demands in young and older adults Attentional costs of walking are not affected by variations in lateralbalance Gait & Posture

Attentional

costs

of

walking

are

not

affected

by

variations

in

lateral

balance

demands

in

young

and

older

adults

Masood

Mazaheri

a,

*,

Melvyn

Roerdink

a

,

Jacques

Duysens

b

,

Peter

J.

Beek

a

,

C.

(Lieke)

E.

Peper

a

a

MOVEResearchInstituteAmsterdam,DepartmentofHumanMovementSciences,VrijeUniversiteitAmsterdam,Amsterdam,TheNetherlands

b

DepartmentofKinesiology,KU-Leuven,Belgium

1. Introduction

Limited abilitytoadjust walkingto taskand environmental demandsincreasesfallriskin elderlypersons [1].Animportant aspectofwalkingadaptabilityistheabilitytocopewithvariations of attentional demands associated with performing secondary tasks while walking [1,2]. An age-related increase in the attentional demands of walking may hamper an individual’s ability to respond to environmental hazards with potentially serious consequences. Indeed, such age-related changes are associatedwithlesssafegait,poormobility,increaseddependence inactivitiesofdailylivingandparticularlyincreasedfallrisk[3].

The interaction between walking and attention has most commonly been assessed using dual-task paradigms in which walkingisperformedsimultaneouslywithasecondarycognitive task[4].Competitionforlimitedattentionalresourcesbetweenthe primary and secondary task may result in interference or decrement in performance of either one or both tasks when comparedwiththeirbaselinesingle-taskperformances. Lundin-Olsson et al.[3] showedthat older adultswho are notable to continue walking while talking are moreprone tofalling than thosewhocanperformthetwotaskssimultaneously.Morerecent studies[5–7]supportincreaseddual-taskinterferencewithaging, suggestingthatwalkingismoreattentionallydemandinginolder thaninyoungadults.

Increasedattentionalcostsofwalkingamongolderadultsmay be attributed to subtle brain impairments or disorders in the coordination of sensory and motor information required for performingcomplexabilities,suchasbalanceregulationduring walking. Previous research revealed that in older adults both

ARTICLE INFO Articlehistory: Received13July2015

Receivedinrevisedform4February2016 Accepted4March2016 Keywords: Attention Aging Balance Walking Visuomotorcontrol ABSTRACT

Increasedattentionalcostsofwalkinginolderadultshavebeenattributedtoage-relatedchangesin visuomotorand/orbalancecontrolofwalking.Thepresentexperimentwasconductedtoexaminethe hypothesisthatattentionalcostsofwalkingvarywithlateralbalancedemandsduringwalkinginyoung andolderadults.Twentyyoungandtwentyolderadultswalkedonatreadmillattheirpreferredwalking speed under five conditions: unconstrained normal walking, walking on projected visual lines correspondingtoeithertheparticipant’spreferredstepwidthor50%thereof(i.e.increasedbalance demand),andwalkingwithinlow-andhigh-stiffnesslateralstabilizationframes(i.e.lowerbalance demands).Attentionalcostswereassessedusingaprobereaction-timetaskduringthesefivewalking conditions,normalizedtobaselineperformanceasobtainedduringsitting.Bothimposedstep-width conditionsweremoreattentionallydemandingthanthethreeotherconditions,intheabsenceofany othersignificantdifferencesbetweenconditions.Theseeffectsweresimilarinthetwogroups.The resultsindicatethattheattentionalcostsofwalkingwere,incontrasttowhathasbeenpostulated previously,notinfluencedbylateralbalancedemands.Theobserveddifferenceinattentionalcosts betweennormalwalkingandbothvisuallinesconditionssuggeststhatvisuomotorcontrolprocesses, ratherthanbalancecontrol,stronglyaffecttheattentionalcostsofwalking.Atentativeexplanationof theseresultsmaybethatvisuomotorcontrolprocessesaremainlygovernedbyattention-demanding corticalprocesses,whereasbalanceisregulatedpredominantlysubcortically.

