma power over all subjects was identified and was selected as the seed parcel, marked here in black. (A, B) Coherence was computed for each subject between this seed parcel and the remaining 399 parcels. These coherence values were subsequently averaged across all subjects for each pair and each frequency. Finally for each pair, the coherence values within the gamma band (40 to 75 Hz) were averaged. (C, D) GC from the seed to the rest of the parcels for the same frequency range. Figure 4. Granger Causality in feedforward and feedback directions (A) GC spectrum averaged across all pairs of areas. GC values were averaged across all area pairs and subjects, separately per hemisphere. Significance relative to the surrogate (random) data was computed by comparing for all subjects the GC between the original and the surrogate conditions using a permutation-based non-parametric statistical test. (B, C) GC spectra averaged separately for inter-areal influences corresponding to feedforward (green) or feedback (black) projections. Two non-parametric statistical tests were performed. One between the actual and surrogate average GC spectra and one between the feedforward and feedback average GC spectra. This analysis was performed separately for the left (B) and right (C) hemisphere. (D) GC spectra, averaged over subjects, separately per area pair and hemisphere. Again two non-parametric statistical tests were performed for each area pair: One between the actual and surrogate data and one between the feedforward and feedback direction. Tests were corrected for multiple comparisons across Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 15 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. frequencies and across area pairs within each hemisphere. Left (right) hemisphere data are shown in lower (upper) triangle. The two arrows above each subplot signify the feedforward (green arrow) and feedback (black arrow) characteristic of the anatomical projections, as defined by the anatomical SLN values (listed above and below the arrows). GC spectra in the feedforward direction are shown in green, GC spectra in the feedback direction in black. See information at the bottom of the plot for details on each subplot. Figure 5. Granger Causality in feedforward and feedback directions after peak alignment In order to account for inter-subject variability, two new frequency axes were devised, aligned to the individual alpha-beta and gamma peaks. The alpha-beta frequency axis spanned from -5 to +5 Hz around the alpha-beta peak. The gamma frequency axis spanned from -25 to +70 around the gamma peak. There was no overlap between these two frequency ranges for any subject. Otherwise, analyses and figure format is as in Figure 4. Figure 6. DAI-SLN correlation spectra reveal systematic relation between GC and anatomy Spearman correlation between DAI values (from human MEG) and SLN values (from macaque retrograde tracing) across pairs of visual areas. The DAI-SLN correlation was determined per frequency of the GC spectrum. Significance relative to zero was computed using a permutation-based non-parametric statistical test against the null hypothesis of zero correlation, with cluster-based multiple comparisons correction. This analysis was performed separately for the left (A) and right (B) hemisphere. In both cases, a frequency cluster of negative correlation was identified in the alphabeta range and a cluster of positive correlation was identified in the gamma range. Figure 7. Granger-Causality-based hierarchy correlates with anatomical hierarchy The relation between the functional hierarchy of human visual areas and the anatomical hierarchy of homologous macaque areas was determined separately for left and right hemispheres. The functional hierarchy of the human visual areas was computed from the DAI values (see Experimental procedures). Three cases were investigated, namely with the functional hierarchy computed from (A) DAI in the gamma band, (B) DAI in the alpha-beta band and (C) the combined DAI in alpha-beta and gamma bands. Early visual areas are depicted in black color, ventral-stream areas in green and dorsal-stream areas in red. Inset brackets on bottom right of each subplot report Pearson correlation between human functional and macaque anatomical hierarchical levels. Notice the push towards lower hierarchical levels of midlevel ventral stream areas by the inclusion of feedback causal influences in the alpha-beta band (compare (C) to (A)). Figure 8. Functional hierarchy of 26 human visual areas The functional hierarchy of 26 human visual areas was computed based on (A) DAI in the gamma band, (B) DAI in the alpha-beta band and (C) the combined DAI in alpha-beta and gamma bands. The functional hierarchy in each case is the average of the functional hierarchies in the left and right hemisphere. Ventral-stream areas are depicted in green, dorsal-stream areas in red and all other areas in black. In (A) the 26 areas have been ranked along the x-axis based on their average hierarchical level depicted on the y-axis. In (B) and (C), the ranking from (A) has been preserved on the x-axis in order to illustrate changes in the hierarchy for DAIs of different frequency bands. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 16 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Author contributions Conceptualization, G.M., J.V., and P.F.; Methodology, G.M., J.V. and J.S.; Formal Analysis, G.M., and J.V.; Investigation, S.P., and H.K.; Resources, P.F.; Writing – Original Draft, G.M., and P.F.; Writing – Review & Editing, G.M., J.V., S.P., J.S., H.K., and P.F.; Visualization, G.M.; Supervision, P.F.; Project Administration, P.F.; Funding Acquisition, P.F. Acknowledgements We thank Marieke van de Nieuwenhuijzen for assisting in the recording of MEG datasets and Eric Maris for help with non-parametric statistics. This work was supported by the Human Connectome Project (WU-Minn Consortium, NIH grant 1U54MH091657 to P.F.), a European Young Investigator Award (P.F.), the European Union (HEALTH-F2-2008-200728 to P.F.), the LOEWE program (“Neuronale Koordination Forschungsschwerpunkt Frankfurt” to P.F.), NWO (VIDI grant to J.S.), the Smart Mix Program of the Netherlands Ministry of Economic Affairs and the Netherlands Ministry of Education, Culture and Science (BrainGain to P.F. and J.S.), ANR-11-BSV4-501 (H.K.), and LabEx CORTEX (ANR-11-LABX-0042 to H.K.). Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 17 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. References Abdollahi, R.O., Kolster, H., Glasser, M.F., Robinson, E.C., Coalson, T.S., Dierker, D., Jenkinson, M., Van Essen, D.C., and Orban, G.A. (2014). Correspondences between retinotopic areas and myelin maps in human visual cortex. Neuroimage 99, 509-524. Arnal, L.H., Wyart, V., and Giraud, A.