Working Memory Training for Healthy Older Adults: The Role ...

WM training was shown to improve performance not only in the trained tasks (or in tasks similar to the one used in the training), but also in ... DownloadArticle DownloadPDF ReadCube EPUB XML(NLM) totalviews ViewArticleImpact SHAREON SoledadBallesteros NationalUniversityofDistanceEducation(UNED),Spain TiinaSalminen SynaptikonGmbH(NeuroNation),Germany JoséM.Reales NationalUniversityofDistanceEducation(UNED),Spain Theeditorandreviewer'saffiliationsarethelatestprovidedontheirLoopresearchprofilesandmaynotreflecttheirsituationatthetimeofreview. Abstract Introduction Method Materials Results DiscussionandConclusions AuthorContributions Funding ConflictofInterestStatement Footnotes References Checkforupdates Peoplealsolookedat ORIGINALRESEARCHarticle Front.Hum.Neurosci.,22March2017Sec.CognitiveNeuroscience https://doi.org/10.3389/fnhum.2017.00099 WorkingMemoryTrainingforHealthyOlderAdults:TheRoleofIndividualCharacteristicsinExplainingShort-andLong-TermGains ErikaBorella1*,ElenaCarbone1,MassimilianoPastore2,RossanaDeBeni1andBarbaraCarretti1* 1DepartmentofGeneralPsychology,UniversityofPadova,Padova,Italy 2DepartmentofDevelopmentalandSocialPsychology,UniversityofPadova,Padova,Italy Objective:Theaimofthepresentstudywastoexplorewhetherindividualcharacteristicssuchasage,education,vocabulary,andbaselineperformanceinaworkingmemory(WM)task—similartotheoneusedinthetraining(criteriontask)—predicttheshort-andlong-termspecificgainsandtransfereffectsofaverbalWMtrainingforolderadults. Method:FourstudiesthatadoptedtheBorellaetal.(2010)verbalWMtrainingprocedurewerefoundeligibleforouranalysisastheyincluded:healthyolderadultswhoattendedeitherthetrainingsessions(WMtraininggroup),oralternativeactivities(activecontrolgroup);thesamemeasuresforassessingspecificgains(onthecriterionWMtask),andtransfereffects(nearestonavisuo-spatialWMtask,nearonshort-termmemorytasksandfaronameasureoffluidintelligence,ameasureofprocessingspeedandtwoinhibitorymeasures);andafollow-upsession. Results:Linearmixedmodelsconfirmedtheoverallefficacyofthetraining,intheshort-termatleast,andsomemaintenanceeffects.Inthetrainedgroup,theindividualcharacteristicsconsideredwerefoundtocontribute(albeitonlymodestlyinsomecases)toexplainingtheeffectsofthetraining. Conclusions:Overall,ourfindingssuggesttheimportanceoftakingindividualcharacteristicsandindividualdifferencesintoaccountwhenexaminingWMtraininggainsinolderadults. Introduction Workingmemory(WM),i.e.,theabilitytoretainandmanipulateinformationforuseincomplexcognitivetasks,isoneofthecoremechanismsinvolvedinhigher-ordercognitiveabilities(e.g.,fluidintelligence,problem-solving,andreadingcomprehension;deRibaupierre,2001;Borellaetal.,2011).Thoughcharacterizedbyalimitedcapacity,WMisacrucialmechanismincognition.Itisalsooneofthecognitiveprocessesthatsufferaclearandlineardeclinewithaging(e.g.,Borellaetal.,2008;Mammarellaetal.,2013).WMisconsequentlyoneofthegeneralprocessestargetedbythenewgenerationofprocess-basedcognitivetraining.TheassumptionthatWMistrainableisbasedonevidenceoftheplasticityofourcognitivesystemacrossthewholelifespan(i.e.,Hertzogetal.,2008).Further,accordingtosomeWMmodels,suchasthecontinuitymodel(seeCornoldiandVecchi,2003;Cornoldi,2010),WMischaracterizedbydifferentprocessesthatdependonthetypeofcontentprocessed(verbalvs.spatial)andalsoontheinvolvementofexecutivecontrol.Therefore,byimprovingWM,itsrelatedprocessescanalsotheoreticallybeenhanced.TheCornoldiandVecchiWMmodeldistinguishedbetweena“basicstructure”(asortofpersonalbiologicalequipment),anda“usedability”determinedbythewayinwhichindividualsusetheirWM.Onthisbasis,thebenefitsoftrainingmaypresumablyconcernnotonlythebasicstructureofWM,butalsoitsusage. TheaimofWMtraininginagingisthustoimproveolderadults'informationprocessingsystem(e.g.,Zinkeetal.,2012;Bürkietal.,2014),inordertosustaintheircognitivefunctioningforanactiveaging.WMtrainingwasshowntoimproveperformancenotonlyinthetrainedtasks(orintaskssimilartotheoneusedinthetraining),butalsoinuntrainedtasks(transfereffects).Trainingchangesthewayinwhichindividualsprocessinformation,enablingthemtomakemoreflexibleuseoftheirownresources. Therecentmeta-analysisbyKarbachandVerhaeghen(2014),focusingonaging,showedthatWMtrainingforolderpeoplecouldpromotesignificantgainsbothinthetrainedtasksandinothersimilartasks(neartransfereffects).Therealsoseemedtobesomeimprovementsinuntrainedtasksthatsharedsomecognitiveprocesseswiththetaskusedinthetraining(fartransfereffects),thoughtheywereusuallysmallintermsofeffectsize(seeKarbachandVerhaeghen,2014).TherehavebeenmixedreportsonthematteroftheefficacyofWMtraininginaging(seeTable1),however,makingitnecessarytoidentifywhichfactorsareinvolvedingivingrisetotrainingbenefits.Amongthenumerousfactorstoconsider,individualcharacteristics—suchasage,generalcognitiveability,andbaselinecognitiveresources—believedtopredictthebenefitsofmemorytraining(e.g.,VerhaeghenandMarcoen,1996)mayalsohavearoleasmodulatorsofWMtrainingoutcomes(Bürkietal.,2014).Surprisingly,theirrolehasnotbeenthefocusofWMtrainingstudiesasyet. TABLE1 Table1.Peer-reviewedstudiesoftheeffectsofWMtrainingstudiesinvolvingolderadultsandthesample'scharacteristics(i.e.,age). Ageisoneofthecrucialfactorsthatmayexplainwhetherandtowhatextentindividualsmaygainmoreorlessintermsofbothspecifictraininggains(inagiventrainedtask)andtransfereffects(e.g.,vonBastianandOberauer,2014).SomeWMtrainingstudiesexaminedtheroleofageinexplainingthebenefitsoftrainingbycomparingyoungandolderadults,orconsideringolderadultsindifferentagebrackets(seealsoBorellaetal.,2014forareview).Someofthestudiesthatincludedbothyoungandolderadultsanalyzedhowperformancechangedoverthecourseofthetrainingsessions(Dahlinetal.,2008;Lietal.,2008;Richmondetal.,2011;Brehmeretal.,2012;vonBastianetal.,2013;Bürkietal.,2014).Brehmeretal.(2012),forinstance,consideredweeklyWMperformancescores,poolingparticipants'dailyperformancein7WMtrainingtasksintoasinglet-standardizedWMperformancescore.Theyfoundthatyoungadultsgainedmorethanolderadultsfromweek1to2,butthenthetwoagegroupsshowedcomparableimprovementsfromthesecondweektotheendoftraining,fromweek2to4.Bürkietal.(2014)foundthatage-relateddifferencesintheperformanceofyoungandolderadultspersistedover10trainingsessions(withgreaterimprovementsintheformer).Lietal.(2008)foundsignificantimprovementsforbothyoungandolderadultsintwotrainedspatialn-backtasks(thoughthebestperformancereachedbytheolderadultswasstillnotasgoodasthatoftheyoungeradults).Otherstudiesreportedmixedresults,however:age-relateddifferencesinfavorofyoungadultswerefoundinsomeofthetrainedtasks,whileimprovementswerecomparablebetweenthetwoagegroupsinothers. Asconcernsspecifictraininggains(i.e.,inthecriteriontasks),asshowninTable1,mixedresultswerefound:comparablebenefitsinyoungandolderadultsintasksstrictlysimilartothoseusedinthetrainingwereobtainedinfivestudies(Lietal.,2008;Richmondetal.,2011;vonBastianetal.,2013;Bürkietal.,2014;Zając-LamparskaandTrempała,2016);threestudiesshowedgreaterimprovementsinyoungthaninolderadults(Dahlinetal.,2008;Heinzeletal.,2014;Salminemetal.,2015);onestudyobtainedmixedresultswithage-relateddifferencesforsomecriteriontasksbutnotforothers(Brehmeretal.,2012);twostudiesshowedthatolderadultsreachedtheyoungadults'baselineperformancelevelontheWMcriteriontasksimmediatelyafterthetraining-i.e.,atthepost-testassessment-(Lietal.,2008;Salminemetal.,2015),andonestudyfoundthatolderadultsexceededtheyoungparticipants'baselineperformanceinthecriteriontask. Similarly,fornearaswellasfartransfereffects(whenfound),studiesfoundeithernodifferencesbetweenthetwoagegroups,orlargereffectsinyoungadultsthaninolderadults,oragainmixedresults(seeTable1).Asforanylong-termeffects,iftheywereexamined,Brehmeretal.(2012),andDahlinetal.(2008)foundacomparablemaintenanceofspecifictraininggainsbetweenyoungandolderadults.Brehmeretal.(2012)alsoidentifiedthemaintenanceofbothnearandfartransfereffectsinbothagegroups.Partiallyincontrast,Lietal.(2008)foundlargerlong-termspecifictraininggainsforyoungadultsthanforolderones,whilethelong-termneartransfereffectswerecomparablebetweenthetwoagegroups(seeTable1). Amongthestudiesfocusingonlyonolderadults(seeTable1),theonesthatfoundsignificantspecifictraininggainsandtransfereffects,alongwiththeirmaintenancewerethoseinvolvingyoung-oldparticipants(from60to74yearsold).Studiesthatincludedold-oldparticipants(from75to87yearsold),andthoseconsideringabroadagerange(i.e.,from60to82)reportedmixedfindingsintermsofspecificandtransfertraininggainsintheshortterm(seeTable1).Asforthemaintenanceeffects,somefoundlimitedtransferbenefits(Borellaetal.,2010,2013,2017;Zinkeetal.,2013),andothersfoundnone(Buschkuehletal.,2008).Inoneofthesestudies,olderagealsoemergedasanegativepredictoroftraininggainsandatleastsometransfereffects(Zinkeetal.,2013);itisworthnotingthattheeffectsizesfortransfereffectsinthiscaseweremediumtolargefortasksassessingneareffects,butonlysmallforfartransfereffects(seeTable1). Takentogether,theabovestudiesseemtosupportanegativeroleofageindeterminingthebenefitsofWMtraining. Anothervariablethatmayinfluencetheefficacyofcognitivetrainingisgeneralcognitiveability,operationalizedinsomestudieswithcrystallizedintelligence,i.e.,performanceinavocabularytest(Zinkeetal.,2013).Thiscanbeconsideredanindexofgeneralcognitiveability(e.g.,Baltes,1987),andapossiblemoderatorofWMtrainingbenefits.TheonlyWMtrainingstudythatconsideredthisvariablefound,however,thatitdidnotcontributetoexplainingWMtraininggainsandtransfereffects(Zinkeetal.,2013). Individualdifferencesincognitiveresources,suchasWMbaselineperformance,areanotherfactorthatmaypredicttrainingoutcomes(seeJaeggietal.,2014forevidenceinyoungadults),butonlytwoWMtrainingstudiesthatfocusedonolderadults(aged77to96,Zinkeetal.,2012;aged65to80andover:Zinkeetal.,2013)haveconsideredthisvariable.Onestudyfoundanegativecorrelationbetweenspecificgainsandparticipants'baselineWMperformance,i.e.,thosewithaweakerbaselineWMperformancegainedmoreinthetrainedtasksthanthosewhoseWMperformancewasbetter(Zinkeetal.,2012).Theotherconfirmedthisassociation,i.e.,thelowerthebaselineWMperformance,thelargerthespecificgainsinthetrainedtasks(Zinkeetal.,2013).InanotherstudybyBürkietal.