The neurodevelopmental differences of increasing verbal working memory demand in children and adults

ContentslistsavailableatScienceDirect

Developmental

Cognitive

Neuroscience

jou rn a l h om ep ag e : h t t p : / / w w w . e l s e v i e r . c o m / l o c a t e / d c n

The

neurodevelopmental

differences

of

increasing

verbal

working

memory

demand

in

children

and

adults

V.M.

Vogan

,

B.R.

Morgan

_,

_T.L.

_Powell

_,

_M.L.

_Smith

_,

_M.J.

_Taylor

a_{DiagnosticImaging,HospitalforSickChildren,Toronto,ON,Canada}

b_{AppliedPsychologyandHumanDevelopment,OntarioInstituteforStudiesinEducation,UniversityofToronto,Toronto,ON,Canada} c_{DepartmentofPsychology,UniversityofToronto,Toronto,ON,Canada}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received22April2015

Receivedinrevisedform26August2015 Accepted13October2015

Availableonline5November2015 Keywords:

Verbalworkingmemory

Functioningmagneticresonanceimaging Development

a

b

s

t

r

a

c

t

Workingmemory(WM)–temporarystorageandmanipulationofinformationinthemind–isakey componentofcognitivematuration,andstructuralbrainchangesthroughoutdevelopmentare associ-atedwithrefinementsinWM.Recentfunctionalneuroimagingstudieshaveshownthatthereisgreater activationinprefrontalandparietalbrainregionswithincreasingage,withadultsshowingmorerefined, localizedpatternsofactivations.However,fewstudieshaveinvestigatedtheneuralbasisofverbalWM development,asthemajorityofreportsexaminevisuo-spatialWM.

WeusedfMRIanda1-backverbalWMtaskwithsixlevelsofdifficultytoexaminethe neurodevelop-mentalchangesinWMfunctionin40participants,twenty-fourchildren(ages9–15yr)andsixteenyoung adults(ages20–25yr).Childrenandadultsbothdemonstratedanopposingsystemofcognitiveprocesses withincreasingcognitivedemand,whereareasrelatedtoWM(frontalandparietalregions)increasedin activity,andareasassociatedwiththedefaultmodenetworkdecreasedinactivity.Althoughtherewere manysimilaritiesintheneuralactivationpatternsassociatedwithincreasingverbalWMcapacityin chil-drenandadults,significantchangesinthefMRIresponseswereseenwithage.Adultsshowedgreater load-dependentchangesthanchildreninWMinthebilateralsuperiorparietalgyri,inferiorfrontaland leftmiddlefrontalgyriandrightcerebellum.Comparedtochildren,adultsalsoshowedgreater decreas-ingactivationacrossWMloadinthebilateralanteriorcingulate,anteriormedialprefrontalgyrus,right superiorlateraltemporalgyrusandleftposteriorcingulate.Theseresultsdemonstratethatwhile chil-drenandadultsactivatesimilarneuralnetworksinresponsetoverbalWMtasks,theextenttowhichthey relyontheseareasinresponsetoincreasingcognitiveloadevolvesbetweenchildhoodandadulthood.

©2015TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Workingmemory(WM)allowsforthetemporarystorageand manipulationofinformation(Baddeley,1992).WMcapacity,the numberofitemsthatcanbeheldinWMatonetime,playsan impor-tant role inthe developmentof othercomplex cognitiveskills, suchasreadingability(Engle,2002;Cainetal.,2004),math per-formance(DumontheilandKlingberg,2012),socialability(Dennis etal.,2009),aswellasingeneralintelligence(Colometal.,2007; Engleetal.,1999),overalllearning(GathercoleandAlloway,2004) and academic achievement (Gathercole et al., 2004a; Alloway, 2009).ImpairedWMcapacityhasbeenlinkedtoanumberof neu-rodevelopmentaldisorders,suchasAttentionDeficitHyperactivity

∗ Correspondingauthorat:DiagnosticImaging,HospitalforSickChildren,555 UniversityAvenue,Toronto,CanadaM5G1X8.Tel.:+14168137654x304299.

E-mailaddress:[email protected](V.M.Vogan).

Disorder(ADHD;seeMartinussenetal.,2005)andAutism Spec-trumDisorder(ASD;Southwicketal.,2011),andvariouslearning (HwangandHosokawa,2007;WangandLiu,2007)andlanguage processingdifficulties(seeWrightandFergadiotis,2012).

Behavioral studies have documented improvements in WM abilitythroughoutchildhoodtoadulthood(e.g.,Conklinetal.,2007; Gathercoleetal.,2004b; Huizingaetal.,2006;Zaldand Iacono, 1998).Whereasmanyotherexecutivefunctioncomponentsshow developmenttypicallyonlyupuntilmid-adolescence,WM contin-uestoshowprotracted developmentwellintoyoung-adulthood (Huizinga et al., 2006), making it particularly susceptible to developmentaldisturbances.Structuralbrainchangesthroughout developmentareassociatedwithrefinementsinvariouscognitive functions,includingWM(Tamnesetal.,2013).Specifically,changes instructureandfunctionofbrainregionsinvolvedinWM,suchas parietalandfrontalregions,occurlaterthanmanyregions, consis-tentwiththeprotractedmaturationofWMfunctions(Sowelletal., 1999).GiventhekeyroleWMplaysincognitivematuration,itis http://dx.doi.org/10.1016/j.dcn.2015.10.008

1878-9293/©2015The Authors.Publishedby Elsevier Ltd.This is anopen accessarticleunder the CCBY-NC-ND license(http://creativecommons.org/licenses/ by-nc-nd/4.0/).

importanttounderstandandcharacterizetheneuralbasisofthis development.

Recentfunctional neuroimaging studiesthat have examined theneuralunderpinningsofWMacrossdevelopmenthaveshown thatwithincreasedage,childrenandadolescentsexhibitgreater activation in prefrontal (Klingberg et al., 2002; Scherf et al., 2006)andparietal(Nageletal.,2013;Spencer-Smithetal.,2013; Klingbergetal.,2002;Scherfetal.,2006)regionsonvisuo-spatial WM tasks. Adults showed similar neural patterns as children andadolescentsonthesetasks,butwithmorerefined,localized activation (Scherf et al., 2006), and some increased activity in “performance-enhancing”regions, suchasthedorsolateral pre-frontalcortex(DLPFC;Jollesetal.,2011;Scherfetal.,2006).For example, Scherfet al. (2006) foundthat children showed lim-ited recruitmentof critical WM substrates(DLPFC and parietal regions)duringavisuo-spatialWMtask,andinsteadreliedmainly onventromedialprefrontalregions.However,theyobservedmore specialized networks (i.e., DLPFC, ventrolateral prefrontal cor-tex[VLPFC]andsupramarginalgyrus)asadolescentsmovedinto adulthood.Thisdevelopmentalpatternofbrainactivityhasalso beencharacterizedas a shiftfrom posterior toanterior activa-tion, with adults showing increased activity in the DLPFC and VLPFC (Kwon et al., 2002; Scherf et al., 2006).Thus, previous literaturesuggeststhatthedevelopmentofhigher-levelWM func-tioninvolvesacombinationofincreasinglocalizationwithincore WMregions and theirintegrationwithperformance-enhancing regions.

