Planning of spatially-oriented locomotion following focal brain damage in humans: A pilot study

(1)

Planning

of

spatially-oriented

locomotion

following

focal

brain

damage

in

humans:

A

pilot

study

Halim

Hicheur

a,∗

_,

_Carole

_Boujon

a

_,

_Cuebong

_Wong

b

_,

_Quang-Cuong

_Pham

b

_,

Jean-Marie

Annoni

c

_,

_Titus

_Bihl

d

a_Unit_of_Sport_and_Movement_Sciences,_Dept._of_Medicine,_University_of_Fribourg,_Switzerland

b_School_of_Mechanical_and_Aerospace_Engineering,_Nanyang_{Technological}_University,_Singapore

c_Chair_of_Neurology,_Dept._of_Medicine,_University_of_Fribourg_&_HFR,_Switzerland

d_{Neurorehabilitation}_Unit,_Hôpital_Fribourgeois_(HFR),_Switzerland

h

i

g

h

l

i

g

h

t

s

•Motorimpairmentsfollowingbraindamagearewelldocumented.

•Inparticular,hemiparesisisacommonfeatureofstrokepatients.However,thenavigationalabilities(here,theabilitytoplanapathtowardsadistant targetwithvisionorblindfolded)ofsuchseverelydisabledpatientshavebeenpoorlyinvestigated.

•Here,wereportthathemipareticpatientswithsigniﬁcantlesionsofthesensory-motorsystemfollowingfocalbraindamagearenotimpairedduring spatially-orientedlocomotionperformedwithvisionorblindfolded.

•Thissuggeststhatthemotorcortexisinvolvedinthesensorimotorimplementationoflocomotionbutisnotinvolvedinthepathplanningstagein humans.

•Suchﬁndingsopennewavenuesforpost-strokerehabilitationwherespeciﬁctests(i.e.,adaptedversionsofourparadigm)mayallowparallelassessment ofcognitiveandlocomotorfunctionsrecoveryafterstroke.

Motor impairments in human gait following stroke or focalbrain damage are well documented. Here,weinvestigatedwhetherstrokeand/orfocalbraindamagealsoaffectthenavigational compo-nentofspatiallyorientedlocomotion. Tenhealthyadultparticipantsandtenadultbrain-damaged patientshadtowalktowardsdistanttargetsfromdifferentstartingpositions(withvisionor blind-folded). No instructions as to which the path to follow were provided to them. We observed very similar geometrical forms of paths across the two groups of participants and across visual conditions. This spatial stereotypy of whole-body displacements was observed following brain damage, even in the most severely impaired (hemiparetic) patients. This contrasted with much more variability at the temporal level. In particular, healthy participants and non-hemiparetic patients varied their walking speed according to curvaturechanges along the path. On the con-trary,thewalkingspeedproﬁleswerenotstereotypicalandwerenotsystematicallyconstrainedby path geometry inhemiparetic patientswhereitwas associatedwith differentsteppingbehaviors.

Abbreviations: VI,visualwalkingcondition;BF,blindfoldedwalkingcondition;PH,hemipareticpatients;PN,nonhemipareticpatients;CO,controlgroup(healthy

participants);SN,step’snumber;SLR,steplengthratio;SDR,stancedurationratio;ST,straighttrajectories;LC,lowcurvaturetrajectories;MC,moderatecurvaturetrajectories;

HC,highcurvaturetrajectories;ATD,averagetrajectorydeviation(meanofspatialvariabilityaroundthemeantrajectoryacrosstrialsandsubjects);MTD,maximaltrajectory

deviation(maximumofspatialvariabilityaroundthemeantrajectoryacrosstrialsandsubjects);ATS,averagetrajectoryseparation(meandistancebetweenmeantrajectories

oftwovisualconditionsortwogroupsofparticipants);MTS,maximaltrajectoryseparation(maximaldistancebetweenmeantrajectoriesoftwovisualconditionsortwo

groupsofparticipants);AMPVEL,amplitudeofvelocityvariationsduringlocomotorpathcompletion.

∗ Correspondingauthor.Fax:+41264268135.

E-mailaddress:halimhicheur@yahoo.fr(H.Hicheur).

Theseobservationsconﬁrmthedissociationbetweencognitiveandmotoraspectsofgaitrecovery post-stroke.Theimpactoftheseﬁndingsontheunderstandingofthefunctionalandanatomicalorganization ofspatially-orientedlocomotionandforrehabilitationpurposesisdiscussedandcontextualizedinthe lightofrecentadvancesinelectrophysiologicalstudies.

Published in %HKDYLRXUDO%UDLQ5HVHDUFK±

which should be cited to refer to this work.

(2)

1. Introduction

Howdoesthebrainstorespatialinformationaboutourposition andorientationinoursurroundingenvironment?Howdoweﬁnd orplanourwayfromoneparticularspatialpositiontoanother? Answerstothesequestionshavecomemainlyfromanimal stud-ieswheredifferenttypesofcellsintherathippocampusandin neighboringregions(e.g.,para-hippocampusandentorhinal cor-tex)werefoundtosignalvariousspatialattributes[seeRef. [1] forarecentreview].Theseincludethepositionoftheanimalin theroom(theso-called“placecells”),itsspatialorientation(“head directioncells”) and, themetrics or theboundaries ofthe sur-roundingspace (“grid”and “border”cells, respectively).Clinical andexperimentaldatainhumansdohoweversuggesttheroleof amoredistributednetworkduringactivenavigation,asevidenced bytheabsenceofbehavioraldeﬁcitsofhippocampalpatientsin pathintegrationtasks[2,3]orbytheroleplayedbybasalganglia inthesteeringofblindfoldedwalkingincircles[4].Inthecaseof pathintegrationtasks,participantsarerequiredtokeeptrackofa referencelocationusingself-motioncues(forexamplebyasking themtoreturntotheirstartingpositionafterhavingbeen pas-sively/activelydisplacedalongapathwithoutvision).Importantly, theinstructionsprovidedtoparticipantscanaffecttheway naviga-tionalabilitiesaremeasured[5]andthenavigationperformance. Forinstance,theinstructionto“maintainthepathinmind”used insomepathintegrationparadigms[3]duringthelearningphase mayforceparticipantstoupdatetheirpositionwithreferencetothe imaginedpath(a“map”representationoftheirbodydisplacement inspace,anallocentricstrategy).Intheabsenceofsuch(explicit) instructions,participantsmightspontaneouslyestimatetheir start-ingpositionbyupdatingtheirpositionstepbystepwithreference totheinitialstarting position(anegocentricorroutestrategy). Asaconsequence,assessingspatialnavigationperformanceusing pathintegrationtasksisproblematicbecauseofpossible interfer-encesbetweenthesespatialprocessing/memorystrategies,hence makingitunclearwhether,undernaturalconditions,participants wouldhavememorizedspatialattributesoftheenvironment, spa-tialattributesoftheirdisplacementswithintheenvironment,or somecombinationthereof.

