Additive effect of contrast and velocity proves the role ... · 1 Additive effect of contrast and velocity proves the role of strong 2 excitatory drive in suppression of visual gamma

1

Additiveeffectofcontrastandvelocityprovestheroleofstrong1

excitatorydriveinsuppressionofvisualgammaresponse.2

3

OrekhovaE.V.1,2,3,ProkofyevA.O.1,NikolaevaA.Yu.1,SchneidermanJ.F.2,3,StroganovaT.A.14

51.MoscowStateUniversityofPsychologyandEducation,CenterforNeurocognitiveResearch(MEG6

Center),Moscow,Russia.72.UniversityofGothenburg,InstituteofNeuroscience&Physiology,DepartmentofClinical8

Neurophysiology93.ChalmersUniversityofTechnologyandMedTechWest,Gothenburg,Sweden10

11

12

13

14

15

Correspondingauthor16

ElenaVOrekhova.MoscowStateUniversityofPsychologyandEducation(MSUPE),Centerfor17

NeurocognitiveResearch(MEGCenter),Shelepihinskajaembankment2a,123290Moscow,18

Russia19

E-mailaddresses:[email protected]

21

22

Acknowledgments23ThisworkwassupportedbytheMoscowStateUniversityofPsychologyandEducation;the24

CharityFoundation"WayOut";SwedishChildhoodCancer(#MT2014-0007); KnutandAlice25

WallenbergFoundation(#2014.0102)and SwedishResearchCouncil(#2017-0068). Weare26

gratefultoallvolunteerparticipants. 27

28

29

Keywords:30

MEG,gammaoscillations,visualmotion,contrast,excitation-inhibitionbalance 31

.CC-BY-ND 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted December 28, 2018. ; https://doi.org/10.1101/497214doi: bioRxiv preprint

https://doi.org/10.1101/497214

http://creativecommons.org/licenses/by-nd/4.0/

2

Abstract1

2Visualgammaoscillationsaregeneratedthroughinteractionsofexcitatoryandinhibitory3neuronsandarestronglymodulatedbysensoryinput.Amoderateincreaseinexcitatory4drive to the visual cortex via increasing contrast or motion velocity of drifting gratings5results in strengthening of the gamma response (GR). However, increasing the velocity6beyond some ‘transition point’ leads to the suppression of the GR. There are two7theoreticalmodels thatcanexplainsuchsuppression.The ‘excitatorydrive’model infers8that,athighdriftingrates,GRsuppression iscausedbyexcessiveexcitationof inhibitory9neurons. Since contrast and velocity have an additive effect on excitatory drive, this10modelpredictsthattheGR‘transitionpoint’forlow-contrastgratingswouldbereachedat11ahighervelocity,ascomparedtohigh-contrastgratings.Thealternative ’velocitytuning’12modelimpliesthattheGRismaximalwhenthedriftingrateofthegratingcorrespondsto13the preferable velocity of the motion-sensitive V1 neurons. This model predicts that14loweringcontrasteitherwillnotaffectthetransitionpointorwillshiftittoalowerdrifting15rate.Wetestedthesemodelswithmagnetoencephalography-basedrecordingsoftheGR16during presentation of low (50%) and high (100%) contrast gratings drifting at four17velocities. We found that lowering contrast led to a highly reliable shift of the GR18suppression transition point to higher velocities, thus supporting the excitatory drive19model.Noeffectsofcontrastorvelocitywerefoundforthealpha-betaresponsepower.20Theresultshaveimportant implicationsfortheunderstandingoftheneuralmechanisms21underlyinggammaoscillationsandthedevelopmentofgamma-basedbiomarkersofbrain22disorders.23

24

25

Introduction26

Gamma-bandoscillationsarisefromapreciseinterplaybetweenexcitation(E)andinhibition(I)27(BuzsakiandWang,2012;Vincketal.,2013)andtheirparametersareaffectedbyexternaland28

internalfactorsthatcanshifttheE-Ibalance(Chenetal.,2017;Hadjipapasetal.,2015;Jiaet29

al.,2013;Salelkaretal.,2018;Sumneretal.,2018).Theseoscillationsaremostprominent in30

theprimaryvisualcortex(V1)andcanbe induced inbothhumansandanimalsbyarangeof31

visual stimuli. Human visual gamma oscillations recorded with Magnetoencephalography32

(MEG) have high within-subject reproducibility (Hoogenboom et al., 2006;33

Muthukumaraswamy et al., 2010; Tan et al., 2016) and are strongly genetically determined34

(van Pelt et al., 2012). This makes the properties of gamma oscillations potentially useful35

signatures of theconstitutionally up- and down-regulatedE-I balance in the human visual36



https://doi.org/10.1101/497214


3

cortex. Various attempts have been made to relate the individual values of power and1

frequencyof thevisual inducedgamma response (GR) to indirectmeasuresof theexcitatory2

state of visual circuitry, such as concentration of GABA and glutamate neurotransmitters3

(Cousijn et al., 2014; Edden et al., 2009), GABAa receptor density (Kujala et al., 2015), or4

performance in a visual orientation discrimination task (Edden et al., 2009), but the link5

betweengammaactivityandtheexcitatoryandinhibitoryprocessesinthevisualcortexisstill6

farfromclear.7Therelationshipbetweentheexcitatorystateofthevisualcortexandthepropertiesof8

gammaoscillationscanbeclarifiedthroughmanipulationsoftheintensityofexcitatoryinput.9

Interestingly, the intensity of visual stimulation has a different impact on the frequency and10

powerofinducedgammaoscillationsinbothmonkeysandhumans.Thefrequencyofgamma11

oscillations nearly linearly increases with increasing visual input strength (Hadjipapas et al.,12

2015; Jia et al., 2013;Murty et al., 2018; Orekhova et al., 2015; Perry et al., 2015; Ray and13

Maunsell,2010;vanPeltetal.,2018)whilegammapowerundergoesnonlinearchanges(Jiaet14

al.,2013;Orekhovaetal.,2018b;Salelkaretal.,2018).15

Localfieldpotential(LFP)studiesofthemonkeyprimaryvisualcortexhaveshownthat16

theincreaseingammafrequencycausedbyincreaseineitherluminancecontrast(Hadjipapas17

etal.,2015;HenrieandShapley,2005)ortemporal frequencyofmovinggratings (Salelkaret18

al.,2018)wasparalleledbya rise inneuronal spiking rate.This suggests that the increaseof19

gamma frequency is associated with growing neural excitation in V1, and particularly with20excitationofinhibitorycircuitrythathavebeenshowntoplaythemajorroleinregulatingthe21

frequency of gamma oscillations (Anver et al., 2011; Ferando and Mody, 2013; Mann and22

Mody,2010).23

Thesamechangesinthestimulationpropertiesthatinducealinearincreaseingamma24

frequency leadtononlinear ---oftenbell-shaped---changes ingammapower inmonkeyLFP25

(Hadjipapasetal.,2015;Jiaetal.,2011;Jiaetal.,2013;Lowetetal.,2015;Salelkaretal.,2018).26

Ina recentMEGstudy (Orekhovaetal., 2018b),weappliedhigh-contrast gratingsdriftingat27

different rates (motion velocities) and found the same bell-shaped changes in the gamma28

responsepowerasthosedescribedintheLFPstudiesonmonkeys.Themajorityofourhealthy29

subjects increasedvisualGRpowerfromthestatictotheslowlymoving(1.2°/s)stimulusand30

then suppressed the response at higher velocities of 3.6 °/s and 6.0 °/s. These findings are31

consistentwiththoseofvanPeltandco-authors(vanPeltetal.,2018),whofoundacontinuous32

growthinthestrengthofvisualgammaresponsewhilechanginggratingvelocitiesfrom0°/sto330.33°/s and further to 0.66 °/s. Given that the fastest velocity used in their study was still34



https://doi.org/10.1101/497214


4

below 1.2°/s, the combined results from both studies are consistent with our previous1

observation that the effect of stimulus velocity on induced gamma power changed from a2

facilitative to a suppressive one at a stimulus speed of about 1.2°/s (‘suppression transition3

velocity’)(Orekhovaetal.,2018b).4

Seeminglyparadoxically,thenonlinearvelocity-relatedgammapowerchangesmaybe5

at least partly explained by rising neural excitation --- in particular, the tonic excitation of I-6

neurons. According to modeling studies, the excitation of I-neurons beyond some critical7‘suppressiontransition’pointresults intheir lossofsynchronywitheachotherandexcitatory8

pyramidal cells that, in turn, leads to suppressionof gammaoscillations (BorgersandKopell,9

2005; Cannon et al., 2014; Kopell et al., 2010). Thus, the gamma suppression at high10

velocities/temporal-frequencies of visualmotionmight be explained by this ‘excitatory drive11

hypothesis’.12

Iftrue,thelinkbetweenstronginhibitionandgammasuppressionmayhaveimportant13

implications for studying the E and I processes in the healthy and diseased human brain. In14

particular,individualvariationsingammasuppressionmayreflectthecapacityofIcircuitryto15

suppressexcessiveexcitationinthenetwork,thuspreventingsensoryoverload.However,the16

question remained to be answered –whether the observed nonlinear behavior of theMEG17

gammaoscillations is related tochanges in the strengthofexcitatorydriveor toaparticular18

property of the visual stimulation. Indeed, rather than being a consequence of the strong19

excitatorydrive to the visual cortex, the suppressionof gamma responsepower could result20from tuningof theV1neurons to theparticular velocity/temporal frequency.Animal studies21

showthat,whilecellsintheLGNrespondtovisualstimuliofratherhightemporalfrequencies,22

theneurons in theprimaryvisual cortex showband-passproperties (Priebeetal.,2006;Van23

Hooseretal.,2013).Onethereforecannotexcludethatapresenceofan'optimalvelocity'that24

induces themaximal visualGRpower simply reflects thenumberofV1neurons ‘tuned’ toa25

respectivevelocity/temporalfrequency.26

Herein,andinordertotestthesetwoalternativeexplanations,wemodulatedtheGR27

bychangingthevelocityofdriftinggratings(stationaryandmovingat1.2,3.6and6.0°/sthat28

correspondedtotemporalfrequenciesof0,2,6and10Hz)fortwoluminancecontrastvalues:29

100%(highcontrast)and50%(lowcontrast).30

Intermsoftheexcitatorydrivemodel,concomitant increasesincontrastandvelocity31

of the stimulation should have an additive effect. If the transition from enhancement to32

suppressionoftheGRrequiresacertainlevelofexcitatorydrive,thenthislevelwillbereached33at a relatively higher motion velocity in case of lower contrast. The excitatory drive model34



https://doi.org/10.1101/497214


5

therefore predicts that the ‘gamma suppression transition velocity’ would be higher at the1