Fewerstudieshaveexaminedtheneuralbasis ofverbalWM development,asthemajorityofstudiesutilizedvisuo-spatialor othernonverbaltasks.VerbalWMisparticularlyimportant,given itsvitalroleinlinguisticprocessesthatarenecessaryforlanguage and other higher-levelcognitive functions(Smith et al., 1998). Brahmbhattetal.(2008)foundsimilaractivationpatternsbetween adolescentsandadultsduringann-backvisualwordtask,withboth groupsshowingactivationin thebilateralfusiformgyrus, ante-riorcingulate,leftprecentralgyrus,leftsuperioranteriortemporal gyrus,leftDLPFC,premotorcortexandleftthalamus.Age-related changes wereevident in theleft parietal lobe, in which adults showedsignificantlygreateractivitythanadolescents.Inaddition tothenatureoftasks,thepatternofbrainactivationalsodepends ontheamountofinformation(i.e.,load)thatneedstobemaintained inWM.Previousliteratureexploringage-relatedchangesinbrain activityassociatedwithverbalWMunderconditionsofincreasing loadfoundthatadolescentsandadultsshowedagreaterincreasein activationacrossloadinleftparietal(O’Hareetal.,2008;Thomason etal.,2009),leftlateralprefrontal(Thomasonetal.,2009)andright cerebellar(O’Hareetal.,2008)regionsthanchildren.Incontrast, Jollesetal.(2011)didnotfindage-relatedloadsensitivityin chil-drenandadults.Also,previousreportsusedonlyuptothreelevels ofdifficulty.

Thepresentstudyexaminedtheeffectsofincreasingload,with sixdifficultylevelsinaverbalWMtask,andhowload-dependent change in brain function differed in children and adults. Prior developmentalvisualverbalWMstudiesthatincludedloadasa manipulationusedcomplextasksthatrequiredmaintenanceand reordering(Jollesetal.,2011)orchangedstimuliappearanceand size(O’Hareetal.,2008;Thomasonetal.,2009)asdifficultylevel increased.Inatypicaln-backtask,participantsviewaseriesof stim-uliandindicatewhetherthecurrentlypresentedstimulusmatches onepresented‘n’(e.g.,0,1,2or3)trialsprior.Asdifficultylevel increases,thenumberof interferingstimuli betweenthetarget andrelevantstimulusincreases,requiringtheutilizationof dif-ferentmental strategies ateach level (e.g.,0-back:recognition; 1-back:maintenance;2-back:maintenanceandmonitoring).These manipulationsincreasebothmemoryloadandexecutivefunction demands(i.e.,strategyneededtocompletethetask)inanon-linear

fashionfromeachleveltothenext,makingfunction-specific alter-ationsdifficulttoquantifyandrelatewithspecificbrainregions.To avoidtheseconfounds,weuseda1-backlettermatchingtask(LMT) whichmanipulatedmemoryloadwhilekeepingexecutive func-tionuniformacrossthedifficultylevels,allowingustoinvestigate directlytheimpactofcognitiveloadonverbalWM. The execu-tivedemands(i.e.,proceduralstrategiesforsolvingthetask)were constantacrossalllevelsoftheLMT;whatvariedwitheachlevel wasthenumberofitems(letters)thathadtoberemembered.A visuo-spatialanalogueofLMThasbeenusedsuccessfullytoexplore WMinfunctionalneuroimagingstudiesofadults(Arsalidouetal., 2013)andchildren withand withoutASD(Voganet al.,2014). ObservationsfromthesestudiespointtoalinearpatternofWM functionacrossload.Ourtaskcancaptureneuralcorrelates associ-atedwiththislinearpatternofactivationwithcognitiveload,and ourobjectiveistodeterminewhetherpatternsofactivationinWM processingchangeacrossdevelopment.

Understandingtheeffectofageonbrainregionsimplicatedin WMandWMcapacitycanprovideinsightintothe developmen-taltrajectoryofverbalWMnetworks,andenhanceourabilityto determineoptimaltimingforinterventionsinpaediatric popula-tionswithWMdifficulties.Givenpreviousliteratureandfindings fromourrecentworkusingthevisuo-spatialversionofLMT,we expectedfrontalandparietalcorticalareasassociatedwithWM wouldbeunder-recruitedinchildrencomparedtoadults,andthese differenceswouldincreasewithincreasingcognitiveload.Further, neuralactivationinbothgroupswouldbeleft-hemisphere dom-inant,giventheverbalnatureofthetask(e.g.,Brahmbhattetal., 2008),andthislocalizedpatternwouldbelessevidentinchildren whoaremorelikelytodemonstratediffuseactivation(Scherfetal., 2006).

2. Methods 2.1. Participants

Twenty-fourtypicallydevelopingchildrenaged9–15(M=12.8, SD=1.7; 7 female) and 16 young adults aged 20–25 (M=22.7, SD=1.5;8female) wereincludedin theanalyses.Sixteenother children (i.e., 40 weretested in total)were excluded fromthe analysesduetoexcessivemovement(n=1),inadequatetask per-formance (n=13) or a combination of these factors (n=2). A chi-squareanalysisconfirmed thatsex distributiondidnot dif-ferbetweenagegroups,2

(1)=1.78,p=0.18).Cognitivefunction

of childparticipants wasassessedusing theWechsler Abbrevi-ated Scale of Intelligence-II (Wechsler, 2003) toensure IQ>80 (M=114.6,SD=8.1).Allsubjectswerescreenedandnotincluded onthebasisof anycurrent significantpsychiatriccomorbidities (APA,2013),neurologicaldisorders,medicalillnesses,prematurity, uncorrectedvision,aswellasstandardMRIcontraindicators(e.g., ferromagnetic implants). A history of developmental disorders, learningdisabilityorADHDwasalsousedtoexcludeparticipants. Participantswererecruitedthroughemaillists,postersinthe hos-pitalandcommunity,privateschoolsandwordofmouth.Informed consentandMRIscanningwereperformedattheHospitalforSick ChildreninToronto.Experimentalprocedureswereapprovedby theResearchEthicsBoardattheHospitalforSickChildren,andall participantsprovidedinformedconsent;forthechildren;parents gavewrittenconsentwhilechildrengaveverbalassent.