ß2016ElsevierB.V.Allrightsreserved.

* Correspondingauthorat:MOVEResearchInstituteAmsterdam,Departmentof HumanMovementSciences,FacultyofBehaviouralandMovementSciences,Vrije UniversiteitAmsterdam,Amsterdam,TheNetherlands.Tel.:+31205982632.

E-mailaddress:m.mazaheri@vu.nl(M.Mazaheri).

ContentslistsavailableatScienceDirect

Gait

&

Posture

j our na l ho me pa g e : w ww . e l se v i e r . com / l oca t e / ga i t po st

cognitive impairments [8] and elevated visuomotor demands

[9–12]areassociatedwithincreasedattentionalcostsofwalking. Few studies [13,14], however, have specifically addressedhow balancecontrolaffectstheattentionalcostsofwalking,especially inolderadults.Pertinentevidencecomesprimarilyfrom experi-mentswithbase-of-support manipulations,showingthat atten-tional costsarehigher during walkingthan duringstandingor sitting[9,10,13,14].Althoughbalancerequirementsmaychange over the gait cycle, inconsistent results have been reported regardingtheattentionalcostsforspecificphasesinthegaitcycle

[13–15]. However, balance demands were never manipulated systematicallyduringthegaitcycleasawhole,whichprecludes drawingfirmconclusionsabouttheeffectofbalancecontrolonthe attentionalcostsofwalking.Inthepresentstudy,wefocusedon lateralstabilitymanipulationsbecausewalkingisless passively stable in mediolateral direction than in fore-aft direction

[16]. Active sensorimotor control required for lateral balance duringwalkingmaybeexpectedtoelevatetheattentionalcosts ofwalking.

Inparticular, weexamined theeffectof variationsin lateral balancedemandsonattentionalcostsofwalkinginbothyoungand olderadults.Balancedemandsweremanipulatedbymeansoftwo levelsofprescribedstepwidth(SW;preferredvs.narrowerthan preferred, imposedby meansof visuallines projectedonto the walkingsurface)andalateralstabilizationdevice(involvingtwo levelsofmechanicalstabilization[17]).Withthesemanipulations, wecreatedconditionswithhigherandlowerbalancedemands, respectively.The attentionalcosts associatedwith these condi-tionswereassessedwithvibrotactilestimulus-responsereaction times(RT)[9,10].Weexpectedthathigherbalancedemands(as evokedbywalkingwithanarrowbaseofsupport)wouldincrease the attentional costs of walking, particularly in older adults. Likewise,weexpectedthatlowerbalancedemands(asevokedby lateralstabilization)wouldreducetheattentionalcostsofwalking, againparticularlyinolderadults.

2. Materialandmethods

2.1. Participants

Twentyyoungadults(female/male:12/8)and20healthyolder adults (female/male: 12/8) participated in the experiment (Table 1). Participants had no self-reported cardiovascular or cardiopulmonar problems, orthopedic conditions, uncorrected visual or auditory impairments, neurological disease, or other conditionslimitingmobility;theydidnotusewalkingaidsandthe MiniMentalStateExamscorefortheolderparticipantsexceeded 23 (range 24–30). All participants provided written informed consentbeforeparticipation.Thedepartmentalethicscommittee approvedtheexperiment.

2.2. Experimentalset-up

The experimental set-up was designedto inducehigher and lower balance demands, using two separate manipulations: prescribedSW and lateral stabilization.Aforce-platform instru-menteddual-belttreadmill(MotekforceLink, Amsterdam/Culem-borg, The Netherlands) equipped with a projector allowing projection of visual lines onto the belt’s surface was used to measureandimposeSWintheprescribedconditions(Fig.1A).Inthe lateralstabilizationconditions,anexternalstabilizer[17](Fig.1B) wasusedtoenhancelateralstability.Twospring-likerubbercords wereattachedtoaframefastenedtothewaistandanchoredto ball-bearing trolleys that moved freely in for-aft direction within a horizontalrailparalleltotheground,positionedateithersideofthe participant.Theheightoftherailwasadjustedtotheparticipant’s waist height. Cords with two different levels of stiffness (low stiffness:760Nm1_{andhighstiffness:1613Nm}1_{,see[17])were}

used,withthehigh-stiffnesslevelprovidinglargerstability.