-L. (2011). Transitions in neural oscillations reflect prediction errors generated in audiovisual speech. Nat Neurosci 14, 797-801. Baldauf, D., and Desimone, R. (2014). Neural mechanisms of object-based attention. Science 344, 424-427. Barone, P., Batardiere, A., Knoblauch, K., and Kennedy, H. (2000). Laminar distribution of neurons in extrastriate areas projecting to visual areas V1 and V4 correlates with the hierarchical rank and indicates the operation of a distance rule. The Journal of neuroscience : the official journal of the Society for Neuroscience 20, 3263-3281. Bastos, A.M., Usrey, W.M., Adams, R.A., Mangun, G.R., Fries, P., and Friston, K.J. (2012). Canonical microcircuits for predictive coding. Neuron 76, 695-711. Bastos, A.M., Vezoli, J., Bosman, C.A., Schoffelen, J.-M., Oostenveld, R., Dowdall, J.R., De Weerd, P., Kennedy, H., and Fries, P. (2015). Visual areas exert feedforward and feedback influences through distinct frequency channels. Neuron 85, 390-401. Bollimunta, A., Chen, Y., Schroeder, C.E., and Ding, M. (2008). Neuronal mechanisms of cortical alpha oscillations in awake-behaving macaques. J Neurosci 28, 9976-9988. Bollimunta, A., Mo, J., Schroeder, C.E., and Ding, M. (2011). Neuronal mechanisms and attentional modulation of corticothalamic alpha oscillations. The Journal of Neuroscience 31, 4935-4943. Bosman, C.A., Schoffelen, J.-M., Brunet, N., Oostenveld, R., Bastos, A.M., Womelsdorf, T., Rubehn, B., Stieglitz, T., De Weerd, P., and Fries, P. (2012). Attentional stimulus selection through selective synchronization between monkey visual areas. Neuron 75, 875-888. Bressler, S.L., and Richter, C.G. (2014). Interareal oscillatory synchronization in top-down neocortical processing. Curr Opin Neurobiol 31C, 62-66. Buffalo, E.A., Fries, P., Landman, R., Buschman, T.J., and Desimone, R. (2011). Laminar differences in gamma and alpha coherence in the ventral stream. Proceedings of the National Academy of Sciences 108, 11262-11267. Buffalo, E.A., Fries, P., Landman, R., Liang, H., and Desimone, R. (2010). A backward progression of attentional effects in the ventral stream. Proc Natl Acad Sci U S A 107, 361-365. Buschman, T.J., and Miller, E.K. (2007). Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices. Science 315, 1860-1862. Cloutman, L.L. (2013). Interaction between dorsal and ventral processing streams: where, when and how? Brain Lang 127, 251-263. Colgin, L.L., Denninger, T., Fyhn, M., Hafting, T., Bonnevie, T., Jensen, O., Moser, M.B., and Moser, E.I. (2009). Frequency of gamma oscillations routes flow of information in the hippocampus. Nature 462, 353-357. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 18 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Dehaene, S., Duhamel, J.R., Hauser, M.D., and Rizzolatti, G. (2005). From monkey brain to human brain: A Fyssen foundation symposium (Cambridge, MA: MIT Press). Desikan, R.S., Ségonne, F., Fischl, B., Quinn, B.T., Dickerson, B.C., Blacker, D., Buckner, R.L., Dale, A.M., Maguire, R.P., Hyman, B.T., et al. (2006). An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage 31, 968-980. Destrieux, C., Fischl, B., Dale, A., and Halgren, E. (2010). Automatic parcellation of human cortical gyri and sulci using standard anatomical nomenclature. Neuroimage 53, 1-15. Dhamala, M., Rangarajan, G., and Ding, M. (2008). Estimating Granger causality from Fourier and wavelet transforms of time series data. Physical review letters 100, 018701-018701. Felleman, D.J., and Essen, D.C.V. (1991). Distributed hierarchical processing in the primate cerebral cortex. Cereb Cortex 1, 1-47. Fischl, B., van der Kouwe, A., Destrieux, C., Halgren, E., Ségonne, F., Salat, D.H., Busa, E., Seidman, L.J., Goldstein, J., Kennedy, D., et al. (2004). Automatically parcellating the human cerebral cortex. Cereb Cortex 14, 11-22. Fontolan, L., Morillon, B., Liegeois-Chauvel, C., and Giraud, A.-L. (2014). The contribution of frequency-specific activity to hierarchical information processing in the human auditory cortex. Nat Commun 5, 4694-4694. Fries, P. (2005). A mechanism for cognitive dynamics: neuronal communication through neuronal coherence. Trends Cogn Sci 9, 474-480. Fries, P. (2015). Rhythms for Cognition: Communication through Coherence. Neuron 88, 220-235. Geweke, J. (1982). Measurement of linear dependence and feedback between multiple time series. Journal of the American Statistical Association 77, 304-313. Glasser, M.F., and Van Essen, D.C. (2011). Mapping human cortical areas in vivo based on myelin content as revealed by T1- and T2-weighted MRI. J Neurosci 31, 11597-11616. Granger, C.W.J. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica: Journal of the Econometric Society, 424-438. Gregoriou, G.G., Gotts, S.J., Zhou, H., and Desimone, R. (2009). High-frequency, long-range coupling between prefrontal and visual cortex during attention. Science 324, 1207-1210. Gregoriou, G.G., Paneri, S., and Sapountzis, P. (2015). Oscillatory synchrony as a mechanism of attentional processing. Brain Res. Haegens, S., Cousijn, H., Wallis, G., Harrison, P.J., and Nobre, A.C. (2014). Inter- and intra-individual variability in alpha peak frequency. Neuroimage 92, 46-55. Hoogenboom, N., Schoffelen, J.-M., Oostenveld, R., Parkes, L.M., and Fries, P. (2006). Localizing human visual gamma-band activity in frequency, time and space. Neuroimage 29, 764-773. Jutai, J.W., Gruzelier, J.H., and Connolly, J.F. (1984). Spectral analysis of the visual evoked potential (VEP): effects of stimulus luminance. Psychophysiology 21, 665-672. Lee, J.H., Whittington, M.A., and Kopell, N.J. (2013). Top-down beta rhythms support selective attention via interlaminar interaction: a model. PLoS Comput Biol 9, e1003164-e1003164. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 19 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Livingstone, M.S. (1996). Oscillatory firing and interneuronal correlations in squirrel monkey striate cortex. Journal of neurophysiology 75, 2467-2485. Maris, E., and Oostenveld, R. (2007). Nonparametric statistical testing of EEG-and MEG-data. Journal of neuroscience methods 164, 177-190. Markov, N.T., Vezoli, J., Chameau, P., Falchier, A., Quilodran, R., Huissoud, C., Lamy, C., Misery, P., Giroud, P., Ullman, S., et al. (2014). Anatomy of hierarchy: feedforward and feedback pathways in macaque visual cortex. J Comp Neurol 522, 225-259. Mitra, P.P., and Pesaran, B. (1999). Analysis of dynamic brain imaging data. Biophysical journal 76, 691-708. Muckli, L., De Martino, F., Vizioli, L., Petro, L.S., Smith, F.W., Ugurbil, K., Goebel, R., and Yacoub, E. (2015). Contextual Feedback to Superficial Layers of V1. Curr Biol. Nolte, G. (2003). The magnetic lead field theorem in the quasi-static approximation and its use for magnetoencephalography forward calculation in realistic volume conductors. Physics in medicine and biology 48, 3637-3637. Olman, C.A., Harel, N., Feinberg, D.A., He, S., Zhang, P., Ugurbil, K., and Yacoub, E. (2012). Layerspecific fMRI reflects different neuronal computations at different depths in human V1. PloS one 7, e32536. Oostenveld, R., Fries, P., Maris, E., and Schoffelen, J.