2.2. Thelettermatchingtask(LMT)

In the current study, LMT is considered a verbal WMtask; althoughitisnotauditoryinnature,itusescompletelyverbal(i.e., linguistic)stimuli.Participantswererequiredtoattendtoletters

Fig.1.ProtocoldescriptionoftheLetterMatchingTask(LMT).(A)Therewere6levelsofdifficultywherethenumberofrelevantletters(A,B,E,K,N,M,N,T)increasedbyoneto increasethedifficultylevel.Difficulty=(#ofrelevantletters)+2.Participantswereaskedtoignoretheglobal“A”figure,letterlocation,letterrepetitionandirrelevantletters (OandP).(B)Itwasablockdesigntask,whereeachrunconsistedofsix32staskblocks(foreachdifficultylevel)followedby20sbaselineblockswherefiguresarepresented withonlyOandP(irrelevantletters).Taskblockswerepresentedpseudo-randomlywithineachrun.(C)Exampleofpartofasequenceinabaselineblock.Participantswere instructednottorespond.(D)Exampleofpartofasequenceinataskblock;participantsindicatedifthecurrentfigure“A”containedofthesameordifferentlettersasthe previousfigure.Allstimuliwerepresentedfor3sfollowedbya1sinter-stimulusfixationcross.

embeddedinafiguralletter‘A’.Participantsweretaughttoignore theglobalfigure‘A’andtwooftheembeddedletters(‘O’and‘P’)that weredefinedasirrelevant,andfocusontheotherletters,defined asrelevant.Thetaskincludedeightrelevantletters(A,B,E,H,K, M,N,T),allwerepresentedinuppercase.Weincludedrelevant andirrelevantletters,astasksthatcontainmisleadingorirrelevant factorscauseinterferenceandincreasedcognitivecontrol;these typesoftaskshavebeenshowntoprovidemorereliablemeasures ofWMcapacity andestimatesofdevelopment(Arsalidouetal., 2010;Powelletal.,2014).Inadditiontoirrelevantletters,thelarge figural‘A’wasalsousedtoevokeinterferenceandelicitcognitive control.Thenumberof‘n’relevantletters(capacity)inthefigures wasincreasedbyoneforeachincreaseinlevelofdifficulty.LMT hastwofunctionsthateachrequireattention:participantsmust firstidentifytherelevantletter(s)embeddedinthe‘A’figureand second,rememberiftheyarethesameordifferentlettersfromthe previousstimulus.Assuch,itemswithonerelevantletter(e.g.n=1) haveadifficultylevelofn+2(e.g.,3;Fig.1A)toaccountforthe addi-tionalfunctionsmentionedabove.Thestimuliwerepresentedone atatime.Participantsindicatedaftereachstimuluswhetherthe

relevantlettersofthecurrentfigurematchedthosefromthe imme-diatelyprecedingfigure(i.e.1-back),disregardingletterlocation andrepetition.Usingadual-keyMRIcompatiblekeypadwiththe righthand,participantsrespondedinthescannerbypushingone buttonfor‘same’whenthestimuluscontainedthesameembedded lettersastheimmediatelypreviousstimulusandanotherbuttonfor ‘different’whenthestimulushaddifferentembeddedletters. Non-matchingstimuli(i.e.,‘different’stimuli)differedfromtheprevious stimulibyonlyoneletterforalllevels.Priortoscanning,all partic-ipantsweretrainedandsuccessfullycompletedpracticetrialsona computeroutsidethescannerwithaccuracyof80%orgreater.

Twenty-fourtaskblocks(withatotalof168tasktrials)and24 baselineblockswerepresentedacrossfourruns.Eachrunincluded six32-sblocks,oneforeachofthesixdifficultylevels;eachtask blockconsistedofeightstimuliofthesamedifficultylevelandthe sixdifficultylevelswererandomizedwithineachrun(Fig.1B).Task blocksalternatedwith20-sbaselineblocks(Fig.1C),wherethe‘A’ figurescontainedonly‘O’and‘P’,theirrelevantletters,and partic-ipantswereinstructedtolookatthefiguresbutnotrespond.The samefourrunswerepresentedtoallparticipantsinthesameorder.

Participantshad3stoviewastimulusandrespond,followedby a1-sinter-stimulusintervalwhereafixationcrosswaspresented (Fig.1D).ThefMRItasktookapproximately22minofscantime.

Behavioraldatawererecorded foraccuracy (proportion cor-rect)andreactiontime;itemswerecorrectifsubjectsresponded correctlywithin3-sofstimulusonset.Participantswereexcluded iftheydidnotreachatleast70%accuracy(averagedacrossfour runs)onthetwoeasiestlevels,andwerealsoexcludediftheydid nothaveatleasttwooutofthefourrunswhere50%ormoreof theblocks wereacceptableinterms ofperformance(70% accu-racy)andmotion.A70%accuracycriterionwaschosentoensure thatparticipantswereperformingbetterthanchance(50%).Motion wasconsideredacceptableifparticipantsmovedlessthan1.5mm fromtheirmedianheadpositionin atleast60%ofthevolumes withinataskblock.SeeSection2.5foradescriptionofdisplacement calculations.

2.3. Imageacquisition

All imaging data were acquired using a 3T Siemens Trio MRI scanner with a 12-channel head coil. Motion restric-tion and head stabilization were achieved with foam padding. A high-resolution T1-weighted 3D MP-RAGE structural scan (Sagittal; FOV=192×240×256mm; 1mm isometric voxels; TR/TE/TI/FA=2300/2.96/900/9),wasusedasanindividual anatom-ical reference for the fMRI images. During structural MRI acquisition, participants watched their choice of movie using MRcompatible gogglesand earphones.Functionalimageswere acquiredwithasingle-shotechoplanarimagingsequence(Axial; FOV=192×192;Res=64×64;30slices5mmthick;3×3×5mm voxels;TR/TE/FA=2000/30/70). Visual stimuli for the LMTtask were displayed on the MR compatible goggles. Stimuli were displayedandbehaviouralperformancewasrecordedusing Pre-sentation(NeurobehavioralSystemsInc.,Berkeley,CA,USA). 2.4. Behavioraldataanalyses

Behavioralaccuracy (%correct)and reactiontimeswere cal-culatedfor each difficulty level,averaging acrossruns foreach individual.Childrenshowedpooraccuracyondifficultylevels7 and8(D7:M=0.64,SD=0.12;D8:M=0.57,SD=0.13),wherethey performedonlymarginallyabovechance(50%).Thusanalysesof onlythefirstfourdifficultylevels(D3toD6)werecompletedto eliminateperformanceasaconfoundingfactorinbrainactivation. Thedatawereanalyzedusingrepeatedmeasuresanalysesof vari-ance(ANOVA),withabetween-subjectfactorofgroup(children andadults)andawithin-subjectfactorofdifficultylevel(i.e.,load; D3,D4,D5,D6).

2.5. fMRIdataanalyses

ImagepreprocessingoffMRIdatawascompletedusinga com-bination of tools from AFNI(Cox, 1996) and FMRIB’s Software Library(FSL; Worsley,2001).We discardedthefirst three vol-umesofeachrunforscannerstabilization.Afterslice-timingand motioncorrection(using3dvolreg),thedataweresmoothed in-planeusingaFWHMGaussiankernel(6mm),temporallyfiltered (lowerand upper cut-off frequencies of 0.01Hz &0.2Hz), and thenconverted topercentsignalchangefromthebaseline vol-umes.ThemaximumEuclideandisplacement(MD)thatanyvoxel movedwithinthebrainwascalculatedusing3dvolregfromAFNI, and wasperformedduring themotioncorrectionstage of pre-processing.WeusedtheMDmetrictoflagthose volumeswith unacceptablemotion(greaterthan1.5mmfromthemedianhead position). If more than 1/3 of volumes exceeded these criteria withinataskblock,theblockwasconsideredtohaveexcessive

motion.Weexploredgroupdifferencesinheadmotionusingthe averageMDforeachsubject.Althoughmoremotionwasfound inchildren(M=0.32mm,SD=0.21mm)thanadults(M=0.14mm, SD=0.11mm),t(38)=3.50,p=0.001,bothgroupshadminimal

aver-agemotionofunder0.35mm.Tofurthercontrolformotion,theMD signalwasincludedasano-interestcovariateintheGLM.Finally, priortogroup-levelanalyses,theimageswereregisteredtothe MNI-152template.