50%,ascomparedtothe100%,contrast.2

The velocity-tuning model predicts a different outcome from that of the excitatory3

drive model. In this case, lowering contrast from 100% to 50% either will not affect the4

'transitionpoint’or shift it to lowervelocity.The shift toa lowervelocity in this casecanbe5

expected,becauseadecreaseinthestimuluscontrastshiftsthetuningofthemotion-sensitive6

V1neurons to lower temporal frequencies (Alitto andUsrey, 2004; Livingstone andConway,72007;Priebeetal.,2006).8

Similar to gamma, the oscillations within the alpha-beta range are also thought to9

reflect the level of cortical activation (Romei et al., 2008; Yuan et al., 2010). Whereas10

suppressionofthealpha-betaoscillationshasrepeatedlybeenshowntoaccompanyanactive11

state of cortical networks related to processing of incoming sensory information, the ‘high-12

alpha’statesareassociatedwithcorticalidlingoractiveinhibitionoftask-irrelevantprocesses13

thatmightotherwiseinterferewithtaskperformance(PalvaandPalva,2011;Pfurtschellerand14

daSilva,1999).Usingthevisualmotionparadigmsimilartothatusedinourstudy,Scheeringa15

et al have shown that the power of alpha-beta and gamma oscillations was inversely16

proportionaltochangesinthebloodoxygenlevel-dependent(BOLD)signalinthevisualcortex,17

althoughatdifferentcorticaldepths(Scheeringaetal.,2016).However,thepowersofalpha-18

beta and gamma oscillations are not interchangeable indexes of cortical activation. In19

particular, changes in the velocity of visual motion affected alpha and gamma oscillations20differently in theLFP inmonkeys (Salelkaretal.,2018).There isevidence thatvisualgamma21

mainlyreflectsfeed-forwardprocesses(Robertsetal.,2013;vanKerkoerleetal.,2014),while22

alphaandbetapowerinthevisualcortexarestronglymodulatedbytop-downinfluencesfrom23

upstreamcorticalareas(Liuetal.,2016;SnyderandFoxe,2010).Therefore,weexpectedthat24

the bell-shaped pattern of power changes would be specific for theMEG gamma band and25

would not be mirrored by opposing changes in alpha-beta activity (i.e. gamma response26

suppression–alphafacilitation).27

Overall, our current MEG study pursued these two main goals. Firstly, we tested28

whether the strength of excitatory drive is the main factor contributing to the initially29

facilitative, and then suppressive, effects of velocity on the visual gamma response power.30

Secondly, by comparing the effects of increasing excitatory drive on gamma and alpha31

oscillations,wesoughttounraveltheuniqueinformationregardingactivationofthevisualthat32

isconveyedbyeachofthethesetwofrequencybands.3334



https://doi.org/10.1101/497214


6

1

Methods2

3

Participants.4

Seventeen neurologically healthy subjects (age 18-39, mean=27.2, sd=6.1; 7 males) were5

recruitedtothestudy.Allparticipantshadnormalorcorrectedtonormalvision.Theinformed6consent form was obtained from all the participants. The study has been approved by the7

ethicalcommitteeofMSUPE.8

9

Experimentaltask.10

ThevisualstimuliweregeneratedusingPresentationsoftware(NeurobehavioralSystemsInc.,11

USA). We used a PT-D7700E-K DLP projector to present images with a 1280 x 1024 screen12

resolutionanda60Hzrefreshrate.Theexperimentalparadigmisschematicallypresentedin13

figure 1. The stimuli were gray-scale sinusoidally modulated annular gratings presented at14

~100% or 50% of Michelson contrast. Luminance of the display, measured from the eye15

position, was 46 Lux during presentation of both 50% and 100% contrast stimuli and 2 Lux16

duringinter-stimulusintervals.Thegratingshadaspatialfrequencyof1.66cyclesperdegreeof17

visualangleandcovered18×18degreesofvisualangle. Theyappearedinthecenterofthe18

screenoverablackbackgroundanddriftedtothecenterwithoneoffourvelocities:0,1.2,3.6,19or6.0°/s, (0,2,6and10Hztemporal frequency)referredtoas ‘static’, ‘slow’, ‘medium’,and20

‘fast’.Eachtrialbeganwith1200mspresentationofafixationcrossinthecenterofthedisplay21

over a black background that was followed by the presentation of the grating that either22

remainedstaticordriftedwithoneofthethreevelocities.Afterarandomlyselectedperiodof23

1200–3000ms,themovementstoppedorthestaticstimulusdisappeared.Tokeepparticipants24

alert,weaskedthemtorespondtothechangeinthestimulationflow(stopofthemotionor25

disappearance of the static stimulus) with a button press. If no response occurred within 126

second,adiscouragingmessage,“toolate!”appearedandremainedonthescreenfor2000ms,27

afterwhich a new trial began. The static stimulus and those drifting at different rateswere28

intermixed and appeared in a random order within each of three experimental blocks. The29

participants responded with either the right or the left hand in a sequence that was30

counterbalancedbetweenblocksandparticipants.Eachofthestimulitypeswaspresented3031

timeswithineachexperimentalblock. Inordertominimizevisualfatigueandboredom,short32(3–6 s) animated cartoon characters were presented between every 2–5 stimuli. The high33



https://doi.org/10.1101/497214


7

(100%)andlow(50%)contrastgratingswerepresentedindifferentexperimentalsessionsand1

theorderofthesesessionswascounterbalancedbetweensubjects.2

3

4

Fig. 1.Experimentaldesign.Eachtrialbeganwiththepresentationofafixationcross that5wasfollowedbyanannulargratingthatremainedstatic(0°/s)ordriftedinwardfor1.2-3s6at one of the three velocities: 1.2, 3.6, 6.0°/s. The 100% and 50% contrast gratingswere7presented in separate sessions. Participants responded to the change in the stimulation8flow (disappearance of the static grating or termination of motion) with a button press.9Shortanimatedcartooncharacterswerepresentedrandomlybetweenevery2-5stimulito10maintainvigilanceandreducevisualfatigue.11

12

Datarecording.13

Neuromagnetic brain activity was recorded using a 306-channel detector array (Vectorview;14

Neuromag,Helsinki,Finland).Thesubjects’headpositionswerecontinuouslymonitoredduring15MEGrecordings.Fourelectro-oculogram(EOG)electrodeswereusedtorecordhorizontaland16

verticaleyemovements.EOGelectrodeswereplacedattheoutercantioftheeyesandabove17

andbelowthelefteye.Tomonitortheheartbeats,oneelectrocardiogram(ECG)electrodewas18

placedatthemanubriumsterniandtheotheroneatthemid-axillaryline(V6ECGlead).MEG,19

+

time

baseline1.2 s

drifting or static grating

1.2-3 s

+

+

rest 3-6 s

+

time

baseline1.2 s

+

+

rest 3-6 s

100% contrast

50% contrast

drifting or static grating

1.2-3 s



https://doi.org/10.1101/497214


8

EOG,andECGsignalswererecordedwithaband-passfilterof0.03–330Hz,digitizedat10001

Hz,andstoredforoff-lineanalysis.2

3

MEGdatapreprocessing.4

Thedatawasfirstde-noisedusingtheTemporalSignal-SpaceSeparation(tSSS)method(Taulu5

andHari,2009)implementedinMaxFilter™(v2.2)withparameters:‘-st’=4and‘-corr’=0.90.For6

all three experimental blocks, the head origin positionwas adjusted to a common standard7position.Forfurtherpre-processing,weusedtheMNE-pythontoolbox(Gramfortetal.,2013)8

aswellascustomPythonandMATLAB®(TheMathWorks,Natick,MA)scripts.9

Thede-noiseddatawasfilteredbetween1and200Hzanddownsampledto500Hz.10

To remove biological artifacts (blinks, heart beats, and in some casesmyogenic activity),we11

then applied independent component analysis (ICA). TheMEG periods with too high (4e-1012

fT/cm for gradiometers and 4e-12 fT for magnetometers) or too low (1e-13 fT/cm for13

gradiometers and1e-13 fT formagnetometers) amplitudeswereexcluded from theanalysis.14

The number of independent components was set to the dimensionality of the raw ‘SSS-15

processed’data(usuallyaround70).WefurtherusedanautomatedMNE-pythonprocedureto16

detectEOGandECGcomponents,whichwecomplementedwithvisual inspectionof the ICA17

results. The number of rejected artifact components was usually 1-2 for vertical eye18

movements,0-3forcardiac,and0-6formyogenicartifacts.19

The ICA-correcteddatawas then filteredwithadiscreteFourier transform filter to20remove power-line noise (50 and 100 Hz) and epoched from −1 to 1.2 sec relative to the21

stimulusonset.Wethenperformedtime-frequencymultitaperanalysis(2.5Hzstep;numberof22

cycles = frequency/2.5) and excluded epochs contaminated by strong muscle artifacts via23

thresholding the high-frequency (70 - 122.5 Hz) power. For each epoch, the 70 - 122.5 Hz24

powerwas averaged over sensors and time points and the thresholdwas set at 3 standard25

deviationsofthisvalue.Theepochsleftwerevisuallyinspectedforthepresenceofundetected26

high-amplitude bursts of myogenic activity and those contaminated by such artifacts were27

manually marked and excluded from the analysis. After rejection of artifacts the average28

number of epochs for the 100% contrast condition was 79, 78, 78, and 79 for the ‘static’,29

‘slow’,‘medium’and‘fast’conditions,respectively.Forthe50%contrast,therespectivevalues30

were79,80,80,and82.31

32

StructuralMRI33



https://doi.org/10.1101/497214


9

Structural brain MRIs (1 mm3 T1-weighted) were obtained for all participants and used for1

sourcereconstruction.2

3

Time-frequencyanalysisoftheMEGdata4

The following steps of the data analyses were performed using Fieldtrip Toolbox functions5

(http://fieldtrip.fcdonders.nl; (Oostenveld et al., 2011)) and custom scripts developedwithin6

