functional connectivity in neuromuscular system underlying

Functionalconnectivityinneuromuscularsystemunderlying

bimanualmusclesynergies

IngmarE.J.deVriesa,AndreasDaffertshofera,DickF.Stegemana,bandTjeerdW.

Boonstraa,c,*

aFacultyofBehaviouralandMovementSciences,VUUniversity,Amsterdam,TheNetherlands

bDondersInstituteforBrain,Cognition&Behavior,RadboudUniversity,Nijmegen,The

NetherlandscBlackDogInstitute,UniversityofNewSouthWales,Sydney,Australia

Runninghead:Theneuralbasisofmusclesynergies

* Corresponding author Black Dog Institute, Hospital Road, Randwick, NSW 2031, Australia. Email: [email protected], Fax: +61 2 9382 8208, Phone: +61 2 9382 8375.

.CC-BY-NC-ND 4.0 International licensenot peer-reviewed) is the author/funder. It is made available under aThe copyright holder for this preprint (which was. http://dx.doi.org/10.1101/056671doi: bioRxiv preprint first posted online Jun. 2, 2016;

http://dx.doi.org/10.1101/056671

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

Abstract

Neuralsynchronyhasbeensuggestedasmechanismforintegratingdistributed

sensorimotorsystemsinvolvedincoordinatedmovement.Totesttheroleof

corticomuscularandintermuscularcoherenceintheformationofbimanualmuscle

synergies,weexperimentallymanipulatedthedegreeofcoordinationbetweenhand

musclesbyvaryingthesensitivityofthevisualfeedbacktodifferencesinbilateralforce.

In16healthyparticipants,corticalactivitywasmeasuredusing64-channel

electroencephalography(EEG)andmuscleactivityoftheflexorpollicisbrevismuscleof

bothhandsusing8×8-channelhigh-densityelectromyography(HDsEMG).Usingthe

uncontrolledmanifoldframework,coordinationbetweenbilateralforceswasquantified

bythesynergyindexRVinthetimeandfrequencydomain.Functionalconnectivitywas

assedusingcorticomuscularcoherencebetweenmuscleactivityandcorticalsource

activityandintermuscularcoherencebetweenbilateralEMGactivity.Asexpected,

bimanualsynergieswerestrongerinthehighcoordinationcondition.RVwashigherin

thehighcoordinationconditioninfrequenciesbetween0and0.5Hz,andabove2Hz.

Forthe0.5–2Hzfrequencybandthispatternwasinverted.Corticomuscularcoherence

inthebetaband(16-30Hz)wasmaximalinthecontralateralmotorcortexandwas

reducedinthehighcoordinationcondition.Incontrast,intermuscularcoherencewas

observedat5-12Hzandincreasedwithbimanualcoordination.Within-subject

comparisonsrevealedanegativecorrelationbetweenRVandcorticomuscularcoherence

andapositivecorrelationbetweenRVandintermuscularcoherence.Ourfindings

suggesttwodistinctneuralpathways:(1)Corticomuscularcoherencereflectsdirect

corticospinalprojectionsinvolvedincontrollingindividualmuscles;(2)intermuscular

coherencereflectsdivergingpathwaysinvolvedinthecoordinationofmultiplemuscles.

Keywords:Motorcoordination,neuralsynchrony,corticomuscularcoherence,

intermuscularcoherence,uncontrolledmanifold


http://dx.doi.org/10.1101/056671


1

1.Introduction

Themanydegreesoffreedom(DOFs)ofthemusculoskeletalsystemprovidegreat

flexibility,buttheymakethecorrespondingcontrolproblemrathercomplex.Thereare

numerouswaystoperformamovementtoachievethesamegoal.Howdoesthenervous

systemcoordinatetheabundantDOFs?Toreducethedimensionalityofthecontrol

problem,ithasbeenproposedthatagroupofmusclesspanningseveraldifferentjoints

canbecomefunctionallylinkedsoastobehaveasasingletask-specificunit(Bernstein

1967;Latashetal.2007;Turvey1990).Thenervoussystemcancontrolthesemuscle

synergies–ratherthanindividualmuscles–andbycombiningmusclesynergiesin

differentways,asmallnumberofsynergiescanaccountforawidevarietyof

movements(d'Avellaetal.2003;Ivanenkoetal.2004;TingandMacpherson2005;

Treschetal.1999).However,relativelylittleisknownabouthowthesemuscle

synergiesareimplementedinthecentralnervoussystem(TreschandJarc2009).The

spinalcircuitryandthepathwaysconnectingthemotorcortexandthespinal

interneuronsappearcentralinunderstandingtheneuralbasisofmusclesynergies

(BizziandCheung2013;Levineetal.2014).

Neuralsynchronizationmayofferawindowintotheneuralcorrelatesofmuscle

synergies(Boonstra2013).Equivalenttoitsroleinperceptualprocesses(Engeletal.

2001;Singer1999),neuralsynchronymayprovideamechanismforintegratingthe

distributedmotorandsensorysystemsinvolvedincoordinatedmovementandposture

(Farmer1998).Ifneuralsynchronyisageneralmechanismforcoordinatingneural

activity,onemightexpectthatmusclestakingpartinasynergyreceivecorrelatedneural

input.Infact,neuralsynchronyhasbeenwidelyobservedinthemotorsystem

(SchnitzlerandGross2005;vanWijketal.2012).Inparticular,corticomuscular

coherencehasbeenobservedbetweenmotorcortexandmuscleactivity(Bakeretal.

1997;Conwayetal.1995;Saleniusetal.1997),andintermuscularcoherencebetween

differentsetsofmuscles(Boonstraetal.2008;Farmeretal.1993;Kilneretal.1999).

Interestingly,musclescontributingtoasynergygenerallyrevealintermuscular

coherencesuggestingtheyreceivecorrelatedneuralinput(Boonstraetal.2009a;

Danna-Dos-Santosetal.2014;DeLucaandErim2002;Laineetal.2015).

Inthisstudywetestedtherelationshipbetweenmusclesynergiesandintermuscular

andcorticomuscularcoherencebyexperimentallymanipulatingtherequiredlevelof


http://dx.doi.org/10.1101/056671


2

coordinationbetweenbilateralhandmuscles.Participantsperformedabimanual

precision-griptaskwhileEEGandEMGactivitywererecorded(cf.Boonstraetal.

2009b).SimilarlytoNazarpouretal.(2012),wecontrolledthedegreeofbimanual

musclesynergythroughvisualfeedbackpresentedtotheparticipantsbyvaryingthe

sensitivityoftheto-be-controlledcursortodifferencesinthetwohandforces,making

thisdifferencetherelevantdimensionfortasksuccess,whilemakingthesumofthetwo

handforcesthe(relatively)irrelevantdimensionfortasksuccess.Thatwaywecould

experimentallymanipulatetheamountofvariabilityintherelevantversusthe

irrelevantdimensionfortasksuccess(cf.ScholzandSchoner1999).