Data were analyzedwith theFSL fMRI ExpertAnalysisTool (FEAT;Worsley,2001).Datawerefitfirsttoablock-designgeneral linearmodelconvolvedwithagammafunctiontomodel hemody-namicresponse,usingthetaskparameters(D3toD6).Toexamine areasthat showedincreasingactivationwithincreasingloadin eachgroup,lineartrendanalyseswereconductedfromD3toD6 usingfixed-effectshigherlevelmodelingforeachgroupseparately. Individualresultswerethenaveragedacrossrunsforeach sub-jectina secondlevelanalysis.Toexamine groupdifferencesin load-dependentgrowthsofactivation,amixed-designANOVAwas used,testingforaGroup×Loadinteraction.Between-group com-parisonswerecarriedoutusingFMRIB’sLocalAnalysisofMixed Effects-1(FLAME1;Worsley,2001)toobtainanaccurate between-subjectvarianceestimation,whichincreasedourabilitytodetect realactivation(Woolrichetal.,2009).Significantactivationswere reportedusingcluster-basedthresholdingdeterminedbyZ>|2.3|, andacorrectedclustersignificancethresholdofp<0.05.

2.6. Lateralityindexcalculation

Laterality indices (LI) were calculated between all bilateral regions ontheAAL90-region atlas.A 5mm radiussphere was drawnaroundeachcoordinateintheatlasandwastransformedto eachsubjects’functionalspace.Theaveragestatisticalvaluewithin eachspherewascalculatedandusedtocalculateLIasfollows: LI= left_left−right

+right

Anegativeindexrepresentsright-lateralizedactivationwhereas positive LIrepresents left-lateralizedactivation.Falsediscovery rate(FDR)wasusedtocorrectformultiplecomparisons.

2.7. Brain-behaviorexploratoryanalyses

Anexploratoryanalysiswasconductedtounderstandthelink betweenLMTperformance(i.e.,accuracy)andneuralactivation. Pearson’srwascomputedtoexplorewhetherincreasingactivation fromD3toD6(i.e.,D6>D3contrast)wasrelatedtoperformance atthemostdifficultlevel(D6)inareasthatdemonstrateda signif-icantLoad×GroupinteractioninthefMRIanalysisabove.AnFDR thresholdofp=0.05wasusedtocorrectformultiplecomparisons. 3. Results

3.1. Behavioraldata

To ensure that there was no sex distribution difference in oursample,weanalyzedbehavioralperformanceacrossdifficulty for males and females. Findings demonstrated that there was nosex×loadinteraction(F(1)=0.691,p=0.495);bothmalesand

femaleswereshowingsimilarperformancetrendsacrossdifficulty, regardlessoftheirgroup.

Duetoviolationsofsphericity,weonlyreportsignificantresults that also survived Greenhouse–Geisser correction. To test age andloadeffects onLMTaccuracy,a repeatedmeasuresANOVA was performed with load (D3–D6) as within-subjects variable andwithgroup(childrenandadults)asabetween-subjects fac-tor.TheANOVAdemonstratedamaineffectofloadonaccuracy,

Table1

LMTbehavioralperformance:accuracyandresponsetimes.

Difficultylevel Children Adults

Accuracy(%) S.D.(%) Accuracy(%) S.D.(%) D3 0.94 0.08 0.96 0.06 D4 0.90 0.09 0.96 0.04 D5 0.83 0.14 0.92 0.10 D6 0.75 0.16 0.86 0.13 RT(s) S.D.(s) RT(s) S.D.(s) D3 1.21 0.20 1.09 0.24 D4 1.48 0.23 1.29 0.25 D5 1.74 0.21 1.57 0.27 D6 1.89 0.22 1.74 0.21

F(3)=24.84, p<0.001, p2=0.40, withaccuracy decreasing with increasing load in both children and adults. Post hoc pairwise comparisons,adjustedformultiplecomparisonsusingBonferroni, revealedthataccuracysignificantlydecreasedwithincreasingload (p<0.05), except between the two easiest and two most diffi-cultlevelsinchildren.Inadults,accuracysignificantlydecreased betweenD3andD6andbetweenD4andD6(p<0.05).Therewas alsoasignificanteffectofgrouponaccuracy,F(1)=6.66,p=0.01, p2=0.15,withadultsperformingsignificantlybetteronD4,D5 andD6(seeTable1andFig.2A).However,childrenperformedwell above-chance(>75%)onalldifficultylevels,andbetweengroup dif-ferencesinbrainactivitywerenotconfoundedbyinadequatetask performanceforeithergroup.AGroup×Loadinteractionwasnot foundinaccuracyscores,F(3)=2.20,p=0.12.

AsimilarANOVAwasconductedtotestageandloadeffectson LMTresponsetimes.Overall,there wasamaineffectofloadon responsetimes,F(3)=182.90,p<0.001,p2=0.83,withresponse

timesincreasingasfunctionofloadforbothchildrenandadults. Posthoccomparisonsconfirmedthatresponsetimesincreased sig-nificantlyateachdifficultylevelinchildren(p<0.01)andadults (p<0.05). Therewas alsoa significantmain effect of group on

Fig.2.LMTbehavioralperformance.(A)MeanproportioncorrectforD3toD6and standarderrorbars.*p<0.05.(B)MeanresponsetimesforD3toD6andstandard errorbars.*p<0.05.

responsetimes,F(1)=6.28,p<0.05,p2=0.14,withchildren

show-inglongerresponsetimethanadultsatD4,D5andD6(seeTable1 andFig.2B).AGroup×Loadinteractionwasnotfoundinresponse times.

3.2. Functionalimagingresults

ThefMRIanalysesdeterminedifthepatternofbrainactivity dif-feredasafunctionoftheWMload(i.e.,difficultylevel)andbetween childrenandadults.Lineartrendanalyses(D3toD6)showedthat somebrainareasshowedincreasesinactivitywithdifficultylevel, whileothersdecreased.Weuse‘increasingactivation’torefertoan increaseintheBOLDsignalwithanincreaseinWMload(i.e.,a pos-itivelinearrelationbetweenBOLDactivityandtaskdifficultylevel) and‘decreasingactivation’torefertoadecreaseintheBOLD sig-nalwithanincreaseinload(i.e.,anegativelinearrelationbetween BOLDactivityandtaskdifficulty).