MATLAB.7Wesubtractedtheaveragedevokedresponses fromeachdataepoch.Thisdecreases8

thecontributionofphase-lockedactivitythatcouldberelatedtotheappearanceofthevisual9

stimulusonthescreenandtheeffectofphoticdrivingrelatedtothetemporalfrequencyofthe10

stimulation,screenrefreshrate,ortheirinteraction.11

The lead field was calculated using individual 'single shell' headmodels and cubic 612

mm-spaced grids linearly warped to theMNI-atlas-based template grid. The time-frequency13

analysisoftheMEGdatawasthenperformedinthefollowingtwosteps.14

First,we tested for the presence of reliable clusters of gamma facilitation and alpha15

suppressionat the source levelusing theDICS inverse-solutionalgorithm (Grosset al., 2001)16

andthecluster-basedpermutationtest(NicholsandHolmes,2002).Thefrequencyofinterest17

wasdefinedindividuallyforeachsubjectandconditionbasedonthesensordata(onlythedata18

from gradiometers were used at this stage). For the gamma range, we performed time-19

frequencydecompositionofpre-stimulus(‘pre’:-0.9to0s)andpost-stimulus(‘post’:0.3to1,220s)MEGsignalsusing8discreteprolatespheroidalsequences(DPSS)taperswith±5Hzspectral21

smoothing.Forthelowfrequencyrange,weused2DPSStapersand±2Hzspectralsmoothing.22

Wethencalculatedtheaverage(post-pre)/preratioandfoundthe4posteriorsensorswiththe23

maximalpost-stimulusincreaseingammapower(45-90Hz)andthosewiththemaximalpost-24

stimuluspowerdecreaseinthelow-frequencyband(7-15Hz).Byaveragingthose4channels,25

we have found the frequencies corresponding to themaximal post-stimulus power increase26

(forgamma)ordecrease(forlowfrequencyrange).Thefrequencyrangesofinterestwerethen27

established within 35-110 Hz and 5-20 Hz limits, where the ratios exceeded 2/3 of the28

respective peak value. The center of gravity of the power over these frequency ranges was29

used as the peak frequency of the gamma and the slow responses, respectively. The time-30

frequency decomposition was then repeated while centered at these peak frequencies.31

Common source analysis filters were derived for combined pre- and post-stimulus intervals32

usingDICSbeamformingwitha5%lambdaparameterandfixeddipoleorientation(i.e,onlythe33largest of the three dipole directions per spatial filterwas kept). The filterwas then applied34



https://doi.org/10.1101/497214


10

separately to the pre- and post-stimulus signals. Subsequently, we calculated univariate1

probabilities of pre- topost-stimulusdifferences in single trial power for each voxel using T-2

statistics. We then performed bootstrap resampling (with 10 000 Monte Carlo repetitions)3

between ‘pre’ and ‘post’ timewindows todetermine individual participants'maximal source4

statisticsbasedonthesumofT-valuesineachcluster.5

Second,inordertoanalyzetheresponseparametersatthesourcemaximum,weused6

linearly constrained minimum variance (LCMV) beamformers (VanVeen et al., 1997). The7spatialfilterswerecomputedbasedonthecovariancematrixobtainedfromthewholeepoch8

and for the three experimental conditions with a lambda parameter of 5%. Prior to9

beamforming,theMEGsignalwasband-passedbetween30and120Hzforthegamma-range10

andlow-passedbelow40Hzforthelowfrequencyrange.The‘virtualsensors'timeserieswere11

extracted for each brain voxel and time-frequency analysis with DPSS multitapers (~1 Hz12

frequencyresolution)wasperformedonthevirtual sensorsignals.For thegammarange,we13

used±5Hzsmoothing.Forthelow-frequencyrange,thisparameterwas±2Hz.The‘maximally14

induced’gammavoxelwasdefinedasthevoxelwiththehighestrelativepost-stimulusincrease15

of45-90Hzpower inthe ‘slow’motionvelocityconditionwithinthevisualcorticalareas(i.e.16

L/R cuneus, lingual, occipital superior,middle occipital, inferior occipital, and calcarine areas17

according to theAALatlas (Tzourio-Mazoyeretal., 2002)). The ‘slow’ conditionwas selected18

because reliability of the gamma response was highest in this condition. The ‘maximally19

suppressed’voxelinthelow-frequencyrangewasthendefinedasthevoxelwiththestrongest20relativesuppressionof7-15Hzpower.Theweightedpeakparametersforthegammaandthe21

low-frequency activity were calculated for the average spectra of the virtual sensors in 2522

voxels closest to, and including, the ‘maximally induced’ and ‘maximally suppressed’ voxels,23

respectively,usingtheapproachdescribedforthesensorlevelanalysis.Foreachindividualand24

condition,wealsoassessedthereliabilityofthepre-topost-stimuluspowerchangesatthese25

25 averaged voxels. The change was considered reliable if its absolute peak value was26

significantwithp<0.0001(Wilcoxonsignedranktest).Coordinatesofthevoxelsdemonstrating27

greatestpowerchangesweredefinedinMNIcoordinates(SupplementaryTables).28

29

30

Results31

32

WemeasuredMEGbrain responses induced by concentric visual grating drifting inwardly at33

fourvelocities(0,1.2,3.6,6.0°/s)andtwocontrastlevels(50%and100%).Belowwepresent34



https://doi.org/10.1101/497214


11

theresultsseparatelyforthegammaandthealpha-betaranges.1

2

Gammafrequencyrange3

Localizationandreliability4

Figure2showsgrandaveragesourcelocalizationoftheinducedgammaresponses.5

6

7Fig. 2. Grand average statistical maps of the cortical GRs to drifting visual8gratings.Themapsaregivenfortheweightedpeakgammapower,separatelyfor9the two luminance contrasts and fourmotion velocities. Positive sign of the T-10statistics corresponds to stimulation-related increase in gamma power. Note11different scales for the50%and100%contrastsand that themagnitudeof the12gammaresponsewasaffectedbybothcontrastandvelocity.13

14

For the 100% contrast stimuli, significant (p<0.05) activation clusterswithmaxima in15

visualcorticalareaswerefoundinallparticipantsandmotionvelocityconditions.Forthe50%16

contrast, the significant activation clusters were absent in one participant in all velocity17

conditions, in two participants in the static condition, and in one participant in the 6.0°/s18

condition.19

Thegrandaverageandindividualgammaresponsespectraatthe‘maximallyinduced’20

groupofvoxelsareshown in figures3and4.TheparticipantsA007,A008,andA016hadno21reliable increase in gammapower (p>0.0001, seeMethods for details) at 50% contrast in at22

-10

10

0

100%

con

tras

t

0°/s 1.2°/s 3.6°/s 6.0°/s

50%

con

tras

t

-8

8

0

T-statistic

T-statistic



https://doi.org/10.1101/497214


12

least one of the velocity conditions (A007: all velocities; A008 and A016: 0°/s). In all other1

participants,theincreasesingammapowerattheselectedvoxelsweresignificantaccordingto2

ourcriteriainallconditions.3

Thepeaksthatwereunreliableaccordingtoatleastoneofthecriteria(i.e.eitherthe4

absence of an activated cluster or low probability (p>0.0001) of gamma increase at the5

selectedvoxels)werefurtherexcludedfromanalysesofthepeakfrequencyandthepositionof6

themaximallyinducedvoxel.Notethatexclusionoftheunreliablegammapeaksaffectedthe7degreesoffreedominthecorrespondingANOVAspresentedbelow.8

9

10

11

12

1314

Fig. 3. Grand average spectra of cortical GRs to the gratings of 100% (left15panel)and50%(rightpanel)contrastsdriftingatfourmotionvelocities.Here16andhereafter,theGRpoweriscalculatedas(post-pre)/preratio,where‘pre’17and‘post’arethespectralpowervaluesinthe-0.9to0and0.3to1.2stime18windowsrelativetothestimulusonset.19

20

21

30 50 70 90 0

1

2

3

30 50 70 90

0°/s1.2°/s3.6°/s6.0°/s

Frequency [Hz] Frequency [Hz]

Powe

r cha

nge

[a.u

.]

100% contrast 50% contrast



https://doi.org/10.1101/497214


13

1

23

Fig.4.IndividualspectraofcorticalGRs.TheplotsarethesameasinFig.3,seeMethodsfor4furtherdetails.5

6

7

In themajority of cases, the voxel with themaximal increase in gamma power was8

locatedinthecalcarinesulcus(seetheSupplementaryTable1Aand1Bforcoordinatesofthe9‘maximally induced’ voxel). There was a significant effect of Velocity for the ‘z’ coordinates10

(F(3,39)=4.7, e=0.84, p<0.05). In the ‘fast’ condition, the voxel with the maximal increase in11

100% 50%

Frequency [Hz]

30 50 70 900246

Powe

r cha

nge

[a.u

.] case A001

30 50 70 90

0°/s1.2°/s3.6°/s6.0°/s

30 50 70 900246

case A002

30 50 70 90Powe

r cha

nge

30 50 70 9002468

case A003

30 50 70 90Powe

r cha

nge

30 50 70 900

2

4case A004

30 50 70 90Powe

r cha

nge

30 50 70 900

1

2case A005

30 50 70 90

Powe

r cha

nge

30 50 70 900123

case A006

30 50 70 90

Powe

r cha

nge

30 50 70 900

1

2case A013

30 50 70 90

30 50 70 900246

case A014

30 50 70 90

30 50 70 90024

case A015

30 50 70 90

30 50 70 90

0

1 case A016

30 50 70 90

30 50 70 900

2

4case A017

30 50 70 90

30 50 70 90

0

1 case A007

30 50 70 90

30 50 70 900123

case A008

30 50 70 90

30 50 70 900

2

4case A009

30 50 70 90

30 50 70 900

2

case A010

30 50 70 90

30 50 70 900

2

4case A011

30 50 70 90

30 50 70 900246 case A012

30 50 70 90Frequency [Hz]