-M. (2011). FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput Intell Neurosci 2011, 156869-156869. Percival, D.B., and Walden, A.T. (2000). Wavelet Methods for Time Series Analysis (Cambridge Series in Statistical and Probabilistic Mathematics) (Cambridge University Press). Plomp, G., Quairiaux, C., Kiss, J.Z., Astolfi, L., and Michel, C.M. (2014). Dynamic connectivity among cortical layers in local and large-scale sensory processing. Eur J Neurosci 40, 3215-3223. Reuter, M., Schmansky, N.J., Rosas, H.D., and Fischl, B. (2012). Within-subject template estimation for unbiased longitudinal image analysis. Neuroimage 61, 1402-1418. Roe, A.W., Chelazzi, L., Connor, C.E., Conway, B.R., Fujita, I., Gallant, J.L., Lu, H., and Vanduffel, W. (2012). Toward a unified theory of visual area V4. Neuron 74, 12-29. Salin, P.A., and Bullier, J. (1995). Corticocortical connections in the visual system: structure and function. Physiol Rev 75, 107-154. Schoffelen, J.-M., and Gross, J. (2009). Source connectivity analysis with MEG and EEG. Hum Brain Mapp 30, 1857-1865. Schomburg, E.W., Fernández-Ruiz, A., Mizuseki, K., Berényi, A., Anastassiou, C.A., Koch, C., and Buzsáki, G. (2014). Theta phase segregation of input-specific gamma patterns in entorhinalhippocampal networks. Neuron 84, 470-485. Sethian, J.A. (1999). Level set methods and fast marching methods: evolving interfaces in computational geometry, fluid mechanics, computer vision, and materials science, Vol 3 (Cambridge university press). Van Essen, D.C. (2005). Corticocortical and thalamocortical information flow in the primate visual system. Prog Brain Res 149, 173-185. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 20 This is a post-print version of the following article: Alpha-beta and Gamma Rhythms Subserve Feedback and Feedforward Influences among Human Visual Cortical Areas. Michalareas, G., Vezoli, J., van Pelt, S., Schoffelen, J.M., Kennedy, H. and Fries, P. Neuron 89, January 20, 2016. http://dx.doi.org/10.1016/j.neuron.2015.12.018. Van Essen, D.C., Glasser, M.F., Dierker, D.L., Harwell, J., and Coalson, T. (2012). Parcellations and hemispheric asymmetries of human cerebral cortex analyzed on surface-based atlases. Cereb Cortex 22, 2241-2262. van Kerkoerle, T., Self, M.W., Dagnino, B., Gariel-Mathis, M.-A., Poort, J., van der Togt, C., and Roelfsema, P.R. (2014). Alpha and gamma oscillations characterize feedback and feedforward processing in monkey visual cortex. Proc Natl Acad Sci U S A 111, 14332-14341. van Pelt, S., Boomsma, D.I., and Fries, P. (2012). Magnetoencephalography in twins reveals a strong genetic determination of the peak frequency of visually induced gamma-band synchronization. J Neurosci 32, 3388-3392. von Stein, A., Chiang, C., and König, P. (2000). Top-down processing mediated by interareal synchronization. Proceedings of the National Academy of Sciences of the United States of America 97, 14748-14753. Wang, X.-J. (2010). Neurophysiological and computational principles of cortical rhythms in cognition. Physiol Rev 90, 1195-1268. Wang, X., Chen, Y., and Ding, M. (2008). Estimating Granger causality after stimulus onset: a cautionary note. Neuroimage 41, 767-776. Womelsdorf, T., Valiante, T.A., Sahin, N.T., Miller, K.J., and Tiesinga, P. (2014). Dynamic circuit motifs underlying rhythmic gain control, gating and integration. Nat Neurosci 17, 1031-1039. Xing, D., Yeh, C.-I., Burns, S., and Shapley, R.M. (2012). Laminar analysis of visually evoked activity in the primary visual cortex. Proc Natl Acad Sci U S A 109, 13871-13876. Michalareas et al., Feedforward and Feedback Influences Among Human Visual Areas. 21 A Experimental paradigm OK Baseline fixation 1.5 sec B Feedback 0.5 sec time Power change spectrum 140 Power change (%) Change, Visual Stimulation Response 0.75-3 sec in < 0.5 sec 0 -40 C 10 30 50 70 90 Frequency(Hz) 110 130 Gamma-power change topography 80 0 -40 Power change (%) 40 -80 D Time-Frequency analysis Frequency(Hz) 90 70 50 30 10 -0.2 0 0.2 stimulus onset 0.4 0.6 0.8 1 Time(sec) Figure 1 A Gamma-power change topography 80 0 -40 Power change (%) 40 -80 B Selected homologous areas Macaque Human injection sites Atlas Labels V1 V1 V2 V2 hV4 V4 MT MT V7 DP phPIT(ventral&dorsal) G_pariet_inf-Angular C TEO 7A Vertices within the selected areas Figure 2 B 0.6 0.5 0.4 0.3 0.2 Coherence A 0.1 0 C D 0.024 0.020 0.012 GC 0.016 0.008 0.004 0 seed parcel Figure 3 A B C 1 20 40 D 120 0.772 0.433 0.772 V2 0.433 0 1 0.321 V4 3.40 0 50 100 freq(Hz) 0.890 0.159 MT 3.14 NA 0.104 DP 4.37 V2 0.942 0.306 MT V4 NA 0.119 TEO 3.78 V2 0.135 DP V4 0.291 1 NA 0.000 7A 4.73 50 100 freq(Hz) 0.306 MT 0.946 0.054 TEO 0.364 DP DP 0.511 MT 0.024 0.662 0.439 TEO 0 0.135 0.802 DP 0.364 0.177 0.333 TEO 0 0.439 TEO V4 MT 0.486 50 100 freq(Hz) 0.341 MT 0.253 1.000 0.661 7A 50 100 freq(Hz) 0.333 TEO MT 0 0.802 0.177 7A 1.000 0.661 7A 0 0.341 0.253 DP 7A 0.337 0.232 DP 1.68 1 50 100 freq(Hz) 0 TEO 0 7A 0 7A 0.706 0.000 7A 1.62 0.337 0.232 DP 1.85 1 0.024 1.90 TEO DP NA 0 0.424 1.56 1 7A 3.11 0 MT 50 100 freq(Hz) TEO V2 0 DP 0.000 0 0.662 2.48 1.47 1 0.054 TEO 2.62 0.838 NA 3.63 V4 0 0.424 V1 0 0.946 2.68 MT 7A TEO 0 0.498 0 MT 0 V4 50 100 freq(Hz) 0.486 2.15 3.35 1 0.838 0.119 140 7A 3.04 V4 50 100 freq(Hz) NA 120 5.05 V2 0 V4 7A V1 DP 3.33 1 60 80 100 Frequency (Hz) 0 0.909 1.61 0 NA DP 0.104 0 0.615 40 3.89 V2 0 0.498 2.56 V2 0 0.511 0 NA 3.87 V4 0 V2 3.86 1 0.942 0 MT 20 TEO 1.78 V4 0 V1 0.615 1 0 V2 50 100 freq(Hz) 0 140 4.86 2.90 3.27 3.11 0 V1 0 0.909 0 V1 MT 2.46 3.73 0 0.159 2.40 0 V1 0 0.291 0 V4 2.52 0 7A 0.936 0.890 0 0.936 V2 0 V1 TEO V2 120 DP 3.75 2.40 0 DP V1 2.84 0.981 60 80 100 Frequency (Hz) V4 0.321 3.99 2.94 V1 MT 40 MT 0.981 V1 3.37 V2 20 V4 0 V4 1 Granger Causality Feedforward(FF) Feedback(FB) Random FF Random FB Significant Frequencies FF vs Random FF FB vs Random FB FF vs FB RIGHT HEMISPHERE V1 GC axes : x10-² V1 0 140 V2 V1 V1 V2 60 80 100 Frequency (Hz) 3.32 x10-² Granger Causality Feedforward(FF) Feedback(FB) Random FF Random FB Significant Frequencies FF vs Random FF FB vs Random FB FF vs FB Granger Causality Granger Causality 0 3.08 x10-² Granger Causality Actual : Left Actual : Right Random : Left Random : Right Significant Frequencies Actual vs Random Left Right Granger Causality 2.33 x10-² GC over FF and FB connections - Right Hemisphere GC over FF and FB connections - Left Hemisphere GC over all connections TEO 0.706 0.000 7A 1 50 100 freq(Hz) 1.82 1 50 100 freq(Hz) 0 1 50 100 freq(Hz) LEFT HEMISPHERE Lower Area A (accord. to SLN) Feed-forward SLN Feed-back SLN Higher Area B (accord. to SLN) spectrum of GC from area A to area B spectrum of GC from area B to area A spectrum of RANDOM GC from area A to area B spectrum of RANDOM GC from area B to area A frequencies of significant GC from A to B w.r.t corresponding RANDOM GC. frequencies of significant GC