Weexpectedtofindhigherlevelsofintermuscularcoherencebetweenhomologous

handmusclesbutlowerlevelsofcorticomuscularcoherencewhenastrongersynergyis

required.Corticomuscularisthoughttoreflectdirectcorticospinalprojections(Williams

andBaker2009),involvedintheselectivecontrolofindividualmuscles(Lemon2008).

Wehenceexpectthatcorticomuscularcoherenceismoredominantwhennobilateral

coordinationisrequiredandbothhandscanbecontrolledindependently.Thiswould

implythatournervoussystemcanflexiblyregulatethecontributionofmusclesynergies

dependingontaskdemandsandthattheneuralbasisofmusclesynergiescanbe

assessedusingfunctionalconnectivityanalysisbetweenmuscleandcorticalactivities.

Thelevelofbilateralmusclesynergywasquantifiedfromtheforcetrajectoriesusinga

measureofsynergyproposedbyLatashetal.(2002).Weextendedthismeasuretothe

frequencydomaintoinvestigatethetimescaleatwhichthebilateralsynergyis

established.Establishingalinkbetweenfunctionalconnectivityandmusclesynergies

willhelptodelineatetheneuralpathwaysinvolvedinmotorcoordination.

2.MaterialsandMethods

2.1Participants

Sixteenhealthyright-handedparticipants(8female,age=25.5±4.3)volunteeredinthe

study.ThelocalethicscommitteeattheFacultyofHumanMovementSciences,VU

UniversityAmsterdam,approvedtheexperimentalprotocolandeachparticipant

providedwritteninformedconsentpriortoparticipatingintheexperiment.


http://dx.doi.org/10.1101/056671


3

2.2Experimentalprotocol

Participantsperformedabimanualprecision-griptaskwhileEEGandEMGactivitywas

recorded.Bimanualforcesweregeneratedagainsttwocompliantforcesensorsheldin

eachhandusingapinchgrip(Fig.1A).

Figure1.Experimentalprotocol.A)High-densityelectrodegridisplacedovertheflexor

pollicisbreviswhileparticipantsperformapinchgriponcustom-madecompliantforcesensor

B)Forcelevelprojection.Cursorandtargetarethevisualfeedbackdisplayedonthemonitor.

Participantswereinstructedtotrackavisualtargetbymovingthecursordisplayedona

monitor(Fig.1B).Atthestartofeachtrial,thetargetwasinthestartingposition.During

thefirst5s,thetargetmovedlinearlytothefinaltargetposition(forceramp),whereit

stayedfortheremaining10softhetrial(constantforce).Participantscouldmovethe

cursorbyapplyingforcetobothsensorsinordertokeepthecursorwithinthetarget.

Thepositionofthecursorwasdeterminedbyalinearweightingoftheforcesgenerated

bytheparticipant:𝑥𝑦 = 𝑐11 𝑐12

𝑐21 𝑐22 ∗ 𝐹!𝐹!

where𝑥and𝑦arethehorizontalandverticalscreencoordinatesand𝐹! and𝐹! arethe

forcelevelsgeneratedbytheleftandrighthand,respectively(Fig.1B).The2×2

projectionmatrix𝑐convertsthegeneratedforcelevelstoscreencoordinates.The

requiredlevelofbilateralmusclesynergywasexperimentallymanipulatedbychanging

theprojectionmatrix𝑐(Nazarpouretal.2012).Forexample,when

𝑐 = −10 101 1

thex-positionofthecursorisrelativelysensitivetodifferencesbetweenthetwoforce

levelssoarelativelystrongcoordinationbetweenbothhandsisrequiredtoperformthe


http://dx.doi.org/10.1101/056671


4

tasksuccessfully(i.e.keepthecursorinthecirculartarget).Thatis,evenasmall

differencebetween𝐹! and𝐹! willmovethecursorsideways.Incontrast,when

𝑐 = −0.01 0.011 1

thex-positionofthecursorisrelativelyinsensitivetodifferencesbetweenthetwoforce

levels,i.e.arelativelyweakcoordinationbetweenbothhandsisrequired.Forthelatter,

thetotalforceneedstobecontrolledtomovethecursorupordown,butalarge

variabilityintheratioofthetwoforcelevelsistolerated(bilateralforcesareallowedto

differ).Therequiredlevelofbilateralcoordinationwasvariedinthreeconditions(low,

medium,high)byvaryingtheprojectionmatrixc:

𝑐1 = −0.01 0.011 1 , 𝑐2 = −0.33 0.33

1 1 , and 𝑐3 = −10 101 1 .

Eachprojectionmatrixwasusedinablockof20trials,i.e.thedurationofablockwas

400sorabout7minutes.Inadditiontotherequiredlevelofsynergy,thetotalforce

levelwasvariedtoensurethatparticipantsweredependentonvisualfeedbackfor

successfultaskperformance.Theforcelevelnecessaryduringtheconstantforcepart

variedbetweentwolevels(1.7and2.1Nperhand)andtrialswithdifferentforcelevels

wererandomizedwithinablock.Theorderofthe3blockswascounterbalanced

betweenparticipants.Theinter-trialintervalwas5sduringwhichtheperformance

scorewaspresented(percentageoftimethecursorwaswithinthetarget)andthetotal

durationofeachtrialwashence20s.Therewasatwo-minutebreakbetweentheblocks.

2.3Dataacquisition

Corticomuscularcoherenceandintermuscularcoherencewereassessedusing

electroencephalography(EEG)andhigh-densitysurfaceelectromyography(HDsEMG).

EEGwasrecordedusinga64-electrodenyloncapplacingtheelectrodesaccordingtothe

extended10-20systemandamplifiedusinga64-channelRefaamplifier(TMSi,

Enschede,TheNetherlands;samplingrate1024Hz).HDsEMGwasrecordedusingtwo

64-channel(8×8,IED4mm)electrodegrids(Fig.1A)andamplifiedusinga128-channel

Refaamplifier(TMSi,Enschede,TheNetherlands;samplingrate2048Hz).Theelectrode

gridswereplacedovertheflexorpollicisbrevis(FPB)muscleofbothhands.Pinchgrip

forcewasrecordedusingacustom-madeforcesensor(Fig.1A;Boonstraetal.2005)and

amplifiedusingaNationalInstrumentsamplifier(NISCXI-1121,samplerate1000Hz).


http://dx.doi.org/10.1101/056671


5

Atthestartofeachtrial,theforcesensoramplifiersentasynchronizationpulsetoboth

Refaamplifiers,whichwasusedtoaligndataoffline.