Table1

Participants’demographicandclinicalcharacteristicspergroup. Youngadults (f/m:12/8) Olderadults (f/m:12/8) Groupcomparisons Statistics p-value Age(year) 23.23.3 72.94.6 t38=39.13 <0.001 Height(cm) 174.59.6 170.910.2 t38=1.15 0.26 Weight(kg) 64.610.8 66.610.2 t37=0.60 0.56 CWS(km/h) 4.20.6 3.70.7 t38=2.43 0.02 FRD(cm) 35.27.6 29.46.1 t38=2.64 0.01 BaselineRT(ms) 233.525.0 297.331.8 t38=7.05 <0.001

Notes:valuesaremeanSD.CWS,comfortablewalkingspeed;FRD,functionalreach distance;RT,reactiontime;andf/m,female/male.

Fig.1.Schematicoftheexperimentalconditionstovarylateralbalancedemands. (A) Walking on visual lines projected onto the treadmill belt; (B) Walking withexternallateralstabilizerwithaspring-likecordattachedtothelight-weight framefastenedtothewaistbeltononeendandontheotherendtothe ball-bearingtrolley.

Stimulus-response RT was measured using a custom-made stimulusvibrator(pulseduration:300ms; attachedtothe non-dominanthand’s wrist) and a response button (sampling rate: 1000Hz;heldinthedominanthand;[9,10]).Asafetyharnessto protectparticipantsfromfallingwasusedinallwalkingconditions thatdidnotinvolvetheexternalstabilizationframe.

Ahorizontallyorientedtapemeasureattachedtothewallatthe height of the participant’s acromion process was used in the FunctionalReachTest(FRT).

2.3. Procedure 2.3.1. Preparation

Participantsfirstpracticedtreadmillwalkingfor10min.Next, comfortable walking speed (CWS) was determined by first increasingtreadmillspeed(in0.1km/hsteps)untiltheparticipant reportedhis/her CWSwasreached.After a 1.5km/hincrement, walking speed was decreased (0.1km/h steps) until CWS was indicatedagain[10].Theaverageofthesetwosubjectiveestimates servedastheparticipant’sCWSandwasusedforallsubsequent walkingconditions.Next,eachparticipant’spreferredSWathis/ herCWSwasdetermined(1min).

Participants’ balanceabilitywasquantifiedasthefunctional reachdistance(FRD)usingtheFRT[18].

2.3.2. Experiment

ParticipantswalkedundertwoprescribedSWconditionsand two lateral stabilization conditions. For the prescribed SW conditions,thedistancebetweenthetwo visuallinesprojected ontothetreadmillbelt’ssurface wassetto100% (preferredSW condition) or 50% (narrow SW condition) of each participant’s preferred SW as determined in the pre-experimental trial. Participantswereinstructed toalignthemidline oftheir shoes withthevisuallines.Thelateralstabilizationconditionsinvolved walking witheither low- or high-stiffness stabilizers. In these conditions, no visual lines were presented. Participants were familiarizedwithwalkingonthevisuallines(1min)andwalking withthelateralstabilizer(5–10min)priortothecorresponding experimentaltrials.

DuringeachwalkingtrialRTwasassessedusing21vibratory stimuli.Tocontrolgaitcycleeffectsonattentionalcosts[13,14], RTswerepresentedatthemomentofheelstrikeofeithertheleftor therightfoot(equallydistributed,randomorder)[10].Thefirst stimulusservedaswarningcueandappearedatleast5safterthe trial had started. Inter-stimulus intervals varied randomly between3and17s.Participantshadtopressthebuttonassoon astheyfeltthevibration,butwereaskedtoprioritizethewalking task[4].