3.2.1. Task-relatedactivationwithingroup:Children

The influence ofload onfunctional activity wasfirst exam-inedseparatelyinchildrenandadults.AsshowninFig.3,areas thatexhibitedincreasingactivation(i.e.,showedapositivelinear relation)underconditionsofincreasingWMloadamongchildren were:bilateralinsula,anteriorcingulateextendingtothemedial prefrontalcortex,middlefrontalgyrus,precuneus,middle occip-italgyrus,temporaloccipitalfusiformcortex,cerebellumandleft medialfrontalgyrusextendingtothecingulate.Areasthat exhib-iteddecreasingactivation(i.e.,showedanegativelinearrelation) underconditionsofincreasingWMloadwere:leftanteriormedial frontalcortex,rightmiddletemporalgyrus,leftangulargyrusand leftposteriorcingulate(seeSupplementarymaterialTable1). 3.2.2. Task-relatedactivationwithingroup:Adults

Amongadults,areasthatexhibitedincreasingactivationasa functionofWMloadwere:bilateralinsula,inferiorfrontalgyrus, precuneus,middleoccipitalgyrus,cerebellum,middlefrontalgyrus andleftsuperiorfrontalgyrusextendingtothecingulate.Areas thatdisplayeddecreasingactivationasafunctionofWMloadwere: bilateralanteriormedialprefrontalcortex,superiortemporalgyrus, leftangulargyusandleftposteriorcingulate(seeSupplementary materialTable2).

3.2.3. GroupdifferencesinWMload-dependentactivation

A Group×Load analysis tested for age effects in WM load-dependentactivation.Groupdifferenceswereobservedforboth negativeandpositivelinearload-dependentrelations.Asshown in Fig.4, withincreasing WM load, BOLD activitywas signifi-cantly greater in adults than children in the bilateralsuperior parietalgyrusextendingtotheprecuneus,inferiorfrontalgyrus, leftmiddlefontalgyrusandrightcerebellum.Comparedto chil-dren,adultsshowedsignificantdecreasesinactivationacrossWM loadinthebilateralanteriorcingulate,anteriormedialprefrontal gyrus,rightsuperiorlateraltemporalgyrusandleftposterior cin-gulate(Table2).

3.2.4. Lateralityanalyses

Forbothchildrenandadults,noneofthe45bilateralAALregions testedexhibitedanLIthatwassignificantlydifferentfrom0(usinga 2-tailedt-test),suggestingthatBOLDactivitywassymmetricacross hemispheres.

3.2.5. Brain-behaviorcorrelations

Brain-behavior relations were explored in frontal and pari-etalareasthatdemonstratedsignificantLoad_×Groupinteractions. Whencollapsing adultand child groups, accuracy atD6 (high-estload)wassignificantlypositivelycorrelatedwiththeD6>D3

Fig.3. GroupactivationmapsforthelineartrendanalysisinchildrenandadultsduringLMT.Significantactivationsusingcluster-basedthresholdingdeterminedbyZ>|2.3| andacorrectedclustersignificantthresholdofp=0.05(usingtheFEATtoolboxofFSL).Areasinreddepictregionsofincreasingactivationasafunctionofdifficultylevel, andareasinbluedepictregionsofdecreasingactivationasafunctionofdifficultylevel.mPFC=medialprefrontalcortex,MidFG=middlefrontalgyrus,PCC=posterior cingulatecortex,Cing=cingulate,AngG=AngularGyrus;Prec=precuneus,ACC=anteriorcingulatecortex,IFG=inferiorfrontalgyrus,amPFC=anteriormedialprefrontal cortex,SupTempG=superiortemporalgyrus,MidTempG=middletemporalgyrus,TempOcFusG=temporaloccipitalfusiformgyrus.(Forinterpretationofthereferencesto colorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

contrastofbrainactivationinthebilateralsuperiorparietalgyrus (Left:r(38)=0.42,p=0.006;Right:r(38)=0.44,p=0.004),leftinferior

frontalgyrus(r(38)=0.51,p=0.001)andleftmiddlefrontalgyrus

(r(38)=0.40, p=0.011), FDR corrected for multiple comparisons

usingathresholdofp=0.05.Inotherwords,asD6performance increased,thechangeinactivationfromD3toD6alsoincreased in these frontal and parietal areas. However, when exploring withingroupbrain-behaviorcorrelationsatD6,childrenshowedno regionsinwhichD6performancewascorrelatedwithneural acti-vation.However,adultsshowedasignificantcorrelationbetween D6accuracyandchangeinactivationfromD3toD6intheleft infe-riorfrontalgyrus(r(14)=0.67,p=0.005),FDR-correctedformultiple

comparisonswithathresholdofp=0.05. 4. Discussion

The present study examined age-related differences in the neuralcorrelatesof verbalworkingmemory,and theimpactof increasingcognitiveload,usingataskthatisolatedWMfrom exec-utivefunctiondemands.Althoughadultsperformedbetterthan

childrenonthethreehigherdifficultlylevelsofthetask(LMT), dif-ferenceswereonlymoderateandbothgroupshadatleasta75% accuracyrateonalllevels.Therefore,functionalneuralactivation inresponsetoLMTwasnotconfoundedbyinadequatetask perfor-manceforchildrenoradults.Neuroimagingresultsdemonstrate manysimilaritiesinchildren andadultsintheneuralactivation patternsassociatedwithincreasingverbalWMcapacity.Bothage groupsalsodemonstratedanopposingsystemofcognitive pro-cesseswhereareasrelated toWM(frontaland parietalregions) increasedinactivity,andareasassociatedwiththedefaultmode network (DMN) decreased in activity withincreasing cognitive demand.However,therewereagegroupdifferencesobservedfor linear-loaddependentfunctionalactivation.

Althoughchildrenexhibitedtask-relatedactivityinmanyofthe samebrainregionsasadults,theyfailedtoexhibitthesamedegree ofincreasingactivationacrosscognitiveloadasadultsin multi-plefrontalandparietalcorticalregions,consistentwithprevious studies(Thomasonetal.,2009;O’Hareetal.,2008).Theseareas includedthebilateralsuperiorparietygyrusextendingtothe pre-cuneus,bilateralinferiorfrontalgyrus,leftmiddlefrontalgyrusand

Fig.4.ResultsfromtheGroup×Loadinteractionanalysis.Significantactivationsusingcluster-basedthresholdingdeterminedbyZ>|2.3|andacorrectedclustersignificant thresholdofp=0.05.Areasinreddepictregionswhereadultsshowedgreaterincreasingactivationasafunctionofdifficultylevelcomparedtochildren.Areasinbluedepict regionsofwhereadultsshowedgreaterdecreasingactivationasafunctionofdifficultylevelcomparedtochildren.MidFG=middlefrontalgyrus,PCC=posteriorcingulate cortex,Cing=cingulate,ACC=anteriorcingulatecortex,IFG=inferiorfrontalgyrus,amPFC=anteriormedialprefrontalcortex,SupParietalGyrus=superiorparietalgyrus; Prec=precuneus,SupTempG=superiortemporalgyrus,Cereb=cerebellum.(Forinterpretationofthereferencestocolorinthisfigurelegend,thereaderisreferredtothe webversionofthisarticle.)

Table2

Areasthatshowedasignificantgroup×loadinteraction.