Thisproblemcanbeovercomebyaskingparticipantsto gener-ate“spontaneous”walkingbehaviors.Inpreviousyears,wetesteda simplegoal-orientedtaskofwalkingtowarddistanttargets(either doorwaysorarrowsplacedontheground).Importantly,nospeciﬁc constraintwasimposedonhealthyparticipantsintermsofthepath theyweretofollow.Participantshadtoperformthesetasksusing theirvisionorblindfolded[6,7].Strikingly,weobservedthatthe generatedlocomotorpathsweresimilaracrossvisualconditions (withvisionorblindfolded)and thatneitherwalkingspeednor walkingdirection(forwardorbackward)signiﬁcantlyaffectedthe shapeofthesepaths[8].Therecordedbodytrajectoriescouldbe predictedbyacombinationoffeed-forwardandfeed-back mech-anisms,dedicatedtoaccountingforthe“global”(path-planning) and“on-line”contributions(visualguidance)tolocomotorpaths formation[7,9].Thissuggestsadissociationbetweenspatial cogni-tionandsensorimotorcontrolmechanismsatworkduringspatial navigation.Stereotyped locomotortrajectorieswerereportedin

adolescentsandadults butnot inchildren under11years [10], showingthatpathplanningdevelopsinlatechildhood(wellafter gaitmaturation),whichsuggestsadistinctdevelopmentofpath planningvsgaitmotorstabilityabilities.

Understandingthepotentialinterferences betweencognitive andmotorprocessesisofparticularimportancefollowingstroke [11].Here, weproposeaproofofconcept analysisdedicatedto measuringthenavigationalperformanceofpatientswithmotor andcognitivedeficitsfollowingbraindamage.Giventhe dissocia-tionbetweenmotorandnavigationalcomponentssuggestedbyour previousfindings,wewereexpectingthattheshapeofbody tra-jectoriesinspace,whichmostlyreflectspatialcognitionprocessing (pathplanning),wouldremainunaffectedbysuchmotordeficits. Incontrast,themotorimplementationofthesetrajectories(e.g., thesteppingbehavior)wouldbeaffectedbysuchdeficitswhich, inourmodel,wouldresultingreatervariabilityaroundthemean trajectory.

2. Materialsandmethods 2.1. Participants

Tenpatients(agedbetween28and68years,sixfemales/four males)withchronicbrainlesionsfollowingcerebrovascularevents, andtenage/gender-matchedhealthypeople(agedbetween28and 70years,withoutanyhistoryofneurologicaldisease)volunteered toparticipateinthisexperiment.Allpatientsfulfillingthe exclu-sion/inclusioncriteria defined prior toexperimental recordings wereaskedtoparticipateinthestudy.Theinclusioncriteriawere asfollows:survivorsofafirst-timecerebrovasculareventresulting instructuralsupratentorialor infratentorialcorticallesions(see Table1fordetailedinformationaboutpatientsandSupplementary materialforMRItemplates),admittedtothein-patient rehabilita-tionunitoftheFribourgCantonalHospital(HFR),agedbetween18 and80years,abletounderstandthemeaningofthestudyandto followinstructions,andwithgoodwalkingability.Exclusion crite-riawereasfollows:acutehealthproblemswhichwouldinterfere withthereliabilityofthetask(infections,decompensateddiabetes, etc.),questionablecardio-pulmonarystatus(cardiacfailure, pul-monaryembolism,oxygentherapy),patientswithacutevigilance andspatialdisorders(confusion,disorientation,etc.)atthetime oftheexperimentalrecordings,patientswitheyedisorders and non-correctedvisionproblems(advancedmaculardegeneration, blindness,etc.),andpregnancy.

Allselected patientsexhibited paresis in the first week fol-lowingbraindamage.Five exhibitedhemiparesisatthetimeof the experimental recordings which took place more than two monthsaftertheonsetofstroke.Theycouldwalkatleast1000m without the need to stop for rest. All participants had nor-malorcorrected-to-normalvision.Allpatientsexhibitedresidual cognitive deficits at the time of evaluation: four had aphasia, one had spatial neglect, two had memoryimpairment and six haddysexecutivesyndrome.Longstandingattentionaldeficitwas present in three patients. Nevertheless,all of themunderstood theinformationandordersgivenduringtheenrolmentandtest processes.Participantsgavetheirinformedconsentpriortotheir

(3)

Table1

Demographicandneurologicaldataofthe10patients.‘Side’standsforthesideoflesion:L= Left,R=Right.Delaymeanspoststrokedelayuntilgaitevaluationandis

giveninmonths.ED=Executivedysfunction.Concerningsiteoflesion:PFC=prefrontalcortex;M1=PrimarymotorCortex,PO=Parieto-occipitalcortex,BG=Basalganglia;

TO=Temporo-occipitalcortex;PTO=Parieto-temporo-occipitalCortex;FP=Fronto-parietalcortex;CRB=Cerebellum.

PatientsGenderAgeAetiology DelayParesis FACscoreSideCognitivedeﬁcit Siteoflesion

P01 F 60 Haemorrhagicstroke 4 No 6 R Spatialneglect,attention,ED PFC,M1,insula,parietalcortex,BG

P02 F 28 Ischemicstroke 8 No 6 R Attention,anterogradeamnesia TO,BG,CRB

P03 M 66 Ischemicstroke 8 Yes 5 L Fluentaphasia,apraxia,ED M1,insula,BG

P04 M 49 Ischemicstroke 5 Yes 4-5 L Non-ﬂuentaphasia,apraxia,ED M1,insula,PTO

P05 M 67 Meningiomasurgery 8 No 6 L Attention FP,involvementofM1

P06 F 57 Subarachnoidhaemorrhage 8 No 5 R Motoraphasia,attention,anterogradeamnesia,EDFP

P07 M 68 Haemorrhagicstroke 2 Yes 6 L Mildlearningdeﬁcit M1/insula

P08 F 60 Vasculitis 13 Yes 4 L Anxiety,ED BG

P09 F 30 Ischemicstroke 11 Yes 6 L Globalaphasia,lefthandapraxia, M1,PO,BG

P10 F 51 Subarachnoid 2.5 No 6 R ED PFC,insulaandleftBG

inclusioninthestudy.ExperimentsconformedtotheCodeofEthics of theDeclaration ofHelsinki and wereapproved bythe Com-mission cantonaled’Ethique dela Recherche surl’Etre humain (VD,Switzerland)and registeredunderreferenceNCT02263560 ontheNIHClinicalTrials.govClinicalTrials.govdatabase.Importantly, all patientscouldwalkautonomously (one hemipareticpatient walkedwithacane)asmeasuredbyaFunctionalAmbulation Cat-egories(FAC)walkingtest[12],forwhichallpatientshadscores equalorsuperiorto4(onthe6-pointFACscale).Inordertotest theirabilitytowalkblindfolded,weaskedpatientstowalk sev-eralstepswhilstblindfoldedinaroom,toturnaroundandtowalk backwithandwithoutphysicalassistance(allpatientssucceeded inthistest).Thepatientswereincludedinthestudyonlywhen theyexpressedthefeelingofsafety(subjective)andwhenno near-falleventoccurred(objective).Allpatientsbutonestayedatleast oneweekatHFR,andreturnedhomeatleastfourweeksbefore theexperimentalrecordings.P07wastestedfewdaysbeforehe returnedhomehavingspenttwomonthsatHFR.Importantly,all patientswerefollowedbythesameneurorehabilitationteam. 2.2. Protocol