4

100% 50% 100% 50%



https://doi.org/10.1101/497214


14

gammapowerwaslocatedonaverage2.6mmmoresuperiorrelativetotheother3conditions1

(Zcoordinateforthestatic:0.61cm,slow:0.54cm,medium:0.52cm,fast:0.82cm).Therewas2

also Velocity*Contrast interaction for the ‘y’ coordinate (F(3,39)=3.6, e=0.59, p<0.05), that is3

explainedbyarelativelymoreposteriorsource locationof thegammamaximuminthe ‘fast’4

low-contrastconditionthan inthe ‘fast’high-contrastcondition(-9.5vs-9.2cm).Considering5

thesmallcondition-relateddifferencesinthepositionofthevoxelwiththemaximalincreasein6

gammapower(<0.6cm,whichwasthesizeofthevoxelusedforthesourceanalysis),all the7stimuliactivatedlargelyoverlappingpartsoftheprimaryvisualcortex.8

9

Effectsofcontrastandvelocity10

Frequency. Figure5Ashowsgroupmeangammapeakfrequencyvalues for thetwocontrast11

andfourvelocityconditions.Gammafrequencywasstronglyaffectedbymotionvelocityofthe12

grating(F(3,39)=78.2,e=0.62p<1e-6)and,toalesserextent,byitscontrast(F(1,13)=14.3,p<0.01).13

Inspection of figure 5A shows that increase in velocity resulted in substantial increase of14

gammafrequency(~12-17Hz).Forthefull-contrastgratingthefrequencyincreasedfrom51.615

Hzincaseofthestaticto68.6Hzincaseofthefastmovingstimulus.Forthe50%contrastthe16

respectivevalueswere51.1and63.0Hz. Increasingcontrast,on theotherhand, led to5Hz17

increaseinfrequencyofgammaresponsetomovingstimuli.18

19

2021

Fig.5. Grandaveragepeak frequency (A) andpower (B)ofcorticalGRsasa22functionofContrastandmotionVelocity.Verticalbarsdenote0.95confidence23

intervals.24

25

Power. Figure 5B shows group mean GR peak power values for the two contrast and four26

velocity conditions. Therewas highly significant effect of Contrast (F(1,16)=66.2,p<1e-6) onGR27



https://doi.org/10.1101/497214


15

peak power, which is explained by a generally higher gamma power in case of high, as1

comparedtolowcontrast.ThesignificanteffectofVelocity(F(3,48)=9.3,e=0.46,p<0.01)isdueto2

aninitialincreaseinGRpowerfromthestatictoslowcondition(F(1,16)=43.1,p<1e-5)followed3

by a decrease from the medium to the fast velocity condition (F(1,16)=14.2, p<1e-4).4

Importantly, there was highly significant Contrast*Velocity interaction effect (F(3,48)=12.4,5

e=0.78, p<1e-4). The contrast had a stronger effect on GR power for the static and slowly6

moving stimuli, as compared to the faster (medium and fast) velocities. As a result, the7maximumoftheinvertedbell-shapedpowervs.motionvelocitycurvemovedtohighervelocity8

inthelowcontrastcondition,ascomparedtothefullcontrastone.9

Generally, the results confirmed our previous finding of an inverted bell-shaped10

dependencyofGRpoweronthevelocityofvisualmotionandextendedthemtotwocontrast11

conditions. The significant Contrast*Velocity interaction showed that the form of this bell-12

shapedcurvewasmodulatedbytheluminancecontrastinsuchawaythatthetransitiontoGR13

suppressionatlowcontrastoccurredatahighervelocity,aspredictedbytheexcitatorydrive14

model.15

16

Suppression transition velocity.We further sought to investigatewhether the shift to higher17

velocities intheGRsuppressiontransitionpointunder lowcontrast isarobustphenomenon,18

which can be detected at the individual subject level. For each subject, we calculated the19

‘suppressiontransitionvelocity’(STVel)asthecentreofgravityvisualmotionvelocity:2021

STVel=(Pow0*0+Pow1.2*1.2+Pow3.6*3.6+Pow6.0*6.0)/(Pow0+Pow1.2+Pow3.6+Pow6.0),22

23

where‘Pow’istheweightedGRpeakpowerintherespectivevelocitycondition(i.e.0,1.2,3.624

or 6.0 °/s). Although the STVel may not precisely correspond to the velocity condition for25

which the GR was at maximum, it reflects the distribution of power between velocity26

conditions.Moreimportantly,itallowsustocharacterizecontrast-dependentshiftsinamore27

rigorousmannerthanrelyingexclusivelyonthediscretevaluesofmotionvelocitiesforwhich28

theGRwasmaximal.DecreasingluminancecontrastresultedinareliableincreaseinSTVel,and29

thustheGRsuppressiontransitionpoint,inall17participants(F(1,16)=202.0;p<1e-6;Fig.5A).30

Lowering contrast led to slowing of gamma oscillations (Fig. 4A); we therefore also31

tested if, irrespectiveof contrast, thepointof transition toGR suppression corresponds toa32

certaingamma frequency.Todo this,weapproximated the suppression transition frequency33(STFreq,i.e.thegammapeakfrequencythatcorrespondstothemaximalgammaresponse)as34



https://doi.org/10.1101/497214


16

thecenterofgravityoftheGRpowerforeachsubject(i.e.,inthesamewayaswedidforthe1

STVel):2

3STFreq=(Pow0*Freq0+Pow1.2*Freq1.2+Pow3.6*Freq3.6+Pow6.0*Freq6.0)/(Pow0+Pow1.2+Pow3.6+Pow6.0),45

where‘Freq’and‘Pow’arethepeakgammaresponsefrequencyandpowerintherespective6

velocityconditions.7

Thesuppressiontransitionfrequencywasslightly,butsignificantly, lower forthe50%8

thanforthe100%contrast(60.5Hzvs63.1Hz;F(1,13)=10.2,p<0.01).9

10

11Fig.6.Effectofcontrastonthegammasuppressiontransitionpoint:individualvariability.A.12The centre of gravity of visual motion velocity that approximates the GR suppression13transitionpoint (‘suppression transition velocity’ - STVel). In every single subject, the STVel14increasedwithdecreasingcontrast. B.Thecentreofgravityofgammapeakfrequencythat15approximates the peak frequency of the GR at the gamma suppression transition point16(‘suppressiontransitionfrequency’-STFreq).TheSTFreqslightly,butsignificantly,decreased17withdecreasingthecontrast.18

19

Rank-orderconsistencyofgammaparameters20

Considering the drastic stimulation-related changes in gamma parameters, their intra-21

individual stability across experimental conditions (i.e. ‘rank-order consistency’) remained an22

open question. In order to investigate if the power and frequency of GR measured in one23

experimentalconditionpredictsthesubject’srankpositioninanotherexperimentalcondition,24

we calculated two types of correlations: 1) between contrasts for the respective velocity25

conditionsand2)betweenvelocities,separatelyforeachofthecontrasts.Sincefrequency,but26

Suppression transition velocity

Contrast100% 50%

1

2

3

4

Mot

ion

velo

city

[°/s

]

Suppression transition frequency

Contrast

80

100% 50%40

50

60

70

Peak

freq

uenc

y [H

z]

A B



https://doi.org/10.1101/497214


17

notpower,ofthevisualgammaoscillations isstronglyaffectedbyageinadults (Gaetzetal.,1

2012; Orekhova et al., 2018b), we calculated partial correlations taking age as a nuisance2

variablewithfrequency.3

The correlations between contrast conditions were generally very high for gamma4

frequency (all R’s>0.89; all p’s <0.0001),with the noticeable exception of the static stimulus5

(R=0.36,n.s.).Forpower,allthebetween-contrastcorrelationsweremodestlytohighlyreliable6

(Fig.7A).ThecorrelationbetweenSTVelvaluesmeasuredat the50%and100%contrastwas7high (Fig. 6B, STVel: R(17)=0.91, p<1e-6), suggesting that thismeasure reliably characterises a8

subject’srankpositioninthegroup,irrespectiveofcontrastchanges.9

The correlations between velocity conditions were also generally high for frequency10

(SupplementaryTable3),butvaried incaseofpower.Thepowerof theGRelicitedby static11

stimulididnotpredictthepoweroftheGRelicitedbygratingsmovingwith‘medium’of‘fast’12

velocities (Fig.7B).Theresultswereverysimilar forthetwocontrasts (SupplementaryFigure13

1).14

Tosumup,rank-orderconsistencyforgammafrequencywasgenerallyhighacrossall15

conditionsexceptforstationarystimuli.Forgammapower,arank-orderconsistencybetween16

stationaryandmedium-to-fastmovinggratingswaslacking.17



https://doi.org/10.1101/497214


18

12Fig. 7.Rank-orderconsistencyof theGRpoweracrossconditions. A.CorrelationsbetweenGRpower3measuredat50%and100%contrasts.B.CorrelationsbetweenGRpowervaluesmeasuredindifferent4velocity conditionsat the100%contrast (seeSupplementary Figure1 for a similar figure for the50%5contraststimuli).BluedotsinAandBdenoteindividualgammaresponsepowervalues.C.Correlations6betweenSTVelvaluesatthetwocontrasts.Greensquaresdenote individualSTVelvalues. The linear7regressionisshowninred.Dashedlinesinallplotscorrespondtotheaxisofsymmetry.8

1 3 5100%, 0°/s

0,0

0,5

1,0

50%

, 0°/s

R=0.7**0 2 4 6

100%, 1.2°/s

1

2

3

4

50%

, 1.2

°/s R=0.64**

0 2 4 6 8100%, 3.6°/s

2

4

6

50%

, 3.6

°/s

R=0.87***0 2 4 6

100%, 6.0°/s

1234

50%

, 6.0

°/s

R=0.96***

R=0.51*0 2 4

0°/s

2

4

6

6.0°

/s

0 2 4 0°/s

2

4

6

83.