from B to A w.r.t corresponding RANDOM GC. frequencies of significant difference between GC from A to B and GC from B to A Figure 4 A B C GC over all connections α/β +5 −20 peak D +20 γ peak 0.772 0.433 0.772 0.981 V2 V1 3.22 0 V2 0.321 V4 3.51 0.890 0.159 MT 3.86 NA V2 0.104 DP 5.14 0.942 0.306 MT V4 NA 0.119 TEO 3.90 0.909 0.135 DP V4 0.946 0.054 TEO NA 0.000 7A 5.43 −5 αβ +5 0 0.364 DP 0.511 MT NA 0.024 −5 αβ +5 DP 0.662 0.439 TEO 0.486 0 MT 0.802 0.177 7A −5 αβ +5 MT 0.486 0.333 TEO TEO 0.253 0 DP 1.000 0.661 7A −5 αβ +5 7A 0 −5 αβ +5 7A V4 0.802 0.177 7A 0.333 TEO MT 1.000 0.661 7A 1.96 0 0.341 0.253 DP 7A 0.337 0.232 DP 1.61 −5 αβ +5 0 γ +20 +60 freq(Hz) TEO 0.706 0.000 7A 1.84 0.337 0.232 DP TEO 1.86 γ +20 +60 freq(Hz) 0.024 0 0.424 MT 0 γ +20 +60 freq(Hz) TEO 0.341 NA 3.55 2.29 1.38 0 0.439 TEO 1.69 −5 αβ +5 7A 0 0.662 0 0.838 0.000 4.01 2.52 0.424 V2 0 MT γ +20 +60 freq(Hz) 0.054 TEO 2.72 DP MT 2.29 4.14 0 0.838 0 V4 γ +20 +60 freq(Hz) 0.364 NA 0 0.946 V4 0 V4 7A DP 0 γ +20 +60 freq(Hz) V1 0 0.498 V4 3.50 −5 αβ +5 TEO 3.42 1.59 2.90 4.45 γ +20 +60 freq(Hz) 0.498 0 V2 0.135 +60 5.38 V2 1.74 0 V2 0 V1 0.511 0 MT 3.68 3.27 0 0.615 DP 0 0.615 V4 0.119 +40 Frequency (Hz) 0 0.909 4.05 0 γ +20 +60 freq(Hz) +20 7A 4.17 V2 0 V2 0 V1 0.306 NA V1 0 MT 2.97 4.30 0 0 −5 αβ +5 DP 0.104 4.89 0.942 γ peak TEO NA V1 2.47 0 V1 7A 0.291 0 V4 2.97 0 TEO 0.936 MT 3.12 0 V1 DP V2 0.159 V2 2.52 0 MT 0.291 V4 α/β +5 −20 peak DP 0 0.936 0 −5 +60 Frequency (Hz) 0.890 V1 3.21 0.981 +40 3.84 V2 0 V4 V4 0 γ +20 +60 freq(Hz) 3.14 V1 0.321 4.39 −5 αβ +5 +20 MT V4 0.433 γ peak Granger Causality Feedforward(FF) Feedback(FB) Random FF Random FB Significant Frequencies FF vs Random FF FB vs Random FB FF vs FB x RIGHT HEMISPHERE V1 α/β +5 −20 peak V2 V1 GC axes : x10-² 0 −5 +60 Frequency (Hz) V1 V1 +40 3.60 10-² Granger Causality Feedforward(FF) Feedback(FB) Random FF Random FB Significant Frequencies FF vs Random FF FB vs Random FB FF vs FB Granger Causality Granger Causality 0 −5 V2 3.46 x10-² Granger Causality Actual : Left Actual : Right Random : Left Random : Right Significant Frequencies Actual vs Random Left Right GC over FF and FB connections - Right hemisphere Granger Causality 2.50 x10-² GC over FF and FB connections - Left hemisphere 0.706 0.000 0 7A −5 αβ +5 γ +20 +60 freq(Hz) 1.86 γ +20 +60 freq(Hz) 0 −5 αβ +5 γ +20 +60 freq(Hz) LEFT HEMISPHERE Lower Area A (accord. to SLN) Feed-forward SLN Feed-back SLN Higher Area B (accord. to SLN) frequencies of significant GC from A to B w.r.t corresponding RANDOM GC. frequencies of significant GC from B to A w.r.t corresponding RANDOM GC. spectrum of GC from area A to area B spectrum of GC from area B to area A spectrum of RANDOM GC from area A to area B spectrum of RANDOM GC from area B to area A frequencies of significant difference between GC from A to B and GC from B to A αβ γ peak frequency in the alpha-beta range peak frequency in the gamma range Figure 5 A B DAI-SLN correlation - Left Hemisphere 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 DAI-SLN correlation - Right Hemisphere 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.226 −5 α/β +5−20 peak γ peak +20 +40 +60 Frequency (Hz) pos.cluster neg.cluster 0.467 0.5 Correlation Coefficient Correlation Coefficient pos.cluster neg.cluster 0.465 0.5 -0.178 −5 α/β +5−20 peak γ peak +20 +40 +60 Frequency (Hz) Figure 6 A DAI from Gamma band Functional hierarchy level Left hemisphere Right hemisphere 7 6 5 4 3 V2 2 1 TEO MT V4 V1 1 2 Ventral Dorsal 7A DP 4 5 7 8 9 Anatomical hierarchy level B V4 V2 6 DP MT V1 [R=0.962, P=0.0005] 3 7A TEO 1 [R=0.945, P=0.0013] 2 3 4 5 6 7 8 9 Anatomical hierarchy level Functional hierarchy level DAI from Alpha-Beta band Left hemisphere 7 6 TEO 3 V1 V2 V1 V4 2 TEO V2 V4 [R=0.750, P=0.0320] 1 2 3 4 5 6 7 8 9 Anatomical hierarchy level 7A DP MT MT 4 C 7A DP 5 1 Right hemisphere Ventral Dorsal [R=0.812, P=0.0267] 1 2 3 4 5 6 7 8 9 Anatomical hierarchy level Functional hierarchy level DAI from combined Gamma and Alpha-Beta bands Left hemisphere 7 6 5 V2 3 1 TEO V4 V1 2 7A DP MT TEO 4 Right hemisphere Ventral Dorsal V1 [R=0.901, P=0.0057] 1 2 3 4 5 6 7 8 9 Anatomical hierarchy level 1 2 V2 7A DP MT V4 [R=0.923, P=0.0030] 3 4 5 6 7 8 9 Anatomical hierarchy level Figure 7 A Hierarchy from DAI in Gamma band Average Rank in Hierarchy 26 Ventral stream Dorsal stream ar.46 7A 21 IPS4 ER IPS2 FST 16 V3C 11 6 V2 LO1 V3A V3 V7 FEF PITd MT V4t IPS1 IPS3 MST LO2 V4 V3B VO1 PITv V3D V1 1 Rank B Hierarchy from DAI in Beta band Average Rank in Hierarchy 26 Ventral stream Dorsal stream Hierarchical level in (A) 21 IPS4 MST FEF IPS2 ER ar.46 V7 IPS3 V3D V3C 16 IPS1 FST LO2 11 V1 LO1 V2 6 V3 PITv MT V4t V3B V3A VO1 PITd V4 1 Rank based on DAI in gamma band (A) C Hierarchy from combined DAI in Gamma and Alpha/Beta bands 26 Average Rank in Hierarchy 7A Ventral stream Dorsal stream Hierarchical level in (A) 21 ar.46 MST FEF 16 LO2 11 V3B 6 V1 1 V2 V3 LO1 V4t MT V7 IPS4 V3D V3C IPS3 IPS1 IPS2 ER 7A FST PITv VO1 PITd V3A V4 Rank based on DAI in gamma band (A) Figure 8 1 Supplemental Data Figure S1. Distributions of individual peak frequencies in Gamma and Alpha-Beta bands For each subject, Gaussian functions were fit to the GC spectrum averaged over all area pairs of both hemispheres, separately for the frequency ranges 4-20 Hz (A, C) and 30-100 Hz (B, D). (A) Histogram of peak frequencies across the 43 subjects in the 4-20 Hz range. (B) Histogram of goodness-of-fit measure (r-square) of the Gaussian function in the 4-20 Hz range across the 43 subjects. (C) Same as (A), but for 30-100 Hz range. (D) Same as (B), but for 30-100 Hz range. Notice that the goodness-offit distribution is heavily skewed towards 1, indicating good fitting results. Figure S2. DAI-SLN correlation spectra versus randomized surrogate data and without peak-frequency alignment. In order to test whether there was any bias in the DAI-SLN correlation presented in Fig. 6, epochs were shuffled prior to computing GC and the subsequent DAI-SLN correlation. In (A) and (B), for left and right hemispheres respectively, this randomized surrogate correlation is depicted by the yellow line. Significance was again assessed with the same non-parametric permutation based tests. Very similar positive and negative frequency clusters were found in the gamma and alpha-beta ranges respectively. (C,D) DAI-SLN correlation spectra without peak-frequency alignment. For these analyses, the spectrum of each subject has not been aligned relative to the individual peaks in the gamma and alpha-beta bands, as was done in Figure 6. Rather the DAI-SLN correlation spectra are averaged over subjects without alignment. Significance is assessed relative to zero correlation as in Fig. 6. (E,F) DAI-SLN correlation without peak-frequency alignment as in (C,D), and testing against random correlation as in (A, B). Note that without peak-frequency alignment, peak correlation values are slightly lower. Figure