Theexperimentstookplaceinalaboratory,thedimensionsof whichwere8.7×6×3.3m(length,widthandheightrespectively). Theprotocolwassimilartotheoneusedinourpreviousstudies [seeRefs.[6,7]andSupplementarymaterial].Briefly,participants hadtostartfromoneofthreefixedpositionsinthelaboratory(left, centerorright)andtowalktowardadistanttargetindicatedby anarrowplacedontheground(seeFig.1A).Thedimensionsof thearrowwere1.20×0.25m(lengthandwidth,respectively).The arrowwasplacedataspecific(x,y)positionintheroomwitha particularorientation(South,East,North andWest,respectively S,E,NandW).Intheblindfoldedcondition,theparticipantfirst observedthearrow whilestandingatthestartingposition.This observationperiodtypicallylastedlessthanthreeseconds.When he(orshe)wasready,heclosedhiseyesandattemptedto com-pletethetaskwithoutvision.Thestartingsignalwasgivenbythe experimenterbytouchingtheparticipant’sshoulderwithhishand (forboth“visual”and“blindfolded”conditions).

2.3. Experimentalconditions

Everyparticipantgenerated114trajectories.Forthe“straight targets”,participantshadtoperform18trials:onecentral start-ingposition(C)×threearrowpositions(1,2and3)×onearrow orientation(N)×two visualconditions(visual VIor blindfolded BF)×three repetitions=18 trajectories. Concerning the “angled targets”,theyhadtoperform96trials:twostartingpositions(L andR)×twoarrowpositions(2and3)×fourarrowdirections(S, E,WandN)×twovisualconditions(VIandBF)×threerepetitions. Thetrialswererandomized(intermsofstartingposition,targets’

position and orientation, visual condition and repetitions) in order toavoid anylearningeffect. Thesewererecorded intwo experimentalsessionsondifferentdays(P04tookpartinthree experimentalsessions).Inadditiontotheinter-trialinterval(which typicallylasted10–20s),arestperiodofﬁveminutesoccurredat themiddleoftheexperimentalsession.Atotalof2280trajectories (1140inpatientsand1440incontrols)wererecorded.Becauseof problemsindataacquisition(withmarkersmissingforatleastone second),22trials(outof2280)wereexcludedfromtheanalysis; thisinvolvedsixtrialsfromhealthyparticipantsand16trialsfrom patients.

2.4. Experimentalrecordings

Three-dimensional positions oflight reﬂectivemarkers were recordedata120Hzsamplingfrequencyusinganoptoelectronic Optitrackmotioncapturesystem(NaturalPointInc.,Oregon,USA) wiredto15cameras.Sixmarkerswereattachedtomotioncapture suitsorfootwraps(respectively)throughvelcro-friendlysurfaces (Optitrack).Twowereplacedontheleftandrightshouldersatthe leveloftheleftandrightacromions.Theywereusedtostudywhole bodytrajectoriesinspace[6,7].Twomarkerswereplacedatthe level oftheheelandthird toeofeach foot.Participantsworea headsetwhichpreventedhearingsoundsfromoutside.

2.5. Dataanalysis

Most of thefollowing methods were presented in previous studies [6,7]. We will describe here the main methodological procedures which allowed us to analyze and compare trajec-tories produced in the different conditions and acrosshealthy (control) participants (CO) and patients (hemiparetic PH and non-hemiparetic PN).The readeris referredto [6,7]and tothe Supplementarymaterialforfurtherdetails.

2.6. Generalparametersandsteppingbehavior

The length of the whole-body trajectories in space, the movement execution duration and the steps’parameters were computed.Weusedheelstrikeandtoeoffeventstodeﬁnesteps [24].Theseeventswerederivedfromthetimecourseofheeland toe Zpositionproﬁlesandcorrespondedtothelocalminimaof thesetwosignals.Weconsideredonestepastheintervalseparating twosuccessiveheelstrikesofthesamefootandcomputedthefeet positionsattheseparticularevents.Thenumber,lengthandstance durationofleft(non-pareticlimbinPHpatients)andright(paretic limb)stepswerecomputedseparately.Thetotalnumberofsteps (SN) and the Step Length/Stance Duration(non-paretic/paretic) Ratios(SLRandSDR,respectively)werecomputedtodocumentthe steppingbehaviorandthepotentialgaitasymmetries(expectedin PHpatientsinparticular).

(4)

Fig.1.Illustrationoftheexperimentalprotocol:A—Participantshadtostartfromoneofthreestartingpositions(L,CandR),toenterthearrowbytheshaftandtostop

walkingwhenatthetipofthe(visibleormemorized)arrowforbothvisualconditions(withvisionorblindfolded).Emptycirclesontheshouldersandthefeetrepresent

markers.Thearrowcouldbeplacedatthreedifferentlocations1–3andorientedalongE,W,NorSdirections(seeB).ForthelateralstartingpositionsLandR,participantshad

toperformthetaskinalldirectionsforpositions2and3.ForthecentralstartingpositionC,theyhadtoperformtargets1N,2Nand3Nonly.Inthisexample,theparticipant

hadtowalktowardthe3StargetfromthestartingpositionL(L3S).