6°/s

0 2 4 0°/s

2

4

6

1.2

°/s

0 2 4 6 1.2°/s

2

4

6

8

3.6

°/s

0 2 4 6 1.2°/s

2

4

6 6

.0°/

s

0 2 4 6 8 3.6°/s

2

4

6

6.0

°/s

R=0.51*

R=0.62*

R=0.97***

R=0.0 R=-0.09

R=0.51*

1 2 3 4 5STVel 100%-contrast

2

3

4

5

STVe

l 50%

-con

trast

A

B

C

R=0.91***



https://doi.org/10.1101/497214


19

1

2

Alpha-betafrequencyrange3

Localizationandreliability4

Figure8showsgrandaveragesourcelocalizationofthealpha-betaresponsetothemoving5

visualgratings.6

7

89

Fig. 8. Grand average power changes in the alpha-beta range. A. Statisticalmaps of the10stimulation-related changes in the two luminance contrasts and four motion velocity11conditions. Blue color corresponds to suppression of the alpha-beta power. B. Spectral12changesataselectionof‘maximallysuppressed’voxels.Dashedlinesshowabsolutepower13spectra(righty-axis,arbitraryunits)correspondingtotheperiodsofvisualstimulationwith14movinggratings.Thedot-dashedlinesofthesamecolorsshowthepowerspectraforthe15respectivepre-stimulusintervals(alsorighty-axis).Solidlinesshowrespectivestimulation-16relatedchangesinpower([post-pre]/preratio,lefty-axis,arbitraryunits).Neithercontrast17norvelocityhadsignificanteffectsonthemagnitudeofthealpha-betasuppression.18

19

20

Clustersofsignificant(p<0.05)alpha-betapowersuppressionwerefoundinnearlyall21

subjectsandconditions.Forthe100%contraststimuli,theclusterswereabsentinonesubject22

inallvelocityconditionsandinonsubjectinthestaticcondition.Thesuppressionmeasuredat23

the selection of the ‘maximally suppressed’ voxels was significant in all motion velocity24

-6

0

100%

con

tras

t

0°/s 1.2°/s 3.6°/s 6.0°/s

50%

con

tras

t

6T-statistic

5 10 15 20-0.7

-0.5

-0.3

-0.1

1

3

5

710-12100% contrast

Powe

r cha

nge

[a.u

.]

Frequency [Hz]

LCMV power [a.u.]

0°/s1.2°/s3.6°/s6.0°/s

5 10 15 20-0.7

-0.5

-0.3

-0.1

1

3

5

710-1250% contrast

Powe

r cha

nge

[a.u

.]

Frequency [Hz]

LCMV power [a.u.]

A B



https://doi.org/10.1101/497214


20

conditionsat100%contrast. Forthe50%contraststimuli,asignificantclusterforalpha-beta1

suppressionwasabsentinonlyonesubjectandonlyinresponsetothestaticstimulus.Inyet2

another subject, the alpha-beta suppression measured at the voxels’ selection was not3

significant (p>0.0001) for the 1.2 °/s condition. The grand average spectra in the alpha-beta4

rangeforthemaximallysuppressedvoxelsareshowninfigure8.5

ANOVAwithfactorsContrastandVelocityrevealedneithersignificantmaineffectsof6

ContrastandVelocitynorContrast*Velocityinteractionforthe‘x’,‘y’,or‘z’coordinatesofthe7‘maximallysuppressed’voxel(allp’s>0.15).8

WethenusedANOVAwithfactorsBand,Contrast,andVelocitytocomparepositions9

of thevoxelswithmaximalalpha-betasuppressionand thosewithmaximalgamma increase.10

The effect of Band was highly significant for the absolute value of the ‘x’ coordinate11

(F(1,16)=81.9,p<1e-6).Themaximallysuppressedalphavoxelwaspositionedsubstantiallymore12

laterally(onaverage25mmfromthemidline)thanthatofthemaximallyinducedgammavoxel13

(on average 8mm from themidline). As can be seen from Supplementary Table 2(A,B), the14

maximally suppressed alpha voxel wasmore frequently located in the lateral surface of the15

occipitallobethanincalcarinesulcusorcuneusregion.NoBandrelateddifferencesintheyor16

zcoordinateswerefound.17

18

Effectsofcontrastandvelocity19

Tocheckifthemagnitudeorpeakfrequencyofthealpha-betasuppressionresponsedepends20onthepropertiesofthestimulation,weperformedANOVAwithfactorsContrastandVelocity.21

22

Power.TherewasnosignificanteffectsofContrastorVelocityortheirinteractioneffectforthe23

alpha-betasuppressionmagnitude(allp’s>0.5).24

25

Frequency. The peak frequency of the alpha-beta suppression sightly, but significantly,26

increasedwithincreasingmotionvelocity(Velocity:F(3,36)=7.1,e=0.57,p<0.01;11.8Hzfor0°/s,27

11.9Hzfor1.2°/s,12.0Hzfor3.6°/s,12.2Hzfor6.0°/s).28

29

30

Discussion31

32Wetestedforanadditiveeffectofcontrastandvelocityofdriftingvisualgratingsonthepower33

of induced MEG gamma oscillations. For both full-contrast and low-contrast gratings, the34



https://doi.org/10.1101/497214


21

power of gamma response initially increased with increasing motion velocity and was then1

suppressedatyethighervelocities.Mostimportantly,wefoundthatloweringcontrastledtoa2

highlyreliableshiftinthegammasuppressiontransitionpointtohighervelocities.Thisfinding3

provides an experimental proof for the crucial role of intensive excitatory input in GR4

suppression. The systematic relationship between the intensity of the visual input and the5

strengthoftheneuralresponsewasclearforgammaoscillationsandabsentinthealpha-beta6

frequencyrange.78

Theroleofexcitatorydriveininducedgammaresponse9

In linewith theprevious studies (Jiaetal.,2013;Perryetal.,2015;RayandMaunsell,2010;10

Roberts et al., 2013; van Pelt et al., 2018), we found that the peak frequency of the visual11

gamma response increased with increasing luminance contrast (Fig. 4A). As an increase in12

luminancecontrastcausesariseofneuronalspikingrate(Bartoloetal.,2011;Hadjipapasetal.,13

2015; Ray and Maunsell, 2010), contrast can be considered as a proxy for excitatory drive14

(Lowet et al., 2015). Therefore, contrast-dependent changes in gamma frequency and15

amplitudecanbeattributedtochangesinexcitatoryinputtothevisualcortex.16

Also in accordance with our previous results—and those of others—17

(Muthukumaraswamy and Singh, 2013; Orekhova et al., 2015; Orekhova et al., 2018b;18

Swettenhametal.,2009;vanPeltetal.,2018),wefoundthatthepeakfrequencyoftheMEG19

GR prominently increased (17 Hz on average for the 100% contrast) with increasingmotion20velocityfrom0to6°/s,whichcorrespondsto0-10Hzoftemporalfrequencyinourstudy(Fig21

4A). No saturation effect has been observed for the gamma frequency even at the highest22

speeds of stimulation.Most probably, such noticeable acceleration of gamma oscillations at23

the high stimulus velocity reflects a high excitatory state of the visual cortex caused by fast24

visualmotion. This assumption is supportedby studiesonmonkeys showing that increasing25

thetemporalfrequencyofdriftinggratingsfrom0Hzuptoapproximately10Hzleadstoarise26

inneuronalspikingandanincreaseingammafrequencyinV1(Salelkaretal.,2018).Fastvisual27

motion might engage more attention, which may also affect the excitatory state of V128

(ReynoldsandChelazzi,2004)and thusmodulate thegammafrequencyduringvisualmotion29

stimulation(vanPeltetal.,2018).Theeffectofattentionaloneis,however,unlikelytoexplain30

theobserved12-17Hzincreaseingammafrequencywithtransitionfromstatictofast-moving31

stimuli presented herein. Indeed, studies inmonkeys have shown that attention has aweak32

effectongammafrequency(approximately3Hz)(Bosmanetal.,2012).Therefore,changesin33gamma frequencywith increasingmotion velocity aremost likely accounted for by a rise in34



https://doi.org/10.1101/497214


22

bottom-upexcitatoryinputtothevisualcortex.1

Themostimportantfindingofthecurrentstudyistheeffectofluminancecontraston2

thebell-shapedinput-outputrelationshipsbetweenvisualmotionvelocityandthemagnitude3

of induced GR . Although at both 50% and 100% contrast, the initially facilitative effect of4

increasing velocity on gamma response strength was followed by a suppressive one, the5

maximumofthisbell-shapedcurveatthe50%contrastshiftedtohighervelocity,ascompared6

tothe100%contrast(Fig.3,5).Inspectionoftheindividualdatahasshownthatwithlowering7contrast, the approximated velocity of the gamma suppression transition increased in every8

singlesubject(Fig.6A,7C).Thisfindingsuggeststhatacertain levelofexcitation isnecessary9

for the gamma suppression to occur, which in case of the low contrast is achieved at a10

relativelyhighvelocity.Thus,ourexperimental resultsstronglysupport the“excitatorydrive”11

model thatpredictsanadditiveeffectof increasingcontrastandvelocity/temporal-frequency12

onthegammasuppressiontransitionpoint.13

The shift of the GR suppression transition point to higher velocities with lowering14

luminancecontrastcontradictsthe‘velocitytuning’model.Animalstudieshaveshownthatat15

lowercontrast, themajorityof responsiveneurons inV1 (AlittoandUsrey,2004;Livingstone16

and Conway, 2007; Priebe et al., 2006), LGN (Alitto and Usrey, 2004), and even the retina17

(ShapleyandVictor,1978,1981)are turned to lower temporal frequencies.Therefore, if the18

maximalGRwouldcorrespondtoan‘optimal’velocity,thesuppressiontransitionpointwould19

rathershifttoalowerdriftingratewhencontrastisreduced.20Another and more general argument against the ‘velocity tuning’ hypothesis is the21

mismatchinvelocitytuningoftheV1multiunitactivityandthatoftheV1gammaresponse.A22

recent study in nonhuman primates have shown that the gamma response power starts to23

suppressat lower temporal frequencies than thoseelicitingmaximal spiking rate inmultiunit24

activity. Specifically, at the temporal frequencies of visual motion corresponding to our25

‘medium’and‘fast’velocities,thegammaresponsealreadypasseditsmaximumandstartedto26

decrease,whileneuralspikingcontinuedtoincreaseorsaturated(seeFig.4in(Salelkaretal.,27

2018)).28

Althoughourfindingsfavortheexcitatorydrivemodel,apartfromtheexcitatorydrive,29

theremaybeother factors that also affect gammaparameters at the suppression transition30

point. The low-contrast stimuli, as comparedwith the high-contrast ones, generally induced31

lower-amplitudegammaresponsesandtriggeredthetransitiontothegammasuppressionat32

lower amplitude and frequencyof theGR (Fig. 5A, 6B). Lower contrast stimuli are known to33activate fewer neurons and to produce lower spiking rates in neural circuitry involved in34



https://doi.org/10.1101/497214


23

gamma synchronization (Henrie and Shapley, 2005). This explains the lower power and1

frequency of the gamma response at the suppression transition point in case of the lower2

contrast.3

4

Stateandtraitdependencyofvisualgammaparameters.5

Previous studies show that parameters of visual gamma oscillations are strongly genetically6

determined (van Pelt et al., 2012) and highly reproducible when measured across time7