(2014),thepre-testscoreobtainedinareasoningmeasurewasconsideredinstead,andtheresultsindicatedthattheeffectsofthetrainingwerepredictednotbythisreasoningscore,butbyagegroup. Overall,thepatternofresultsconcerningtheroleofindividualcharacteristicsandindividualdifferencesintraining-relatedperformancegainsandtransfereffectsisrathermixed.Itisalsoworthnotingthat,despitetheimportanceofanalyzingindividualfactorswhenassessingthebenefitsofWMtraining,onlythreestudieshavesofaraddressedthisissuedirectlyinrelationtoaging(Zinkeetal.,2012,2013;Bürkietal.,2014). Hencethepresentstudy,theaimofwhichwastoexaminetheroleofindividualdifferences(orindividualcharacteristics)byjointlyconsideringdifferentfactorstoidentifythosecapableofinfluencingshort-andlonger-termtraining-inducedplasticity,measuredintermsofbothspecifictraininggainsandtransfereffects.Thefactorsconsideredaspotentialmediatorsoftheefficacyoftraining(e.g.,vonBastianandOberauer,2014)weredemographiccharacteristics(i.e.,age),baselineWMperformance(Zinkeetal.,2012)andgeneralcognitiveability(i.e.,crystallizedintelligencemeasuredwithavocabularytest;Zinkeetal.,2013).Theroleofeducationwasalsoexaminedbecauseeducationisconsideredanindexofcognitiveefficiencythatcanpreservecognitivefunctioning,andbecauseitisalsousedasaproxyofcognitivereserve(e.g.,Stern,2002;Staffetal.,2004),althoughnostudieshaveexaminedwhetheritinteractswiththetrainabilityofWM. WeinvestigatedtheroleofthesevariablesbyanalyzingdataemergingfromstudiesthatadoptedthesameWMtrainingprocedure,developedbyBorellaetal.(2010).Thisisoneofthefewprocedurestohavebeenusedacrossdifferentstudies,generatingconsistentandpromisingresultsintermsofshort-andlong-termbenefits(Borellaetal.,2010,2017)—alsointasksrelatedtoeverydayabilities(Carrettietal.,2013b;Cantarellaetal.,2017)—innormalandpathologicalaging(inhealthyyoung-oldandold-old,Borellaetal.,2013,2014;inamnesticMildCognitiveImpairment,Carrettietal.,2013a).TheeffectivenessofthetraininghasbeenattributedtothefactthatitinvolvesparticipantspracticingwithacomplexWMspantask,combininganadaptiveprocedurewithasystematicvariationofthedemandsofthetask,sothatitremainsconstantlynovelandchallenging,keepingparticipantsinterestedandmotivatedduringtheproposedactivities.Accordingtotheauthors,thetrainingalsoengagesnumerousdifferentprocessesthatincludeencoding,maintainingandinhibitinginformation,simultaneouslymanagingtwotasks,sustainingandshiftingattention.Together,theseaspectsarebelievedtopromotelearningandparticularlytoenablethetrainingtofavortransfereffects(seeBorellaetal.,2010).Todate,sevenstudieshaveadoptedthisprocedure(seeTable2forasummary),andfourwereselectedforthepresentanalysisbecause:(i)thesameverbalprocedurewasadopted;(ii)thesamemeasureswereusedtoassesstraininggainsandtransfereffects;(iii)afollow-upsessionwasincluded;and(iv)asampleofhealthyolderadultswasconsidered(seeTables2,3). TABLE2 Table2.SummaryofthecharacteristicsandresultsofthesevenstudiesthatemployedtheBorellaetal.(2010)trainingprogram. TABLE3 Table3.SummaryofthesamplesizesandoutcomemeasuresconsideredinthefourstudiesconductedusingtheBorellaetal.(2010)trainingprogramandanalyzedinthepresentstudy. Specifictraininggainsandtransfereffectswerecategorizedalongaconceptually-basedcontinuumofnearesttofartransfertasks(i.e.,Noacketal.,2009).ThecomplexWMtask(theCategorizationWorkingMemorySpantask,CWMS)wasusedtoassessspecifictraininggainsbecauseitissimilartothetaskadministeredtoparticipantsduringthetrainingsessions.AnothercomplexWMtaskmeasuringthesamenarrowability(WM),andalsoinvolvingactiveprocesses(seeCornoldiandVecchi,2003),butwithadifferenttypeofmaterial(visuo-spatial,theDotMatrixTask)wasadministeredtoassesswhatwedescribehereasnearesttransfereffects.Measuresofthesamebroadability(memory),butwithdifferentdemandsfromthoseoftheothercomplexWMtasks(theForwardandBackwardDigitSpantests;seemeta-analysesbyBoppandVerhaeghen,2005)wereusedtoassessneartransfereffects.Finally,tasksassessingfluidintelligence(theCattelltest),processingspeed(thePatternComparisontask),andinhibitorymechanisms(StroopColortestandintrusionerrorsintheCWMS),i.e.,mechanismsdifferingfromWMbutknowntocorrelatewithWMandtohelpexplainingtheage-relateddeclineinWM(e.g.,deRibaupierreandLecerf,2006),wereusedtomeasurefartransfereffects. Linearmixedeffects(LME)modelswereusedtoexaminetheroleofindividualcharacteristics(demographicvariables)andindividualdifferencesinpredictingimprovementsinthemeasuresusedtoassesstheeffectsofthetraining(intermsoftraininggainsandtransfereffects).Thesemodelsaffordamorerobustanalyticalapproachforaddressingproblemsassociatedwithhierarchicalandcorrelateddatathanthetraditionalanalysesgenerallyconductedintrainingstudies(e.g.,ANOVA,t-test).Inparticular,LMEmodelsallowforamoreflexibleapproachindealingwithindividualchangesovertimewhenrepeatedmeasuresareconsidered(e.g.,GueorguievaandKrystal,2004;Wainwrightetal.,2007;Baayenetal.,2008). Ingeneral,weexpectedtoconfirmthebeneficialeffectoftheWMtrainingintermsofshort-andlong-termgainsinthecriteriontask,andatleastshort-termtransfereffectsforallthemeasuresconsidered.Theadvantageofperformingsuchananalysisonallfourstudiessharingthesameprocedurelayinenablingustoestablishthestrengthoftheeffects(i.e.,effectsizes)onalargersample. Concerningthemainobjectiveofthestudy,weusedLMEmodelstoanalyzeparticipants'individualcharacteristicsanddifferencesvis-à-vistheshort-andlong-termeffectsoftheirtraining.Analyzingthesepotentialpredictorswillenableustotestthetwoproposedtheoreticalexplanationsforindividualdifferencesintraining-relatedperformancegains,i.e.,acompensationoramagnificationeffectofprocess-basedtrainingoncognitioninolderadults(seeforexampleTitzandKarbach,2014;seealsoLövdénetal.,2012).Ifthereisamagnificationeffect,thenindividualswhoalreadyperformwellwillbenefitthemostfromtheWMtraining.Inotherwords,high-performingparticipantsmayhavemoreefficientcognitiveresourcesandthereforebeinabetterpositiontolearnandimplementnewabilities.TheWMtrainingshouldthereforeresultinamagnificationofage-related(inolderadults)andindividualdifferences;baselinecognitiveperformanceshouldalsobepositivelyassociatedwithtraining-relatedgainsandtransfereffects.Ifthereisacompensationeffect,ontheotherhand,thenhigh-performingindividualswillbenefitlessfromthetrainingbecausetheyarealreadyfunctioningattheiroptimallevel,andthushavelessroomforimprovement.Inthiscase,age-relatedandindividualdifferencesshouldbereducedafterthetraining,andbaselinecognitiveperformanceshouldbenegativelyassociatedwithtraining-inducedgains. Themagnificationandcompensationeffectswouldthusleadinoppositedirections.Amongtheolderadults,theyoungerparticipantswithagoodcognitivestatus,asrepresentedbyameasureofcrystallizedintelligence(vocabulary),agoodWM(revealedbytheirbaselineCWMSperformance),andagoodeducationmightprofitmorefromanadaptivetrainingonacomplexaspectofcognition—WM—becausearelativelyhighleveloffunctioningisrequiredtoactivelyengageinandbenefitfromtheactivitiesproposedinthetraining(BissigandLustig,2007;Lustigetal.,2009),whichwould,inturns,magnifytheirabilities.Ontheotherhand,olderparticipantswithaworsecognitivestatusmightbenefitmorefromtheWMtraining(Zinkeetal.,2012,2013)becauseitcouldcounteractthesuboptimaluseofresourcestypicalofagingbypromptingamoreflexibleuseoftheseresources,morereliantoncontrolledthanonautomaticprocesses,thatwouldre-activatetheolderparticipants'potential,havingacompensatoryeffect. Thereisalsothepossibility,asemergesfromtheresultsobtainedbyZinkeetal.(2013),thatthefactorsthoughttopredicttraining-relatedgainsmightdependonthemeasuresconsidered,becausethesefactorsmayalsotakeeffectindependently.Infact,theindividualcharacteristicsexaminedmayexplainthetraininggainsdifferently,asthetransfertasksvaryinseveralaspects—notonlyintermsoftheirrelationshipwithWM,butalsointermsoftheprocessesinvolved,suchasthetypeofcontrol(passiveasintheshort-termmemorytasksvs.activeasinthereasoningtask),orthemoreorlessstronginvolvementoffluidabilities(strongerinreasoningandinprocessingspeedthaninshort-termmemorytasks)and/orthoserelatedtoknowledge. Method Table1liststhecharacteristicsandthemainresultsofthesevenstudiesthatusedtheverbalWMtrainingproceduredevelopedbyBorellaetal.(2010).AsmentionedintheIntroduction,threestudies(inthelastrowsofTable2)werenotconsideredbecause:oneinvolvedasampleofolderadultswithmildcognitiveimpairment(Carrettietal.,2013a);oneusedavisuo-spatialversionofthetrainingprogram(Borellaetal.,2014);andonedidnotincludeafollow-upassessment(Cantarellaetal.,2017). Thefourstudiesconsideredeligibleforthepresentanalysishadincommon:(i)thesameverbalprocedure;(ii)thesamemeasuresforassessingtraininggainsandtransfereffects(seeTable3);and(iii)afollow-upassessment. Participants Allthefourstudiesconsideredhereincludedasampleofhealthyolderadults(allnativeItalianspeakers)recruitedfromtheUniversityoftheThirdAge,atsocialclubsinnorth-easternItaly,orbywordofmouth,whoallvolunteeredforthestudy. Theyweretold,eitherindividually(Borellaetal.,2010,2017)orataplenarysession(Borellaetal.,2013;Carrettietal.,2013b),thattheywouldbeinvolvedinoneoftwodifferentprogramseachconsistingoffiveindividualsessions,plusafinaloneatalaterdate(follow-up).Theywerealsotoldthattheactivitiesproposedinoneprogramwouldconcerntheircognitivefunctioning(i.e.,practicingwithmemorytasks),whileintheotheronetheywouldbeaskedtoreflectonaspectsofmemory(e.g.,autobiographicalrecall)andcompletesomequestionnaires. Dependingonthestudy,participantshadtomeetthefollowinginclusioncriteria:(i)goodphysicalandmentalhealth,assessedbymeansofaquestionnaireandasemi-structuredinterview,respectively(asinBorellaetal.,2010;Carrettietal.,2013b);(ii)noneoftheexclusioncriteriaproposedbyCrooketal.(1986)asinBorellaetal.