2.4Preprocessing

ThedataanalysiswasperformedinMatlab(2012a,TheMathWorks,Natwick,MA,USA)

usingtheFieldtriptoolboxforM/EEGanalysis

(http://www.ru.nl/neuroimaging/fieldtrip;Oostenveldetal.2011).EMG,EEG,andforce

signalswerealignedusingtheaforementionedsynchronizationpulseandsegmented

intoseparatetrials.Forcesignalswerelinearlyinterpolatedto2048Hz.Onlythe

constantforcepartofeachtrialwasselectedtoassesbilateralmusclesynergies,i.e.

fromt=6tot=15s(excludingthefirstsecondofconstantforcetoavoidtransient

activity).EEGsignalswerelinearlyinterpolatedto2048Hzandhigh-passfiltered

(Butterworth,secondorder,cutoff1Hz).Thesignalswereband-stopfilteredat50Hz

anditshigherharmonicstoremovelinenoise.ICAwasusedtoremoveeye-blinks,eye-

movementsandmuscleactivityfromtheEEGdata(Jungetal.2000).

WeusedHDsEMGtoimprovetheestimationofcorticomuscularandintermuscular

coherence(vandeSteegetal.2014).BadHDsEMGchannelswereremovedandon

average59±4channelsperhandwereusedforfurtheranalysis.TheHDsEMGsignals

werere-referencedtothereferenceelectrode(placedontheprocessstyloidofthe

radius),high-passfiltered(Butterworth,secondorder,cutoff10Hz)andband-stop

filteredat50Hzanditshigherharmonicstoremovelinenoise.Principalcomponent

analysiswasappliedtotheHDsEMGsignalstoreducetheeffectsofamplitude

cancellationandheterogeneityinthemotorunitactionpotentialsandtoimprovethe

estimationofcorrelatedinput(Staudenmannetal.2006;vandeSteegetal.2014).On

average,theleadingfourprincipalcomponentswereremoved(lefthand3.6±7.2,right

hand4.5±6.0),andsignalswerereconstructedusingtheremainingPCs.Afterthis,the

EMGenvelopeswereextractedusingtheHilberttransform.Thisfull-waverectification

providesaninstantaneousamplitudeestimateoftheEMGsignal,whichreflectsthenet

inputtothemotorunitpool(BoonstraandBreakspear2012).EMGenvelopeswere

averagedacrossallgridelectrodesforeachhandseparately(vandeSteegetal.2014).


http://dx.doi.org/10.1101/056671


6

2.5DataAnalysis

2.5.1Musclesynergies

Musclesynergiescanbequantifiedaccordingtotheuncontrolledmanifold(UCM)

methodas

𝑅! = 𝑉!"#𝑉!"#

,

(1)

where𝑉!"#isthevariabilityinthedimensionirrelevantfortasksuccess(i.e.theUCM)

and𝑉!"# thevariabilityinthedimensionorthogonaltotheUCM,i.e.thedimension

relevantfortasksuccess(Latashetal.2002).Strictlyspeakingwedidnothavea

completelyuncontrolledmanifoldhere,sinceparticipantshadtocontrolbothDOFsin

thetaskparameters(x-andy-positionofcursor).Nevertheless,wemanipulatedthe

degreewithwhichbothDOFsneedtobecontrolled(Nazarpouretal.2012),andused

theUCMmethodtoquantifychangesinbimanualcoordination.Hereweusedthe

variancewithinatrialasameasureofmovementvariability(Scholzetal.2003).This

approachispreferablewhenthenumberoftrialsissmall.Movementerrorwas

computedbysubtractingthetargetforcesfromthemeasuredforcesignals.Fluctuations

intheUCMandORTdimensionswereestimatedbythesumandthedifferenceofthe

errorsignalsofbothhands,respectively.Herethedifferenceofthetwoforceswasthe

manipulatedperformancevariableandhencecouldbeconsideredtheORTdimension,

i.e.thecontrolledmanifold.

Thevariances𝑉!"#and𝑉!"# weredeterminedwithineachtrial.Theratioofthese

variancesgivesthequantitativemeasureofsynergy(RV)accordingtoequation(1).

RV<1indicatesnosynergyispresent,whereasRV>1indicatesasynergybetween

bilateralhandmuscles(Latashetal.2002).Tocorrectfortheinherentskewednessof

ratiodata,theRVvalueswerelogtransformed(logRV>0indicatesasynergyandlogRV

<0nosynergy)(Verrel2010).TheRVvalueswereaveragedoverthe10trialswithin

eachcondition.

PreviousstudieshavegenerallyinvestigatedthemeasureofsynergyRVinthetime

domain.Toinvestigatethetimescaleatwhichthebilateralsynergywasestablishedand

toinvestigatethepotentialrelationshipswithcorticomuscularcoherenceand

intermuscularcoherence,weextendedthemeasureofsynergyRVtothefrequency


http://dx.doi.org/10.1101/056671


7

domain.Scholzetal.(2003)alreadyshowedthatthereseemtobedifferenttimescales

involvedinthestructureofmovementvariability.Thepowerspectraldensity(PSD)

describesthedistributionofthevarianceacrossfrequencies,suchthattheintegralof

thePSDacrossfrequenciesisequaltovarianceofasignal.Hence,equivalenttoequation

(1)themeasureofsynergyRVinthefrequencydomainisgivenby:

𝑅! 𝑓 =

𝑃!"# 𝑓𝑃!"# 𝑓

,

(2)

where𝑃!"# 𝑓 and𝑃!"# 𝑓 arethePSDinUCMandORTdimension,respectively.The

PSDsintheUCMandORTdirectionwereestimatedusingamulti-tapermethodbased

ondiscreteprolatespheroidalsequences(Slepiansequences)astapers(Mitraand

Pesaran1999).Theamountofsmoothingthroughmulti-taperingwassetat±0.2Hz.

WeassessedRVinthetimeandfrequencydomaintocomparebothapproaches.

2.5.2Corticomuscularandintermuscularcoherence

Wefirstlocalizedthesourcesinthebrainshowingmaximalcorticomuscularcoherence

withtheEMGrecordedfromtheleftandrighthandusingDICSbeamformers(Grosset

al.2001).SincewedidnothaveaccesstoindividualsubjectsanatomicalMRIdata,we

usedthevolumeconductionmodel,sourcemodelandMRItemplatesintegratedinthe

Fieldtriptoolbox(Oostenveldetal.2011).Usingthismethod,firstthecross-spectral

density(CSD)betweenallEEGsignalpairsandbetweenallEEGsignalsandtheHDsEMG

signalswerecalculated,againusingamulti-tapermethodwithacenterfrequencyof23

Hzand±7Hzofspectralsmoothing(sensorlevelanalysisshowedsignificant

corticomuscularcoherenceinthe16-30Hzrange).WeusedtheaveragedCSDoverthe

z-scoreddataofallparticipantstolocalizetwoaveragesources.Thecoordinatesofthe

sourceswereconvertedtotheTalairachcoordinatesystemtodeterminethelocation

labelsusingtheTalairachClient(Lancasteretal.2000).