(Hoogenboom et al., 2006; Muthukumaraswamy et al., 2010). As such, they represent8

constitutional traits that aredeterminedbyneurophysiological andneuro-anatomical factors9

(Schwarzkopfetal.,2012;vanPeltetal.,2018).10

While being individually stable in a single experimental condition, the gamma11

parameters are strongly affected by the properties of the visual stimulation. This raises the12

question of whether gamma measured in response to different visual stimuli --- e.g. those13

having different contrasts or moving with different speeds --- reflect the same or different14

neurophysiologicaltraits.15Previous studies that sought to find a link between visual gamma oscillations and16

processesofneuralexcitationandinhibitioninthehumanbrain(Cousijnetal.,2014;Eddenet17

al.,2009;Kujalaetal.,2015)measuredparametersoftheinducedgammaresponseinasingle18

experimental condition. Such an approach implicitly assumes that the features of gamma19

response do not substantially vary with stimulus properties in term of the rank-order20

consistencyoftheindividualvalues.Recently,vanPeltetal(vanPeltetal.,2018)reportedhigh21

between-condition correlations for gamma peak frequency and power measured at two22

contrasts (50%and100%)andat three velocities (0, 0.33°/s, 0.66°/s) that support this view.23

Importantly,thegammaresponsepowerinthatstudymonotonouslyincreasedwithincreasing24

velocity.ThismeansthatallofthestimulationvelocitieschosenbyvanPeltetalwerebelow25

the‘gammasuppressiontransitionvelocity’andcorrespondedtotheascendingbranchofthe26

bell–shapedcurvecharacterizingtherelationbetweenvelocityandgammapower.27

When we used a broader range of velocities, we found no reliable correlations28between the amplitudes of gamma responses caused by static (0°/s) and rapidly moving29

(3.2°/s,6.0°/s)stimuli(Fig.7B).Thus,theintra-individualstabilityofgammapoweracrossthese30

conditionswaslacking,suggestingthatthestrengthoftheGRattheascendinganddescending31

branches of the bell-curve is mediated by distinct neural processes. These findings clearly32

indicate that the analysis of visually-induced changes in gamma oscillations should extend33

beyondtheclassicalgammafrequencyandpowermetricsinsingleexperimentalconditions.34



https://doi.org/10.1101/497214


24

We found a remarkably strong and highly reliable correlation between gamma1

suppressiontransitionvelocitiesatthe100%and50%contrasts(SpearmanR(17)=0.91),pointing2

to the high within-subject reliability of individual STVel values across different experimental3

conditions. As we will argue in the following discussion, the STVel is a physiologically4

meaningfulmeasure thatmayhelp to characterize thecapacityof thenetwork to regulate a5

balancebetweenexcitationandinhibition.6

7

Magnitudeofthealpha-betasuppressionisnotsensitivetothestrengthofvisualinput.8

Althoughstationaryanddriftingvisualgratingsleadalsotoastrongandreliablesuppressionof9

theMEGpower in the alpha-beta rangeover theposterior cortical areas (Fig.3), the visually10

inducedalpha-betaresponsewasclearlydifferentfromthatinthegammarange.First,despite11

apparentoverlap(Fig.2,8),thespatial localizationofthealpha-betasuppressionandgamma12

facilitationsignificantlydiffered.Whereasthemaximumofthegammaresponsewaslocalized13

near the calcarine sulcus, the alpha-beta suppression occurred in more lateral areas of the14

occipital lobes (see also Supplementary Table 1 and 2 for localization of themaximal effect15

voxels).Thisfindingisinlinewithresultsofpreviousstudiesthatshowmorelateralpositions16

of the sources of alpha-beta suppression relative to those of gamma enhancement17

(Hoogenboometal.,2006;Koelewijnetal.,2011;Muthukumaraswamyetal.,2016).Second,18

and more importantly, in sharp contrast with the gamma response, neither contrast nor19

velocity affected power suppression in the alpha-beta frequency range. This lack of a20relationship between the intensity of visual stimulation and the degree of the alpha-beta21

suppression questions the commonly held notion that the magnitude of visual alpha22

suppression indexestheexcitatorystateof thevisualcortex.Ontheotherhand,our findings23

donot contradict thenumerous results associating relativedecreasesand increases in visual24

alpha-bandactivitywithattention-relatedinfluencesmodulatingexcitabilityofvisualcortex(de25

Pestersetal.,2016;Jensenetal.,2014;Klimesch,2012;Stormeretal.,2016).26

Of note, the weighted peak frequency of the alpha-beta suppression slightly, but27

significantly, increased with increasing velocity of visual motion, most probably because of28

strongerbetasuppressionassociatedwithhighervelocity(Fig.8B). Itwouldbeinterestingto29

investigate in more detail how activity in the beta and low gamma ranges is affected by30

increasingvelocityofthevisualmotion,butthistopicisbeyondthescopeofthepresentstudy.31

32

GammasuppressiontransitionpointandE-Ibalanceinthevisualcortex.33



https://doi.org/10.1101/497214


25

Overall, our findings indicate that the strengthof excitatorydrive, but not velocity tuningof1

visualneurons, is themain factor contributing toan initially facilitativeand thensuppressive2

effectofvisualmotionvelocityonthegammaresponsepower.Thisfactisfullyconsistentwith3

predictionsofcomputationalmodelingstudies(BorgersandKopell,2005;BorgersandWalker,4

2013;Cannonetal.,2014;Kingetal.,2013;Kopelletal.,2010),and, takenonestepfurther,5

favors the idea that the gamma suppression transition point could serve as a non-invasive6

biomarkeroftheefficiencyoftheE/Ibalanceregulationinthevisualcortex.7Itiscommonlyacceptedthattheactivestatesofthebrainrelatedtotheprocessingof8

sensory information are characterized by a ‘desynchronized’ pattern of electromagnetic9

activity, i.e. by a relative predominance of high-frequency (gamma) oscillations and relative10

suppression of those of lower frequencies (alpha, beta). Indeed, we have found that a11

moderate rise in sensory input drive produced an increase of gammapower,most probably12

due toaprogressivelygreater recruitmentof visualneurons intogammasynchronization.At13

first glance, the suppressionof thegamma responseata yethigher levelof inputdrivemay14

appearcounterintuitive. Ithasbeen,howeverpredictedthatgammamaydesynchronizeata15

high level of excitatory drive due to an overexcited state of inhibitory neurons (Borgers and16

Kopell,2005;Kingetal.,2013).Ithasbeenarguedthattheasynchronousactivityofthehighly17

excitedinhibitoryneuronsisparticularlyeffectiveindown-regulatingactivity intheexcitatory18

principlecells,andthereforehasanimportantroleinhomeostaticregulationoftheneuralE/I19

balance(BorgersandKopell,2005).Incaseofstrongexcitabilityoftheprincipalcells,stronger20inhibitory influences would be required to achieve a breakdown of gamma synchrony. This21

meansthattheconstitutivefactorsthatincreaseordecreasetheexcitabilityoftheE-orI-cells22

mayresultinshiftingthegammasuppressionpointtohigherorlowerintensitiesofexcitatory23

drive.24

Specifically,thetransitiontosuppressionofthegammaresponseatarelativelyhigher25

level of excitatory drive and/or the shallower slope of this suppression would reflect less26

effectiveinhibition.Inindirectsupportofthisassumption,wehaverecentlydescribedalackof27

velocity-relatedgammasuppressioninasubjectwithepilepsyandoccipitalspikes(Orekhovaet28

al.,2018b),whomighthaveanelevatedE-Iratiointhevisualcortex.Thecausallinkbetween29

the E-I ratio and the gamma suppression transition point could, in the future, be proved30

directly using pharmacological or other (e.g. tDCS, TMS) approaches tomanipulating the E-I31

balance.32

Gamma synchronization facilitates neural communication and increases the effective33outputofpresynaptic(Bastosetal.,2015;Fries,2005,2009,2015;Nietal.,2016;Rigoulotet34



https://doi.org/10.1101/497214


26

al., 2017; Vinck et al., 2013). Therefore, synchronization of V1 neural activity in the gamma1

range is thought toenhancetheeffectiveoutput fromtheprimaryvisualcortex toupstream2

cortical areas, thus affecting visual perception. Aweak suppression of gammaoscillations at3

highstimulationintensitiesmaythenresultinastrongerimpactofthesensorystimulationon4

theneural network, and lead to sensoryoverload. Indeed, in a recent study,wehave found5

that a weaker suppression of gamma response at fast motion velocities (of high-contrast6

gratings) correlated with enhanced sensory sensitivity, especially in the visual domain7(Orekhovaetal.,2018a).8