(2010,2013);(iii)aMini-MentalStateScore(Folsteinetal.,1975)higherthan27(asinBorellaetal.,2013);(iv)amaximumscoreontheItalianChecklistforMultidimensionalAssessments(SVAMA;Gallinaetal.,2006),i.e.,nosignsofincipientdementia(asinBorellaetal.,2017).Inallfourstudies,participantswererandomlyassignedtoeitherthetrainedgrouportheactivecontrolgroup. Overall,148participantswereinvolvedinthefourstudiesconsidered,with73formingthetrainedgroups,and75theactivecontrolgroups.Thepooledtrainedandcontrolgroupswerecomparableintermsofage(agerange:61–87;trainedgroup:M=71.63,SD=5.53;controlgroup:M=71.61SD=5.67),F(1,146)<1,yearsofformaleducation(from8to24years;trainedgroup:M=9.42SD=4.54;controlgroup:M=9.97SD=4.72),F(1,146)<1,andvocabularyscoreintheWechslerAdultIntelligenceScale—Revised(WAIS–R;Wechsler,1981;max70;trainedgroup:M=49.21SD=10.89;controlgroup:M=47.04SD=11.87),F(1,146)=1.33,p=0.25. Ascommonoutcomemeasuresusedtoassesstransfereffectsvariedwithinthefourstudiesconsidered,pooledtrainedandcontrolgroupswerecomparedwithrespecttodemographiccharacteristicsandvocabularyscore.Thepooledtrainedandcontrolgroupswerenotstatisticallydifferentintermsofage,yearsofformaleducation,andvocabularyscore1. Materials CriterionTask CategorizationWorkingMemorySpan(CWMS)task(DeBenietal.,2008) Thetaskconsistedof10setsofwordlists,eachincluding20listsofwords(dividedintogroupscontainingfrom2to6lists).Participantslistenedtoasetofwordlistsaudio-recordedatarateof1wordpersecondandtheyhadtotapwiththeirhandonthetablewheneverananimalnounwasheard(processingphase).Theintervalbetweenwordlistswas2s.Attheendofaset,participantsrecalledthelastwordoneachlist(maintenancephase)—i.e.,theyneededtorememberfrom2to6wordsaltogether,dependingonthedifficultyoftheset. ThetotalnumberofwordsrecalledwasusedasthemeasureofWMperformance(maximum20). NearestTransferEffects Visuo-SpatialWMTask DotMatrixtask(adaptedfromMiyakeetal.,2001) Inthistaskparticipantshadtocheckamatrixequationconsistingofanadditionorasubtractionpresentedaslinesdrawnona3×3matrix,andtomemorizesequencesofdotspresentedona5×5grid.Theyweregivenamaximumof4.5stocheckeachequationandsay“True”or“False.”Immediatelyaftertheygavetheiranswer,theywereshowna5×5gridcontainingadotinoneofthesquaresfor3s.Afterseeingsetsoftwotosixpairsofequationsandgrids,theyhadtoindicatethepositionsofthedotsonablank5×5grid.Therewasonepracticetrialwithtwoequations,eachwithonedot.Thenumberofdotlocationstorecallincreasedfromtwotosix.Atotalof28equationsand28matriceswerepresented.Thetotalnumberofdotpositionscorrectlyrecalledwasconsideredasthedependentvariable(maximumscore14). NearTransferEffects Short-TermMemoryTasks ForwardandBackwardDigitSpantasks(DeBenietal.,2008) Participantshadtorepeatseriesofdigitsinthesame(forward)orreverse(backward)order.Eachlevel(from3to9digitsfortheforwardtask,from2to8digitsforthebackwardtask)containedtwoseriesofdigits.Aftertwoconsecutiverecallerrors,thetaskwasdiscontinued.Onepointwasawardedforeachcorrectlyrecalledseries.Thefinalscorecorrespondedtothetotalnumberofseriesrecalledcorrectly(maximumscoreof14forbothtasks). FarTransferEffects FluidIntelligence CultureFairtest(Cattelltest;CattellandCattell,1963) Thistaskconsistedof4subtests(tobecompletedin2.5–4min.,dependingonthesubtest)inwhichparticipantswereaskedto:(1)choosefromamongsixdifferentoptionswhichonescorrectlycompletedaseriesoffigures;(2)identifyfiguresorshapesthatdidnotbelonginasequence;(3)chooseitemsthatcorrectlycompletedmatricesofabstractfigures;(4)assessrelationshipsbetweensetsofitems.Thedependentvariablewasthenumberofcorrectanswersacrossthefoursubtests(maximumscore50). ProcessingSpeed Patterncomparisontask(adpatedfromSalthouseandBabcock,1991) Inthistask,participantshadtodecidewhetherarrangementsoflinesegmentswereidenticalornot.Theitemstobecomparedweresetoutontwopageseachcontaining30items.ResponsesconsistedofwritingS(forSi[Yes],foridenticalitems)orN(forNo,fordifferentitems)onthelinebetweenthetwoitemsineachpair.Theexperimenterusedastopwatchtorecordthetimetakentocompleteeachpage.Threepracticetrialswererunbeforetheexperimentstarted.Thedependentvariablewasthetotaltimetakentocompletethetask. Inhibition StroopColortask(adaptedfromTrenerryetal.,1989) Inthistaskparticipantswereshownsixcards.Thefirsttwocontainednamesofcolorsprintedinanincongruentinkcolor(Incongruentcondition);thethirdandfourthcontainednamesofcolorsprintedinacongruousinkcolor(Congruentcondition);andthelasttwocontainedcolorpatches(Controlcondition).Participantshadtonametheinkcolorofeachstimulusandwereaskedtoprocessthestimuliasquicklyandaccuratelyaspossible.Theexperimenterrecordedresponselatenciesforallconditionsbyusingastopwatchtotimetheintervalbetweennamingthefirstandlaststimuli—astypicallydoneinotherstudiesusingthepaperversion(e.g.,WestandAlain,2000;VanderElstetal.,2006;Troyeretal.,2006)—andnotedtherespondents'accuracybyhandonapreparedform.Thedependentvariable—inordertocontrolforindividualdifferencesatthebaseline(e.g.,Borellaetal.,2009)—wastheinterferenceindexcomputedintermsofrelativedifferencebetweenthetimetakentocompletethetaskintheincongruentandcontrolconditions,thatis[(incongruentcondition−controlcondition)/controlcondition].Ahigherscorethusimpliedagreaterdifficultyincontrollingtheprepotentresponseintheincongruentcondition. IntrusionerrorsintheCWMS-CWMSintrusionerrors-(DeBenietal.,2008) ThetotalintrusionerrorsmadeintheCWMS,i.e.,wordsthatwerenotactuallythelastinofeachstringofwordspresented,werealsoconsideredasameasureofinhibition,representingaparticipant'sabilitytoinhibitnolongerrelevantinformation(Borellaetal.,2007). Foreachtask,twoparallelversionsweredevisedandadministeredinacounterbalancedorderacrossthetestingsessions. Procedure Participantsattendedsixindividualsessions:thefirstandfifthwerethepre-testandthepost-testsessions,andthesixthwasforthefollow-up(held6–8monthslater).Fortheotherthreesessions,thetrainedparticipantsattendedthetrainingprogram,whiletheactivecontrolswereinvolvedinalternativeactivities.Forbothgroups,allactivitieswerecompletedwithina2-weektimeframe,withafixed2-daybreakbetweensessions.Thedurationofthesessionsandtheamountofinteractionwiththeexperimenterweremuchthesameforthetwogroups. ThethreesessionsofWMtraining(sessions2-3-4)lastedabout30–40min.Participantswerepresentedwithlistsofwordsaudio-recordedandorganizedinthesamewayasfortheCWMStask,andaskedtorecalltargetwords,andtotapwiththeirhandonthetablewhentheyheardananimalnoun.ThemaintenancedemandoftheCWMStaskwasmanipulatedbyincreasingthenumberofwordsthatsuccessfulparticipantswereaskedtorecall,andbypresentingthelowestmemoryloadtoparticipantswhowereunsuccessful(session2).Thedemandsofthetaskalsovariedand,dependingonthesession,theycouldinvolvehavingtorecall:(i)thelastorfirstwordineachlist;(ii)wordsprecededbyabeepsound.Theprocessingdemand(tappingonthetablewhenananimalnounoccurred)wasalsomanipulatedbyvaryingthefrequencyoftheanimalwordsinthelists(session3).Participantsintheactivecontrolgroupwereaskedtocompletequestionnairesonmemory(session1:AutobiographicalMemoryquestionnaire;session2:MemorySensitivityquestionnaire—DeBenietal.,2008),andonpsychologicalwell-being(DeBenietal.,2008;seeBorellaetal.,2010formoredetailsofthetrainingprogramandtheactivecontrolgroup'sactivities). TheprocedurewascompletedinaccordancewiththeDeclarationofHelsinki(2008). Results LinearMixedModels Linearmixedeffects(LME)modelswereusedtoanalyzethedatabecause,unlikethemoreclassicalandfrequently-usedmethods,theyenableestimatestobeadjustedforrepeatsampling(whenmorethanoneobservationarisesfromthesameindividual),andforsamplingimbalance(whensomeindividualsaresampledmorethanothers),andbecausetheyallowforvariationamongindividualswithinthedata(McElreath,2016).AdoptingtheBayesianapproachwhenestimatingparametersenabledustoexploitalltheadvantagesofLMEmodeling,focusdirectlyontheprobabilityofaneffect,giventheobserveddata(posteriorprobability),andcomputetheevidenceofourresults. AnalyticalPlan Foreachoftheeightmeasuresofinterest(i.e.,thetotalnumberofwordsrecalledintheCWMS,thetotalnumberofdotsrecalledintheDotMatrixtask,thetotalnumberofseriescorrectlyrecalledintheForwardandintheBackwardSpantasks,thetotalnumberofcorrectlyanswereditemsintheCattelltest,thetotaltimetakentocompletethetaskinthePatternComparisontest,theinterferenceindexintheStroopColortask,andthetotalnumberofintrusionerrorsintheCWMS),wetestedseveralmixedeffectsmodelsincludingallcombinationsofpredictors,i.e.,group(trainedvs.control),age,education,vocabulary,baselineperformanceintheverbalWMtask(theCWMS),andsubjectsasrandomeffects.Moreprecisely,westartedfromthenullmodel(i.e.,themodelwithnopredictors),consideringonlythelongitudinaleffect(pre-test,post-test,follow-up)andsubsequentlyintroducedallthepredictorsandtheinteractionsofallthepredictorswiththesessions. Fordataanalysis,weproceededasfollows: (i)Afirstgraphicalinspectionoftheunivariateandbivariatedistributionsofalltheoutcomevariablesconsidered,thentheirsummarizingwithdescriptivestatistics(seeTable4).Thisstepwasdoneto:(i)checkthedatadistributionsandidentifyanyerrors/anomalies(e.g.,wronglabelcodesorpotentialoutliers);andto(ii)ascertainwhichmodeltoadoptforabetterfitofourdataandchooseappropriatepriorsfortheparameters. (ii)Modelfittingandparameterestimation:eachmodelwasfitted(separatelyforeachmeasureofinterest)usingtheBayesianMCMCestimationmethodimplementedintheSTANprobabilisticprogramminglanguage(StanDevelopmentTeam,2015)withtheRpackagesrstanarm(GabryandGoodrich,2016)andbrms(Buerkner,2016). Fortheregressionparameters(ß)weusednormalpriors(M=0andSD=10),andforstandarddeviationparametersweusedhalf-Studentt(df=3,M=0,SD=10);convergenceswereassessedbyexaminingthepotentialscalereductionfactor(PSRF;GelmanandRubin,1992). (iii)Comparisonbetweenthemodels(separatelyforeachmeasureofinterest)toidentifythebestone.WeconsideredtheWidelyApplicableInformationCriterion(WAIC;Watanabe,2010),wherelowervaluesindicateabetterfit,andtheAkaikeWeight,i.e.,anestimateofthelikelihoodofamodelmakingthebestpredictiononnewdata,conditionalonthesetofmodelsconsidered(Burnhametal.,2011;McElreath,2016). (iv)Analysisofthebestmodelusingposteriordistributionsofparameters.Parameterestimatesweresummarizedbyusingposteriormeansand95%CredibilityIntervals(CI;Kruschke,2011). TABLE4 Table4.Descriptivestatisticsfortheoutcomemeasuresbygroup(trainedvs.controls)andbyassessmentsession(pre-test,post-test,follow-up). DataInspection Table3containsthedescriptivestatisticsforeachofthemeasuresofinterestbygroup(trainedandcontrol),andbyassessmentsession(pre-test,post-test,andfollow-up). ModelFittingandParameterEstimation Inall,1,026models,58fortheCWMS,and121foreachoftheothermeasuresofinterest,werefitted.Thefitindicesofthe5bestmodelsforeachmeasureofinterestaregiveninTable5. TABLE5 Table5.Fitindicesofthefivebestmodelsforeachvariableofinterest. ModelComparisonandBestModelAnalysis Foreachoutcomemeasurewecomparedthefitindicesofthe5bestmodels.Then,focusingonthebestone(theonewiththelowestWAIC),weproceededwithagraphicalinspectionoftheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interactiontoassesstheeffectivenessofthetraining(seeFigure1). FIGURE1 