ForthetwolocationsshowingmaximalcorticomuscularcoherencewiththeEMG

envelopesoftherightandlefthand,weconstructedaspatialfilterforeachparticipant

toreconstructvirtualsourcesignals.Thevirtualsourcesignalscontainamagnitudeand

adirectionandhenceyieldthreesignals(inx-,y-,andz-direction)foreachsolution

point.Singularvaluedecompositionwasusedtoselectthedirectionin3Dspacethat

explainsmostvariance,thusleavinguswithonesourcesignalperparticipantperhand.

Magnitudesquaredcoherencebetweenthevirtualsourcesignalsandtheaveraged


http://dx.doi.org/10.1101/056671


8

HDsEMGsignals(corticomuscularcoherence)andbetweentheaveragedHDsEMG

signalsofthetwohands(intermuscularcoherence)werecalculatedovertheconstant

forcepart(t=6-15s).Again,amulti-tapermethodwasusedforspectraldecomposition.

Theamountofsmoothingtroughmulti-taperingwassetat±1.5Hz,whichmeansa3Hz

smoothingboxwasusedaroundeachfrequencyofinterest.Basedonthegrandaverages

thefrequencyrangesofinterest(i.e.showingsignificantcoherence)wereselectedfor

statisticalanalysis.

2.6Statisticalanalysis

StatisticalanalysiswasperformedinIBMSPSSStatisticsVersion21.A3×2repeated

measuresANOVAwithwithin-subjectfactorscoordinationlevelandforcelevelservedto

testforsignificantdifferencesinRVinthetimedomain.Likewise,coherencevaluesin

thefrequencybandsofinterestwerecomparedusingthesame3×2repeatedmeasure

ANOVA.Hence,four3×2repeatedmeasuresANOVAwereperformed:RV,

corticomuscularcoherencebetweenthelefthandandit’svirtualsourcesignal,between

therighthandandit’svirtualsourcesignalandintermuscularcoherencebetweenboth

hands.Thesame3×2repeatedmeasuresANOVAwasusedforeachfrequencybandof

interesttotestforsignificantdifferencesinRV(f).The95%confidenceintervalsofthe

coherenceestimatesweredeterminedthroughphaserandomization.Inaddition,

Spearman'srank-ordercorrelationwasestimatedbetweenRVandthecoherencevalues

acrossthe6(3×2)conditionsforeachparticipantseparately,todirectlytestthe

relationshipbetweenneuralsynchronyandmusclesynergies.Thecorrelation

coefficientsofthe16participantswerecomparedagainstzerousingaone-samplet-test

totesttherelationshipatgrouplevel.

3.Results

Figure2showsforcedataofarepresentativeparticipantanddemonstratesthe

effectivenessoftheexperimentalmanipulation:Inthelowcoordinationconditionthe

twoforceswereallowedtodifferfromeachother(Fig.3AandB)andthevariabilitywas

mostpronouncedintheORTdimension(Fig.3C).Incontrast,inthehighcoordination

conditionthetwoforcesdifferedrelativelylittlefromeachotherandvariabilitywas

mostlyobservedintheUCMdimension.


http://dx.doi.org/10.1101/056671


9

Figure2.Forcedataofexemplarparticipant.Therowsshowthethreecoordinationlevels

(blue=low,black=medium,red=high).A)Timeseriesofrawforcesignalsofbothhandsfor

threerepresentativetrials.B)Timeseriesoferrorsignals(forcesignalsrelativetotarget

signals).C)Clusterplotwithonthehorizontalandverticalaxesforcelevelsoftheleftandright

hand,respectively,att=12sforall16participants.ValuesinpanelCarerelativetotheaverage

overthosetrials.Ellipsesshowthe95%confidenceinterval.

RViscalculatedaccordingtoEquation(1)toquantifythelevelofmusclesynergyfor

eachconditionseparately(Fig.3).LogRVrevealedastronglysignificantmaineffectof

synergylevel(F(2,30)=182.36,p<.001).Nosignificantmaineffectofabsoluteforce

levelandnointeractioneffectwerefound.Asstated,logRV<0andlogRV>0quantifyno

synergyandsynergy,respectively.One-samplet-testsrevealedthatlogRVwas

significantlysmallerthan0inthelowcoordinationconditionforbothforcelevels(p<

.001),significantlylargerthan0inthehighsynergyconditionforbothforcelevels(p<

.001),andnotsignificantlydifferentfrom0inthemediumsynergyconditionforboth

low(p=.15)andhigh(p=.33)absoluteforcelevels.Theseresultsconfirmedthatthe

experimentalmanipulationresultedintheintendeddifferencesinbimanualsynergy

betweenconditions.

−0.4 0 0.4

0

0.4

UCM ORT

A B C

F1F2

All trials at t = 12 s

lowmedhigh

0 5 150

1

2

F [N

]Raw

0 5 15−0.4

0

0.4

Relative

0 5 150

1

2

F [N

]

0 5 15−0.4

0

0.4

0 5 150

1

2

F [N

]

t [s]0 5 15

−0.4

0

0.4

t [s]


http://dx.doi.org/10.1101/056671


10

Figure3.ThelevelofmusclesynergyRVacrossexperimentalconditions.Valuesare

averagedoverthe10trialspercondition(blue=low,black=medium,andred=high

coordination).Errorbarsshowthestandarderrorofthemean.

Figure4showsthepowerinbothdimensionsandRVasafunctionoffrequency.As

expected,mostofthepowerwasobservedinthelowerfrequencies(Fig.4A,B).Forlow

frequencies(<0.5Hz)thepowerintheorthogonaldimensiondecreasedwithincreasing

coordination(highcondition).Incontrast,thepowerintheUCMdimensionincreased

withincreasingcoordinationlevel.Consequently,RVwashigherinthehigher

coordinationconditionsfortheselowfrequencies(Fig.4C).Figure4Dshowsadifferent

frequencyprofilefordifferentcoordinationconditions:ForlowandmediumRVwas

negativeforverylowfrequencies(<0.5Hz),positiveforfrequenciesbetween0.5and2

Hz,andthenconvergedto0athigherfrequencies.Thefrequencyspectrumforthehigh

coordinationconditionwaslargelyopposite.