Regionswhereadultsshowedgreateractivationasafunctionofdifficultylevelthanchildren

Region Hem MNICoordinates ClusterSize(voxels) Zvalue P-value

x y z

SuperiorParietalGyrusextendingtoPrecuneus L −30 −64 46 2962 6.35 2.96×10−10

SuperiorParietalGyrusextendingtoPrecuneus R 36 −60 54 2595 6.19 2.68×10−9

InferiorFrontalGyrus L −48 26 18 924 5.05 3.94×10−4

InferiorFrontalGyrus R 38 30 14 682 4.64 3.56×10−3

Cerebellum R 24 −64 −24 634 4.54 5.66×10−3

MiddleFrontalGyrus L −34 6 58 456 4.65 0.04

Regionswhereadultsshowedlessactivationasafunctionofdifficultylevelthanchildren

Region Hem MNICoordinates Clustersize(Voxels) Zvalue p-Value

x y z

Anteriormedialprefrontalcortex L −4 62 22 3155 6.49 9.62×10−11

Anteriorcingulate/cingulategyrus L −4 18 24 X 5.51

Anteriorcingulate/cingulategyrus R 4 14 26 X 4.69

Anteriormedialprefrontalcortex R 12 68 10 X 5.05

Heschl’sgyrus R 42 −20 14 1707 4.88 8.94×10−7

Posteriorcingulatecortex L −12 −50 36 1094 4.24 9.42×10−5

Note:thep-valueshownforeachclusterrepresentstheestimatedsignificanceasdeterminedbytheFEATtoolboxofFSL;X=peaklocalmaximaswithincluster.

rightcerebellum.Thus,althoughthenumberofbrainregions show-ingload-dependentactivationdidnotchangeacrossage,theextent towhichparticipantsreliedontheseareasinresponsetoincreasing cognitiveloadchangedbetweenchildhoodandadulthood.

Increasedrecruitmentoffrontalandparietalareasduring work-ingmemorytaskshasbeenshowntounderlieimprovementsin workingmemoryandcognitivecontroloverthecourseof devel-opment(seeBungeandWright,2007,for areview).TheDLPFC (i.e.,middlefrontalgyrus)isreferredtoasacore “performance-enhancing” region believed to play a critical role in holding information ‘online’ (Powell and Voeller, 2004) and mediating strategic organization and data compression processes (Rypma etal.,2002;Boretal.,2003).Thisisconsistentwiththedorsolateral

prefrontalcortexbeingsensitivetoincreasingcognitivedemand in oursand other studies(Scherfet al.,2006; Thomasonetal., 2009).Theinferiorfrontalgyrushasalsobeenidentifiedasanarea involvedinhighercognitivemonitoring(e.g.choosing,comparing, judgingandretrieving),aswellaslanguageprocessing,specifically inthelefthemisphere(seeLiakakisetal.,2011forreview;Strand etal.,2008).Findingsfromourstudysuggestthatadultsare increas-inglyrecruitingtheseregions,moresothanchildren,toadjustfor increasingtaskdemand.

Incontrastwithsomedevelopmentalneuroimagingstudiesof verbalWM,(Thomasonetal.,2009;Nageletal.,2013),wedidnot findtheexpectedlefthemispherelateralizationofneural activa-tion.Similar toTamnesetal. (2013)who alsodidnotfindleft

lateralizationinaverbalWM‘KeepTrackTask’,weusedacomplex taskwithvisuallypresentedstimuliandvisual-searchcomponent thatmightdependonawider,moregeneralneuralnetwork. Partic-ipantsmaynothavetransformedvisuallypresentedverbalstimuli (i.e.,letters)intoaphonologicalcode,andinsteadmayhaverelied onvisual-search strategies. Adults may have been more likely thanchildrentorelyonsuchvisual-searchstrategiesascognitive demandincreased,giventhegreaterrelianceontheprecuneuswith increasingcognitiveload–anarearesponsivetospatialvisual pro-cessing(UngerleiderandMishkin,1982)andwidelyimplicatedasa hubregioninthebrain(e.g.,MishkinandUngerleider,1982;Smith etal.,1998;vandenHeuvelandSporns,2011)–astaskdifficulty increased.

Inadditiontotheincreasedactivationwithtaskdifficulty,we alsofoundtheinverse,withdecreasesasafunctionofloadinareas typicallyassociatedwiththedefault-modenetwork(Raichleetal., 2001), includingthemedialprefrontal,precuneusand posterior cingulate.Thispatternofdecreaseswasalsoseeninthe visual-spatialanalogueofthistaskinadults(Arsalidouetal.,2013)and children(Voganetal.,2014).Inthecurrentstudy,bothchildren andadultsshowedthesedecreaseswithincreasingload,butthe effectsweregreaterintheadults,suggestingthattheyarebetter abletomodulatemind-wandering(Buckneretal.,2008),during performanceofademandingtask.Bothgroupsshoweddecreases intheleftangulargyrus,closetoWernicke’sarea,involvedin lan-guagefunction.Thisareaisimplicatedinreadingandsuggested thatbothchildrenandadults,neededtosupresstheactivityinthis area,tostopreadingtheletters,tocompletetheharderlevelsof thistask.

Cerebellaractivationwasseeninbothchildrenandadults,with theinteractionshowinggreater rightcerebellarinvolvement in adultswithincreasingmemoryload.Many studieshave shown thecerebellumisimplicatedincognitiveandexecutivefunctions (forreviews see De Smet et al., 2013; Strick et al.,2009), and deficitsassociatedwithcerebellardysfunctionareoftenreported in working memory and verbal fluency in adults and children (BellebaumandDaum,2007;Gottwaldetal.,2004;Levisohnetal., 2000).Consistentwithourfindings,particularlyrightcerebellar activationsareseeninadultsinverbaltasks(ChenandDesmond, 2005;Kirschenetal.,2005;O’Hareetal.,2008).Thespecificrole ofthecerebellum incognitivefunctionsisstilltobedelineated (TimmannandDaum,2007;Stoodley,2012);however,our find-ingsofincreasingcerebellaractivationwithincreasingverbalWM load,withage-relateddifferences,alignswiththemodelofitsrole inexecutivefunctionswhichshowmarkeddevelopmentbetween mid-childhoodandadulthood.

Toavoidperformancebeingaconfoundingfactorintheneural activationduringLMT,individuals withinadequate task perfor-mancewereexcludedfromthepresentstudy.Consequently,our samplemay beless representative ofchildren withlowerWM capacity.Further,duetothecomplexnatureofLMT,wecouldnot analyzethehighesttwotaskdifficultylevels(D7andD8)where childrenperformedonlyslightlyabovechance.UnderstandingWM processingat very highcognitive demands is importantin our understandingthedevelopmentofWMcapacity.Inaddition,as manycognitivetasks,selectiveattentionisrequiredwhen complet-ingLMT.However,wedidnotadministeranindependentmeasure ofselectiveattention,andthereforeitsspecificinfluenceonthis taskcannotbeanalyzed. Thecurrent studyextends knowledge ofverbal WMdevelopmentbyusing a noveltaskthat isolated cognitiveloadonWM,holdingexecutivefunctionconstant,and utilizingmultipledifficultylevelstoexaminetheimpactof cogni-tiveload.Performanceonthiscomplextaskdependedonageneral neuralnetworkrather thana leftlateralized neuralsystem,and ourfindings demonstratethatwithdevelopment,thedegreeto whichindividuals rely onfrontaland parietalregions increases

withcognitivedemands.Futurestudieswillcomparenormative developmentalpatternsofverbalWMprocessingtoclinical popu-lationswhohavecognitiveandmemorydysfunction.