Figure1.Group(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interactionfromthebestmodelforeachmeasureofinterest.CategorizationWorkingMemorySpantask(A),DotMatrixtask(B),Forward(C)andBackward(D)digitspantasks,Cattelltest(E),PatternComparisontask(F),StroopColortask(G)andintrusionerrorsintheCategorizationWorkingMemorySpantask(H).CWMS,CategorizationWorkingMemorySpanTask;CWMSintrusions,intrusionerrorsintheCategorizationWorkingMemorySpanTask;RTs,ResponseTimes.Segmentsrepresentthe95%credibilityintervals. Togainabetterunderstandingoftheextentofthebenefitsoftrainingonthetrainedgroup'sperformance,theeffectsizewascomputedonthedifferencesbetweenthetwogroups(trainedandcontrol)atpre-test,post-test,andfollow-up(seeTable6).Inaddition,toascertainthedimensionoftheimmediate(pre-vs.post-test)andlong-term(pre-testvs.follow-up)gainsobtainedinthetrainedgroup,Cohen'sdwascomputedusingthefollowingformula:{(Post-testorfollow-upforthetrainedgroup−Pre-testforthetrainedgroup)−(Post-testorfollow-upforthecontrols−Pre-testforthecontrols)}/(PooledSDofthedifference;seeWeiszandHawley,2001).Thisenabledustoadjustthegainsshownbythetrainedgroupinrelationtothegainobtainedbytheactivecontrolgroup(seeTable6). TABLE6 Table6.Effectsizes. Then,forthetrainedgroup,weconductedagraphicalinspection(separatelyforeachsession)ofthevaluesfittedforthesignificanteffectsofthebestmodel(seeFigure2),supportedbytheevidenceratioforthehypothesisinvolvingtheßcoefficientsconsidered(seeTable7). FIGURE2 Figure2.Fittedvaluesfromthebestmodelforthetrainedgroupaloneforeachvariableofinterest.CategorizationWorkingMemorySpantask(A),DotMatrixtask(B),Forward(C)andBackward(D)digitspantasks,Cattelltest(E),PatternComparisontask(F),StroopColortask(G)andintrusionerrorsintheCategorizationWorkingMemorySpantask(H).CWMS,CategorizationWorkingMemorySpanTask;CWMSintrusions,intrusionerrorsintheCategorizationWorkingMemorySpanTask;RTs,ResponseTimes;cwmspre,CWMSbaselineperformancelevel. TABLE7 Table7.Evidenceratioofthedifferencesbetweenpre-andpost-testslopes,andbetweenpost-testandfollow-upslopes,foreachvariableofinterest,byfitindex. Theevidenceratiorepresentstheevidenceofthetargetedhypothesis(e.g.,ß>0)withrespecttotheoppositehypothesis(e.g.,ß<0).Iftheevidenceratioequals1,thenthetwohypothesesareequallyplausible.Anevidenceratiolargerthan1indicatesthatthetargethypothesisismoreplausiblethantheoppositeone,andanevidenceratioof<1meansthattheoppositehypothesisismoreplausiblethanthetargetedone.Inthepresentstudy,theevidenceratiowasusedtoassessthedifferencesbetweenthepre-andpost-testslopes,andbetweenthepost-testandfollow-upslopes. CriterionTask CWMS FortheCWMSweconsideredatotalof58models2.ThebestmodelwastheonewiththeSessionXGroupXVocabularyinteraction,withaprobabilityof59.1%(seeTable5). Figure1AshowstheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interactionfromthebestmodel:thetrainedgroupperformedbetteratpost-testthanatpre-test,andmaintaineditsbetterperformancefrompost-testtofollow-up.Theeffectsizesforgroupdifferenceswerelargeatbothpost-testandfollow-up(seeTable6).Nodifferenceswerefoundfortheactivecontrolgroup.Thetrainedgroupoutperformedthecontrolgroupatpost-testandfollow-up(seeFigure1A).Theneteffectsizeindexforthetrainedgroup,adjustedonthecontrolgroup'sperformance,waslargeforimmediategains(pre-vs.post-test)andforlong-termsgains(seeTable6). Figure2Ashowsthefittedvaluesfromthebestmodelforthetrainedgroupalone,asafunctionofvocabularyscoreatthethreeassessmentsessions,withtherelativeestimatedlineartrend.Theregressionslopedecreasedfromsession0(pre-test)tosession1(post-test),withanevidenceratioof89.91(seeTable7),anditbecameflatatfollow-up,asshownbytheevidenceratio(seeTable7).TheseresultsindicatethatparticipantswithlowvocabularyscoresweretheoneswhoshowedanimprovementinperformanceinthecriterionWMtaskfrompre-testtopost-test,andmaintainedthisgainatfollow-up. NearestTransferEffects Visuo-SpatialWorkingMemory DotMatrixtask FortheDotMatrixtaskweconsideredatotalof121models.ThebestmodelwasgivenbytheSessionXAgeXGroupXVocabularyinteractionwithaprobabilityof84.7%.Thismodelwasabout17timesmoreevidentthanthenextone,whichachievedaprobabilityofaround5.1%(seeTable5). AsfortheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interaction,thetwogroupsdidnotdifferatpre-test.Thetrainedgroupperformedbetteratpost-testthanatpre-test,butthisgainwasnotmaintainedatfollow-up,whenperformancewasnotasgoodasatpost-test(seeFigure1B).Effectsizesforgroupdifferenceswerelargeatpost-testandbecamesmallatfollow-up(seeTable6).Nodifferenceswerefoundforthecontrolgroup.Thetrainedgrouponlyoutperformedthecontrolgroupatpost-test(seeFigure1B).Theneteffectsizeforthetrainedgroup,adjustedonthecontrolgroup'sperformance,waslargefortheimmediategains(pre-vs.post-test),butbecamesmallforthelong-termsgains(seeTable6). Figure2Bshowsthevaluesfittedfromthebestmodelforthetrainedgroupalone,asafunctionofage,andofvocabularyscoreatthethreeassessmentsessions,withthecorrespondingestimatedlineartrend.Forage,theregressionslopesuggestsachange—withanevidenceratioof32.06(seeTable7)—fromsession0(pre-test)tosession1(post-test),andaslightdeteriorationfrompost-testtofollow-up(evidenceratioof7.05;seeTable7):itwastheyoungerparticipantswhoseperformanceimprovedfrompre-testtopost-test,andthendroppedbackatfollow-uptomuchthesameastheirpre-testperformance. Asforvocabulary,theregressionsloperosefromsession0(pre-test)tosession1(post-test),withanevidenceratioof5.02(seeTable7),andclearlydroppedagain,becomingflatatfollow-up(seeevidenceratio):itwastheparticipantswithhighvocabularyscoreswhoseperformanceimprovedintermsofthenumberofdotscorrectlyrecalledfrompre-testtopost-test,butnotatfollow-upwhentheirperformanceclearlydeteriorated. NearTransferEffects Short-TermMemory ForwardDigitSpantask Atotalof121models,oneofwhichdidnotconverge,wereconsideredfortheForwardDigitSpantask.ThebestmodelwasrepresentedbytheSessionXEducationXGroupXVocabularyinteractionwithaprobabilityof27.3%.Thismodeldidnotseemmuchmoreevidentthanthenexttwo,forwhichtheprobabilitywasaround23.5and17.7%,respectively(seeTable4). AsfortheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interaction,thetwogroupsdidnotdifferatpre-test.Thetrainedgroupperformedbetteratpost-testthanatpre-test,butthisgainwasnotmaintainedatfollow-up,whenperformancewasworsethanatpost-test(seeFigure1C).Theeffectsizesforgroupdifferenceswerelargeatpost-testandbecamesmallatfollow-up(seeTable6).Nodifferenceswereseenforthecontrolgroup.Thetrainedgrouponlyoutperformedthecontrolgroupatpost-test(seeFigure1C).Theneteffectsizeforthetrainedgroup,adjustedonthecontrolgroup'sperformance,waslargeforimmediategains(pre-vs.post-test),butbecamesmallforlong-termsgains(seeTable6). Figure2Cshowsthevaluesfittedfromthebestmodelforthetrainedgroupaloneasafunctionofeducation,vocabulary,andpre-testperformanceintheWMcriteriontask,atthethreeassessmentsessions,withthecorrespondingestimatedlineartrend.Aminimalchangefrompre-testtopost-testemergedforallthevariables:theevidenceratiobetweensession0(pre-test)andsession1(post-test)was7.39foreducation,199forvocabulary,and0.63forpre-testperformanceintheWMcriteriontask(seeTable7).Atfollow-up,therewasachangefrompost-test(seeevidenceratio),withperformancedroppingbacktopre-testlevels.Inparticular,itwastheparticipantswhohadalimitededucation,lowvocabularyscores,andapoorpre-testperformanceintheWMcriteriontaskwhoexperiencedaslightimprovementintheirperformance,butonlyatpost-test. Backwarddigitspantask FortheBackwardDigitSpantaskweconsideredatotalof121models,oneofwhichdidnotconverge.ThebestmodelemergedfortheSessionXAgeXEducationXGroupinteraction,withaprobabilityof48.3%.Thismodelwasabouttwotimesmoreevidentthanthenextone,forwhichtheprobabilitywasaround29%(seeTable5). FortheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interaction,thetwogroupsdidnotdifferatpre-test.Thetrainedgroupperformedbetteratpost-testthanatpre-test,andthenitsperformancedeterioratedfrompost-testtofollow-up(seeFigure1D).Theeffectsizesforgroupdifferenceswerelargeatpost-testandbecamesmallatfollow-up(seeTable6).Nodifferenceswereidentifiedinthecontrolgroup.Thetrainedgrouponlyoutperformedthecontrolgroupatpost-test(seeFigure1D).Theneteffectsizeforthetrainedgroup,adjustedonthecontrolgroup'sperformance,waslargeforimmediategains(pre-vs.post-test),butbecamesmallforlong-termsgains(seeTable6). Figure2Dshowsthevaluesfittedfromthebestmodelforthetrainedgroupaloneasafunctionofeducation,andofageatthethreeassessmentsessions,withthecorrespondingestimatedlineartrend.Foreducation,therewasnochangeintheslope.Forage,theyoungertheparticipants,thegreatertheimprovementinperformancefrompre-testtopost-test,withanevidenceratioof77.43(seeTable7).Frompost-testtofollow-up,therewasadeclineintheslope,withanevidenceratioof665.67(seeTable7),meaningthatperformancereturnedtothelevelsseenatpre-test. FarTransferEffects FluidIntelligence Cattelltest FortheCattelltestweconsideredatotalof121models.ThebestmodelwasobtainedwiththeSessionXAgeXCWMSbaselineXGroupinteraction,withaprobabilityof43.1%.Thismodelwasaboutthreetimesmoreevidentthanthenext,forwhichtheprobabilitywasaround16.6%(seeTable5). FortheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interaction,thetwogroupsdidnotdifferatpre-test.Thetrainedgroupperformedbetteratpost-testthanatpre-test,andthenitsperformancedeclinedfrompost-testtofollow-up(seeFigure1E).Theeffectsizesforgroupdifferencesweremediumatpost-testandbecamesmallatfollow-up(seeTable6).Nodifferencescametolightforthecontrolgroup.Thetrainedgroupoutperformedthecontrolgroupatbothpost-testandfollow-up(seeFigure1E).Theneteffectsizeforthetrainedgroup,adjustedonthecontrolgroup'sperformance,wasmediumforbothimmediategains(pre-vs.post-test)andlong-termgains(seeTable6). Figure2Eshowsthevaluesfittedfromthebestmodelforthetrainedgroupaloneasafunctionofage,andofpre-testperformanceintheWMcriteriontaskatthethreeassessmentsessions,withthecorrespondingestimatedlineartrend.Forage,itwastheyoungerparticipantswhoseperformancewasbetteratpost-testthanatpre-test,withanevidenceratioof443.44(seeTable7),andtheyalsomaintainedtheirbetterlevelofperformanceatfollow-up,withanevidenceratioof41.11(seeTable7). Therewasnodifferencesintheslopesforpre-testperformanceintheWMcriteriontask,asconfirmedbytheevidenceratio.ItisworthnotingthattherewasaslightlyhighervariabilityintheCattelltestatpost-testforparticipantswithlowscoresforpre-testperformanceintheWMcriteriontask. ProcessingSpeed PatternComparisontask