Theexperimentconsistedoftwoblocks,whichwere counter-balancedacrossparticipants:onewithprescribedSWs(preferred andnarrow), theother withlateralstabilization(low and high stiffness).In each block, conditions were presented in random order,withtwoconsecutivetrialspercondition.Inaddition,each blockcomprised onecontrol condition involving unconstrained walking,yieldingfivedual-taskwalkingtrialsintotalperblock. PriortothefirstandafterthesecondblockoftrialsabaselineRT trialwasconducted,measuringRTwhilesittingonachair.Two single-taskunconstrainedwalkingtrials(i.e.withoutRT)werealso conducted,onepriortothefirstsittingtrialandonebetweenthe twoblocks.Alltrialslasted2.5min.Sufficientrestperiodswere administeredbetweentrialsandblockstopreventfatigue.

2.4. Dataanalysis

All data were analyzed using custom-made Matlab (Math-works, Natick, MA, USA) scripts. The RT obtained for the first

stimulusineachtrial(warningcue)waseliminated,aswereRTs <120msand>1100ms[10].Accordingly,9stimulus-response pairs were discarded in the older group (i.e. <0.01%; lateral stabilization:2;prescribedSW:7).Noresponsewasdetectedfor 28stimuli(i.e.<0.01%;young:7, older:21;predominantlyfor prescribed SW: 14). RT was defined as the median of the remainingtimeintervalsbetweenstimulusandresponseonsets per trial, and subsequently averaged over the two trials per condition. To eliminate individual baseline differences, atten-tional costs were characterized as difference scores (

D

RT=RTwalking conditionRTsitting) andproportional difference

scores(

D

RTprop=[RTwalkingconditionRTsitting]/RTsitting).

ForeachprescribedSWtrial,theactuallyperformedSWwas determinedfromtheforce-platedatabytakingthemedianofthe absolutedifferencesbetweenleftandrightmediolateral center-of-pressurepositionsatmid-stance(i.e.halfwaybetweenfootcontact andfootoff).SWwasaveragedoverthetwotrialspercondition and normalized to theimposed SW. SW could not be reliably determinedforoneolderparticipantintheimposedSWconditions becausegait events werenot welldemarcated. For the uncon-strainedwalkingtasks,stepwidth,stridelength,stridetimeand cadenceweredetermined.

2.5. Statisticalanalysis

Age,height,weight,CWS,FRDandbaselineRTwerecompared betweenthetwogroupsusingindependentt-tests.Toexaminethe adherence to the imposed SW conditions, normalized SW was subjected to a 2 (group: young vs. older adults) by 2 (task: preferredvs.narrowSW)mixed-modelANOVA.Theeffectsofage andlateralbalancedemandson

D

RTand

D

RTpropwereexamined

using 2 (group) by 5 (task: narrow SW, preferred SW, uncon-strainedwalking,low-stiffnessstabilizer,high-stiffnessstabilizer) mixed-modelANOVAs.ToexaminewhethertheRTtaskaffected gait, gait parameters were compared between unconstrained walkingwithandwithoutRT,usinga2(group)by2(task:withvs. without RT) mixed-model ANOVA. Alpha level was set at 0.05. Paired t-tests (with Bonferroni correction) were used for post hoc pair-wise comparisons. Partial eta squared (

h

2

p) and

Hedges’gav(gav)wereusedtodetermineeffectsize[19].

3. Results

Table 1 presents the participants’ demographic and test characteristics.CWSand FRDscoresweresignificantlylowerin olderadults.BaselineRTwassignificantlyhigherinolderadults. TheANOVAonnormalizedSWyieldedasignificantmaineffect oftask(F1,37=244.94,p<0.001;

h

2p¼0:87):normalizedSWwas

largerfor thepreferred SW condition than for the narrowSW condition (88%9% vs. 55%12%). The absence ofa significant group effect (F1,37=0.22, p=0.64;

h

2p¼0:01) or group by task

interaction (F1,37=0.72, p=0.40;

h

2p¼0:02) indicated that both

groupsadheredtothetaskinasimilarfashion.