2.7. Categorizationandcomputationofthetrajectories

Here,thetested trajectorieswereclassifiedaccordingtothe amountofbodyrotationtheyrequired.Fourcategorieswere dis-tinguished:quasi-straighttrajectoriesST,lowLC,moderateMCand highlyHCcurvedtrajectories.Thebeginning(t=0)ofeach trajec-torywassettothetimeinstantwhentheparticipantcrossedthe X-axisaty=0.5(the averagelengthofthefirst“straight-ahead” step).InordertohavethesamecriterionfortheVIandBF condi-tions,theendofeachtrajectory(t=1)wassettothetimeinstant when theparticipant’sspeed became less than 0.06ms−1 (this valuewaslessthan5%oftheaveragenominalwalkingspeed).We chosethisstrictlypositivethresholdbecauseofthesmallresidual movementsoftheupperbodyoccurringafterparticipantsstopped walking.Whenaderivativeofthepositionwasneeded(tocompute velocityprofile,for instance), a second-order Butterworthfilter withcut-offfrequency6.25Hzwasappliedbeforethederivation. 2.8. Spatialandtemporalattributesofthelocomotortrajectories

Thereaderisreferredtothesupplementarymaterialfordetails ofthecalculationofthefollowingparameters.Thespatial variabil-ityoftheactualtrajectoryaroundtheaveragetrajectory(averaged acrossrepetitionsandsubjects)wasmeasuredusingtheaverage

andmaximaltrajectorydeviationparameters(ATDandMTD),for everyparticulartarget.Thecomparisonof theaverage trajecto-riesrecordedintwoconditions(VIandBF) orbetweenpatients and healthyparticipants inthe sameconditionwere measured usingtheaverageandmaximaltrajectoryseparationparameters (ATSand MTS).These parametersareexpressedincentimeters. Atthetemporallevel,thecomputationofthevariabilityaround theaveragevelocityprofilesfocusedoninvestigatingwhether par-ticipantsvariedtheirwalkingspeedatsimilarinstants/positions along thetrajectory.Thiswasquantifiedusingtheaverageand maximalvelocitydeviationparametersAVDandMVD(expressed in meters/s). Theobservation ofquasi-constant walking veloci-tiesinhemipareticpatientsobviouslyresultedinhighAVD/MVD values(whichmeasuredeviationsfromaveragevelocityprofiles) acrosssub-groupsofpatients.Therefore,wealsomeasuredhow muchwalkingvelocityvariedduringpathcompletion.Thiswas doneby computing the amplitude of walking speed variations (AMPVEL) during path completion for every trial. Importantly here,high-frequencyoscillationsinducedbysteppingactivitywere removedtofocusspecificallyontheglobalvariationofthewalking speedinducedbycurvaturevariationsalongthepath[13,14].This wasdonebyindividuallyadjustingthecut-offfrequenciesofthe second-orderButterworthfilterusedwhencomputingthevelocity

(5)

proﬁles(withaﬁxedvalueof0.5Hzforhealthyparticipantsandin the0.2–0.5Hzrangeforpatients).

2.9. Statisticaltests

AllstatisticalcomparisonsweredoneusingtheStatistica8.0 software package(Statsoft®_)._We _performed_repeated measure-ment analysesof variance(ANOVA)tocomparetheparameters (ATD, MTD, AVD and MVD) calculated for the 2258 recorded trials (dependent variables: four categories×two visual condi-tions×threerepetitions;categoricalvariables:twogroups).These comparisonsallowedustoquantifytheeffectsofthemagnitudeof curvature(categoriesST,LC,MCandHC),thevisualcondition(VIvs BF)andthegroup(patientsPH+PNvsCOparticipants)onthe spa-tialandtemporalattributesofthetrajectories.Asecondseriesof ANOVAanalyseswereperformedtocomparegeneral(e.g.,walking speed,travelleddistancesandwalkingdurationwhencompleting apath)andlocalstepparameters(e.g.,numberofstepsto com-pleteapath,steplength/durationratios).Indeed,theseparameters (exceptforthesteplength/durationratios)wereobviouslymore dependentuponinitialdistances(betweenthestartingpointand thetargetposition/orientation)thanonthemagnitudeof curva-ture.Thus, trials werecategorized according tothesedistances into11categoriesoftrajectories(11categories×twovisual condi-tions×threerepetitions;categoricalvariables:twogroups).Before performingeachoftheserepeated-measuresANOVAcomparisons onthe patients’ data, we ﬁrst checkedfor the homogeneityof variancesbyperformingaMauchley’ssphericitytest.The normal-ityofeachofthecomputeddistributionswasthentestedusing Kolgomorov–Smirnovtests.Ifbothhypotheses(varianceand nor-mality)wererespected,weperformedtheANOVAcomparisons.

When statistically signiﬁcant differences (p<0.05) were observed between sub-groups of participants (e.g., PH and PN patients),theANOVAcomparisonswereperformedasecondtime onthepopulationofpatientsonly(N=10),totestfortheeffectof hemiparesisonthecomputedvariables.Here,thetestsofsphericity andnormalitywereperformedonthepatients’data.Anyatypical behaviorobservedforaparticularpatientorhealthyparticipant (detectedornotbythesecomparisons)willbementionedinthe text.

3. Results

3.1. Generalparametersandsteppingbehavior

Onaverage,hemipareticpatientswalkmoreslowlythanhealthy controls and non hemiparetic brain lesion patients [15]. We observedthesamegroupeffecthere:themeanwalkingspeedof COwasequalto0.98±0.11and0.85±0.12m/s(visualand blind-foldedtrials,respectively)while itwasequalto0.88±0.15and 0.74±0.15m/sinPNandto0.59±0.19and0.47±0.17m/sinPH (F(1,16)=14,879,p<0.01).Wealsoobservedasignificanteffect ofthecategoryoftrajectories(F(10,160)=58,394,p<0.01)onthe walkingspeed(seeFig.6AoftheSupplementarymaterialformean valuesforallgroupsandcategories),withthehighestspeedsbeing reached for the second and third straight targets (ST)and the lowestspeedsbeingreachedfor thefourmostangled(HC) tar-gets.Thevisualconditionalsosignificantlyaffectedthewalking speed(F(1,16)=121,8,p<0.01).The(category×vision)interaction effectwassignificant(F(10,160)=8,0425,p<0.01)whilethe (cat-egory×group)interactioneffectwasnotsignificant(p>0.05).We didnotobserveanyotherinteractioneffect.Theeffectof hemipare-sisonthewalkingspeedwasassessedbyperformingANOVAonthe patients’group.Thiseffectwassignificant(F(1,8)=6,87,p=0.03): PHwalkedatasignificantlylowerspeedthanPN.Weobserved

hereasigniﬁcant(category×hemiparesis)interactioneffect(F(10, 80)=2149,p=,029)withwalkingspeedsbeingnearlyconstantin thePHsub-groupacrossthe11testedcategoriesoftrajectories. However,theindividualanalysisofthewalkingspeedswithineach sub-groupofpatientsrevealedthatPHpatientsP07andP09had walkingspeedscomparabletothatofPN.Thiseffectof hemipare-siscanthusbeattributedtoPHpatientsP03,P04andP08.Thus, hemiparesissigniﬁcantlyreducedthewalkingspeed(comparedto PN)whichwasconstantacrossstraightandangledtargetsforthese threePHpatientsonly.

Thesechangesobservedatthelevelofthewalkingspeedswere associatedwithcorrespondingchangesattheleveloftraveled dis-tancesanddurationtimes.Thedetailedresultsofthisanalysisas wellasthoseofthesteppingbehavioraredescribedinthe supple-mentarymaterial.Brieﬂy,weobservedthatthetraveleddistances werecomparable acrossgroupsof participants.Thus, thelower averagewalkingspeedresultedinlongermovementdurationsin PHpatients.