Weconsideredatotalof121modelsforthePatternComparisontask.ThebestmodelwastheonewiththeSessionXCWMSbaselineXGroupXVocabularyinteraction,reachingaprobabilityof38.5%.Thismodelwasabouttwotimesmoreevidentthanthenext,whichreachedaprobabilityofaround20.5%(seeTable5). FortheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interaction,thetwogroupsdidnotdifferatpre-test.Thetrainedgroupperformedbetter(takinglesstimetocompletethetask)atpost-testthanatpre-test,andmaintainedthisimprovementfrompost-testtofollow-up(seeFigure1F).Theeffectsizesforgroupdifferenceswerelargeatpost-testandbecamemediumatfollow-up(seeTable6).Nodifferenceswerefoundforthecontrolgroup.Thetrainedgroupoutperformedthecontrolgroupatpost-testand,toacertainextentat,atfollow-uptoo(seeFigure1F).Theneteffectsizeforthetrainedgroup,adjustedonthecontrolgroup'sperformance,wasmediumforbothimmediategains(pre-vs.post-test)andlong-termgains(seeTable6). Figure2Fshowsthevaluesfittedfromthebestmodelforthetrainedgroupalone,asafunctionofvocabulary,andpre-testperformanceintheWMcriteriontaskatthethreeassessmentsessions,withthecorrespondingestimatedlineartrend.Therewasaveryweakeffectforvocabulary—theevidenceratiowas13.13(seeTable7)—anditwastheparticipantswhohadahigherpre-testvocabularyscorewhoimprovedintheprocessingspeedmeasurefrompre-topost-test.NodifferenceswerefoundintheslopewhentheWMcriteriontaskatpre-testwasconsidered;ahighindividualvariabilityalsoemerged(seeFigure2). Inhibition Stroopcolortask Weconsideredatotalof121modelsfortheStroopcolorindexonresponsetimes(RTs).ThebestmodelcoincidedwiththeSessionXAgeXGroupinteraction,withaprobabilityof22%,whichisratherlow,thoughthismodelwasabout4timesmoreevidentthanthenextone,whichreachedaprobabilityofaround6.3%(seeTable5).AsshowninFigure1G,noGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interactionwasfound.Thisisinlinewiththenulleffectsizeidentifiedonthedifferencesbetweenthegroups,bothimmediatelyafterthetrainingandatfollow-up(seeTable6),andalsoontheeffectsizecomputedfortheimmediateandlong-termtraininggainsobtainedbythetrainedgroup(seeTable6). Thebestmodelwasonlyexaminedforthetrainedgroup.Figure2Gshowsthevaluesfittedfromthebestmodelasafunctionofageatthethreeassessmentsessions,withthecorrespondingestimatedlineartrend.Theresultssuggestthatyoungerparticipantsweremoresensitivetointerference,andthatitdecreasedfrompre-testtopost-test,withanevidenceratioof665.67(seeTable7),butthenroseagaintothepre-testlevelatfollow-up. CWMSintrusions Weconsideredatotalof121modelsfortheCWMSintrusionerrors.ThebestmodelwastheonewiththeSessionXAgeXCWMSbaselineXGroupinteraction,withaprobabilityof93%.Thismodelshowedahigherevidencethantheothers,sincetheassociatedprobabilitywas18timeshigherthanthatofthenextmodel(seeTable5). ThedifferenceswerenotverylargefortheGroup(trainedvs.control)XSession(pre-testvs.post-testvs.follow-up)interaction.Performanceseemedtodeterioratefrompre-testtopost-testinthetrainedgroup,butnotinthecontrolgroup,whilethenumberoferrorsincreasedatfollow-up(seeFigure1H).Theeffectsizesforgroupdifferencesweresmallatpost-testandbecamemediumatfollow-up(seeTable6),butthetwogroupsdidnotdiffer.Theneteffectsizeforthetrainedgroup,adjustedonthecontrolgroup'sperformance,wassmallforimmediategains(pre-vs.post-test),butbecamemediumforlong-termgains(seeTable6). Figure2Hthevaluesfittedfromthebestmodelforthetrainedgroupaloneasafunctionofage,andofpre-testperformanceintheWMcriteriontask,atthethreeassessmentsessions,withthecorrespondingestimatedlineartrend.Theregressionslopeforagedecreasedfrompost-testtofollow-up,butnotfrompre-testtopost-test,assupportedbytheevidenceratio,indicatingthatitwastheolderindividualswhomademoremistakesatpre-test,anditwasonlyatfollow-upthattheywerelikelytoexperiencefewerintrusionerrors.TheregressionslopesforbaselineperformanceintheWMcriteriontaskdecreasedfromsession0(pre-test)tosession1(post-test),withanevidenceratioof799(seeTable7),andthenbecameflat,assupportedbytheevidenceratio(seeTable7):alltheparticipantswithapoorWMperformancemademoremistakes,intermsofintrusionerrors,atpre-test,andfewermistakesatpost-test,andthisimprovementwasmaintainedatfollow-up. DiscussionandConclusions OuraiminthepresentstudywastodelineatehowcertainindividualcharacteristicscontributetoexplainingWMtraininggainsandtransfereffects.Despitetheimportanceofindividualcharacteristicsincognition,onlythreestudiesintheagingliteraturehaveconsideredthisissue.Here,age,formaleducation,generalcognitiveability(operationalizedwiththevocabularyscore),andWMbaselineperformancelevelwereconsideredasapredictoroftheshort-andlong-termspecifictraininggainsandtransfereffectsofaverbalWMtraininginasampleofhealthyolderadults. Toelucidatethisissue,ananalysiswasconductedusinglineargeneralizedmixedeffectsmodelsondatafromfourpreviousstudiesonhealthyolderadultsthatadoptedtheverbalWMtrainingprogramdevelopedbyBorellaetal.(2010).Partoftheinterestofsuchananalysisliesinthat—forthefirsttime,toourknowledgeatleast—allthestudiesexaminedwerebasedonthesameprocedureandthesameassessmentmeasures,andtheyallincludedafollow-upsession.TheBorellaetal.(2010)trainingprogramseemstobetheonlyWMproceduretohavebeenappliedrepeatedlyinolderadultswithconsistentresultsacrossstudies.Itisworthaddingthatanotheradvantageofthestudiesselectedforthepresentanalysisistheinclusionofanactivecontrolgroup,andparallelversionsofthetaskswerepresented(asrecommended,butrarelydone,intheliterature;Zinkeetal.,2012).Theeffectsidentifiedthereforecannotbeattributedtotheinfluenceofitem-specificpractice. Overall,ourfindingsconfirmedtheefficacyoftheverbalWMtrainingprocedureproposedbyBorellaetal.(2010):thetrainedgroupshowedspecificgains,performingbetterinthecriteriontaskthantheactivecontrolgroupimmediatelyafterthetraining,andmaintainingthisbenefitatfollow-up.PositiveeffectsoftheWMtrainingwerealsogenerallyapparentintermsoftransfereffects,intheshorttermatleast(atpost-test),sincethetrainedgroupoutperformedtheactivecontrolsinalltheneartransfermeasuresconsidered.Asforthefartransfermeasures,thetrainedgroupagainoutperformedtheactivecontrolsinalltasks,butnotintermsoftheStroopColorindexonRTsorCWMSintrusionerrors. Thispatternofresultswasconfirmedbythegenerallylargepost-testeffectsizes(over0.80)computedonthedifferencesbetweenthetrainedandactivecontrolgroups,withtheexceptionofthereasoningtask–theCattelltest-(mediumeffectsizes)andintrusionerrorsintheCWMS(smalleffectsizes).Atfollow-up,thedifferencesremainedlargeforthecriteriontask,butbecamemediumintheprocessingspeedtask—thePatternComparisontask—andforintrusionerrorsintheCWMS,andsmallintheothertasks(ForwardDigitandBackwardDigitSpantasks,DotMatrixtask,andCattelltest).TherewerenochangesintheeffectsizeoftheStroopColorindexonRTs,whichconfirmstheabsenceofaneffectofthetraining—thefactthattherewerenogroupdifferences—betweenthetrainedandcontrolgroups.Alsobyconsideringtheneteffectsizeofthetrainingactivitiesonparticipants'performance,thatischangesinthetrainedgroupacrosssessions—pre-test,post-testandfollow-up—(seeTable6)afteradjustingthevalueforanychangeinthecontrolgroup,thetrainingbenefitswereconfirmed.Itisconsequentlyreasonabletosaythatthetrainingproducedsomemaintenanceeffectsonthetrainedgroup'sperformance. Theseoverallfindingsareconsistentwiththepreviously-publishedresultsobtainedwiththesameWMtrainingprogram.Althoughthepresenttrainingregimenisquiteshort(onlythreetrainingsessions)nearandfartraininggainswerefound,confirmingthatthisWMtrainingprocedureiseffective.Asalsosuggestedbythemeta-analysisconductedbyKarbachandVerhaeghen(2014),thelengthofatrainingdoesnotseemacrucialfactorindeterminingitsefficacy:infact,mostoftheWMtrainingproceduresforolderadultsfailedtodocumentanybenefitsalthoughtheyweremuchlongerthantheoneconsideredhere(seeBorellaetal.,2017;seealsoTable1).Theadaptiveregimenadoptedmaywellhavefavoredtraininggainsby:(i)ensuringthatthetaskswerealwayschallenging,cognitivelydemandingandnovel,consequentlyinducingparticipantstoadheretothetask;(ii)producingachangeinparticipants'allocationofattentionalresourcesbecausethetrainingengagesseveralprocesses(includingencoding,retaininginformation,inhibitingnolongerrelevantinformation,managingtwotaskssimultaneously,shiftingattention,andattentionalcontrol)foranefficienthandlingofthedifferentdemandsofthetasks.Ontheotherhand,thelackofanyshort-termtransfereffectsforthetwoinhibitorymeasuresmaymeanthatinhibitorymechanismsarelessamenabletotraining(seeZavagninandRiboldi,2012).SomedegreeofcautionisrequiredininterpretingthefindingsobtainedwiththeStroopColortaskbecausetheywerebasedonRTs,whicharenotaveryreliableindicator(e.g.,deRibaupierreetal.,2003;seeLudwigetal.,2010;Borellaetal.,2017),andthesamplewasreducedforthisparticularmeasure(seeTable3).Itisalsopossible,asdiscussedbelow,thatitwouldtakelongertopromptanydetectablechangefortheinhibitorymeasures,orsomeofthematleast. Concerninglong-termeffectsofthetraining,therewasevidenceofthemaintenanceofthespecifictraininggain(inthecriterionWMtask),inlinewithalltheWMtrainingstudiesintheagingliterature.Inthetransfertasks,thetraininggainswereonlymaintainedfortheCattelltestandthePatternComparisontask,asseeninotherstudiesusingthesametrainingprocedure:theadvantageofthetrainedgroupoverthecontrolslayintherangeofamediumeffectsize(ornear-mediumforthePatternComparisontask).Suchaselectivemaintenanceofthetraininggainsmaybeattributabletothewell-documentedstrongrelationshipbetweenWMand(i)processingspeed(measuredwiththePatternComparisontask),and(ii)reasoningability(foracomprehensivediscussionseeBorellaetal.,2010). ImprovingWMperformancemakescognitiveoperationsmoreefficient,thusfosteringtheabilitytomoveamongthebasicinformationprocesses.Theothertasksmaycalluponmoretask-specificprocessesandabilitiesinstead,leadingtoonlytransient(immediate)transfereffects(foradiscussion,seeBorellaetal.,2010).Oneoftheinhibitorymeasuresexamined,intrusionerrorsintheCWMS,seemsparticularlyintriguinginthatitonlyshowedaclearimprovement(fewerintrusionerrorsinthecriteriontask)atfollow-up.Thismaymeanthatitwouldtakelongertoseeabenefitofthetrainingforsomemeasures(inhibitorymechanismsinthepresentcase).Thisphenomenon(i.e.,cleartransfereffectsonlyatfollow-up)hasbeenfoundinothertrainingstudiesinagingtoo(e.g.,Borellaetal.,2017),andhasbeencalledthe“sleeper”effect.Althoughitsnatureneedstobefurtherinvestigated(seeJaeggietal.,2014),itmayindicatethatcertainabilitiestakelongertoshowasignificantimprovementinperformance.Futurestudiesshouldmakeanefforttoexaminethisissue. SucharesultontheintrusionerrorsandnotintheStroopColortask(leavingasidetheproblemsassociatedwithmeasuringRTsandthereducedsamplesize)mayevenindicatethatWMtrainingismorebeneficialforsomeinhibitoryfunctionsthanforothers.Infact,intrusionerrorsintheCWMSandtheStroopColortaskdobelongtotwodifferentinhibitoryfunctions.CWMSintrusionerrors—aninternalmeasureoftheWMtaskandthereforecloselyrelatedthereto(seeRobertetal.,2009)—representtheresistancetoproactiveinterferencefunctionofinhibition,whichhelpsattentiontobefocusedonrelevantitemsandsimultaneously-presentedirrelevantitemstobeignored;theStroopColortaskmeasurestheresistancetoprepotentresponsefunctionofinhibition,whichblocksdominantandprepotentcognitiveresponsesautomaticallyactivatedbythestimuluspresented(e.g.,Borellaetal.,inpress).Resistancetoproactiveinterferenceisalsoconsideredtheonlyinhibitoryfunctionrelatedtothecontrolofinformationcomingfrommemorycontent(FriedmanandMiyake,2004). Inlightofthepresentfindings,thequestionsfollowingare:1.Doanyindividualcharacteristicshaveaparttoplayinthesefindings?Aretheeffectsoftrainingsupportedbymagnificationorcompensationeffects,orboth?Ofcourse,therecouldbeseveralaspects,suchasmethodologicalissues,butalsoparticipants'individualcharacteristics,capableofexplainingthetraininggainsandsupportingtheresults.Here,weparticularlyanalyzedtheroleofcertaindemographicvariables(ageandeducationallevel),cognitiveabilitiesinWM(i.e.,pre-testperformanceinthecriteriontask),andavocabularytestscoreasanindicatorofcrystallizedintelligence. Ourfindingsshowedthattheroleoftheindividualcharacteristicsconsidereddependedonthetypeofmeasureexamined,andtheeffectofthesevariableswasverymodestforsometasks.Themostinterestingaspectseemstobethatthefactorsconsideredwouldsupporteitheracompensationoramagnificationeffectimmediatelyafterthetraining,dependingonwhichmeasurewasanalyzed.Inparticular,irrespectiveofthenearorfartransfereffects,themorethetasksdemandedactiveinformationprocessing(i.e.,theDotMatrix,BackwardDigitSpanandPatternComparisontasks,theCattelltest,andtheStroopColortask),themorethefactorsexaminedseemedtosupportamagnificationeffect(ofvariablerobustness).Inotherwords,participantswhohadahigherinitialperformanceinthecrystallizedmeasureusedhereand/orwereyounger,weremorelikelytoimproveafterthetraining.Formorepassivetasks,ontheotherhand(i.e.,theForwardDigitSpantask,whichisashort-termmemorytask),ourresultssupportedacompensationeffect:participantswithlowerbaselinevocabularyscores,anolderage,andaweakerWMperformancebenefitedmorefromthetraining.Aparticularpatternemergedforthecriteriontask(i.e.,thetasksimilartotheoneusedinthetraining)andforacloselyrelatedmeasure(CWMSintrusionerrors):althoughthecriteriontaskiscomplex,participantswithalowerperformanceinataskofcrystallizedintelligence,asassessedwiththevocabularytest,gainedmorefromthetrainingthanthosewithhighervocabularyscores.Thatvocabularyshouldhavesucharolesuggeststhatknowledgecancounteractage-relateddecline(e.g.,Baltes,1987).Further,theroleofvocabularyinexplainingspecifictraininggainsmayalsosuggestthatparticipantsexhibitingtransfereffectswerethosewhoacquirednewknowledge,ratherthanagreaterprocessingefficiency.Afinely-gradedanalysisatindividuallevelmaybeabletoclarifythisissue. AsforintrusionerrorstherewasevidenceofacompensationeffectrelatedtoageandbaselineWMperformance,witholderparticipantsandthosewithalowerbaselineWMperformanceimprovingthemost.Theseresultsmaymeanthattheexercisesusedinthetrainingenabledindividualswithalowercrystallizedabilitytoadapttothedemandsofthetasks,engagebetter-controlledprocesses,andmakemoreefficientuseoftheirresources.Itishardlysurprisingthatsuchapatternofresultsshouldemergewithatrainingtaskthatinvolvesanadaptiveprocedure,whichmayhaveenabledprogresstobemadeduringthetraining,leadingtoabetterperformanceandfewerintrusionerrorsatpost-testanditsmaintenanceatfollow-up.Similarly,participantswithalowbaselineWMperformancealsobecamebetterabletomanagenolongerrelevantinformation(CWMSintrusionerrors). Onewaytointerprettheseresultscouldbebyreferringtothesupply-demandmismatchconceptualizedbyLövdénetal.(2010).Accordingtotheseauthors,changesincognitiveperformanceareinducedbyamismatchbetweenavailableresourcesandtaskdemands:tocopewiththismismatch,individualsengageinactivitiesthatpromoteflexibility,andconsequentlyalsoplasticity.Thishypothesisenablesustopredicthowindividualdifferencesmightaffectthebenefitsgainedfromatrainingregimen,dependingonatask'scomplexity.Thecompensationeffectseenforthecriteriontask(whichcloselyresembledthetaskusedinthetraining)maybeduetothefactthatusinganadaptiveprocedurewhilepracticingwiththetrainingtaskfavoredthe“rightamount”ofsupply-demandmismatch(i.e.,demandsexceedingthantheavailablecapacity)forindividualswithaweakerprofileintermsoftheirgeneralcognitiveabilitiestore-activatetheirpotential,andthusbenefitfromthetrainingintermsofabetterperformanceinthecriteriontask.Itmighthavebeeneasiertosupportthisinterpretationifourtrainingprocedurehadbeendesignedtoenableustotesthowperformancechangedfromonetrainingsessiontothenext.Suchananalysiswouldalsoshedlightonwhathappenstoindividualswiththeoppositeprofile(goodgeneralcognitiveabilities—highvocabularyscoresinthepresentstudy),whowouldonlyexperiencethemismatchiftheWMtasksusedinthetrainingweremoredifficult,soinsteadofbenefitingintermsofperformanceinthetrainedtasktheywouldgainintermsofplasticityinthetransfertasks. Accordingtothemismatchconcept,themagnificationeffectfoundforthemoredemandingtransfertasksmayindicatethatinparticipantswithahigherprofile,intermsofage(i.e.,youngerindividuals),orcrystallizedintelligence(i.e.,thosewithhighervocabularyscores)thetraininginducedasupply-demandmismatchthatgaveanimpetustochange,thusengenderingaflexiblebehavior.Inparticipantswithalowerprofile,ontheotherhand,thehighdemandsofthetasksusedinthetrainingmightpreventanysupply-demandmismatchbecausetheseindividualsmightabandonanyattemptorbeunabletoapplyresourcesandprocessessuitedtothetask. ThemixedresultsemergingfromthepresentanalysisasconcernstheroleofindividualcharacteristicsinexplainingthecompensationormagnificationeffectsareconsistentwithareportfromZinkeetal.(2013):theauthorsfoundthatparticipantswithweakertransfereffectswereolder(magnification),andthatthosewithsmallertraininggainshadstrongertransfereffects(compensation).TheroleofageandWMperformancevaried,however,dependingonthetransfertasksconsidered.Theroleofthepredictorswasexaminedtoo,buttheeffectsizewassmallforsomeofthetransfertasks,andthislimitsthevalueoftheresultsobtained. Ithastobesaidthattheresultsfoundinthepresentstudyweremodesttoo,sosomedegreeofcautioniswarrantedininterpretingthem.Further,suchapatternoffindingswasfoundatpost-testand,exceptforthecriteriontaskandtheintrusionerrors,theroleofthepredictorswasnotmaintainedatfollow-upfortheothermeasures.Sucharesultcouldbeinterpretedintwodifferentways:onestemsfromontheideathat,becauseofthetraining,theindividualcharacteristicsarenolongersignificantbecausesomethingbeyondthemhasbeenmodifiedduringthetraining,suchasthewayinwhichparticipantsprocessinformation;theothersimplyattributestheresulttothefactthattheeffectsofthetrainingwerenotmaintained.ItthereforeseemsimportanttoanalyzetheinfluenceofotherindividualcharacteristicsontheeffectsofWMtraining.Thishasbeendone,forthefirsttime,atleasttoourknowledge,inboththeshortandthelongterm;theotherthreestudiesthatapproachedsuchanissuedidnotconsidertheroleofthepredictorsinthelongterm. Ourfindingsalsosuggestthatcompensatoryandmagnificationeffectsarenotmutuallyexclusiveinexplainingtraininggains;theymaybothcontributetocharacterizingandexplainingtheoutcomeoftraining.ItwouldthereforebeimportantforfutureWMtrainingstudiesingeneral(andinaging,inparticular)tomaketheefforttoexaminetheroleofindividualfactors.Weareawarethatlargesamplesareneededforsuchanalyses,butitisonlybytryingtoovercomesuchpracticalproblemsthatresearchcanadvanceandenableustoascertaintherealusefulnessofinterventiontopromoteanactiveaging. Anumberoflimitationsofthepresentstudyhavetobeacknowledged.First,singlemeasureswereusedtorepresenttheconstructsofinterest,whereasusingmultipleindicatorsofthesameprocess(e.g.,Shipsteadetal.,2010)intrainingstudieswouldenableustodrawstrongerconclusions.Second,wewereunabletoconsidertraininggainsper-se,i.e.,theroleofindividualdifferencesinimprovementsinducedbytrainingacrossthesessions,therateoflearning,inpredictingtheoutcomeofthetraining,becausetheparticularprocedureuseddidnotallowfortheseaspectstobeanalyzed.Examiningwhatprogressagivenindividualmakes,andhowitrelatestotheeffectsofthetrainingcouldshedfurtherlightonthevalueofthepresenttrainingprogram(Zinkeetal.,2013;Bürkietal.,2014).Athirdlimitationliesinthatweanalyzedtheroleofalimitednumberoffactorspotentiallyinfluencingtraininggains.Infuturestudiesweplanttoconductamorecompleteassessmentofgeneralcognitiveability,andassessingthemwithtasksnotusedtotesttransfereffects(toavoidmulti-collinearityproblems).Sinceweacknowledgetheexploratorynatureofourresults,wehavediscussedoveralltrendsforthetransfertasks,ratherthantheinfluenceofeachofthespecificpredictorsoneachtask.Therearealsootherfactors,ofcourse,thatmayhaveinfluencedthetraininggains,andthathaveyettobeconsidered,suchasmetacognitive(motivational)variables,mood,andpsychologicalwell-being(e.g.,vonBastianandOberauer,2014;KönenandKarbach,2015).Itmightalsobeofinteresttoanalyzetheinfluenceofgenetics(suchasdopamineavailability)ontraininggains(vonBastianandOberauer,2014).Futurestudiesshouldthereforestrivetoincludeabroadarrayoffactors,withlargerandmorehomogeneoussamples(i.e.,withsamesize)thantheoneusedhere,inanefforttodelineatealltheconditionscapableofshapingtheeffectsofWMtraininginolderadults. Toconclude,thepresentstudyprovidesfurtherevidenceoftheelderlygainingincognitiveflexibilityandplasticityfromaverbalWMtraining.ItalsohighlightstheimportanceofanalyzingthefactorsinfluencingWMtraininggainsinaging.Also,althoughweshowedthatolderpeople'sWMcanbeimprovedthankstoaplasticitythatpersistswithaging,wefoundthattheroleofindividualcharacteristicsdependedonthetransfermeasureexamined.Itisconsequentlyimportanttoascertain“who”gainsfromthetraining,butalso“whogainsinwhichtasks,”inordertobeabletodesignthemosteffectiveWMtrainingtosuitanindividual'scognitiveprofile.Thisstudycouldthusbeconsideredasoneofthefirstpromisingstepstowardclarifyingtheimpactofindividualcharacteristicsontheshort-andlong-termefficacyofWMtraining. AuthorContributions EBdesignedthestudy,assistedincarryingouttheanalyses,andwrotethepaper.ECwrotethepaperandassistedincarryingouttheanalyses.MPcarriedoutthestatisticalanalyses.RDassistedinwritingthepaper.BCdesignedthestudy,assistedincarryingouttheanalyses,andwrotethepaper. Funding ThestudywassupportedbythegrantCPDA141092/14awardedbytheUniversityofPadovatoBC. ConflictofInterestStatement Theauthorsdeclarethattheresearchwasconductedintheabsenceofanycommercialorfinancialrelationshipsthatcouldbeconstruedasapotentialconflictofinterest. ThereviewerJMRandhandlingEditordeclaredtheirsharedaffiliation,andthehandlingEditorstatesthattheprocessneverthelessmetthestandardsofafairandobjectivereview Footnotes 1.