Futurestudiesinclinicalpopulationscharacterizedbyelevatedcorticalexcitability(e.g.9

epilepsy,migraine, some forms of ASD) would help to assess the potential value of gamma10

suppressionasabiomarkerforimpairedregulationoftheE/Ibalance.11

12

13

CONCLUSIONS14

15

To summarize, our current findings provide strong support for the hypothesis linking16

magnitude of the visual gamma response caused by drifting gratings to the strength of17

excitatory drive. At the same time, our results indicate that ‘velocity turning’ of V1 neurons18

doesnotplayaprimaryroleinregulatingthemagnitudeofthegammaresponse.19

In both low- and high-contrast conditions the power of visual gamma oscillations20

demonstrated bell-shaped dependency on velocity of visual motion: it first increased with21

increasingvelocityandthendecreaseswithfurtherincreaseinthedriftingrate.Aspredicted22

by the ‘excitatorydrivemodel’, reducing intensityof thevisual inputby loweringcontrastof23the grating shifted the maximum of the bell-shaped curve --- the ‘suppression transition24

velocity’’ --- to higher velocity values. Unlike induced gamma oscillations, the visual alpha25

responsewasnotsensitivetothechangesinthegratings'velocityorcontrast.26

Our present findings, which support the causative role of strong excitatory drive in27

visual gamma response suppression, have important theoretical implications. The28

computationalmodelingstudiesandourpreviousexperimentalresults implythatsuchcausal29

relationships may reflect the capacity of inhibitory neurons to down-regulate growing30

excitation and to maintain the E-I balance. Therefore, the shape of the gamma response31

modulationcurveand, inparticular,the‘suppressiontransitionvelocity’parameterevaluated32

inthepresentstudy,mayprovideimportantinformationaboutregulationoftheE-Ibalance.33



https://doi.org/10.1101/497214


27

Weanticipatethatsuchacombinatoryindexas‘suppressiontransitionvelocity’willbe1

abettermeasureoftheE-Idysfunctioninbraindisordersthanvisualgammaparametersthat2

areextractedinasingleexperimentalcondition.Apartfromhavingtheoreticalrelevance,this3

index evades several caveats of using peak gamma amplitude and frequency. Firstly, it is4

inherently relational, and is therefore less sensitive to inter-individual differences in cortical5

anatomy and SNR. Secondly it allows to avoid ambiguity related to single-condition6

assessmentsofgammaamplitudes,whichdonotalwayscorrelatewitheachother(e.g.,when7measured with static vs. ‘fast movement’ conditions); Thirdly, it's individual values have8

remarkablerank-orderconsistencywhenmeasuredatdifferentcontrasts,andthereforeshould9

have high test–reteststability. Altogether, these considerations suggest that the ‘gamma10

suppressiontransitionvelocity’hasatranslationalpotentialasan indexoftheE-Ibalancefor11

informingclinicalpracticeandtrials.12

13

14

15



https://doi.org/10.1101/497214


28

1

Referenses2

3

Alitto, H.J., Usrey,W.M., 2004. Influence of contrast on orientation and temporal frequency4

tuninginferretprimaryvisualcortex.JournalofNeurophysiology91,2797-2808.5

Anver,H.,Ward,P.D.,Magony,A.,Vreugdenhil,M.,2011.NMDAReceptorHypofunctionPhase6

Couples Independent gamma-Oscillations in the Rat Visual Cortex.7

Neuropsychopharmacology36,519-528.8

Bartolo, M.J., Gieselmann,M.A., Vuksanovic, V., Hunter, D., Sun, L., Chen, X., Delicato, L.S.,9

Thiele, A., 2011. Stimulus-induced dissociation of neuronal firing rates and local field10potential gamma power and its relationship to the resonance blood oxygen level-11

dependentsignalinmacaqueprimaryvisualcortex.EuropeanJournalofNeuroscience34,12

1857-1870.13

Bastos,A.M.,Vezoli,J.,Bosman,C.A.,Schoffelen,J.M.,Oostenveld,R.,Dowdall,J.R.,DeWeerd,14

P.,Kennedy,H.,Fries,P.,2015.VisualAreasExertFeedforwardandFeedback Influences15

throughDistinctFrequencyChannels.Neuron85,390-401.16

Borgers,C., Kopell,N., 2005. Effectsofnoisydriveon rhythms innetworksofexcitatoryand17

inhibitoryneurons.NeuralComputation17,557-608.18

Borgers,C.,Walker,B.,2013.Togglingbetweengamma-frequencyactivityandsuppressionof19

cellassemblies.FrontComputNeurosc.20

Bosman, C.A., Schoffelen, J.M., Brunet, N., Oostenveld, R., Bastos, A.M., Womelsdorf, T.,21

Rubehn, B., Stieglitz, T., De Weerd, P., Fries, P., 2012. Attentional Stimulus Selection22

throughSelectiveSynchronizationbetweenMonkeyVisualAreas.Neuron75,875-888.23Buzsaki, G., Wang, X.J., 2012. Mechanisms of Gamma Oscillations. Annual Review of24

Neuroscience,Vol3535,203-225.25

Cannon, J.,McCarthy,M.M., Lee, S., Lee, J., Borgers, C.,Whittington,M.A., Kopell,N., 2014.26

Neurosystems:brainrhythmsandcognitiveprocessing.EuropeanJournalofNeuroscience27

39,705-719.28

Chen, G., Zhang, Y., Li, X., Zhao, X., Ye, Q., Lin, Y., Tao, H.W., Rasch, M.J., Zhang, X., 2017.29

DistinctInhibitoryCircuitsOrchestrateCorticalbetaandgammaBandOscillations.Neuron30

96,1403-1418e1406.31

Cousijn, H., Haegens, S., Wallis, G., Near, J., Stokes, M.G., Harrison, P.J., Nobre, A.C., 2014.32

Resting GABA and glutamate concentrations do not predict visual gamma frequency or33



https://doi.org/10.1101/497214


29

amplitude. Proceedings of the National Academy of Sciences of the United States of1

America111,9301-9306.2

dePesters,A.,Coon,W.G.,Brunner,P.,Gunduz,A.,Ritaccio,A.L.,Brunet,N.M.,deWeerd,P.,3

Roberts,M.J.,Oostenveld,R.,Fries,P.,Schalk,G.,2016.Alphapowerindexestask-related4

networksonlargeandsmallscales:AmultimodalECoGstudyinhumansandanon-human5

primate.Neuroimage134,122-131.6

Edden, R.A., Muthukumaraswamy, S.D., Freeman, T.C., Singh, K.D., 2009. Orientation7discrimination performance is predicted by GABA concentration and gamma oscillation8

frequencyinhumanprimaryvisualcortex.JNeurosci29,15721-15726.9

Ferando,I.,Mody,I.,2013.Alteredgammaoscillationsduringpregnancythroughlossofdelta10

subunit-containing GABA(A) receptors on parvalbumin interneurons. Frontiers in Neural11

Circuits,p.144.12

Fries, P., 2005. A mechanism for cognitive dynamics: neuronal communication through13

neuronalcoherence.TrendsinCognitiveSciences9,474-480.14

Fries, P., 2009. Neuronal gamma-band synchronization as a fundamental process in cortical15

computation.AnnualReviewofNeuroscience,Vol3532,209-224.16

Fries, P., 2015. Rhythms for Cognition: Communication through Coherence.Neuron 88, 220-17

235.18

Gaetz, W., Roberts, T.P.L., Singh, K.D., Muthukumaraswamy, S.D., 2012. Functional and19

structuralcorrelatesoftheagingbrain:Relatingvisualcortex(V1)gammabandresponses20toage-relatedstructuralchange.HumanBrainMapping33,2035-2046.21

Gramfort,A.,Luessi,M.,Larson,E.,Engemann,D.A.,Strohmeier,D.,Brodbeck,C.,Goj,R.,Jas,22

M., Brooks, T., Parkkonen, L., Hamalainen, M., 2013. MEG and EEG data analysis with23

MNE-Python.FrontNeurosci.24

Gross,J.,Kujala,J.,Hamalainen,M.,Timmermann,L.,Schnitzler,A.,Salmelin,R.,2001.Dynamic25

imagingofcoherentsources:Studyingneuralinteractionsinthehumanbrain.Proceedings26

oftheNationalAcademyofSciencesoftheUnitedStatesofAmerica98,694-699.27

Hadjipapas,A.,Lowet,E.,Roberts,M.J.,Peter,A.,DeWeerd,P.,2015.Parametricvariationof28

gamma frequency and power with luminance contrast: A comparative study of human29

MEGandmonkeyLFPandspikeresponses.Neuroimage112,327-340.30

Henrie, J.A., Shapley,R.,2005. LFPpower spectra inV1cortex: thegradedeffectof stimulus31

contrast.JNeurophysiol94,479-490.32

Hoogenboom, N., Schoffelen, J.M., Oostenveld, R., Parkes, L.M., Fries, P., 2006. Localizing33humanvisualgamma-bandactivityinfrequency,timeandspace.Neuroimage29,764-773.34



https://doi.org/10.1101/497214


30

Jensen, O., Gips, B., Bergmann, T.O., Bonnefond, M., 2014. Temporal coding organized by1

coupled alpha and gamma oscillations prioritize visual processing. Trends Neurosci 37,2

357-369.3

Jia, X.X., Smith,M.A., Kohn, A., 2011. Stimulus Selectivity and Spatial Coherence of Gamma4

ComponentsoftheLocalFieldPotential.JNeurosci31,9390-9403.5

Jia, X.X., Xing, D.J., Kohn, A., 2013. No Consistent Relationship between Gamma Power and6

PeakFrequencyinMacaquePrimaryVisualCortex.JNeurosci33,17-U421.7King,P.D.,Zylberberg,J.,DeWeese,M.R.,2013.Inhibitoryinterneuronsdecorrelateexcitatory8

cellstodrivesparsecodeformationinaspikingmodelofV1.JNeurosci33,5475-5485.9

Klimesch, W., 2012. alpha-band oscillations, attention, and controlled access to stored10

information.TrendsCognSci16,606-617.11

Koelewijn, L., Dumont, J.R.,Muthukumaraswamy, S.D., Rich, A.N., Singh, K.D., 2011. Induced12

andevokedneuralcorrelatesoforientationselectivityinhumanvisualcortex.Neuroimage13

54,2983-2993.14

Kopell,N.,Börgers,C.,Pervouchine,D.,Malerba,P.,Tort,A.,2010.GammaandThetaRhythms15

in Biophysical Models of Hippocampal Circuits. Springer Series in Computational16

NeuroscienceSpringerScience+BusinessMedia.17

Kujala,J.,Jung,J.,Bouvard,S.,Lecaignard,F.,Lothe,A.,Bouet,R.,Ciumas,C.,Ryvlin,P.,Jerbi,18

K.,2015.GammaoscillationsinV1arecorrelatedwithGABA(A)receptordensity:Amulti-19

modalMEGandFlumazenil-PETstudy.SciRep5,16347.20Liu,Y.,Bengson,J.,Huang,H.,Mangun,G.R.,Ding,M.,2016.Top-downModulationofNeural21

Activity in Anticipatory Visual Attention: ControlMechanisms Revealed by Simultaneous22

EEG-fMRI.CerebCortex26,517-529.23

Livingstone, M.S., Conway, B.R., 2007. Contrast affects speed tuning, space-time slant, and24

receptive-fieldorganizationofsimplecellsinmacaqueV1.JNeurophysiol97,849-857.25

Lowet,E.,Roberts,M.,Hadjipapas,A.,Peter,A.,vanderEerden,J.,DeWeerd,P.,2015.Input-26