Asignificantmain effectoftaskon both

D

RTmeasureswas observed (

D

RT: F4,152=83.58, p<0.001;

h

2p¼0:69;

D

RTprop:

F4,152=89.69, p<0.001;

h

p2¼0:70; Fig. 2). Post hoc analysis

showedthat

D

RTand

D

RTpropwerelargerforthetwoprescribed

SWconditionsthantheotherthreeconditions(

D

RT:preferredSW: gav’s>1.35; narrow SW: gav’s>1.36;

D

RTprop: preferred SW:

gav’s>1.23;narrowSW:gav’s>1.19),whereasneitherthetwoSW

conditions(

D

RT:gav=0.05;

D

RTprop:gav=0)northeotherthree

conditions (

D

RT: gav’s<0.12;

D

RTprop: gav’s<0.08) differed

significantly from each other. The main effect of group (

D

RT: F1,38=0.97, p=0.33;

h

2p¼0:03;

D

RTprop: F1,38=1.38, p=0.25;

h

2

p=0.52;

h

2

p¼0:02;

D

RTprop: F4,152=0.24, p=0.92;

h

2p¼0:01)

werenotsignificant.

TheANOVAsongaitparametersasobtainedforunconstrained walking(Table2)revealedsignificantlywiderSWandshorterstride lengthinolderadultsthanyoungadults.Stridelengthandstride timedecreasedwithdualtasking,whereascadenceincreased.

4. Discussion

Wetestedtheassumptionthattheattentionalcostsofwalking vary with lateral balance control requirements [13,14,20] in a

head-on fashion.Wehypothesizedthat lowerbalancedemands (lateral stabilization) would reduce the attentional costs of walking, whereas higher balance demands (narrow base of support)wouldincreasetheattentionalcosts.Theseeffectswere expectedtobemorepronouncedinolderadults.

However,theobtained

D

RTmeasureswerenotinfluencedby loweredbalancedemands:walkingwithlateralstabilizationdid notresultinlower

D

RTand

D

RTpropcomparedtounconstrained

walking,andneitherdidvariationsinstiffnessofthestabilization deviceaffectthe

D

RTmeasures.ComparisonbetweenthetwoSW conditions(narrowvs.preferred)revealednoeffectofincreased balancedemandsontheattentionalcostseither,astheexpected increasein

D

RTmeasuresforthenarrowSWconditionwasnot observed.Theseresultssuggestthatthecontributionofbalance control to attentional costs of walking was rather limited. Interestingly, however, the attentional costs increased when steps were adjusted to visuallines (i.e. in the prescribedSW conditions). In particular, the observed difference between unconstrained walking and walking on visual lines at the individual’spreferredSW(i.e.thetwoconditionswith compara-ble balance demands) indicatedthat the required visuomotor control in the latter situation resulted in elevated attentional demands. These results suggest that the attentional costs of walking depend more on visuomotor factors thanon balance demands.

Ourfindingsarenotconsistentwithstudiesreportingvariation of attentional costs with changes in balance requirements

[9,10,13,14].Inthosestudies,base-of-supportmanipulationswere used tovarybalancedemands (walking vs.standingor sitting; single-supportvs.double-supportstancephases),whereasinthe presentstudythebalance-demandsmanipulationwaseffectuated throughouttheentiregaitcycle.Thesedifferenceshamperdirect comparisonwithpreviousresults.