Acknowledgements

Theauthorswouldliketothankallofthefamiliesandchildren fortheirsupportandparticipation.Wewouldalsoliketothank RachelLeungandBeckyBaatjesforadministeringtheADOS,and Dr.JessicaBrianforreviewingallassessments.Sincerethanksto ourMRItechnicians,RuthWeissandTammyRayner,foralltheir supportindataacquisition.Theauthorswouldalsoliketothank MarieArsalidoufordevelopingtheLMTtask,andallowingustouse it.Lastly,thanksChiaraBorrelliinherassistancewithdataanalysis. ManythankstoRinaGoukonforherhelpwithparticipant recruit-mentanddatacollection.ThisresearchwasfundedbyCanadian InstitutesofHealthResearch(MOP-106582).

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.dcn.2015.10.008. References

Alloway,T.P.,2009.Workingmemory,butnotIQ,predictssubsequentlearningin childrenwithlearningdifficulties.Eur.J.Psychol.Assess.25(2),92–98.

AmericanPsychiatricAssociation,2013.DiagnosticandStatisticalManualof MentalDisorders,fifthed.AmericanPsychiatricPublishing,Arlington,VA.

Arsalidou,M.,Pascual-Leone,J.,Johnson,J.,2010.Misleadingcuesimprove developmentalassessmentofworkingmemorycapacity:thecolormatching tasks.Cogn.Dev.25(3),262–277.

Arsalidou,M.,Pascual-Leone,J.,Johnson,J.,Morris,D.,Taylor,M.J.,2013.A balancingactofthebrain:activationsanddeactivationsasafunctionof cognitiveload.BrainBehav.3(3),273–285.

Baddeley,A.,1992.Workingmemory.Science255(5044),556–559.

Bellebaum,C.,Daum,I.,2007.Cerebellarinvolvementinexecutivecontrol. Cerebellum6(3),184–192.

Bor,D.,Duncan,J.,Wiseman,R.J.,Owen,A.M.,2003.Encodingstrategiesdissociate prefrontalactivityfromworkingmemorydemand.Neuron37(2),361–367.

Brahmbhatt,S.B.,McAuley,T.,Barch,D.M.,2008.Functionaldevelopmental similaritiesanddifferencesintheneuralcorrelatesofverbalandnonverbal workingmemorytasks.Neuropsychologia46(4),1020–1031.

Buckner,R.L.,Andrews-Hanna,J.R.,Schacter,D.L.,2008.Thebrain’sdefault network.Ann.N.Y.Acad.Sci.1124(1),1–38.

Bunge,S.A.,Wright,S.B.,2007.Neurodevelopmentalchangesinworkingmemory andcognitivecontrol.Curr.Opin.Neurobiol.17(2),243–250.

Cain,K.,Oakhill,J.,Bryant,P.,2004.Children’sreadingcomprehensionability: concurrentpredictionbyworkingmemory,verbalability,andcomponent skills.J.Educ.Psychol.96(1),31–42.

Chen,S.H.A.,Desmond,J.E.,2005.Temporaldynamicsofcerebro-cerebellar networkrecruitmentduringacognitivetask.Neuropsychologia43(9), 1227–1237.

Colom,R.,Jung,R.E.,Haier,R.J.,2007.Generalintelligenceandmemoryspan: evidenceforacommonneuroanatomicframework.Cogn.Neuropsychol.24 (8),867–878.

Conklin,H.M.,Luciana,M.,Hooper,C.J.,Yarger,R.S.,2007.Workingmemory performanceintypicallydevelopingchildrenandadolescents:behavioral evidenceofprotractedfrontallobedevelopment.Dev.Neuropsychol.31(1), 103–128.

Cox,R.W.,1996.AFNI:Softwareforanalysisandvisulizationoffunctional magneticresonanceneuroimages.Comput.Biomed.Res.29,162–173.

DeSmet,H.J.,Paquier,P.,Verhoeven,J.,Mariën,P.,2013.Thecerebellum:itsrolein languageandrelatedcognitiveandaffectivefunctions.BrainLang.127(3), 334–342.

Dennis,M.,Agostino,A.,Roncadin,C.,Levin,H.,2009.Theoryofminddependson domain-generalexecutivefunctionsofworkingmemoryandcognitive inhibitioninchildrenwithtraumaticbraininjury.J.Clin.Exp.Neuropsychol. 31(7),835–847.

Dumontheil,I.,Klingberg,T.,2012.Brainactivityduringavisuospatialworking memorytaskpredictsarithmeticalperformance2yearslater.CerebralCortex 22(5),1078–1085.

Engle,R.W.,2002.Workingmemorycapacityasexecutiveattention.Curr.Dir. Psychol.Sci.11(1),19–23.

Engle,R.W.,Tuholski,S.W.,Laughlin,J.E.,Conway,A.R.A.,1999.Workingmemory, short-termmemory,andgeneralfluidintelligence:alatent-variableapproach. J.Exp.Psychol.:Gen.128(3),309–331.

Gathercole,S.E.,Alloway,T.P.,2004.Workingmemoryandclassroomlearning.J. Prof.Teach.StudentsSpecificLearn.Difficulties17,2–12.

Gathercole,S.E.,Pickering,S.J.,Knight,C.,Stegmann,Z.,2004a.Workingmemory skillsandeducationalattainment:evidencefromnationalcurriculum assessmentsat7and14yearsofage.Appl.Cogn.Psychol.18(1),1–16.

Gathercole,S.E.,Pickering,S.J.,Ambridge,B.,Wearing,H.,2004b.Thestructureof workingmemoryfrom4to15yearsofage.Dev.Psychol.40(2),177–190.

Gottwald,B.,Wilde,B.,Mihajlovic,Z.,Mehdorn,H.M.,2004.Evidencefordistinct cognitivedeficitsafterfocalcerebellarlesions,JournalofNeurology.Neurosurg. Psychiatry75(11),1524–1531.

Huizinga,M.,Dolan,C.V.,van,d.M.,2006.Age-relatedchangeinexecutive function:developmentaltrendsandalatentvariableanalysis. Neuropsychologia44(11),2017–2036.

Hwang,Y.,Hosokawa,T.,2007.WorkingmemoryinJapanesechildrenwith learningdisabilities:assessmentusingtheJapaneseshortformoftheSwanson cognitiveprocessingtest.Jpn.J.SpecialEduc.44(6),463–472.

Jolles,D.D.,Kleibeuker,S.W.,Rombouts,S.A.R.B.,Crone,E.A.,2011.Developmental differencesinprefrontalactivationduringworkingmemorymaintenanceand manipulationfordifferentmemoryloads.Dev.Sci.14(4),713–724.

Kirschen,M.P.,Chen,S.A.,Schraedley-Desmond,P.,Desmond,J.E.,2005.Load-and practice-dependentincreasesincerebro-cerebellaractivationinverbal workingmemory:anfMRIstudy.Neuroimage24(2),462–472.

Klingberg,T.,Forssberg,H.,Westerberg,H.,2002.Increasedbrainactivityinfrontal andparietalcortexunderliesthedevelopmentofvisuospatialworking memorycapacityduringchildhood.J.Cogn.Neurosci.14(1),1–10.