S3. Granger-Causality-based hierarchy versus anatomical hierarchy without peakfrequency alignment. This figure presents the relation of the human functional hierarchy derived for DAI values versus the macaque anatomical hierarchy from SLN values, similar to Figure 7, but without alignment of individual peak frequencies. In the same fashion as in Figure 7, three cases were investigated, namely with the functional hierarchy computed from (A) DAI in the gamma band, (B) DAI in the alpha-beta band and (C) the combined DAI in alpha-beta and gamma bands. Early visual areas are depicted in black, ventral-stream areas in green, and dorsal-stream areas in red. Inset brackets on bottom right of each subplot report Pearson correlation between human functional and macaque anatomical hierarchical levels. Results were qualitatively the same as in Fig. 7. Table S1. Number of dipoles within each of the 26 visual areas in the left and right hemispheres of the MEG source model. A B Hist. of GC Peak frequency in gamma range 10 15 N subjects N subjects 8 6 4 0 52 C 54 56 58 60 62 64 Peak frequency(Hz) 66 68 0 70 0 D Hist. of GC Peak frequency in alpha/beta range 20 0.2 0.4 R−square 0.6 0.8 1 Hist. of Goodness-of-fit of Gaussian in alpha/beta range 25 20 N subjects 15 N subjects 10 5 2 10 5 0 Hist. of Goodness-of-fit of Gaussian in gamma range 20 15 10 5 6 8 10 12 14 Peak frequency(Hz) 16 18 20 0 0 0.2 0.4 R−square 0.6 0.8 1 Supplemental Figure 1 Correlation between DAI and SLN Left Hemisphere B Aligned Spectrum - Test vs. Random actual random pos.cluster neg.cluster 0.465 0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.226 -0.3 −5 α/β +5−20 peak γ peak +20 +40 Correlation Coefficient Correlation Coefficient A Right Hemisphere 0.3 0.2 0.1 0 -0.1 -0.2 -0.178 −5 α/β +5−20 peak Frequency (Hz) 0.5 +20 +40 +60 Frequency (Hz) 0.410 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 Non-Aligned Spectrum - Test vs. Zero 0.5 pos.cluster neg.cluster Correlatino Coefficient Correlation Coefficient γ peak D Non-Aligned Spectrum - Test vs. Zero 20 pos.cluster neg.cluster 0.421 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.166 -0.193 1 40 60 80 100 120 140 Frequency (Hz) E 1 20 40 60 80 100 120 140 Frequency (Hz) F Non-Aligned Spectrum - Test vs. Random actual random pos.cluster neg.cluster 0.410 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 20 actual random pos.cluster neg.cluster 0.421 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.166 -0.193 1 Non-Aligned Spectrum - Test vs. Random 0.5 Correlation Coefficient 0.5 Correlation Coefficient actual random pos.cluster neg.cluster 0.467 0.4 +60 C Aligned Spectrum - Test vs. Random 0.5 40 60 80 100 120 140 Frequency (Hz) 1 20 40 60 80 100 120 140 Frequency (Hz) Supplemental Figure 2 A DAI from Gamma band Left hemisphere Functional hierarchy level 7 6 4 DP 2 1 V2 V1 [R=0.975, P=0.0002] 2 DP MT V4 V2 V1 7A TEO V4 3 1 Ventral Dorsal 7A TEO MT 5 Right hemisphere 3 4 5 6 7 8 9 Anatomical hierarchy level B 1 [R=0.952, P=0.0010] 2 3 4 5 6 7 8 9 Anatomical hierarchy level DAI from Alpha-Beta band Left hemisphere Functional hierarchy level 7 6 7A DP 5 4 MT V1 3 V2 V1 V2 TEO V4 [R=0.693, P=0.0414] 1 C 2 3 4 5 6 7 8 9 Anatomical hierarchy level 7A DP MT V4 2 1 Right hemisphere Ventral Dorsal TEO [R=0.770, P=0.0328] 1 2 3 4 5 6 7 8 9 Anatomical hierarchy level Functional hierarchy level DAI from combined Gamma and Alpha-Beta bands Left hemisphere 7 6 7A DP 5 MT 4 3 2 V1 1 Right hemisphere V2 Ventral Dorsal MT TEO V4 V1 V2 2 3 4 5 6 7 8 9 Anatomical hierarchy level TEO V4 [R=0.933, P=0.0020] 1 7A DP [R=0.930, P=0.0027] 1 2 3 4 5 6 7 8 9 Anatomical hierarchy level Supplemental Figure 3 Area Name V1 V2 V3 V3A V3B V3C V3D V4 V4t LO1 LO2 PITd PITv MT VO1 MST FST ER V7 IPS1 IPS2 IPS3 IPS4 7A FEF N dipoles in Le Hemisphere 31 40 49 9 12 11 8 9 2 14 11 3 3 3 11 2 2 22 5 5 6 15 12 56 24 N dipoles in Right Hemisphere 28 38 44 10 8 11 8 10 3 15 9 2 2 8 7 2 2 22 6 5 6 15 12 58 20 Supplemental Table 1 2 Supplemental Experimental Procedures Data Acquisition MEG: Recordings were performed with a whole-head MEG system (CTF Systems Inc.) comprising 275 axial gradiometers, located at the Donders Institute of Radboud University Nijmegen (Nijmegen, Netherlands). In addition, horizontal and vertical Electrooculograms (EOG) were recorded using bipolar electrodes for off-line eye-movement artifact rejection. All signals were low-pass filtered at 300 Hz, sampled at 1200 Hz, and stored. Head position relative to the gradiometer array was determined prior and after the MEG recording using 3 coils attached to subject’s nasion, left and right ear canals. MRI: Structural MR images were acquired using an Avanto 1.5 T whole body MRI scanner (Siemens, Germany). A volume head coil was used for RF transmission and signal reception. The scan was performed with a 3D MPRAGE sequence with 1 mm isotropic resolution. Subjects Datasets of 43 subjects were selected from a pool of 160 subjects, which had been recorded for genotyping of parameters derived from visually induced gamma-band activity. The genotyping study required a large number of subjects and was fully focused on gamma, and it therefore accepted subjects, in which sub-gamma activity was confounded by artifacts. For the present study, subjects were excluded if 1) their body carried ferromagnetic elements causing low-frequency MEG artifacts (e.g. metallic orthodontic retainers), 2) they did not have normal vision, 3) they used medication, 4) they exhibited head movements larger than 10 mm during the experiment. This left 43 subjects (36 right-handed) with an average age of 23.1 (range: 18 to 34). All subjects gave written informed consent according to guidelines of the local ethics committee (Commissie Mensgebonden Onderzoek Regio Arnhem-Nijmegen, The Netherlands). Human Brain Model For source analysis, structural MR images were acquired for each individual subject, and the cortical mantle was extracted with the FreeSurfer longitudinal stream pipeline (Reuter et al., 2012). Cortical parcellation and area definition used a recent visuotopic atlas provided by (Abdollahi et al., 2014), the FreeSurfer software suite (http://FreeSurfer.net/) (Fischl et al., 2004; Desikan et al., 2006; Destrieux et al., 2010) and the Caret software suite (http://brainvis.wustl.edu/) (Glasser and Van Essen, 2011; Van Essen et al., 2012). The cortical sheet extracted from FreeSurfer was transformed and registered into the Caret representation through the Caret pipeline “freesurfer_to_fs_LR”. In Caret, the Conte69 atlas is defined and contains a composite parcellation (Van Essen et al., 2012) which includes multiple visuotopic areas in the occipital, parietal and temporal lobe, which have been mapped based on a number of studies. The recent visuotopic atlas provided by (Abdollahi et al., 2014) is based on Maximum Probability Maps (MPMs) for 18 human visual areas, that have been built based on recent Multimodal Surface Matching (MSM) algorithms for inter-subject registration, which utilize as registration features not only structural cortical characteristics but also retinotopic activation maps. The resulting parcellation provides a more refined cortical representation of these 18 cortical areas as compared to previous studies. This parcellation atlas is represented on the fs_LR cortical surface 3 template, similarly to the Conte69 visuotopic atlas. Hereafter, for convenience, this parcellation will be referred to in this text as the “Abdollahi2014” parcellation. Finally, one of the parcellation schemes of FreeSurfer, “aparc.a2009s” (Destrieux et al., 2010), provides parcels segregating all the main gyri and sulci of the human cortex and was used in addition to the visuotopic Conte69 and the parcellations from Caret, as explained in more detail below. Both, the FreeSurfer and Caret cortical sheet representations contain more than 105 vertices per hemisphere and were subsampled for MEG source analysis with the MNE software suite (Gramfort et al., 2014). Through recursive icosahedron subdivision, this software subsamples the dense FreeSurfer cortical representation by a fixed number of 4098 vertices per hemisphere. Those vertices were mapped to their nearest neighbors in the Caret common template surface representation and thereby to the Conte69 atlas. The vertex positions were then transformed into the head coordinate system, where the MEG measurements were acquired and the inverse solution was computed. Selection of 7 homologous brain areas between human and macaque V1, V2: In both macaque and human brains, V1 and V2 are the largest cortical areas and have the clearest homology between these species (Van Essen, 2005). They are both available in the Abdollahi2014 atlas labelled as V1 and V2 respectively. MT: In the macaque, area MT is located on the posterior bank of the superior temporal sulcus, is characterized by its distinct heavy myelination, and contains a complete topographic representation of the contralateral visual field (Van Essen et al., 1981). In the human brain, area MT has been consistently mapped at a corresponding location, has a similar visuotopic organization and is characterized by its motion-selective activation (Huk et al., 2002; Fischl et al., 2008; Kolster et al., 2010). The area used here is area MT/V5 in the so called MT/V5+ complex identified in the human brain (Kolster et al., 2010). This area is available in the Abdollahi2014 atlas, labelled as MT. V4 (hV4): In the macaque, area V4 has been mapped both physiologically and with functional MRI. It consists of 2 parts, a dorsal lower field representation and a ventral upper-field representation. The horizontal meridian is represented anterior to both ventral and dorsal V4 in the macaque brain (Fize et al., 2003; Janssens et al., 2014; Kolster et al., 2014). In the human brain the strongest evidence of homology comes mainly from the ventral part of human V4, which has been mapped by various studies in a corresponding location and with similar visuotopic organization (Brewer et al., 2005; Van Essen, 2005; Swisher et al., 2007; Kolster et al., 2010). This area is available in the Abdollahi2014 atlas, labelled as hV4. DP (V7): In the macaque brain, visual area DP occupies the dorsal aspect of the prelunate gyrus immediately anterior to area V3A. In the human brain, an area located also just anteriorly to area V3A (as defined in (Tootell et al., 1998) or V3D as defined in (Georgieva et al., 2009) and (Abdollahi et al., 2014)) with a putative homology in its retinotopic representation has been identified by (Tootell et al., 1998) and given the name V7. This area is available in the Abdollahi2014 atlas, labelled as V7. TEO (PIT): In the macaque brain, according to (Felleman and Essen, 1991), the posterior part of the inferotemporal cortex is divided into two regions labelled as PITd (dorsal) and PITv (ventral). These areas coincide largely with the architectonic Temporal Occipital area (TEO). In the human brain, the putative homologous areas have been mapped using functional MRI, have been shown to be 4 involved in processing two-dimensional shape, and they have been given the same names as in the macaque (Kolster et al., 2010). These areas are available in the Abdollahi2014 atlas labelled as phPITd and phPITv. Here, these ventral and dorsal parcels have been collapsed into a single parcel called PIT. 7A (Gyrus parietalis inferior – Angular gyrus): In the macaque brain, area 7A occupies the caudal part of the Inferior Parietal Lobule (IPL) and extends only a short distance into the intraparietal sulcus (Felleman and Essen, 1991). In an alternative, widely used parcellation scheme of the macaque IPL by (Pandya and Seltzer, 1982) this area was attributed to 2 areas of the caudal IPL, namely the areas labelled as PG and Opt. In the human brain, (Caspers et al., 2011) identified 2 homologous areas through diffusion-weighted MRI, spanning the caudal part of human IPL and showing strong resemblance in their connectivity patterns with that of the macaque. In the human visuotopic parcellation scheme available in the Conte69 atlas, no parcel covers the caudal part of IPL, where this homology has been identified. In FreeSurfer’s “aparc.a2009s” parcellation scheme (Destrieux et al., 2010) the caudal IPL is covered by the parcel labelled as “G_pariet_inf-Angular”, which primarily encapsulates the angular gyrus and is segregated from rostral IPL by the sulcus intermedius premus. Dorsally it extends into the intrapariatal sulcus (IPS). Based on the above anatomical characteristics, this parcel was selected as homologous to macaque area 7A. As mentioned above, the dense cortical surface representation of the common template “fs_LR” surface was down-sampled for the MEG source analysis. Accordingly, the representation of the selected areas was defined in this subsampled space. Details on the Selection of 26 visual areas from the human brain V1, V2, V4, MT, V7, 7A: These areas are the same that were used in the 7-area analysis (see above and Experimental Procedures).FEF: The Frontal Eye Field is a cortical area located in humans around the intersection of the pre-central sulcus with the dorsal portion of the superior frontal sulcus. It is considered a “hub”, having anatomical connections with visual, parietal, temporal and prefrontal areas. Functionally it is involved in oculomotor behavior, visuo-spatial attention, visual awareness and perceptual modulation (Vernet et al., 2014). In the visuotopic parcellation of the Conte69 atlas, an FEF parcel is not available. The most relevant parcel that is available is Brodmann area 8, in which FEF occupies only a relatively small portion. In order to represent FEF, only the ventral branch of the superior precentral sulcus and its intersection with the superior frontal sulcus were manually selected as the parcel that represents FEF (Amiez and Petrides, 2009). The