ThepatternofRVatfrequenciesbelow0.5Hzhencelargelymirroredthetime-domain

analysisshowninFigure3,whichisexpectedastheselowfrequenciescontainmost

powerandthuslargelydeterminetheoverallvarianceofthesignals.However,theorder

ofRVacrossconditionschangedathigherfrequencies.Forfrequenciesof0.5–2Hzthe

largestRVwasobservedinthelowcoordinationconditionanddecreasedforthe

mediumandhighcoordinationconditions(Fig.4D).Inthisfrequencyband,the

strongestsynergyisthusobservedwhenthelowestbimanualcoordinationisrequired.

Infrequenciesabove2Hztheorderacrossconditionschangedagainandwasmore

similartotheorderseenatthelowestfrequencies,thatis,thestrongestsynergywas

low highforce level

-2

-1

0

1

2

3lo

g R

v

lowmediumhigh


http://dx.doi.org/10.1101/056671


11

observedinthehighcoordinationcondition.ANOVAofRV(f)showedasignificantmain

effectofsynergylevelforfrequenciesbelow0.5Hz(F(2,30)=35.96,p<0.001),for

frequenciesbetween0.5and2Hz(F(1.4,20.8)=8.82,p=0.004)andforfrequencies

above2Hz(F(1.6,23.8)=7.47,p=0.005).Therewasnosignificantmaineffectofforce

levelandnosignificantinteractioneffectforanyofthethreefrequencyranges.

Figure4.UCManalysisinfrequencydomain.Thesolidlinesandbarsrepresentgrand

averages,averagedacrossthe20trialspercoordinationlevelandallparticipants.Inthefigure

weaveragedoverthetwoabsoluteforcelevelswithineachcoordinationlevel.Theshadedareas

surroundingthelinesandtheerrorbarsinthebargraphrepresentthestandarderrorofthe

mean.Eachcolourshowsoneofthethreeconditions(blue=low,black=medium,red=high).

A)Powerspectraldensityintheorthogonaldimension(psdORT)displayedonalogarithmic

scale.B)Powerspectraldensityintheuncontrolledmanifolddimension(psdUCM).C)Rvin

frequencydomain.Verticaldashedlinedelineatethedifferentfrequencybands.D)Rvdividedin

threefrequencybands,i.e.0-0.5Hz,0.5-2Hzand2-10Hz.

DICSrevealedthecorticalsourcesshowingmaximalcorticomuscularcoherenceinthe

betafrequencyrange(23±7Hz)foreachhand.Figure5showsthesourcesvisualized

onageneralMRItemplate(Holmesetal.,1998).Thecoordinatesofthesourcesinthe

MNIcoordinatesystemwere[4.7,0.5,5.5]cmand[-4.2,0.0,6.0]cmfortheleftandright

hand,respectively.Bothsourceswereintheprecentralgyrus(i.e.Brodmannarea6),

whichismainlycomposedofthepremotorcortex.

freq [Hz]0 1 2 3 4 5

log

PSD

ORT

-14

-10

-6A

freq [Hz]0 1 2 3 4 5

log

PSD

UCM

-14

-10

-6B

freq [Hz]0 1 2 3 4 5

log

Rv

-1.5

0

1.5

C

freq [Hz]0 - 0.5 0.5 - 2 2+

log

Rv

-1

-0.5

0

0.5

1

1.5Dlowmediumhigh


http://dx.doi.org/10.1101/056671


12

Figure5.Corticalsourcesshowingmaximalcorticomuscularcoherence.Upperrowshows

thesourcewithmaximalcorticomuscularcoherencewiththeEMGenvelopofthelefthand.

LowerrowshowsthesourcewithmaximalcorticomuscularcoherencewiththeEMGenvelopof

therighthand.Orthogonalbluelinesintersectatthemaximalcorticomuscularcoherencevoxel.

SourcesaredisplayedonanMRItemplate.

Asexpected,corticomuscularcoherencebetweenEMGenvelopesandcontralateral

virtualsourcesignalsduringtheconstantforceintervalwasstatisticallysignificantin

thebetaband(16-30Hz;Fig.6A).Importantly,therewasasignificantmaineffectof

conditionincorticomuscularcoherencebetweenlefthandandrightcortex(F(2,30)=

12.25,p<.001)andbetweenrighthandandleftcortex(F(1.6,23.5)=11.88,p=.001).

Posthoccomparisonsshowedhigherbeta-bandcoherenceinlow(p=.008)andmedium

(p=.004)comparedtohighcoordinationlevelsbetweenlefthandandrightcortex,and

higherbeta-bandcoherenceinlowcomparedtomedium(p=.01)andhigh(p=.004)

coordinationlevelsbetweenrighthandandleftcortex(Fig.6C).Incontrast,

intermuscularcoherencewasonlysignificantinthealpha-band(5-12Hz;Fig.6A)and

mainlyinthehighcoordinationcondition.Intermuscularcoherenceinthealpha-band

showedasignificantmaineffectofcondition(F(1.5,22.5)=10.19,p=0.002).Posthoc

comparisonsrevealedlargeralpha-bandcoherenceinhighcomparedtolow(p=0.015)

andmedium(p=0.002)coordinationlevels(Fig.6C).Therewasasignificantincreasein

corticomuscularcoherencebetweenlefthandandrightcortex(F(1,15)=7.09,p=

0.018)andasignificantdecreaseinintermuscularcoherence(F(1,15)=5.15,p=


http://dx.doi.org/10.1101/056671


13

0.038)athigherforcelevels.Therewasnomaineffectofforcelevelforcoherence

betweenrighthandandleftcortex(F(1,15)=0.014,p=0.908),norwerethereany

significantinteractioneffects.

Figure6.Corticomusuclarandintermuscularcoherence.Linesandbarsareaveragesover

participants,shadedareasanderrorbarsdepictthestandarderrorofthemeananddash-dot

linesarethe95%confidenceinterval.Columnsdepictfromlefttoright:corticomuscular

coherencebetweenlefthandandrightmotorcortex,coherencebetweenrighthandandleft

motorcortexandintermuscularcoherencebetweenbothhands(IMC).A)Grand-average

coherencespectraB)Coherenceinthethreecoordinationconditions(blue=low,black=


http://dx.doi.org/10.1101/056671


14

medium,red=high).TheupperandlowerrowinBshowlowandhighabsoluteforcelevel,

respectively.C)Averagedcoherenceinsignificantfrequencybands(16–30Hzfor

corticomuscularcoherenceand5–12Hzforintermuscularcoherence).Color-codingissameas

inB.*significantlydifferentatp<0.05.

Whencomparingthecorrelationcoefficientsofthe16participantsagainstzero,we

foundnegativecorrelationsbetweenRVandcorticomuscularbeta-bandcoherencewith

theleftEMG(rho=-0.42±0.07,t(15)=-5.99,p<.001)andrightEMG(rho=-0.55±0.09,

t(15)=-6.00,p<.001)andpositivecorrelationsbetweenRVandintermuscularcoherence

(rho=-0.40±0.10,t(15)=3.90,p=.001).