Theminorimpactoflateralbalancedemandsonattentional costs of walking may be related to the neurophysiological mechanisms underlying balance control. The presence of postural responses in decerebrated cats underscores the role ofsubcorticalstructuresinmediatingbalancereactions,atleast inmammals[21].Ithasbeensuggestedthatthisfindingmaybe generalizedtohumansin viewofasimilarrelianceofpostural reactionsonbrainstemstructures[22].Thesubcorticalnatureof thesemechanismsmayexplainwhyinourstudylateralbalance control during walking did not appear to affect higher-level cognitive processes associated with the RT task. In contrast, single-unitrecordingstudiesin animalsshowedmorereliance on cortical activity (e.g., primary motor cortex) in Iocomotor tasksthatarehighlydependentonvisuomotorprocesses,such as precision stepping [23]. Koenraadt et al. [24] reported increased activity in the prefrontal cortex in humans during walkingonvisual targetscompared tounconstrainedwalking. Asthisareaistypicallyinvolvedincomplexgaittasksthatare attentionallydemanding,suchaswalkingwhiletalking[25],this observationsuggeststhatvisuallyguidedwalkingrequiresmore attention than normal walking. Indeed, larger RTs have been

Fig.2.MeanDRT(panelA)andDRTprop(panelB)forwalkingwithnarrowand

preferredprescribedstepwidths,normalwalking,andwalkingwithlow-stiffness andhigh-stiffnessstabilizers.Errorbarsindicatestandarderrorofthemean.

Table2

Effectsofgroupandsinglevs.dualtaskingongaitparameters.

Youngadults Olderadults Maineffects Interactioneffect

Singletaska Dualtaska Singletaska Dualtaska Groupb Taskb GroupTaskb Stepwidth(m) 0.14(0.03) 0.15(0.04) 0.16(0.02) 0.17(0.02) 7.77(0.01;0.17) 2.40(0.13;0.06) 0.60(0.44;0.02) Stridelength(m) 1.28(0.13) 1.25(0.13) 1.12(0.17) 1.11(0.17) 9.99(<0.01;0.21) 28.73(<0.001;0.43) 2.87(0.10;0.07) Stridetime(s) 1.12(0.10) 1.10(0.09) 1.11(0.12) 1.10(0.14) 0.001(0.97;0.00) 16.46(<0.001;0.30) 0.54(0.47;0.01) Cadence(steps/min) 108(9) 110(9) 109(11) 111(12) 0.04(0.86;0.001) 22.15(<0.001;0.34) 1.45(0.24;0.04) a

Valuesarepresentedas:mean(SD).

b

Valuesarepresentedas:F-ratio(p-value;h2

reported for visually cued walking than for unconstrained walking[9,10],eveniftheunconstrainedandvisuallyimposed walking patterns were similar, indicating a relation between visuomotor demands and the attentional costs of walking

[11,12].

Ourhypothesisthatvariationoflateralbalancedemandshasa morepronouncedeffectonattentionalcostsofwalkinginolder adultswasnotsupportedeither.Thisabsenceofagroupbytask interactionmayberelatedtotheprimarilysubcorticalnatureof posturalcontrol.Thefinding thatalsotheelevated attentional demands in the two SW conditions (involving enhanced visuomotorcontrol)didnotdifferbetweenthegroupsmaybe associatedwiththefactthatparticipantswalkedattheirpreferred walkingspeed,whichwasslowerinolderadults(seeTable1). Reduced walking speed may reflect a conservative strategy adopted by older adults to preserve their limited attentional resources(indicatedbylowerbaselineRT;cf.Table1)forother tasks. Recent studies reported slower self-selected walking speedsinvisuolocomotorsituations(e.g.,walkingona narrow path [26]or a sequence ofstepping stones[27])comparedto unconstrainedwalking,foryoungandolderadultsalike.Thiswas interpreted as anadaptive strategy tofavor task performance relativetothevisualcontext[26,27].Givenourcurrentresults,it thusseemslikelythatolderadultssloweddowntheirpreferred walkingspeedtoincreasetheavailabletimeforvisuolocomotor control[28,29].