Dependent Frequency Modulation of Cortical Gamma Oscillations Shapes Spatial27

SynchronizationandEnablesPhaseCoding.PlosComputationalBiology11.28

Mann, E.O., Mody, I., 2010. Control of hippocampal gamma oscillation frequency by tonic29

inhibitionandexcitationofinterneurons.NatureNeuroscience13,205-U290.30

Murty,D.,Shirhatti,V.,Ravishankar,P.,Ray,S.,2018.LargeVisualStimuliInduceTwoDistinct31

GammaOscillationsinPrimateVisualCortex.JNeurosci38,2730-2744.32



https://doi.org/10.1101/497214


31

Muthukumaraswamy,S.D.,Routley,B.,Droog,W.,Singh,K.D.,Hamandi,K.,2016.Theeffects1

ofAMPAblockadeonthespectralprofileofhumanearlyvisualcortexrecordingsstudied2

withnon-invasiveMEG.Cortex81,266-275.3

Muthukumaraswamy,S.D.,Singh,K.D.,2013.Visualgammaoscillations:Theeffectsofstimulus4

type,visualfieldcoverageandstimulusmotiononMEGandEEGrecordings.Neuroimage5

69,223-230.6

Muthukumaraswamy, S.D., Singh, K.D., Swettenham, J.B., Jones, D.K., 2010. Visual gamma7oscillationsandevokedresponses:Variability,repeatabilityandstructuralMRIcorrelates.8

Neuroimage49,3349-3357.9

Ni,J.G.,Wunderle,T.,Lewis,C.M.,Desimone,R.,Diester, I.,Fries,P.,2016.Gamma-Rhythmic10

GainModulation.Neuron92,240-251.11

Nichols, T.E., Holmes, A.P., 2002. Nonparametric permutation tests for functional12

neuroimaging:Aprimerwithexamples.HumanBrainMapping15,1-25.13

Oostenveld,R.,Fries,P.,Maris,E.,Schoffelen,J.M.,2011.FieldTrip:Opensourcesoftwarefor14

advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput Intell15

Neurosci.16

Orekhova,E.V.,Butorina,A.V.,Sysoeva,O.V.,Prokofyev,A.O.,Nikolaeva,A.Y.,Stroganova,T.A.,17

2015. Frequency of gamma oscillations in humans is modulated by velocity of visual18

motion.JournalofNeurophysiology114,244-255.19

Orekhova, E.V., Stroganova, T.A., Schneiderman, J.F., Lundström, S., Riaz, B., Sarovoc, D.,20Sysoeva,O.V.,Brant,G.,Gillberg,C.,Hadjikhani,N.,2018a.Neuralgaincontrolmeasured21

through cortical gamma oscillations is associated with sensory sensitivity. Hum Brain22

Mapp,inpress.23

Orekhova, E.V., Sysoeva, O.V., Schneiderman, J.F., Lundstrom, S., Galuta, I.A., Goiaeva, D.E.,24

Prokofyev,A.O., Riaz, B., Keeler, C.,Hadjikhani,N.,Gillberg, C., Stroganova, T.A., 2018b.25

Input-dependentmodulationofMEGgammaoscillationsreflectsgaincontrolinthevisual26

cortex.SciRep8,8451.27

Palva,S.,Palva, J.M.,2011.Functional rolesofalpha-bandphasesynchronization in localand28

large-scalecorticalnetworks.FrontPsychol2,204.29

Perry,G.,Randle,J.M.,Koelewijn,L.,Routley,B.C.,Singh,K.D.,2015.LinearTuningofGamma30

Amplitude and Frequency to LuminanceContrast: Evidence froma ContinuousMapping31

Paradigm.PLoSOne.32

Pfurtscheller, G., da Silva, F.H.L., 1999. Event-related EEG/MEG synchronization and33desynchronization:basicprinciples.ClinicalNeurophysiology110,1842-1857.34



https://doi.org/10.1101/497214


32

Priebe, N.J., Lisberger, S.G., Movshon, J.A., 2006. Tuning for spatiotemporal frequency and1

speedindirectionallyselectiveneuronsofmacaquestriatecortex.JournalofNeuroscience2

26,2941-2950.3

Ray, S., Maunsell, J.H., 2010. Differences in gamma frequencies across visual cortex restrict4

theirpossibleuseincomputation.Neuron67,885-896.5

Reynolds,J.H.,Chelazzi,L.,2004.Attentionalmodulationofvisualprocessing.AnnualReviewof6

Neuroscience27,611-647.7Rigoulot, S., Knoth, I.S., Lafontaine,M.P., Vannasing, P.,Major, P., Jacquemont, S.,Michaud,8

J.L.,Jerbi,K.,Lippe,S.,2017.AlteredvisualrepetitionsuppressioninFragileXSyndrome:9

NewevidencefromERPsandoscillatoryactivity.IntJDevNeurosci59,52-59.10

Roberts,M.J.,Lowet,E.,Brunet,N.M.,TerWal,M.,Tiesinga,P.,Fries,P.,DeWeerd,P.,2013.11

RobustGammaCoherencebetweenMacaqueV1andV2byDynamicFrequencyMatching.12

Neuron78,523-536.13

Romei,V.,Brodbeck,V.,Michel,C.,Amedi,A.,Pascual-Leone,A.,Thut,G.,2008.Spontaneous14

fluctuationsinposterioralpha-bandEEGactivityreflectvariabilityinexcitabilityofhuman15

visualareas.CerebCortex18,2010-2018.16

Salelkar, S., Somasekhar, G.M., Ray, S., 2018. Distinct frequency bands in the local field17

potentialaredifferentlytunedtostimulusdriftrate.JNeurophysiol.18

Scheeringa,R.,Koopmans,P.J.,vanMourik,T.,Jensen,O.,Norris,D.G.,2016.Therelationship19

betweenoscillatoryEEGactivityandthelaminar-specificBOLDsignal.Proceedingsofthe20NationalAcademyofSciencesoftheUnitedStatesofAmerica113,6761-6766.21

Schwarzkopf, D.S., Robertson, D.J., Song, C., Barnes, G.R., Rees, G., 2012. The frequency of22

visually induced gamma-band oscillations depends on the size of early human visual23

cortex.JNeurosci32,1507-1512.24

Shapley,R.M.,Victor,J.D.,1978.Theeffectofcontrastonthetransferpropertiesofcatretinal25

ganglioncells.JPhysiol285,275-298.26

Shapley, R.M., Victor, J.D., 1981. How the contrast gain control modifies the frequency27

responsesofcatretinalganglioncells.JPhysiol318,161-179.28

Snyder,A.C.,Foxe,J.J.,2010.Anticipatoryattentionalsuppressionofvisualfeaturesindexedby29

oscillatoryalpha-bandpowerincreases:ahigh-densityelectricalmappingstudy.JNeurosci30

30,4024-4032.31

Stormer,V.,Feng,W.,Martinez,A.,McDonald,J.,Hillyard,S.,2016.Salient,IrrelevantSounds32

ReflexivelyInduceAlphaRhythmDesynchronizationinParallelwithSlowPotentialShiftsin33VisualCortex.JCognNeurosci28,433-445.34



https://doi.org/10.1101/497214


33

Sumner,R.L.,McMillan,R.L.,Shaw,A.D.,Singh,K.D.,Sundram,F.,Muthukumaraswamy,S.D.,1

2018. Peak visual gamma frequency is modified across the healthy menstrual cycle.2

HumanBrainMapping39,3187-3202.3

Swettenham,J.B.,Muthukumaraswamy,S.D.,Singh,K.D.,2009.SpectralPropertiesofInduced4

and EvokedGammaOscillations inHumanEarlyVisual Cortex toMoving and Stationary5

Stimuli.JournalofNeurophysiology102,1241-1253.6

Tan,H.R.M.,Gross,J.,Uhlhaas,P.J.,2016.MEGsensorandsourcemeasuresofvisuallyinduced7gamma-bandoscillationsarehighlyreliable.Neuroimage137,34-44.8

Taulu,S.,Hari,R.,2009.Removalofmagnetoencephalographicartifactswithtemporalsignal-9

space separation: demonstrationwith single-trial auditory-evoked responses.HumBrain10

Mapp30,1524-1534.11

Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N.,12

Mazoyer,B.,Joliot,M.,2002.AutomatedanatomicallabelingofactivationsinSPMusinga13

macroscopicanatomicalparcellationoftheMNIMRIsingle-subjectbrain.Neuroimage15,14

273-289.15

Van Hooser, S.D., Roy, A., Rhodes, H.J., Culp, J.H., Fitzpatrick, D., 2013. Transformation of16

Receptive Field Properties from LateralGeniculateNucleus to Superficial V1 in the Tree17

Shrew.JournalofNeuroscience33,11494-11505.18

van Kerkoerle, T., Self, M.W., Dagnino, B., Gariel-Mathis, M.A., Poort, J., van der Togt, C.,19

Roelfsema, P.R., 2014. Alpha and gamma oscillations characterize feedback and20feedforward processing in monkey visual cortex. Proc Natl Acad Sci U S A 111, 14332-21

14341.22

vanPelt,S.,Boomsma,D.I.,Fries,P.,2012.MagnetoencephalographyinTwinsRevealsaStrong23

Genetic Determination of the Peak Frequency of Visually Induced Gamma-Band24

Synchronization.JournalofNeuroscience32,3388-3392.25

van Pelt, S., Shumskaya, E., Fries, P., 2018. Cortical volume and sex influence visual gamma.26

Neuroimage178,702-712.27

VanVeen, B.D., vanDrongelen, W., Yuchtman, M., Suzuki, A., 1997. Localization of brain28

electrical activity via linearly constrained minimum variance spatial filtering. Ieee29

TransactionsonBiomedicalEngineering44,867-880.30

Vinck, M., Womelsdorf, T., Fries, P., 2013. Gamma-band synchronization and information31

transmission. In:Quiroga,R.Q.,Panzeri, S. (Eds.),PrinciplesofNeuralCoding.CRCPress,32

pp.449-469.33



https://doi.org/10.1101/497214


34

Yuan,H.,Liu,T.,Szarkowski,R.,Rios,C.,Ashe, J.,He,B.,2010.Negativecovariationbetween1

task-related responses in alpha/beta-band activity and BOLD in human sensorimotor2

cortex:anEEGandfMRIstudyofmotorimageryandmovements.Neuroimage49,2596-3

2606.4

5



https://doi.org/10.1101/497214


Documents

Additive effect of contrast and velocity proves the role ... · 1 Additive effect of contrast and velocity proves the role of strong 2 excitatory drive in suppression of visual gamma