following areas are available in the atlas dataset of (Abdollahi et al., 2014) which in the following text will be referred to as “Abdollahi2014”: V3: Area V3, in both human and monkey, consists of a lower and an upper field representation, neighboring area V2 dorsally and ventrally respectively. Although these dorsal and ventral parts have been shown to have some asymmetries in their function, architecture and connectivity, they are considered in both human and monkey as the constituents of a single retinotopic area (Van Essen, 2005). V3A, V3B, V3C, V3D: These areas that span the cortical surface adjoined dorsally to area V3 were identified by (Georgieva et al., 2009). Therein, they are jointly referred to as the “V3A complex” and they represent the same cortical areas previously referred to as V3A and V3B (Wandell et al., 2007). The dorsal areas of the complex, V3C and V3D, adjoin ventrally area V7. These areas contain neurons 5 with larger retinotopic fields than areas V1, V2 and V3. These areas have been implicated in motion processing, with areas V3A and V3B (corresponding to area V3B in (Fischer, 2011)) having a higher preference for externally induced retinal motion, and areas V3C and V3D (corresponding to area V3A in (Fischer, 2011)) having a higher preference for real motion. LO1, LO2: Lateral Occipital Areas 1 and 2. These areas are located in the fundus of the lateral occipital sulcus between dorsal V3 and MT, with LO2 anterior to LO1. The retinotopy of these areas is less precise and more variable than early visual areas. It has been suggested that these two areas represent shape information, with LO1 extracting boundary information and LO2 extracting regions and representing shape (Larsson and Heeger, 2006; Wandell et al., 2007). PITv, PITd: These two areas are the constituents of area PIT used in the 7-area analysis. In the 26area analysis PIT is replaced by these two areas. MT, pMSTv, pFST, pV4t: These four areas comprise the human middle temporal MT/V5+ complex, homologous to the MT/V5 complex in the monkey brain. This complex consists of the middle temporal area (MT/V5), the putative ventral part of the medial superior temporal area (pMSTv), the putative fundus of the superior temporal area (pFST) and the putative V4 transitional zone. Area V4t is adjoined ventrally to MT/V5 and dorsally to phPITd. Area pMSTv is located rostrally to area MT/V5. Area pFST is located anteriorly to area V4t and ventrally to area pMSTv. All four areas respond strongly to motion, have much larger receptive fields than V1 and are considered parts of the dorsal stream. Within the MT/V5+ complex, area pMSTv has the largest receptive fields. The more ventral areas of the complex, V4t and pFST, respond stronger to shapes, while the more dorsal areas, MT/V5 and pMSTv respond stronger to 3D structure from motion (Kolster et al., 2010). Area MT here is the same as the one used in the 7-area analysis. VO1: Area VO1 is part of Ventral Occipital areas(VO) located on the fusiform gyrus anteriorly of area hV4 and is considered part of the ventral stream. It has a preference for foveal and parafoveal eccentricities. Due to the fact that VO areas show little object selectivity, it has been hypothesized that they play an intermediate visual processing role between early visual cortex and higher order object-selective cortex (Brewer et al., 2005; Arcaro et al., 2009; Kolster et al., 2010). The following areas are available in the visuotopic parcellation of the Conte69 atlas: IPS1, IPS2, IPS3, IPS4: These four parcels are subdivisions of the posterior medial bank of the intrapariatel sulcus(IPS) with distinct retinotopic maps, numbered from posterior to anterior locations. Numerous functions related to visual processing such as visual attention, saccadic preparation and multisensory integration have been shown to activate these areas (Swisher et al., 2007). ER: The entorhinal cortex, located in the medial temporal lobe, comprises of a set of cortical areas that are known to be a major gateway to and from the hippocampus. Based on its anatomical connectivity and functional properties, it has been categorized as multimodal association cortex. Although it is not retinotopically organized, it has strong connections with some visual areas (Felleman and Essen, 1991) and it has recently been shown to play an important role in spatial navigation (Moser et al., 2008). Area 46: Brodmann area 46 corresponds with the dorsolateral prefrontal cortex in and around the middle third of the principal sulcus. It has strong connections mainly with parietal and temporal cortex (Petrides and Pandya, 1999). There has been strong evidence that functionally it is involved in maintaining and manipulating information in working memory (Krawczyk, 2002). 6 Retrograde tracing database Description of the anatomical dataset acquisition and analysis has been reported in (Markov et al., 2014). The values that we used correspond to multiple injections each into V1, V2, V4 and single injections into areas MT, DP, TEO and 7A. Only SLN values based on at least 10 labeled neurons were included. Three projections had fewer than 10 neurons labeled, and their SLN was therefore considered unreliably quantified. These projections were V1-to-DP, V1-to-7A and V2-to-7A. The V1to-TEO projection was reported as missing i.e. no labeled neurons found in V1 after TEO injection. For these projections, there is strong evidence in the literature that they are present and all have a feedforward characteristic, and this was used in place of SLN evidence. Updates, atlases and additional information concerning the anatomical dataset that was used for this work is available at www.core-nets.org. Inverse solution Subject-wise structural MRI scans were used to derive individual single-shell volume conductor models (Nolte, 2003). The inverse solution was then performed using Linearly Constrained Minimum Variance (LCMV) adaptive spatial filtering, often referred to as “Beamforming” (Van Veen et al., 1997). The beamformer gives, for each of a set of predefined locations, a coefficient matrix such that the multiplication of the matrix with the sensor-level signals provides an estimate of the dipole moment at that location, while minimizing influences of other, linearly uncorrelated sources. We used this approach to scan a set of dipole locations on the cortical mesh. As the beamformer output is simply a linear projection of sensor-level data, it does not only apply to the raw signal of the different MEG sensors but also to any linear decomposition, like the Fourier transform of the raw signal. There are beamformer methods that derive spatial filters for each frequency separately (Gross et al., 2001). They are mostly suited to provide source estimates for selected frequency bands. The current work investigates the entire spectrum, and focuses on GC, whose estimation entails multiple neighboring frequencies. Therefore, we opted to