Kwon,H.,Reiss,A.L.,Menon,V.,2002.Neuralbasisofprotracteddevelopmental changesinvisuo-spatialworkingmemory.Proc.Natl.Acad.Sci.USA99, 13336–13341.

Levisohn,L.,Cronin-Golomb,A.,Schmahmann,J.D.,2000.Neuropsychological consequencesofcerebellartumourresectioninchildren:cerebellarcognitive affectivesyndromeinapaediatricpopulation.Brain:J.Neurol.123(5), 1041–1050.

Liakakis,G.,Nickel,J.,Seitz,R.J.,2011.Diversityoftheinferiorfrontalgyrus—a meta-analysisofneuroimagingstudies.Behav.BrainRes.225(1),341–347.

Martinussen,R.,Hayden,J.,Hogg-Johnson,S.,Tannock,R.,2005.Ameta-analysisof workingmemoryimpairmentsinchildrenwithattention-deficit/hyperactivity disorder.J.Am.Acad.ChildAdolesc.Psychiatry44(4),377–384.

Mishkin,M.,Ungerleider,L.G.,1982.Contributionofstriateinputstothe visuospatialfunctionsofparieto-preoccipitalcortexinmonkeys.Behav.Brain Res.6(1),57–77.

Nagel,B.J.,Herting,M.M.,Maxwell,E.C.,Bruno,R.,Fair,D.,2013.Hemispheric lateralizationofverbalandspatialworkingmemoryduringadolescence.Brain Cogn.82(1),58–68.

O’Hare,E.D.,Lu,L.H.,Houston,S.M.,Bookheimer,S.Y.,Sowell,E.R.,2008.

Neurodevelopmentalchangesinverbalworkingmemoryload-dependency:an fMRIinvestigation.Neuroimage42(4),1678–1685.

Powell,T.L.,Arsalidou,M.,Vogan,V.,Taylor,M.J.,2014.Letterandcolourmatching tasks:parametricmeasureofdevelopmentalworkingmemorycapacity.Child Dev.Res.2014(ArticleID961781)9pages.

Powell,K.B.,Voeller,K.K.S.,2004.Prefrontalexecutivefunctionsyndromesin children.J.ChildNeurol.19(10),785–797.

Raichle,M.E.,MacLeod,A.M.,Snyder,A.Z.,Powers,W.J.,Gusnard,D.A.,Shulman, G.L.,2001.Adefaultmodeofbrainfunction.Proc.Natl.Acad.Sci.U.S.A.98(2), 676–682.

Rypma,B.,Berger,J.S.,D’Esposito,M.,2002.Theinfluenceofworking-memory demandandsubjectperformanceonprefrontalcorticalactivity.J.Cogn. Neurosci.14(5),721–731.

Scherf,K.S.,Sweeney,J.A.,Luna,B.,2006.Brainbasisofdevelopmentalchangein visuospatialworkingmemory.J.Cogn.Neurosci.18(7),1045–1058.

Smith,E.E.,Jonides,J.,Marshuetz,C.,Koeppe,R.A.,1998.Componentsofverbal workingmemory:evidencefromneuroimaging.Proc.Natl.Acad.Sci.U.S.A.95, 876–882.

Southwick,J.S.,Bigler,E.D.,Froehlich,A.,DuBray,M.B.,Alexander,A.L.,Lange,N., Lainhart,J.E.,2011.Memoryfunctioninginchildrenandadolescentswith autism.Neuropsychology25(6),702.

Sowell,E.R.,Thompson,P.M.,Holmes,C.J.,Jernigan,T.L.,Toga,A.W.,1999.Invivo evidenceforpost-adolescentbrainmaturationinfrontalandstriatalregions. Nat.Neurosci.2(10),859–861.

Spencer-Smith,M.,Ritter,B.C.,Mürner-Lavanchy,I.,El-Koussy,M.,Steinlin,M., Everts,R.,2013.Age,sex,andperformanceinfluencethevisuospatialworking memorynetworkinchildhood.Dev.Neuropsychol.38(4),236–255.

Stoodley,C.J.,2012.Thecerebellumandcognition:evidencefromfunctional imagingstudies.Cerebellum11(2),352–365.

Strand,F.,Forssberg,H.,Klingberg,T.,Norrelgen,F.,2008.Phonologicalworking memorywithauditorypresentationofpseudo-words–aneventrelatedfMRI study.BrainRes.1212,48–54.

Strick,P.L.,Dum,R.P.,Fiez,J.A.,2009.Cerebellumandnonmotorfunction.Annu. Rev.Neurosci.32,413–434.

Tamnes,C.K.,Walhovd,K.B.,Grydeland,H.,Holland,G.,Ostby,Y.,Dale,A.M.,etal., 2013.Longitudinalworkingmemorydevelopingisrelatedtostructural maturationoffrontalandparietalcortices.J.Cogn.Neurosci.25,1611–1623.

Thomason,M.E.,Race,E.,Burrows,B.,Whitfield-Gabrieli,S.,Glover,G.H.,Gabrieli, J.D.E.,2009.Developmentofspatialandverbalworkingmemorycapacityin thehumanbrain.J.Cogn.Neurosci.21(2),316–332.

Timmann,D.,Daum,I.,2007.Cerebellarcontributionstocognitivefunctions:a progressreportaftertwodecadesofresearch.Cerebellum6(3),159–162.

Ungerleider,L.G.,Mishkin,M.,1982.Twocorticalvisualsystems.In:Ingle,D.J.I., Goodale,M.A.,Mansfield,R.J.W.(Eds.),AnalysisofVisualBehavior.TheMIT Press,Cambridge,MA,pp.549–586.

vandenHeuvel,M.P.,Sporns,O.,2011.Rich-cluborganizationofthehuman connectome.J.Neurosci.31,15775–15786.

Vogan,V.M.,Morgan,B.R.,Lee,W.,Powell,T.L.,Smith,M.L.,Taylor,M.J.,2014.The neuralcorrelatesofvisuo-spatialworkingmemoryinchildrenwithautism spectrumdisorder:effectsofcognitiveload.J.Neurodev.Disord.6(1),1–15.

Wang,E.,Liu,C.,2007.Workingmemorycapacityofadolescentswithlearning disability.Chin.Ment.HealthJ.21,587–590.

Wechsler,D.,2003.WechslerIntelligenceScaleforChildren.Psychological Corporation,SanAntonia,TX.

Woolrich,M.W.,Jbabdi,S.,Patenaude,B.,Chappell,M.,Makni,S.,Behrens,T., Smith,S.M.,2009.BayesiananalysisofneuroimagingdatainFSL.NeuroImage 45(1),S173–S186.

Worsley,K.J.,2001.Statisticalanalysisofactivationimpacts.In:Jezzard,P., Matthews,P.M.,Smith,S.M.(Eds.),FunctioningMRI:AnIntroductionto Methods.OxfordUniversityPress,Oxford,UK,InCh14.

Wright,H.H.,Fergadiotis,G.,2012.Conceptualisingandmeasuringworking memoryanditsrelationshiptoaphasia.Aphasiology26(3–4),258–278.

Zald,D.H.,Iacono,W.G.,1998.Thedevelopmentofspatialworkingmemory abilities.Dev.Neuropsychol.14(4),563–578.