Because theabsenceofsignificanteffects maybeassociated withlimitedsamplesize,weconductedaposthocpoweranalysis fordetectingagroupbytaskinteraction[30].Givenoursample size(n=40),alphalevel(0.05),andobtainedinteractioneffectsize (Cohen’sd=0.14),thepowertodetectsuchaneffectwas0.09.The requiredsamplesizetoobtainpowerattherecommendedlevelof 0.80 wasthus 636,which would be exceptionally large for an experimentalstudylikethis.Anotherlimitationofthestudymay reside in the reduced ecological validity of treadmill walking, which may have increased attentional costs, introducing the potentialrisk of a ceiling effect obscuring differences between conditions.However,thepronouncedelevationof

D

RTmeasures intheprescribedSWconditionsindicatesthattreadmillwalkingas such did not induce a ceiling effect for attentional demands. AnotherlimitationrelatestoourdecisiontopresentRTstimuliat heel strike, whereas the more attentionallydemanding single-supportstancephasemayhavebeenmoresensitivetoconditionor groupeffects[14].Afinallimitationisthatdual-taskingeffectson gaitparameterswereonlyexaminedforunconstrainedwalking. The RTtask induced significant but small differences in stride length (2cm), stride time (20ms) and cadence (1 step/min), suggestingthattheRTtaskhadalimitedeffectonwalking.Thisis consistentwith otherstudies showing no [13,14] or negligible effects[9]ofRTtasksongaitparametersundertheinstructionto prioritizethewalkingtask.However,aswedidnotinclude single-tasktrialsforthelateral-stabilizationandimposedSWconditions (tolimittheexperiment’sduration),itremainsuncertaintowhat extent the prioritization instructions were successful in those experimentalconditions.

In conclusion, our results indicate that, in healthy adults, attentional demands of walking were not influenced by variations in lateral balance demands. Perhaps the primarily subcortical nature of postural responses [21,22] requires minimal use of attentional resources. The higher attentional costs observed for walking on visual lines indicated that visuolocomotor demands contributed moreto the attentional costsofwalkingthanbalancedemands.Theobservationthatthe wayinwhich

D

RTmeasuresvariedoverconditionsdidnotdiffer overthe age groups may be associated withboth the largely subcortical control of balance and the fact that both groups

walkedattheirpreferredwalkingspeed(whichwasslowerfor olderadults).

Acknowledgments

The authors thank Elma Ouwehand for her assistance in recruitment, data collection and analysis, Bert Coolen for his technicalsupport,andTrienkeIJmkerforsharingherknowledgeof thelateralstabilizationdevice.Thisresearchwasconductedaspart of MOVE-AGE, an Erasmus Mundus Joint Doctorate program (2011–2015)fundedbytheEuropeanCommission.

Conflictofinterest Nonedeclared.

References

[1]BalasubramanianCK,ClarkDJ,FoxEJ.Walkingadaptabilityafterastrokeand itsassessmentinclinicalsettings.StrokeRes.Treat.2014;2014:591013. [2]Shumway-CookA,PatlaAE, StewartA,FerrucciL, CiolMA,GuralnikJM.

Environmentaldemandsassociatedwithcommunitymobilityinolderadults withandwithoutmobilitydisabilities.Phys.Ther.2002;82:670–81. [3]Lundin-OlssonL,NybergL,GustafsonY. Stopswalkingwhentalkingasa

predictoroffallsinelderlypeople.Lancet1997;349:617.

[4]WoollacottM,Shumway-CookA.Attentionandthecontrolofpostureand gait: a review ofan emerging area ofresearch. Gait Posture 2002;16: 1–14.

[5]LindenbergerU,MarsiskeM,BaltesPB.Memorizingwhilewalking:increasein dual-taskcostsfromyoungadulthoodtooldage.Psychol.Aging2000;15: 417–36.

[6]PriestAW,Salamon KB,Hollman JH.Age-relateddifferences indualtask walking:acrosssectionalstudy.J.Neuroeng.Rehabil.2008;5:29.

[7]KrampeRT,SchaeferS,LindenbergerU,BaltesPB.Lifespanchangesin multi-tasking:concurrentwalkingandmemorysearchinchildren,young,andolder adults.GaitPosture2011;33:401–5.

[8]Sheridan PL,Solomont J,Kowall N,HausdorffJM. Influenceof executive functiononlocomotorfunction:dividedattentionincreasesgaitvariability inAlzheimer’sdisease.J.Am.Geriatr.Soc.2003;51:1633–7.