^AsfortheForwardandtheBackwardDigitSpantasksandthePatternComparisontask,thesemeasureswereusedinthree(Borellaetal.,2010,2013,2017)ofthefourstudiesconsidered.Thepooledtrained(n=56)andcontrol(n=56)groupsfromthesethreestudiesdidnotdifferintermsofage(trainedgroup:M=72.43SD=5.81;controlgroup:M=72.04,SD=6.16),F(1,110)<1,yearsofformaleducation(trainedgroup:M=9.482SD=4.72;controlgroup:M=10.30SD=4.65),F(1,110)<1,orvocabularyscoreontheWechslerAdultIntelligenceScale-Revised(WAIS–R;Wechsler,1981;trainedgroup:M=50.23SD=11.27;controlgroup:M=49.07SD=11.45),F(1,110)<1. AsfortheCattelltest,thiswasusedinthree(Borellaetal.,2010,2013;Carrettietal.,2013b)ofthefourstudiesconsidered.Thepooledtrained(n=55)andcontrol(n=57)groupsfromthesetwostudiesdidnotdifferintermsofage(trainedgroup:M=72.35,SD=5.86;controlgroup:M=72.72SD=5.47),F(1,110)<1,yearsofformaleducation(trainedgroup:M=7.80SD=3.73;controlgroup:M=8.49SD=4.07),F(1,110)<1,orvocabularyscoreontheWechslerAdultIntelligenceScale—Revised(WAIS–R;Wechsler,1981;trainedgroup:M=45.11SD=8.89;controlgroup:M=43.16SD=10.25),F(1,110)=1.15,p=0.28. TheDotMatrixtaskandtheStroopColortaskwereusedintwo(Borellaetal.,2010,2013)ofthefourstudiesconsidered.Thepooledtrained(n=38)andcontrol(n=38)groupsfromthesetwostudiesdidnotdifferintermsofage(trainedgroup:M=73.84,SD=6.12;controlgroup:M=73.89SD=5.85),F(1,74)<1,yearsofformaleducation(trainedgroup:M=7.66SD=3.79;controlgroup:M=8.24SD=3.62),F(1,74)<1,orvocabularyscoreontheWechslerAdultIntelligenceScale—Revised(WAIS–R;Wechsler,1981;trainedgroup:M=44.79SD=8.93;controlgroup:M=44.21SD=9.65),F(1,74)<1. 2.^FortheCWMS,thebaselineperformanceintheCWMSwasnotenteredasapredictorbecauseofmulticollinearityproblems. References Baayen,R.H.,Davidson,D.J.,andBates,D.M.(2008).Mixed-effectsmodelingwithcrossedrandomeffectsforsubjectsanditems.J.Mem.Lang.59,390–412.doi:10.1016/j.jml.2007.12.005 CrossRefFullText|GoogleScholar Baltes,P.B.(1987).Theoreticalpropositionsoflife-spandevelopmentalpsychology:onthedynamicsbetweengrowthanddecline.Dev.Psychol.23,611–626.doi:10.1037/0012-1649.23.5.611 CrossRefFullText|GoogleScholar Bissig,D.,andLustig,C.(2007).Trainingmemory:whobenefitsfrommemorytraining?Psychol.Sci.8,720–726.doi:10.1111/j.1467-9280.2007.01966.x CrossRefFullText|GoogleScholar Bopp,K.L.,andVerhaeghen,P.(2005).Agingandverbalmemoryspan:ameta-analysis.J.Gerontol.BPsychol.Sci.Soc.Sci.60,223–233.doi:10.1093/geronb/60.5.P223 PubMedAbstract|CrossRefFullText|GoogleScholar Borella,E.,Carretti,B.,andDeBeni,R.(2008).Workingmemoryandinhibitionacrosstheadultlife-span.ActaPsychol.128,33–44.doi:10.1016/j.actpsy.2007.09.008 PubMedAbstract|CrossRefFullText|GoogleScholar Borella,E.,Carretti,B.,Cantarella,A.,Riboldi,F.,Zavagnin,M.,andDeBeni,R.(2014).Benefitsoftrainingvisuospatialworkingmemoryinyoung-oldandold-old.Dev.Psychol.50,714–727.doi:10.1037/a0034293 PubMedAbstract|CrossRefFullText|GoogleScholar Borella,E.,Carretti,B.,Cornoldi,C.,andDeBeni,R.(2007).Workingmemory,controlofinterferenceandeverydayexperienceofthoughtinterference:whenagemakesthedifference.AgingClin.Exp.Res.19,200–206.doi:10.1007/BF03324690 PubMedAbstract|CrossRefFullText|GoogleScholar Borella,E.,Carretti,B.,Mitolo,M.,Zavagning,M.,Caffarra,P.,Mammarella,N.,etal.(2017).Characterizingcognitiveinhibitorydeficitsinmildcognitiveimpairment.PsychiatryRes.251,342–348.doi:10.1016/j.psychres.2016.12.037 PubMedAbstract|CrossRefFullText|GoogleScholar Borella,E.,Carretti,B.,Riboldi,F.,andDeBeni,R.(2010).Workingmemorytraininginolderadults:oftransferandmaintenanceeffects.Psychol.Aging25,767–778.doi:10.1037/a0020683 PubMedAbstract|CrossRefFullText|GoogleScholar Borella,E.,Carretti,B.,Sciore,R.,Capotosto,E.,Tacownat,L.,Cornoldi,C.,etal.(inpress).Trainingworkingmemoryinolderadults:isthereanadvantagewhenstrategiesareused?Psychol.Aging. PubMedAbstract|GoogleScholar Borella,E.,Carretti,B.,Zanoni,G.,Zavagnin,M.,andDeBeni,R.(2013).Workingmemorytraininginoldage:anexaminationoftransferandmaintenanceeffects.Arch.Clin.Neuropsychol.28,331–347.doi:10.1093/arclin/act020 PubMedAbstract|CrossRefFullText|GoogleScholar Borella,E.,Delaloye,C.,Lecerf,T.,Renaud,O.,anddeRibaupierre,A.(2009).Doagedifferencesbetweenyoungandolderadultsininhibitorytasksdependonthedegreeofactivationofinformation?Eur.J.Cogn.Psychol.21,445–472.doi:10.1080/09541440802613997 CrossRefFullText|GoogleScholar Borella,E.,Ghisletta,P.,anddeRibaupierre,A.(2011).Agedifferencesintextprocessing:theroleofworkingmemory,inhibitionandprocessingspeed.J.Gerontol.BPsychol.Sci.Soc.Sci.66,311–320.doi:10.1093/geronb/gbr002 PubMedAbstract|CrossRefFullText|GoogleScholar Brehmer,Y.,Rieckmann,A.,Bellander,M.,Westerberg,H.,Fischer,H.,andBäckman,L.(2011).Neuralcorrelatesoftraining-relatedworking-memorygainsinoldage.Neuroimage4,1110–1120.doi:10.1016/j.neuroimage.2011.06.079 CrossRefFullText|GoogleScholar Brehmer,Y.,Westerberg,H.,andBackman,L.(2012).Working-memorytraininginyoungerandolderadults:traininggains,transferandmaintenance.Front.Hum.Neurosci.6:63.doi:10.3389/fnhum.2012.00063 PubMedAbstract|CrossRefFullText|GoogleScholar Buerkner,P.-C.(2016).brms:BayesianRegressionModelsusingStan.RPackageVersion0.10.10.Availableonlineat:https://CRAN.R-project.org/package=brms Bürki,C.,Ludwig,C.,Chicherio,C.,anddeRibaupierre,A.(2014).Individualdifferencesincognitiveplasticity:aninvestigationoftrainingcurvesinyoungerandolderadults.Psychol.Res.78,821–835.doi:10.1007/s00426-014-0559-3 PubMedAbstract|CrossRefFullText|GoogleScholar Burnham,K.P.,Anderson,D.R.,andHuyvaert,K.P.(2011).AICmodelselectionandmultimodelinferenceinbehavioralecology:somebackground,observations,andcomparisons.Behav.Ecol.Sociobiol.65,23–35.doi:10.1007/s00265-010-1029-6 CrossRefFullText|GoogleScholar Buschkuehl,M.,Jaeggi,S.M.,Hutchison,S.,Perrig-Chiello,P.,Däpp,C.,Müller,M.,etal.(2008).Impactofworkingmemorytrainingonmemoryperformanceinold-oldadults.Psychol.Aging23,743–753.doi:10.1037/a0014342 PubMedAbstract|CrossRefFullText|GoogleScholar Cantarella,A.,Borella,E.,Carretti,B.,Kliegel,M.,anddeBeni,R.(2017).Benefitsintasksrelatedtoeverydaylifecompetencesafteraworkingmemorytraininginolderadults.Int.J.Geriatr.Psychiatry.32,86–93.doi:10.1002/gps.4448 PubMedAbstract|CrossRefFullText|GoogleScholar Carretti,B.,Belacchi,C.,andCornoldi,C.(2010).Difficultiesinworkingmemoryupdatinginindividualswithintellectualdisability.J.Intellect.Disabil.Res.54,337–345.doi:10.1111/j.1365-2788.2010.01267.x PubMedAbstract|CrossRefFullText|GoogleScholar Carretti,B.,Borella,E.,Fostinelli,S.,andZavagnin,M.(2013a).Benefitsoftrainingworkingmemoryinamnesticmildcognitiveimpairment:specificandtransfereffects.Int.J.Geriatr.Psychiatry25,617–626.doi:10.1017/S1041610212002177 PubMedAbstract|CrossRefFullText|GoogleScholar Carretti,B.,Borella,E.,Zavagnin,M.,andDeBeni,R.(2013b).Gainsinlanguagecomprehensionrelatingtoworkingmemorytraininginhealthyolderadults.Int.J.Geriatr.Psychiatry28,539–546.doi:10.1002/gps.3859 PubMedAbstract|CrossRefFullText|GoogleScholar Cattell,R.B.,andCattell,A.(1963).MeasuringIntelligencewiththeCultureFairTests.Champaign,IL:InstituteforPersonalityandAbilityTesting. GoogleScholar Cornoldi,C.(2010).“Metacognition,intelligence,andacademicperformance,”inMetacognition,StrategyUse,andInstruction,edsH.S.WatersandW.Schneider(NewYork,NY:GuilfordPress),257–277. GoogleScholar Cornoldi,C.,andVecchi,T.E.(2003).Visuo-SpatialWorkingMemoryandIndividualDifferences.Hove:PsychologyPress. GoogleScholar Crook,T.,Bartus,R.T.,Ferris,S.H.,Whitehouse,P.,Cohen,G.D.,andGershon,S.(1986).Age-associatedmemoryimpairment:proposeddiagnosticcriteriaandmeasuresofclinicalchange—reportofaNationalInstituteofMentalHealthworkgroup.Dev.Neuropsychol.2,261–276.doi:10.1080/87565648609540348 CrossRefFullText|GoogleScholar Dahlin,E.,Nyberg,L.,Bäckman,L.,andStigsdotterNeely,A.(2008).Plasticityofexecutivefunctioninginyoungandolderadults:immediategains,transfer,andlong-termmaintenance.Psychol.Aging23,720–730.doi:10.1037/a00114296 PubMedAbstract|CrossRefFullText|GoogleScholar DeBeni,R.,Borella,E.,Carretti,B.,Marigo,C.,andNava,L.(2008).BAC.PortfolioperlaValutazionedelBenessereedelleAbilitàCognitivenell'etàadultaeAvanzata[TheAssessmentofWell-beingandCognitiveAbilitiesinAdulthoodandAging].Firenze:Giunti,OS. GoogleScholar deRibaupierre,A.(2001).“Workingmemoryandattentionalprocessesacrossthelifespan,”inLifeSpanDevelopmentofHumanMemory,edsP.GrafandN.Otha(Cambridge,MA:MITPress),59–80. GoogleScholar deRibaupierre,A.,andLecerf,T.(2006).Relationshipsbetweenworkingmemoryandintelligencefromadevelopmentalperspective:convergentevidencefromaneo-Piagetianandapsychometricapproach.Eur.J.Cogn.Psychol.18,109–137.doi:10.1080/09541440500216127 CrossRefFullText|GoogleScholar deRibaupierre,A.,Borella,E.,andDelaloye,C.(2003).“Inhibitionetvariabilite'individuelle:ge'ne'ralite'ouspe'cificite'desprocessusinhibiteurs[Inhibitionandindividualvariability:generalityorspecificityofinhibitoryprocesses],”inInhibitionNeuraleetCognitive,edS.Moutier(Paris:Herme's),103–124. GoogleScholar Ekstrom,R.B.,French,J.W.,Harman,H.H.,andDermen,D.(1976).ManualforKitofFactor-ReferencedCognitiveTests.Princeton,NJ:EducationalTestingService. GoogleScholar Folstein,M.,Folstein,S.E.,andMcHugh,P.R.(1975).“Mini-MentalState”:apracticalmethodforgradingthecognitivestateofpatientsfortheclinician.J.Psychiatr.Res.12,189–198.doi:10.1016/0022-3956(75)90026-6 PubMedAbstract|CrossRefFullText|GoogleScholar Friedman,N.P.,andMiyake,A.(2004).Therelationsamonginhibitionandinterferencecontrolfunctions:alatent-variableanalysis.J.Exp.Psychol.Gen.133,101–135.doi:10.1037/0096-3445.133.1.101 PubMedAbstract|CrossRefFullText|GoogleScholar Gabry,J.,andGoodrich,B.(2016).rstanarm:BayesianAppliedRegressionModelingviaStan.RPackageVersion2.10.11.Availableonlineat:https://cran.rstudio.com/web/packages/rstanarm/ Gallina,P.,Saugo,M.,Antoniazzi,M.,Fortuna,P.,Toffanin,R.,Maggi,S.,etal.