estimate the source spectrum by means of the LCMV broadband inverse solution. The LCMV solution is based on the broadband covariance matrix of the non-averaged MEG sensor data. When determining the LCMV solution, we used a regularization parameter of 20% of the covariance matrix. The set of brain locations for which the inverse solution was computed comprised of all the points distributed within the human visual cortical areas of interest in each hemisphere. Supplemental References Abdollahi RO, Kolster H, Glasser MF, Robinson EC, Coalson TS, Dierker D, Jenkinson M, Van Essen DC, Orban GA (2014) Correspondences between retinotopic areas and myelin maps in human visual cortex. Neuroimage 99:509-524. Amiez C, Petrides M (2009) Anatomical organization of the eye fields in the human and non-human primate frontal cortex. Prog Neurobiol 89:220-230. Arcaro MJ, McMains SA, Singer BD, Kastner S (2009) Retinotopic organization of human ventral visual cortex. J Neurosci 29:10638-10652. Brewer AA, Liu J, Wade AR, Wandell BA (2005) Visual field maps and stimulus selectivity in human ventral occipital cortex. Nat Neurosci 8:1102-1109. 7 Caspers S, Eickhoff SB, Rick T, von Kapri A, Kuhlen T, Huang R, Shah NJ, Zilles K (2011) Probabilistic fibre tract analysis of cytoarchitectonically defined human inferior parietal lobule areas reveals similarities to macaques. Neuroimage 58:362-380. Desikan RS, Ségonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, Buckner RL, Dale AM, Maguire RP, Hyman BT, Albert MS, Killiany RJ (2006) An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage 31:968-980. Destrieux C, Fischl B, Dale A, Halgren E (2010) Automatic parcellation of human cortical gyri and sulci using standard anatomical nomenclature. Neuroimage 53:1-15. Felleman DJ, Essen DCV (1991) Distributed hierarchical processing in the primate cerebral cortex. Cereb Cortex 1:1-47. Fischer E (2011) Visual motion and self-motion processing in the human brain, MPI Series in Biological Cybernetics, Bd. 31: Logos Verlag Berlin GmbH. Fischl B, Rajendran N, Busa E, Augustinack J, Hinds O, Yeo BTT, Mohlberg H, Amunts K, Zilles K (2008) Cortical folding patterns and predicting cytoarchitecture. Cereb Cortex 18:1973-1980. Fischl B, van der Kouwe A, Destrieux C, Halgren E, Ségonne F, Salat DH, Busa E, Seidman LJ, Goldstein J, Kennedy D, Caviness V, Makris N, Rosen B, Dale AM (2004) Automatically parcellating the human cerebral cortex. Cereb Cortex 14:11-22. Fize D, Vanduffel W, Nelissen K, Denys K, Chef d'Hotel C, Faugeras O, Orban GA (2003) The retinotopic organization of primate dorsal V4 and surrounding areas: A functional magnetic resonance imaging study in awake monkeys. J Neurosci 23:7395-7406. Georgieva S, Peeters R, Kolster H, Todd JT, Orban GA (2009) The processing of three-dimensional shape from disparity in the human brain. J Neurosci 29:727-742. Glasser MF, Van Essen DC (2011) Mapping human cortical areas in vivo based on myelin content as revealed by T1- and T2-weighted MRI. J Neurosci 31:11597-11616. Gramfort A, Luessi M, Larson E, Engemann DA, Strohmeier D, Brodbeck C, Parkkonen L, Hämäläinen MS (2014) MNE software for processing MEG and EEG data. NeuroImage 86:446-460. Gross J, Kujala J, Hämäläinen M, Timmermann L, Schnitzler A, Salmelin R (2001) Dynamic imaging of coherent sources: studying neural interactions in the human brain. Proceedings of the National Academy of Sciences 98:694-699. Huk AC, Dougherty RF, Heeger DJ (2002) Retinotopy and functional subdivision of human areas MT and MST. J Neurosci 22:7195-7205. Janssens T, Zhu Q, Popivanov ID, Vanduffel W (2014) Probabilistic and single-subject retinotopic maps reveal the topographic organization of face patches in the macaque cortex. J Neurosci 34:10156-10167. Kolster H, Peeters R, Orban GA (2010) The retinotopic organization of the human middle temporal area MT/V5 and its cortical neighbors. J Neurosci 30:9801-9820. Kolster H, Janssens T, Orban GA, Vanduffel W (2014) The retinotopic organization of macaque occipitotemporal cortex anterior to V4 and caudoventral to the middle temporal (MT) cluster. J Neurosci 34:10168-10191. Krawczyk DC (2002) Contributions of the prefrontal cortex to the neural basis of human decision making. Neurosci Biobehav Rev 26:631-664. Larsson J, Heeger DJ (2006) Two retinotopic visual areas in human lateral occipital cortex. J Neurosci 26:13128-13142. Markov NT, Vezoli J, Chameau P, Falchier A, Quilodran R, Huissoud C, Lamy C, Misery P, Giroud P, Ullman S, Barone P, Dehay C, Knoblauch K, Kennedy H (2014) Anatomy of hierarchy: feedforward and feedback pathways in macaque visual cortex. J Comp Neurol 522:225-259. Moser EI, Kropff E, Moser M-B (2008) Place cells, grid cells, and the brain's spatial representation system. Annu Rev Neurosci 31:69-89. 8 Nolte G (2003) The magnetic lead field theorem in the quasi-static approximation and its use for magnetoencephalography forward calculation in realistic volume conductors. Physics in medicine and biology 48:3637-3637. Pandya DN, Seltzer B (1982) Intrinsic connections and architectonics of posterior parietal cortex in the rhesus monkey. J Comp Neurol 204:196-210. Petrides M, Pandya DN (1999) Dorsolateral prefrontal cortex: comparative cytoarchitectonic analysis in the human and the macaque brain and corticocortical connection patterns. Eur J Neurosci 11:1011-1036. Reuter M, Schmansky NJ, Rosas HD, Fischl B (2012) Within-subject template estimation for unbiased longitudinal image analysis. Neuroimage 61:1402-1418. Swisher JD, Halko MA, Merabet LB, McMains SA, Somers DC (2007) Visual topography of human intraparietal sulcus. J Neurosci 27:5326-5337. Tootell RB, Hadjikhani N, Hall EK, Marrett S, Vanduffel W, Vaughan JT, Dale AM (1998) The retinotopy of visual spatial attention. Neuron 21:1409-1422. Van Essen DC (2005) Surface-Based Comparisons of Macaque and Human Cortical Organization. In: From Monkey Brain to Human Brain: A Fyssen Foundation Symposium, pp 3-19: Bradford Books. Van Essen DC, Maunsell JH, Bixby JL (1981) The middle temporal visual area in the macaque: myeloarchitecture, connections, functional properties and topographic organization. J Comp Neurol 199:293-326. Van Essen DC, Glasser MF, Dierker DL, Harwell J, Coalson T (2012) Parcellations and hemispheric asymmetries of human cerebral cortex analyzed on surface-based atlases. Cereb Cortex 22:2241-2262. Van Veen BD, van Drongelen W, Yuchtman M, Suzuki A (1997) Localization of brain electrical activity via linearly constrained minimum variance spatial filtering. Biomedical Engineering, IEEE Transactions on 44:867-880. Vernet M, Quentin R, Chanes L, Mitsumasu A, Valero-Cabré A (2014) Frontal eye field, where art thou? Anatomy, function, and non-invasive manipulation of frontal regions involved in eye movements and associated cognitive operations. Front Integr Neurosci 8:66-66. Wandell BA, Dumoulin SO, Brewer AA (2007) Visual field maps in human cortex. Neuron 56:366-383.