[9]PeperCE,OorthuizenJK,RoerdinkM.Attentionaldemandsofcuedwalkingin healthyyoungandelderlyadults.GaitPosture2012;36:378–82.

[10]MazaheriM,RoerdinkM,BoodRJ,DuysensJ,BeekPJ,PeperCE.Attentional costsof visuallyguided walking:effects of age, executivefunction and stepping-taskdemands.GaitPosture2014;40:182–6.

[11]BockO.Dual-taskcostswhilewalkingincreaseinoldageforsome,butnotfor othertasks:anexperimentalstudyofhealthyyoungandelderlypersons.J. Neuroeng.Rehabil.2008;5:27.

[12]MenantJC,SturnieksDL,BrodieMA,SmithST,LordSR.Visuospatialtasks affectlocomotorcontrolmorethannonspatialtasksinolderpeople.PLOSONE 2014;9:e109802.

[13]LajoieY,TeasdaleN,BardC,FleuryM.Attentionaldemandsforstaticand dynamicequilibrium.Exp.BrainRes.1993;97:139–44.

[14]LajoieY,TeasdaleN,BardC,FleuryM.Uprightstandingandgait:arethere changesinattentionalrequirementsrelatedtonormalaging.Exp.AgingRes. 1996;22:185–98.

[15]SajikiN,IsagodaA,MoriaiN,NakamuraR.Reactiontimeduringwalking. Percept.Mot.Skills1989;69:259–62.

[16]BaubyCE,KuoAD.Activecontroloflateralbalanceinhuman walking.J. Biomech.2000;33:1433–40.

[17]IjmkerT,HoudijkH,LamothCJ,BeekPJ,vanderWoudeLH.Energycostof balancecontrolduringwalkingdecreaseswithexternalstabilizerstiffness independentofwalkingspeed.J.Biomech.2013;46:2109–14.

[18]DuncanPW,WeinerDK,ChandlerJ,StudenskiS.Functionalreach:anew clinicalmeasureofbalance.J.Gerontol.1990;45:M192–7.

[19]LakensD.Calculatingand reportingeffectsizestofacilitatecumulative science: a practical primer for t-tests and ANOVAs. Front Psychol. 2013;4:863.

[20]EbersbachG,DimitrijevicMR,PoeweW.Influenceofconcurrenttasksongait: adual-taskapproach.Percept.Mot.Skills1995;81:107–13.

[21]HoneycuttCF,GottschallJS,NicholsTR.Electromyographicresponsesfromthe hindlimbmusclesofthedecerebratecattohorizontalsupportsurface per-turbations.J.Neurophysiol.2009;101:2751–61.

[22]JacobsJV,HorakFB.Corticalcontrolofposturalresponses.J.NeuralTransm. 2007;114:1339–48.

[23]ArmstrongDM.Thesupraspinalcontrolofmammalianlocomotion.J.Physiol. 1988;405:1–37.

[24]KoenraadtKL,RoelofsenEG,DuysensJ,KeijsersNL.Corticalcontrolofnormal gait andprecisionstepping: anfNIRS study.NeuroImage 2014;85(Pt 1): 415–22.

[26]SchaeferS,SchellenbachM,LindenbergerU,WoollacottM.Walkingin high-risksettings:doolderadultsstillprioritizegaitwhendistractedbyacognitive task.Exp.BrainRes.2015;233:79–88.

[27]PeperCE,deDreu MJ,RoerdinkM.Attuningone’s stepstovisualtargets reducescomfortablewalkingspeedinbothyoungandolderadults.Gait Posture2015;41:830–4.

[28]Orcioli-SilvaD,SimieliL,BarbieriFA,StellaF,GobbiLT.Adaptivewalkingin Alzheimer’sdisease.Int.J.AlzheimersDis.2012;2012:674589.

[29]MakiBE.Gaitchangesinolderadults:predictorsoffallsorindicatorsoffear.J. Am.Geriat.Soc.1997;45:313–20.