The analysisof thesteppingbehavior provided observations comparabletothatofotherstudiesonhemipareticgait[16].The strongestmarkerofgaitasymmetryinPHwasalongerstancephase durationofthenon-pareticlimb.Importantly,wedidnotobserve anystatisticallysigniﬁcanteffectoftheturningdirection(withPH turningﬁrstwiththeirpareticornon-pareticlimbsforrightand leftturns,respectively)onallcomputedparameters,asreported inarecentstudy[16].Takentogether,theseresultsrevealedthat, togeneratetrajectoriesofcomparabledistances,differentstepping patternswereimplementedindifferentgroupsofparticipants.

3.2. Spatialattributesofthewhole-bodytrajectories

Typical trajectories observed for healthy participants and patientsaredepictedforfourtargetsinFig.2.Thesimilar geomet-ricalformofthewhole-bodytrajectoriesacrossvisualconditions andgroupsisremarkable.Theabsenceofeffectofcorticaland sub-corticallesions(seeFig.2B5andC5)onthegeometricalformof trajectoriesbothduringVIandBFtrialsshouldbenoticed,even forthehemipareticpatientP04whosuffersfromimportant corti-callesionsinthelefthemisphere.Inthispatientandintheother PH patients,onecouldobservelocaloscillationsduringthe tra-jectory which are due totheparticular steppingactivity ofPH patients(seeabove).Thesimilaritybetweenhealthyparticipants andpatients’averagetrajectoriesononehand,andbetweenVIand BFaveragetrajectoriesontheotherhand,wasquantifiedusing theATSandMTSparameters(Fig.3AandB).Thespatial trajec-toryseparation betweengroups rangedbetween4/5(ATS/MTS) and 9/12cmfortheVItrialsandbetween3/6and22/32cmfor theBFtrials,respectively.TheATSandMTSparameterscomputed between visualconditions were of comparable magnitude. The spatial variability aroundthe averagetrajectory wasquantified usingtheATDandMTDparameters(Fig.3AandB).Thesenever exceeded25cm(ATD),47cm(MTD),48cm(ATD)and84cm(MTD) centimetersforVIandBFtrials,respectively.TheANOVA compar-isons revealednoeffectofthegroup(p>0.05).Theyrevealeda significanteffectofthecategory(F(3,48)=47,23,p<0.01andF(3, 48)=34,53,p<0.01)andofthevisualcondition(F(1,16)=110,3, p<0.01andF(1,16)=142,8,p<0.01)aswellasaninteraction (cat-egory×visualcondition) effect (F(3, 48)=5,68, p<0.01 and F(3, 48)=4,18,p=0.010)ontheATD/MTDparameters,respectively.The variabilityincreasedwithincreasingcurvatureandforBFtrials[as reportedinpreviousstudies,seeRefs.[6,7]].

Takentogether,theseresultsindicatethatallgroupsof partici-pantsgeneratedsimilarformsoflocomotorpathsacrosscategories andvisualconditions.

(6)

Fig.2.Typicaltrajectoriesgeneratedbycontrolparticipants(A),non-hemiparetic(B)andhemiparetic(C)patients.Panels1–4indicatestraight-aheadwalking(ST),low

curvature(LC),mediumcurvature(MC)andhighcurvature(HC)trajectories.Panel5indicatethelesionedbrainareainpatients(seeTable1fordetails).Averagetrajectories

performedforvisualandblindfoldedtrialsarerepresentedbythicklines(redandblackcolors,respectively).Thevariabilityaroundtheaveragetrajectoryisrepresentedby

theshadowregion.High-frequencyoscillationsaroundthetrajectorywereobservedinPHpatientsonlyandareassociatedwiththespeciﬁcsteppingpatternofPHpatients

(seeSupplementarymaterialfordetailsofgaitpatternchangesinallgroups).Notethegreatsimilarityofthelocomotortrajectoriesforallgroupsandvisualconditions.Note

alsothatthevariabilityaroundtheseaveragetrajectories(shadowregion)ishigherwithoutvisionforallgroupsandvisualconditions.

3.3. Temporalattributesofthewhole-bodytrajectories

Thewalkingvelocityvariationsduringtrajectorygenerationare depictedforthesameparticipantsandtargetsasthosepresentedin Fig.2(Fig.3).Thefirstandmainobservationisthedifferent veloc-ityprofilesofPHpatientscomparedtoPNandhealthyparticipants. Indeed,notonlydoPHpatientswalkmoreslowlythantheother groups,buttheyalsomaintainaconstantvelocityduringtheir dis-placementintheroom,forboth“straight”and“angled”targets.This contrastswithsimilarvelocityprofilesbetweenPNpatientsand healthyparticipants.Thesimilaritybetweenthevelocityprofiles acrossgroups,categoriesandvisualconditionshasbeenquantified usingtheAVDandMVDparameters,respectively,(seeFig.4Cand D).ComparablevalueswereobservedbetweenCOandPNgroups whilehighervalueswereobservedforPH.However,theANOVA couldnotreasonablybeperformedforthewholedataset(AVDand MVD)astheMauchleytestforsphericitywaspositive(the vari-ancesofthepatients’groupwerenothomogeneousacrossgroups). WethereforeperformedtheANOVAseparatelyfortheCOandthe patients’groups.TheonlystatisticallysignificanteffectfortheCO wastheoneofthecategoryoftargets(F(3,27)=16,76,p<0.01)on theMVDparameter(whichwassignificantlyhigherforHC com-paredtoSTcategory).Thetestforsphericitywaspositiveforthe patients’groupforboth AVDandMVDparameters.Thiscanbe explainedbylargerAVD/MVDvariabilitywithinthePHsubgroup. ItcanalsobeexplainedbyalackofsensitivityoftheAVD/MVD

parameters(whichwerededicatedtomeasuredeviationsfromthe meanvelocityproﬁleacrossindividuals)indetectingdifferent pat-ternsofvelocityproﬁlesalongindividualpathcompletion.