4.Discussion

Weinvestigatedtherelationshipbetweenfunctionalconnectivityandmusclesynergies

byexperimentallymanipulatingthelevelofbimanualcoordination.Weestimated

corticomuscularcoherencebetweentheFPBmuscleandsourceactivityinthe

contralateralmotorcortexandintermuscularcoherencebetweenbilateralFPBmuscles

atthreelevelsofbimanualcoordination.Wehypothesizedthatintermuscularcoherence

isinvolvedinmotorcoordinationandwouldhenceincreasewithstrongerbimanual

synergylevels,whilecorticomuscularcoherencewoulddecreaseasitismainlyinvolved

inthecontrolofindividualmuscles.Asexpected,corticomuscularcoherenceinthebeta

band(16-30Hz)decreasedwhileintermuscularcoherenceinthealphaband(5-12Hz)

increasedintheconditionrequiringstrongerbimanualcoordination.WeusedtheUCM

methodtoquantifythesynergybetweenbilateralhandmuscles,whichconfirmedthat

theexperimentalmanipulationhadtheintendedeffect.Weextendedthismetrictothe

frequencydomaintoinvestigatethetimescaleatwhichthebilateralsynergyis

established.Ouranalysisrevealeddistinctfrequencybands:forlow(<0.5Hz)andhigh

(>2Hz)frequenciesRVwashighestinthehighcoordinationcondition,whereasfor

frequenciesbetween0.5and2Hzthiseffectwaslargelyinversed.WhencorrelatingRV

withfunctionalconnectivityacrossconditions,wefoundnegativecorrelationwith

corticomuscularcoherenceandpositivecorrelationwithintermuscularcoherence.

Takentogether,ourfindingssuggestthatdistinctmotorpathwaysareinvolvedin

bilateralcoordinationofhandmuscles.


http://dx.doi.org/10.1101/056671


15

Thecurrentstudyconfirmsthatneuralsynchronizationisamechanisminvolvedinthe

formationofmusclesynergiesbyexperimentallymanipulatingtherequiredlevelof

bilateralcoordinationandshowingincreasedintermuscularcoherencewithincreasing

levelsofbimanualsynergy.Theseresultsextendpreviousstudiesthathaveshowna

relationshipbetweenintermuscularcoherenceandmusclesynergies.Synchronized

motorneuronactivityiscommonlyobservedinsynergisticmusclesthatacttogetherto

accomplishthesamejointmovement(DeLucaandErim2002;Laineetal.2015;Poston

etal.2010).Intermuscularcoherencehasalsobeenobservedbetweenmuscleactingon

distinctjoints,forexamplebetweenbilateralhandmuscleswhenbothfingersmovedin-

phasebutnotforanti-phasemovements(EvansandBaker2003).Similarly,homologous

TAmusclesrevealed10-Hzintermuscularcoherencewhentheyweresimultaneously

activatedduringrhythmicposturalsway(Boonstraetal.2009a).Intermuscular

coherencehasalsobeenobservedformulti-musclesynergiesinvolvedinpostural

control(Boonstraetal.2015;Danna-Dos-Santosetal.2014).Whilethesestudiesare

largelyobservational,weexperimentallymanipulatedthevisualfeedbacktostructure

theforcevariabilityoftheleftandrighthand(cf.Nazarpouretal.2012).Itisimportant

tonoticethatwhileweinducedchangesincorrelationsbetweenbilateralforce

fluctuations,themeanforceremainedunchangedexcludingpotentialeffectsofforce

levelandamplitudecancellationonintermuscularcoherence(HerouxandGandevia

2013).

Thepatternsoffunctionalconnectivityinformtheneuralcircuitryinvolvedinmuscle

synergies(Boonstra2013).Corticomuscularcoherencewassignificantinthebetaband

(16-30Hz),whileintermuscularcoherencewasonlyobservedinthealphaband(5-12

Hz;Fig.6).AsbilateralEMGactivityinthealphabandwasnotcoherentwithcortical

activity,itisunlikelythatthecorrelatedinputgeneratingintermuscularcoherencein

thisfrequencybandhadacorticalorigin.Thissuggeststhattwodistinctneural

pathwaysareinvolvedincorticomuscularandintermuscularcoherence(cf.Boonstraet

al.2009b).Thetwopathwaysmayreflectaphylogeneticallynewersystemcontaining

monosynapticinnervationstomotoneuronsofindividualmusclesanda

phylogeneticallyoldersystemcontainingdescendingprojectionsdrivingspinal

interneuronalmodules,respectively(cf.BizziandCheung2013).Thecontributionof

bothpathwaysmighthavebeenvarieddependingontherequiredlevelofbimanual

coordination,whichresultedinoppositechangesincorticomuscularandintermuscular


http://dx.doi.org/10.1101/056671


16

coherence.Byexperimentallymodifyingthedegreeofunimanualsynergies,Nazarpour

etal.(2012)foundintermuscularcoherenceinthebetabandbetweenupperarm

musclesformingamusclesynergy.Intermuscularcoherenceinthebetabandbetween

muscleswithinthesamearmmayreflectdivergentcorticospinalprojections(Farmeret

al.1993).Hencecorrelatedinputtounimanualmusclesmaybegeneratedbythe

phylogeneticallynewersystem.Itappearsthatmultipleparallelpathwaysareinvolved

intheformationofmusclesynergies(cf.TreschandJarc2009),ratherthanastrict

hierarchicalorganization(Loebetal.1999;TingandMcKay2007).Thedescending

pathways(Lemon2008)andspinalcircuitry(BizziandCheung2013)involvevarious

degreesofdivergentprojectionstodistinctmotorunitpools.Itishencepossiblethat

correlatedinputobservedinmusclesformingasynergyresultsfromtheactivationof

dedicatedhardwiredcircuits.However,Nazarpourandcolleaguesfoundthatmuscles

thathavenodirectanatomicalconnectioncouldrapidlyformsynergiesifrequiredbyan

abstractnewtasksuggestingthateveniflow-levelanatomicalsynergiesmaybeused,

thesecanbereadilyoverwritten.Theextensiveanatomicaldivergenceandconvergence

inthemotorsystemmaythusformasubstrateonwhichabstracttask-dependent

functionalsynergiesemerge(Nazarpouretal.2012).Correlatedinputmayresultfrom

multipleparallelprojectionstothemusclesandrequirethecoordinationofmultiple

sourcesthatsimultaneouslyconstrainmuscleactivity.Suchamany-to-many

relationshipwouldinvolvesynchronizationthroughoutthemotorsystemthatcanbe

convenientlystudiedusingcomplexnetworkanalysis(Boonstraetal.2015).