Thatiswhywemeasuredhowmuchwalkingvelocityvaried duringpathcompletionbycomputingtheamplitudeofwalking speed variations(AMPVEL, Fig.4D). Thetest for sphericity was positivewhenappliedtothewholeAMPVELdataset.Wethus per-formedtheANOVAcomparisonsseparatelyforCOandpatients.The velocityvariationsrangedbetween0.4to0.7m/sacrosscategories andconditionsfortheCOgroup.Weobservedasignificanteffect ofthecategory(F(3,27)=33,42,p<0.01),ofthevisualcondition (F(1,9)=19,51,p<0.01)aswellasaninteraction(category×visual condition)effect(F(3,27)=10,85,p<0.01)onAMPVEL.Therange ofvelocityvariationswascomparablebetweenCOandPNgroups (Fig.4D).WethusgroupedtogetherPNdatawiththe(age-gender matched)COdata.Thetestforsphericitywasnegative.TheANOVA comparisonsrevealednosignificanteffectofthegroup(p>0.05) andasignificanteffectofthecategory(F(3,24)=53,67,p<0.01), ofthevisualcondition(F(1,8)=25,49,p<0.01)aswellasan inter-action(category×visualcondition)effect(F(3,24)=8,99,p<0.01) onAMPVEL.Thus,bothCOandPNvariedsignificantlymoretheir walkingspeedalong thepathfor thehighlycurvedtrajectories (HC)andinVFtrials.Thetestforsphericityperformedonpatients’ datawaspositive,revealingthatAMPVELwasnothomogeneousin thepatients’population.Weobservedthatvariabilitywas particu-larlyimportantinPHinBFtrialsandfortheHCcategory(Fig.4D).

(7)

Fig.3.Typicalvelocityproﬁlesgeneratedbycontrolparticipants(A),non-hemiparetic(B)andhemiparetic(C)patientsduringtrajectorycompletion.Samecolorcodeas

Fig.4.Notethequasi-constantwalkingvelocitygeneratedbythePHpatientinperformingthetask,incontrastwiththePNpatientsandthecontrolgroup.Thistypeof

constant-velocitypatternswasobservedinpatientsP03,P04andP08.

WethusrepeatedthesphericitytestfortheST,LCandMC cate-goriesononehandandfortheHCcategoryontheotherhand,and bothtestswerenegativeonpatients’data.TheANOVA compar-isonsperformedforthefirstthreecategoriesrevealedthatAMPVEL wassignificantly smallerforPH compared toPN (F(1,8)=6,32, p=0.036).Wealsoobservedasignificanteffectofthecategory(F(2, 16)=3,83,p=0.044),ofthevisualcondition(F(1,8)=23,56,p<0.01) aswellasaninteraction(category× visualcondition)effect(F(2, 16)=3,6443,p=0.04964)onAMPVEL.Velocityvariedlessduring pathcompletionforSTtargetsandinVFtrials.TheANOVA com-parisonsperformedfortheHCtargetrevealedasignificanteffectof thevisualcondition(F(1,8)=33,77,p<0.01)onAMPVELandaweak (butnotsignificant)effectofthegroup(F(1,8)=4,93, p=0.057). Thus,PHpatientshadlargerintra-groupvariabilitythanCOandPN patients(seealsoSupplementarymaterialforsimilarobservations reportedatthelevelofthesteppingbehavior).Overall,variationsin walkingvelocityduringpathcompletionaresignificantlysmaller inPHpatientscomparedtoPNandCO(Fig.4D).

Takentogether,theseresultsshowthat,onaverage,COandPN groupsgeneratesimilarvelocitypatterns,whilePHgroup gener-atedifferent(andmorevariable)velocitypatternsandvarytheir walking speed signiﬁcantly less along the path (with patients P03-P04-P08 walking atquasi-constant velocities whatever the curvaturevariationsalongthepath).

4. Discussion

Motorimpairmentsingaitsofindividualsfollowingstrokeor brain damage are welldocumented [see Ref. [15] for a recent review].Here, weinvestigated whethertheseimpairmentsalso affectthecognitive(navigational)component ofa spatially ori-entedlocomotortask.Importantly,participantswerefreetochoose any path allowing them to reach a distant (visible or memo-rized)target(withvisionorblindfolded).Aspreviouslyobserved in healthy adult participants [6–8,10], we observed very simi-largeometricalformsofpathsacrosstargetpositionsandvisual conditions.Remarkably,thisspatialstereotypyofthelocomotor trajectories was observed following brain damage, even in the mostseverelyimpaired(hemipareticPH)patients.Thiscontrasted withmuch morevariability atthetemporal level.In particular, healthy participants and non-hemiparetic patients varied their walkingspeedaccordingtocurvaturechangesalongthepath.On thecontrary,thewalkingspeedprofileswerenotstereotypicaland werenotsystematicallyconstrainedbypathgeometryin hemi-pareticpatients(walkingvelocitywasalmostconstantinthreePH patients)whereitwasassociatedwithdifferentstepping behav-iors.Thisextensionofourpreviousobservationstopatientswith significantlesionsofthesensory-motorsystem,andindependently ofthepresenceofnon-navigationalcognitivedeficits,yieldsdirect

(8)

Fig.4.Measuresofthespatial(A–B)andtemporal(C–E)variabilityofthewalkingtrajectoriesacrossgroupsandvisualconditions.ST,LC,MCandHCindicatestraight-ahead

walking,lowcurvature,mediumcurvatureandhighcurvaturetrajectories,respectively.(A–B)Averageandmaximaldistancesbetweentrajectoriesacrossgroupsorvisual

conditions(ATSandMTS,respectively:seeinsertinpanelAforthecodecolorusedtodisplaytheControlsvsPatientsandVisualvsBlindfoldedcomparisons,respectively).

Averageandmaximaldistancesbetweentrajectoriesacrossrepetitionsofallparticipants(ATDandMTD,codecolorsimilartoFig.3).Notethattrajectoryseparationindices

acrossgroupsorvisualconditionsneverexceed,onaverage,0.2m(ATS)andthatallpatientsdonotdifferfromcontrolparticipants’values.(C–D)Similarmeasurements

performedforthevelocityproﬁle:notethesystematicallyhighervariabilityforPHpatientsonly.E—Amplitudeofvelocityvariationsduringpathcompletion.Notethat

velocityvariessigniﬁcantlylessinblindfoldedtrialsandforPHpatientsacrosscategoriesandvisualconditions.

experimentalevidenceinhumansthatthemotorcortexisinvolved in the sensorimotor implementation of locomotion but is not involvedinthepathplanning/navigationalstage.Thishasseveral implicationsthatarediscussedbelow.

4.1. Navigationalabilitiesfollowingstrokeorbraindamage

VanderHametal.[17]observedthat29%ofpost-mildstroke patientsincludedintheirstudycomplainedofspatialnavigation impairmentswhichcouldnotbedetectedusingmostofthe neu-ropsychologicaltestsincommonuse.Suchfrequentcomplaintof strokepatientsisrelatedtothedisorientationproblemsthey expe-riencewhennavigatinginafamiliarspaceinpartialorcomplete absenceofvision(likewalkingfromtheirbedtothekitcheninthe night).Importantly,thiscanoccurevenafterfullmotorrecovery,as recentlyreportedbyHanetal.[18]whodescribedthecaseofa 72-yearoldpatientwhoexperiencedasuddeninabilitytonavigateto therestroom,kitchen,oranyfamiliarplaceaftersufferingastroke. Notably,thispatientwithmultipleacuteischemicbrainlesionsin theparietalandoccipitallobeswasblindfor30yearsanddidnot exhibitanydisorientationproblemspriortothestroke.The disori-entationproblemswereresolvedwithinthreedaysoftreatment. Theabsenceofpersistentnavigationalproblemsevenin hippocam-palpatientsperformingactivelocomotororpointingtasks[2,3]can berelatedtothedifﬁcultyinassessingspatialmemoryand/or nav-igationabilities[5].Thegoal-orientedtaskwetestedherebelongs tothelattertypeofabilitiesinthesensethatallparticipantshad bothtoplan and executea whole-body displacementina new environment. We didnotobserve differencesbetweenpatients and healthy participantsat thespatial (cognitive)level despite important differences at the execution (motor) level. In other

words,wecouldnotfindany“path-planning”impairmentevenin patientshavingexperiencedstrokesonlytwomonthsbefore inclu-sioninthestudy.Inareviewpaper,Krakauer[19]distinguished thepost-strokekinematicalanddynamicalpotentialtroublesfor armmovementsandarguedthattheabnormal(hemiparetic)arm movementsmaysuggestadeficitintransformingaplanned tra-jectoryintotheappropriatejointangles. Thisdistinctionseems alsotoholdtrueforhemipareticwhole-bodymovements. How-ever,wecannotexcludethepossibilitythatpatientswithspecific (e.g.,hippocampal)lesionsorpatientsatearliertimepointsafter strokeswouldalsoexhibitpathplanningdeficits.Thisshouldbe testedinfuturestudies.Besides,itcouldbearguedthattheconstant walkingspeedsobservedinthreehemipareticpatientsreveal plan-ningdeficitsatthetemporallevel.Whilewecannotexcludethis possibility,ourresultsshowacleardissociationbetweenspatial andtemporalcomponentsofwhole-bodymotionplanning.Thisis furtherdiscussedbelow.

4.2. Sensorimotorimplementationoflocomotortrajectories Anotherfunctionalimplicationofthepresentstudyisrelated tothenatureofthemechanismsunderlyinglocomotortrajectory formationandcontrol.Wepreviouslyreportedthatinhealthy par-ticipantsthespatialstereotypyoflocomotortrajectoriesdoesnot relyontheavailabilityof visualinputs [7,9].Our presentstudy extendstheseobservationstopatientsandtohealthyelderly peo-ple.Nevertheless,inpreviousstudiesweobservedthatvisionis involvedinminimizingthevariationsaroundtheaverage (stereo-typed)trajectories.Wecouldpredictbothaveragetrajectoriesand variabilityproﬁlesaroundthesetrajectoriesusingamodel combin-ingtwomodulesaccountingforthe“global”(path-planning)and

(9)

“on-line”contributions(visualguidance)tolocomotorpaths for-mation.Importantly,bothmodulesrelyonoptimalityprinciples alreadydescribedforhandmovementsplanningandcontrol[20].

Otherapproachesdonotassumesuchcomputationalmodules andproposea directuseofopticflow(aloneorincombination withothervisualvariables)toguidelocomotion andalsoallow forthepredictionofalargerangeoflocomotortrajectories[21]. Thespatialstereotypyoftrajectoriesobservedduringblindfolded locomotionquestionstherelevanceofsuchapproaches.Itcould bearguedthatthestereotypedbehaviorobservedduring blind-folded locomotion is the by-product of some preserved visual motor feedbackloops. Inany case,this would require translat-ingthememorized(static)position/orientationofthelocomotor goalintoappropriate(dynamic)visuo-motorpatterns,e.g.,some non-visualcomputationalprocessing.Adaptingourparadigmand testingcongenitallyblindpeoplemighthelptodisambiguatethe possibilityofpreservedvisual-motorfeedbackloops.Besides,the observationofspatialstereotypyinhemipareticpatientsdoesnot supportthehypothesisofpreservedvisual-motorfeedbackloops asthesepatientssufferfromsignificantimpairmentsoftheir loco-motorpatterns;rather,itseemsthatsomehigherorder(cognitive) mechanismexplainsthespatialstereotypyoflocomotor trajecto-rieswhilevisuomotorcontrolloopswouldlikelybeinvolvedinthe on-linecontrolofthesteeringbehavior.Thus,theconstant walk-ingspeedsobservedinhemipareticpatientswouldrepresentan adaptedvisuomotorsteeringstrategyratherthanapath-planning deficit.Atthemodelinglevel,previousstudieshavesuggestedthat, in both hand movements and locomotion, thevelocity profiles alongapredefinedpathareconstrainedbygeometricquantities orprinciples(curvatureoraffineinvariance)oroptimization prin-ciples[14,20,22].Withinthis context,theobservationofsimilar geometricalformsofpaths,butdifferentvelocityprofiles,in hemi-pareticpatientsfurthersupportstheexistenceofa dissociation betweenthespatial(path-planning)andtemporal(sensorimotor implementationlevel)componentsofspatially-oriented locomo-tioninhumans.

4.3. Brainareasinvolvedinthespatialandtemporalaspectsof spatially-orientedwalking

Inarecentreviewpapermainlybasedonanimalstudies,Drew andMarigold[23]proposedananatomicaldistinctionofbrainareas involvedindifferentaspectsoflocomotion.Namely,theyprovided neurophysiologicalevidenceforadifferentcontributionof poste-riorparietalandmotorcorticesintheplanningandexecutionof locomotion,respectively.Inthecitedstudies,theplanninglevel wasmainly related tolimb trajectory planningduringobstacle avoidancetasksincats(withvision).Ourbehavioralobservations showthat,inourpatients,importantbrainlesionsinthe premo-torandmotorcorticesdonotaffectpath-planningmechanisms; this supports the propositions of Drew and Marigold [23] and goesevenfurtheraswesuggestsimilardistinctionsbetween plan-ning(spatial)andexecution(temporal)levelsoflocomotionbut atthelevel ofthewhole-body trajectory.Besides,weobserved that patientswithfocal lesionsofthecerebellum orbasal gan-gliadidnotexhibitnavigationaldeﬁcits.Importantly,thisabsence of behavioral deﬁcitsat theplanning level was observed even intheabsenceofvision. Itshouldalsobenotedthatmore dis-tributedcorticalnetworkisinvolvedintheplanningandexecution ofwhole-bodydisplacement.Takentogether,theseobservations suggestthatmedio-temporalareasknowntobeinvolvedinspatial processing(includingthehippocampus,seeSection1)mayplay acritical roleinprovidinga continuous“routetofollow”signal tothemotorcortexwhichwouldthentranslatethisintomotor commands. Following stroke, plasticity in the motor and somatensory cortices would result in a different sensorimotor

implementationofthelocomotorpath.Whether“thepath plan-ning” stage is encoded in allocentric or egocentric coordinates cannotbeansweredusingthepresentdata.Futurestudies manipu-latingthesedifferenttypesofspatialrepresentationsmayprovide adeeperunderstandingofthefunctionalandanatomical organi-zationofspatially-orientedlocomotion.Anotherlimitationofour study is the relatively small-scale environment (comparable to dailylocomotortasksathome)inwhichparticipantsperformed thetasks.