WequantifiedbilateralsynergiesbyextendingthemeasureofsynergyRVinthe

frequencydomain.Theseanalysesshowedthatwhilethelowfrequencies(0–0.5Hz)

mirroredthetime-domainanalysis,i.e.RVwashighestinthehighcoordination

condition,frequenciesbetween0.5and2Hzrevealedalargelyoppositeeffect.The

increasedfluctuationsinthetask-relevantdirection(VORT)observedinthisfrequency

rangearesomewhatsurprising,asitwouldsuggestthestrongestsynergyisobservedin

thelowcoordinationcondition.ThereversaloftheorderofRVacrossconditions

indicatesthatbimanualmusclecoordinationoperatesatdistincttimescales.This

confirmspreviousfindingsbyScholzetal.(2003)whoalsofounddifferenttimescales

involvedinthestructureofforcevariability.Theysuggestedcontrolontwohierarchical

levels,i.e.high-levelvoluntarycontrolthatreducesthevarianceinthetask-relevant

dimension(VORT)onalargertimescale(>250ms),andlow-levelinvoluntarycontrol


http://dx.doi.org/10.1101/056671


17

thatchannelsthevarianceintothetask-irrelevantdimension(VUCM)onasmaller

timescale(<250ms).Theapparentdifferenceintimescalesbetweenourfindingsand

theirfindings(Scholzetal.2003)likelyreflectstheirchoiceofanarbitrarycut-off

frequencyof4Hz(i.e.250ms),whereasnocut-offfrequencyisrequiredwhen

calculatingRVinthefrequencydomain.Elevatedvarianceat0.5-2Hzmayreflectan

involuntaryerrorcorrectionmechanism,thatis,itmayresultfromdelaysinfeedback

control.Visuallyguidedresponseshaveafeedbackdelayofabout200-300ms,which

wouldcorrespondtooscillationsinforcesignalswithaperiodlengthoftwicethedelay

time(i.e.400-600ms)whenvisuallytrackingatargetforce(Miall1996).Avisual

feedbacksystemmaybeinvolvedinthehigh-frequencyconditiontomorecloselymatch

bilateralforces,whichreducesbilateralvariabilityatlongertimescalesbutincreases

variabilityat0.5-2Hz.ByextendingtheRVinthefrequencydomaintheproposedmetric

enablesinvestigatingthetimescalesofmotorcoordinationinmoredetail.

Althoughourfindingsconfirmarelationbetweenintermuscularcoherenceandmuscle

synergies,thisdoesnotnecessitateacausallinkandhencedoesnotdemonstratethat

neuralsynchronyprovidesaneuralmechanismforestablishingmusclesynergies.This

wouldrequiredirectlyperturbingneuralsynchronyandobservingitseffectonmotor

coordination.Insteadofbeingthemechanismforsynergyformation,intermuscular

coherencemayreflectcorrelatedinputresultingfromdivergingmotorpathwaysthat

areengagedtoestablishmusclesynergies.Intermuscularcoherencewouldthenprovide

thespectralandspatialfingerprintsofthesepathwaysandcanhencebeusedtomap

them(Boonstra2013;Boonstraetal.2015).Thismayhelptofurtherestablishwhether

musclesynergiesareformedbyrecruitinghardwiredneuralcircuitsorbycombing

differentneuralpathwaysinatask-dependentmanner.

5.Grants

ThisresearchwasfinanciallysupportedbytheNetherlandsOrganisationforScientific

Research(NWO#45110-030).


http://dx.doi.org/10.1101/056671


18

6.References

BakerSN,OlivierE,andLemonRN.CoherentoscillationsinmonkeymotorcortexandhandmuscleEMGshowtask-dependentmodulation.JPhysiol501:225-241,1997.

BernsteinN.Thecoordinationandregulationofmovements.London:Pergamon,1967.

BizziE,andCheungVC.Theneuraloriginofmusclesynergies.FrontComputNeurosci7:51,2013.

BoonstraTW.Thepotentialofcorticomuscularandintermuscularcoherenceforresearchonhumanmotorcontrol.FrontHumNeurosci7:855,2013.

BoonstraTW,andBreakspearM.Neuralmechanismsofintermuscularcoherence:implicationsfortherectificationofsurfaceelectromyography.JNeurophysiol107:796-807,2012.

BoonstraTW,ClairboisHE,DaffertshoferA,VerbuntJ,vanDijkBW,andBeekPJ.MEG-compatibleforcesensor.JNeurosciMethods144:193-196,2005.

BoonstraTW,DaffertshoferA,RoerdinkM,FlipseI,GroenewoudK,andBeekPJ.Bilateralmotorunitsynchronizationoflegmusclesduringasimpledynamicbalancetask.EurJNeurosci29:613-622,2009a.

BoonstraTW,DaffertshoferA,vanDitshuizenJC,vandenHeuvelMR,HofmanC,WilligenburgNW,andBeekPJ.Fatigue-relatedchangesinmotor-unitsynchronizationofquadricepsmuscleswithinandacrosslegs.JElectromyogrKinesiol18:717-731,2008.

BoonstraTW,Danna-DosSantosA,XieH-B,RoerdinkM,StinsJF,andBreakspearM.Musclenetworks:ConnectivityanalysisofEMGactivityduringposturalcontrol.SciRep5:178302015.

BoonstraTW,vanWijkBC,PraamstraP,andDaffertshoferA.CorticomuscularandbilateralEMGcoherencereflectdistinctaspectsofneuralsynchronization.NeurosciLett463:17-21,2009b.

ConwayBA,HallidayDM,FarmerSF,ShahaniU,MaasP,WeirAI,andRosenbergJR.Synchronizationbetweenmotorcortexandspinalmotoneuronalpoolduringtheperformanceofamaintainedmotortaskinman.JPhysiol489:917-924,1995.

d'AvellaA,SaltielP,andBizziE.Combinationsofmusclesynergiesintheconstructionofanaturalmotorbehavior.NatNeurosci6:300-308,2003.

Danna-Dos-SantosA,BoonstraTW,DeganiAM,CardosoVS,MagalhaesAT,MochizukiL,andLeonardCT.Multi-musclecontrolduringbipedalstance:anEMG-EMGanalysisapproach.ExpBrainRes232:75-87,2014.

DeLucaCJ,andErimZ.Commondriveinmotorunitsofasynergisticmusclepair.JNeurophysiol87:2200-2204,2002.


http://dx.doi.org/10.1101/056671


19

EngelAK,FriesP,andSingerW.Dynamicpredictions:oscillationsandsynchronyintop-downprocessing.NatRevNeurosci2:704-716,2001.