4.4. Implicationsforrehabilitation

Thepotentialdissociationbetweencognitiveandmotoraspects ofgaitrecoverypost-strokemustbefurtherstudiedatdifferent time points afterstroke and for complex locomotor tasks. The findingsofthispilotstudyarereminiscentofthosereportedby Belmontietal.[10]:cognitiveandmotorcomponentsofhuman locomotionseemtoevolveindependentlyduringalifetime,witha stabilizationofgaitoccurringearlier(around4–5yearsofage)than pathplanning(around11–13yearsofageintheirstudy).When appliedtofocalneurologicaldiseasesaffectinggait,thisemphasizes theneedtodeveloptests(i.e.,adaptedversionsofourparadigm) allowingparallelassessmentofcognitiveandmotorfunctionsafter stroke;thismayhelprehabilitationteamstobetterfocusonthe specificdeficitsofpatients.

Acknowledgments

TheauthorswouldliketothankDrAlanChauvinforhishelp withthestatisticalanalysisandtwoanonymousreviewersfortheir constructivesuggestionsandcomments.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.bbr.2015.12.014. References

[1]G.Buzsaki,E.I.Moser,Memory,navigationandthetarhythminthe hippocampal-entorhinalsystem,Nat.Neurosci.16(2013)130–138.

[2]S.Kim,M.Sapiurka,R.E.Clark,L.R.Squire,Contrastingeffectsonpath integrationafterhippocampaldamageinhumansandrats,Proc.Natl.Acad. Sci.U.S.A.110(2013)4732–4737.

[3]Y.Shrager,C.B.Kirwan,L.R.Squire,Neuralbasisofthecognitivemap:path integrationdoesnotrequirehippocampusorentorhinalcortex,Proc.Natl. Acad.Sci.U.S.A.105(2008)12034–12038.

[4]C.Paquette,E.Franzen,G.M.Jones,F.B.Horak,Walkingincircles:navigation deﬁcitsfromParkinson’sdiseasebutnotfromcerebellarataxia,Neuroscience 190(2011)177–183.

[5]T.Wolbers,M.Hegarty,Whatdeterminesournavigationalabilities?Trends Cogn.Sci.14(2010)138–146.

[6]H.Hicheur,Q.C.Pham,G.Arechavaleta,J.P.Laumond,A.Berthoz,The formationoftrajectoriesduringgoal-orientedlocomotioninhumans.I.A stereotypedbehaviour,Eur.J.Neurosci.26(2007)2376–2390.

[7]Q.C.Pham,H.Hicheur,Ontheopen-loopandfeedbackprocessesthatunderlie theformationoftrajectoriesduringvisualandnonvisuallocomotionin humans,J.Neurophysiol.102(2009)2800–2815.

[8]Q.C.Pham,A.Berthoz,H.Hicheur,Invarianceoflocomotortrajectoriesacross visualandgaitdirectionconditions,Exp.BrainRes.210(2011)207–215.

[9]Q.C.Pham,H.Hicheur,G.Arechavaleta,J.P.Laumond,A.Berthoz,The formationoftrajectoriesduringgoal-orientedlocomotioninhumans.II.A maximumsmoothnessmodel,Eur.J.Neurosci.26(2007)2391–2403.

[10]V.Belmonti,G.Cioni,A.Berthoz,Developmentofanticipatoryorienting strategiesandtrajectoryformationingoal-orientedlocomotion,Exp.Brain Res.227(2013)131–147.

[11]C.Chen,D.Leys,A.Esquenazi,Theinteractionbetweenneuropsychological andmotordeﬁcitsinpatientsafterstroke,Neurology80(2013)S27–S34.

[12]M.K.Holden,K.M.Gill,M.R.Magliozzi,J.Nathan,L.Piehl-Baker,Clinicalgait assessmentintheneurologicallyimpaired.Reliabilityandmeaningfulness, Phys.Ther.64(1984)35–40.

[13]H.Hicheur,A.Berthoz,Howdohumansturn?Headbodymovementsforthe steeringoflocomotion,ProceedingsoftheIEEE-RoboticsandAutomatic SocietyInternationalConferenceonHumanoidRobots(2005)265–270.

(10)

[14]H.Hicheur,S.Vieilledent,M.J.E.Richardson,T.Flash,A.Berthoz,Velocityand curvatureinhumanlocomotionalongcomplexcurvedpaths:acomparison withhandmovements,Exp.BrainRes.162(2)(2005)145–154.

[15]S.Lauziere,M.Betschart,R.Aissaoui,S.Nadeau,Understandingspatialand temporalgaitasymmetriesinindividualspoststroke,Int.J.Phys.Med. Rehabil.2(2014).

[16]M.Godi,A.Nardone,M.Schieppati,Curvedwalkinginhemipareticpatients,J. Rehabil.Med.42(2010)858–865.

[17]I.J.VanderHam,N.Kant,A.Postma,J.M.Visser-Meily,Isnavigationabilitya probleminmildstrokepatients?Insightsfromself-reportednavigation measures,J.Rehabil.Med.45(2013)429–433.

[18]Y.H.Han,M.C.Pai,C.T.Hong,Topographicaldisorientationinapatientwith late-onsetblindnesswithmultipleacuteischemicbrainlesions,J.Clin. Neurosci.18(2011)283–285.

[19]J.W.Krakauer,Armfunctionafterstroke:fromphysiologytorecovery,Semin. Neurol.25(2005)384–395.

[20]E.Todorov,M.I.Jordan,Smoothnessmaximizationalongapredeﬁnedpath accuratelypredictsthespeedproﬁlesofcomplexarmmovements,J. Neurophysiol.80(1998)696–714.

[21]B.R.Fajen,W.H.Warren,Behavioraldynamicsofsteering,obstacleavoidance, androuteselection,J.Exp.Psychol.Hum.Percept.Perform.29(2003) 343–362.

[22]Q.-CPham,D.Bennequin,Afﬁneinvarianceofhumanhandmovements:a

directtest.ArXive-prints(2012)p.1467.

[23]T.Drew,D.S.Marigold,Takingthenextstep:corticalcontributionstothe controloflocomotion,Curr.Opin.Neurobiol.33C(2015)25–33.

[24]H.Hicheur,A.V.Terekhov,A.Berthoz,Intersegmentalcoordinationduring humanlocomotion:doesplanarcovariationofelevationanglesreﬂectcentral constraints?JNeurophysiol.96(2006)1406–1419.