EvansCM,andBakerSN.Task-dependentintermanualcouplingof8-Hzdiscontinuitiesduringslowfingermovements.EurJNeurosci18:453-456,2003.

FarmerSF.Rhythmicity,synchronizationandbindinginhumanandprimatemotorsystems.JPhysiol509:3-14,1998.

FarmerSF,BremnerFD,HallidayDM,RosenbergJR,andStephensJA.Thefrequencycontentofcommonsynapticinputstomotoneuronesstudiedduringvoluntaryisometriccontractioninman.JPhysiol470:127-155,1993.

GrossJ,KujalaJ,HamalainenM,TimmermannL,SchnitzlerA,andSalmelinR.Dynamicimagingofcoherentsources:Studyingneuralinteractionsinthehumanbrain.ProcNatlAcadSciUSA98:694-699,2001.

HerouxME,andGandeviaSC.Humanmusclefatigue,eccentricdamageandcoherenceintheEMG.ActaPhysiol208:294-295,2013.

IvanenkoYP,PoppeleRE,andLacquanitiF.Fivebasicmuscleactivationpatternsaccountformuscleactivityduringhumanlocomotion.JPhysiol556:267-282,2004.

JungTP,MakeigS,HumphriesC,LeeTW,McKeownMJ,IraguiV,andSejnowskiTJ.Removingelectroencephalographicartifactsbyblindsourceseparation.Psychophysiology37:163-178,2000.

KilnerJM,BakerSN,SaleniusS,JousmakiV,HariR,andLemonRN.Task-dependentmodulationof15-30HzcoherencebetweenrectifiedEMGsfromhumanhandandforearmmuscles.JPhysiol516:559-570,1999.

LaineCM,Martinez-ValdesE,FallaD,MayerF,andFarinaD.Motorneuronpoolsofsynergisticthighmusclessharemostoftheirsynapticinput.JNeurosci35:12207-12216,2015.

LancasterJL,WoldorffMG,ParsonsLM,LiottiM,FreitasCS,RaineyL,KochunovPV,NickersonD,MikitenSA,andFoxPT.AutomatedTalairachatlaslabelsforfunctionalbrainmapping.HumBrainMapp10:120-131,2000.

LatashML,ScholzJP,andSchonerG.Motorcontrolstrategiesrevealedinthestructureofmotorvariability.ExercSportSciRev30:26-31,2002.

LatashML,ScholzJP,andSchonerG.Towardanewtheoryofmotorsynergies.MotorControl11:276-308,2007.

LemonRN.Descendingpathwaysinmotorcontrol.AnnuRevNeurosci31:195-218,2008.

LevineAJ,HinckleyCA,HildeKL,DriscollSP,PoonTH,MontgomeryJM,andPfaffSL.Identificationofacellularnodeformotorcontrolpathways.NatNeurosci17:586-593,2014.


http://dx.doi.org/10.1101/056671


20

LoebGE,BrownIE,andChengEJ.Ahierarchicalfoundationformodelsofsensorimotorcontrol.ExpBrainRes126:1-18,1999.

MiallRC.Task-dependentchangesinvisualfeedbackcontrol:afrequencyanalysisofhumanmanualtracking.JMotBehav28:125-135,1996.

MitraPP,andPesaranB.Analysisofdynamicbrainimagingdata.BiophysJ76:691-708,1999.

NazarpourK,BarnardA,andJacksonA.Flexiblecorticalcontroloftask-specificmusclesynergies.JNeurosci32:12349-12360,2012.

OostenveldR,FriesP,MarisE,andSchoffelenJM.FieldTrip:OpensourcesoftwareforadvancedanalysisofMEG,EEG,andinvasiveelectrophysiologicaldata.ComputIntellNeurosci2011:156869,2011.

PostonB,Danna-DosSantosA,JesunathadasM,HammTM,andSantelloM.Force-independentdistributionofcorrelatedneuralinputstohandmusclesduringthree-digitgrasping.JNeurophysiol104:1141-1154,2010.

SaleniusS,PortinK,KajolaM,SalmelinR,andHariR.Corticalcontrolofhumanmotoneuronfiringduringisometriccontraction.JNeurophysiol77:3401-3405,1997.

SchnitzlerA,andGrossJ.Normalandpathologicaloscillatorycommunicationinthebrain.NatRevNeurosci6:285-296,2005.

ScholzJP,KangN,PattersonD,andLatashML.Uncontrolledmanifoldanalysisofsingletrialsduringmulti-fingerforceproductionbypersonswithandwithoutDownsyndrome.ExpBrainRes153:45-58,2003.

ScholzJP,andSchonerG.Theuncontrolledmanifoldconcept:identifyingcontrolvariablesforafunctionaltask.ExpBrainRes126:289-306,1999.

SingerW.Neuronalsynchrony:aversatilecodeforthedefinitionofrelations?Neuron24:49-65,111-125,1999.

StaudenmannD,KingmaI,DaffertshoferA,StegemanDF,andvanDieenJH.ImprovingEMG-basedmuscleforceestimationbyusingahigh-densityEMGgridandprincipalcomponentanalysis.IEEETransBiomedEng53:712-719,2006.

TingLH,andMacphersonJM.Alimitedsetofmusclesynergiesforforcecontrolduringaposturaltask.JNeurophysiol93:609-613,2005.

TingLH,andMcKayJL.Neuromechanicsofmusclesynergiesforpostureandmovement.CurrOpinNeurobiol17:622-628,2007.

TreschMC,andJarcA.Thecaseforandagainstmusclesynergies.CurrOpinNeurobiol19:601-607,2009.

TreschMC,SaltielP,andBizziE.Theconstructionofmovementbythespinalcord.NatNeurosci2:162-167,1999.

TurveyMT.Coordination.AmPsychol45:938-953,1990.


http://dx.doi.org/10.1101/056671


21

vandeSteegC,DaffertshoferA,StegemanDF,andBoonstraTW.High-densitysurfaceelectromyographyimprovestheidentificationofoscillatorysynapticinputstomotoneurons.JApplPhysiol116:1263-1271,2014.

vanWijkBC,BeekPJ,andDaffertshoferA.Neuralsynchronywithinthemotorsystem:whathavewelearnedsofar?FrontHumNeurosci6:2012.

VerrelJ.Distributionalpropertiesandvariance-stabilizingtransformationsformeasuresofuncontrolledmanifoldeffects.JNeurosciMeth191:166-170,2010.

WilliamsER,andBakerSN.Circuitsgeneratingcorticomuscularcoherenceinvestigatedusingabiophysicallybasedcomputationalmodel.I.Descendingsystems.JNeurophysiol101:31-41,2009.


http://dx.doi.org/10.1101/056671


functional connectivity in neuromuscular system underlying

Documents