(2006).ValidazionedellaschedaperlaValutazioneMultidimensionaledell'Anziano(SVAMA)[Validationofthechecklistfortheassessmentmultidimensionaloftheelderly].TendenzeNuove3,229–264. GoogleScholar Gelman,A.,andRubin,D.B.(1992).Inferencefromiterativesimulationusingmultiplesequences(withdiscussion).Stat.Sci.7,457–511.doi:10.1214/ss/1177011136 CrossRefFullText|GoogleScholar Gueorguieva,R.,andKrystal,J.H.(2004).MoveoverANOVA:progressinanalyzingrepeated-measuresdataanditsreflectioninpaperspublishedinthearchivesofgeneralpsychiatry.Arch.Gen.Psychiatr.61,310–317.doi:10.1001/archpsyc.61.3.310 PubMedAbstract|CrossRefFullText|GoogleScholar Heinzel,S.,Schulte,S.,Onken,J.,Duong,Q.-L.,Riemer,T.G.,Heinz,A.,etal.(2014).Workingmemorytrainingimprovementsandgainsinnon-trainedcognitivetasksinyoungandolderadults.Neuropsychol.Dev.Cogn.BAgingNeuropsychol.Cogn.21,146–173.doi:10.1080/13825585.2013.790338 PubMedAbstract|CrossRefFullText|GoogleScholar Hertzog,C.,Kramer,A.F.,Wilson,R.S.,andLindenberger,U.(2008).Enrichmenteffectsonadultcognitivedevelopment.Psychol.Sci.PublicInterest9,1–65.doi:10.1111/j.1539-6053.2009.01034.x PubMedAbstract|CrossRefFullText|GoogleScholar Jaeggi,S.M.,Buschkuehl,M.,Shah,P.,andJonides,J.(2014).Theroleofindividualdifferencesincognitivetrainingandtransfer.Mem.Cogn.42,464–480.doi:10.3758/s13421-013-0364-z PubMedAbstract|CrossRefFullText|GoogleScholar Karbach,J.,andVerhaeghen,P.(2014).Makingworkingmemorywork:ameta-analysisofexecutive-controlandworkingmemorytraininginolderadults.Psychol.Sci.25,2027–2037.doi:10.1177/0956797614548725 PubMedAbstract|CrossRefFullText|GoogleScholar Könen,T.,andKarbach,J.(2015).Thebenefitoflookingatintraindividualdynamicsincognitivetrainingdata.Front.Psychol.6:615.doi:10.3389/fpsyg.2015.00615 PubMedAbstract|CrossRefFullText|GoogleScholar Kruschke,J.K.(2011).DoingBayesianDataAnalysis.Burlington,MA:AcademicPress. GoogleScholar Li,S.-C.,Schmiedeck,F.,Huxhold,O.,Röcke,C.,Smith,J.,andLindenberger,U.(2008).Workingmemoryplasticityinoldage:practicegain,transfer,andmaintenance.Psychol.Aging23,731–742.doi:10.1037/a0014343 PubMedAbstract|CrossRefFullText|GoogleScholar Lövdén,M.,Bäckman,L.,Lindenberger,U.,Schaefer,S.,andSchmiedek,F.(2010).Atheoreticalframeworkforthestudyofadultcognitiveplasticity.Psychol.Bull.136,659–676.doi:10.1037/a0020080 PubMedAbstract|CrossRefFullText|GoogleScholar Lövdén,M.,Brehmer,Y.,Li,S.-C.,andLindenberger,U.(2012).Training-inducedcompensationversusmagnificationofindividualdifferencesinmemoryperformance.Front.Hum.Neurosci.6:141.doi:10.3389/fnhum.2012.00141 PubMedAbstract|CrossRefFullText|GoogleScholar Ludwig,C.,Borella,E.,Tettamanti,M.,anddeRibaupierre,A.(2010).AdultagedifferencesintheStroopcolortest:acomparisonbetweenanitem-by-itemandablockedversion.Arch.Gerontol.Geriat.51,35–142.doi:10.1016/j.archger.2009.09.040 PubMedAbstract|CrossRefFullText|GoogleScholar Lustig,C.,Shah,P.,Seidler,R.,andReuter-Lorenz,P.A.(2009).Aging,training,andthebrain:areviewandfuturedirections.Neuropsychol.Rev.19,504–522.doi:10.1007/s11065-009-9119-9 PubMedAbstract|CrossRefFullText|GoogleScholar Mammarella,I.C.,Borella,E.,Pastore,M.,andPazzaglia,F.(2013).Thestructureofvisuospatialmemoryinadulthood.Learn.Individ.Diff.25,99–110.doi:10.1016/j.lindif.2013.01.014 CrossRefFullText|GoogleScholar McElreath,R.(2016).StatisticalRethinking:ABayesianCoursewithExamplesinRandStan.BocaRaton,FL:ChapmanandHall/CRC. GoogleScholar Miyake,A.,Friedman,N.P.,Rettinger,D.A.,Shah,P.,andHegarty,M.(2001).Howarevisuospatialworkingmemory,executivefunctioning,andspatialabilitiesrelated?Alatent-variableanalysis.J.Exp.Psychol.Gen.130,621–640.doi:10.1037/0096-3445.130.4.621 PubMedAbstract|CrossRefFullText|GoogleScholar Noack,H.,Lövdén,M.,Schmiedek,F.,andLindenberger,U.(2009).Cognitiveplasticityinadulthoodandoldage:gaugingthegeneralityofcognitiveinterventioneffects.Restor.Neurol.Neurosci.27,435–453.doi:10.3233/RNN-2009-0496 PubMedAbstract|CrossRefFullText|GoogleScholar Pazzaglia,F.,DeBeni,R.,andMeneghetti,C.(2007).Theeffectsofverbalandspatialinterferenceintheencodingandretrievalofspatialandnonspatialtexts.Psychol.Res.71,484–494.doi:10.1007/s00426-006-0045-7 PubMedAbstract|CrossRefFullText|GoogleScholar Richmond,L.L.,Morrison,A.B.,Chein,J.M.,andOlson,I.R.(2011).Workingmemorytrainingandtransferinolderadults.Psychol.Aging26,813–822.doi:10.1037/a0023631 PubMedAbstract|CrossRefFullText|GoogleScholar Robert,C.,Borella,E.,Fagot,D.,Lecerf,T.,andDeRibaupierre,A.(2009).Workingmemoryandinhibitorycontrolacrossthelifespan:intrusionerrorsintheReadingSpanTest.Mem.Cog.37,336–345.doi:10.3758/MC.37.3.336 PubMedAbstract|CrossRefFullText|GoogleScholar Salminem,T.,Frensch,P.,Storbach,T.,andSchubert,T.(2015).Age-specificdifferencesofdualn-backtraining.AgingNeuropsychol.Cogn.23,18–39.doi:10.1080/13825585.2015.1031723 CrossRefFullText|GoogleScholar Salthouse,T.A.,andBabcock,R.L.(1991).Decomposingadultagedifferencesinworkingmemory.Dev.Psychol.27,763–776.doi:10.1037/0012-1649.27.5.763 CrossRefFullText|GoogleScholar Shipstead,Z.,Redick,T.,andEngle,R.(2010).Doesworkingmemorytraininggeneralize?Psychol.Belg.50,245–276.doi:10.5334/pb-50-3-4-245 CrossRefFullText|GoogleScholar Staff,R.T.,Murray,A.D.,Deary,I.J.,andWhalley,L.J.(2004).Whatprovidescerebralreserve?Brain127,1191–1199.doi:10.1093/brain/awh144 PubMedAbstract|CrossRefFullText|GoogleScholar StanDevelopmentTeam(2015).StanModelingLanguageUser'sGuideandReferenceManual.Version2.10.10.Availableonlineat:http://mc-stan.org Stepankova,H.,Lukavsky,J.,Buschkuehl,M.,Kopecek,M.,Ripova,D.,andJaeggi,S.M.(2014).Themalleabilityofworkingmemoryandvisuospatialskills:arandomizedcontrolledstudyinolderadults.Dev.Psychol.50,1049–1059.doi:10.1037/a0034913 PubMedAbstract|CrossRefFullText|GoogleScholar Stern,Y.(2002).Whatiscognitivereserve?Theoryandresearchapplicationofthereserveconcept.J.Int.Neuropsychol.Soc.8,448–460.doi:10.1017/S1355617702813248 PubMedAbstract|CrossRefFullText|GoogleScholar Titz,C.,andKarbach,J.(2014).Workingmemoryandexecutivefunctions:effectsoftrainingonacademicachievement.Psychol.Res.78,852–868.doi:10.1007/s00426-013-0537-1 PubMedAbstract|CrossRefFullText|GoogleScholar Trenerry,M.R.,Crosson,B.,DeBoe,J.,andLever,W.R.(1989).StroopNeuropsychologicalScreeningTest.Lutz,FL:PsychologicalAssessmentResources. GoogleScholar Troyer,A.K.,Leach,L.,andStrauss,E.(2006).Agingandresponseinhibition:normativedatafortheVictoriaStrooptest.AgingNeuropsychol.Cogn.13,20–35.doi:10.1080/138255890968187 PubMedAbstract|CrossRefFullText|GoogleScholar VanderElst,W.,VanBoxtel,M.P.J.,VanBreukelen,G.J.P.,andJolles,J.(2006).TheStroopcolor–wordtest:influenceofage,sex,andeducation,andnormativedataforalargesampleacrosstheadultagerange.Assessment13,62–79.doi:10.1177/1073191105283427 PubMedAbstract|CrossRefFullText|GoogleScholar Verhaeghen,P.,andMarcoen,A.(1996).Onthemechanismsofplasticityinyoungandolderadultsafterinstructioninthemethodofloci:evidenceforanamplificationmodel.Psychol.Aging11,164–178.doi:10.1037/0882-7974.11.1.164 PubMedAbstract|CrossRefFullText|GoogleScholar vonBastian,C.C.,andOberauer,K.(2014).Effectsandmechanismsofworkingmemorytraining:areview.Psychol.Res.6,803–820.doi:10.1007/s00426-013-0524-6 CrossRefFullText|GoogleScholar vonBastian,C.C.,Langer,N.,Jäncke,L.,andOberauer,K.(2013).Effectsofworkingmemorytraininginyoungandoldadults.Mem.Cogn.41,611–624.doi:10.3758/s13421-012-0280-7 PubMedAbstract|CrossRefFullText|GoogleScholar Wainwright,P.E.,Leatherdale,S.T.,andDubin,J.A.(2007).AdvantagesofmixedeffectsmodelsovertraditionalANOVAmodelsindevelopmentalstudies:aworkedexampleinamousemodeloffetalalcoholsyndrome.Dev.Psychobiol.49,664–674.doi:10.1002/dev.20245 PubMedAbstract|CrossRefFullText|GoogleScholar Watanabe,S.(2010).AsymptoticequivalenceofBayescrossvalidationandwidelyapplicableinformationcriterioninsingularlearningtheory.J.Mach.Learn.Res.11,3571–3594. GoogleScholar Wechsler,D.(1981).WechslerAdultIntelligenceScale—Revised(WAIS–R).NewYork,NY:PsychologicalCorporation. GoogleScholar Weisz,J.R.,andHawley,K.M.(2001).ProceduralandCodingManualforIdentificationofEvidence-BasedTreatments.LosAngeles,CA:UniversityofCalifornia. GoogleScholar West,R.L.,andAlain,C.(2000).Age-relateddeclineininhibitorycontrolcontributestotheincreasedStroopeffectobservedinolderadults.Psychophysiology37,179–189.doi:10.1017/S0048577200981460 PubMedAbstract|CrossRefFullText|GoogleScholar Xin,Z.,Lai,Z.R.,Li,F.,andMaes,J.H.(2014).Near-andfar-transfereffectsofworkingmemoryupdatingtraininginelderlyadults.Appl.Cogn.Psychol.28,403–408.doi:10.1002/acp.3011 CrossRefFullText|GoogleScholar Zając-Lamparska,L.,andTrempała,J.(2016).Effectsofworkingmemoryandattentionalcontroltrainingandtheirtransferontofluidintelligenceinearlyandlateadulthood.HealthPsychol.Rep.4,41–53.doi:10.5114/hpr.2016.56846 CrossRefFullText|GoogleScholar Zavagnin,M.,andRiboldi,F.(2012).Trainingdipotenziamentodellamemoriadilavoroedellefunzioniesecutivenellapopolazioneanziana:unarassegna.[Trainingofworkingmemoryandexecutivefunctionsinelderlypeople:Areview].GiornaleItalianodiPsicologia4,745–770.doi:10.1421/73141 CrossRefFullText|GoogleScholar Zinke,K.,Zeintl,M.,Eschen,A.,Herzog,C.,andKliegel,M.(2012).Potentialsandlimitsofplasticityinducedbyworkingmemorytraininginold-oldage.Gerontology58,79–87.doi:10.1159/000324240 PubMedAbstract|CrossRefFullText|GoogleScholar Zinke,K.,Zeintl,M.,Rose,N.S.,Putzmann,J.,Pydde,A.,andKliegel,M.(2013).Workingmemorytrainingandtransferinoldage:effectsofage,baselineperformanceandtraininggains.Dev.Psychol.50,304–315.doi:10.1037/a0032982 PubMedAbstract|CrossRefFullText|GoogleScholar Keywords:workingmemorytraining,olderadults,age,workingmemorybaselineperformance,generalcognitiveability,trainingbenefits,individualdifferences,individualcharacteristics Citation:BorellaE,CarboneE,PastoreM,DeBeniRandCarrettiB(2017)WorkingMemoryTrainingforHealthyOlderAdults:TheRoleofIndividualCharacteristicsinExplainingShort-andLong-TermGains.Front.Hum.Neurosci.11:99.doi:10.3389/fnhum.2017.00099 Received:21October2016;Accepted:20February2017;Published:22March2017. Editedby:SoledadBallesteros,UniversidadNacionaldeEducaciónaDistancia,Spain Reviewedby:JoséManuelReales,UniversidadNacionaldeEducaciónaDistancia,SpainTiinaSalminen,SynaptikonGmbH(NeuroNation),Germany Copyright©2017Borella,Carbone,Pastore,DeBeniandCarretti.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(CCBY).Theuse,distributionorreproductioninotherforumsispermitted,providedtheoriginalauthor(s)orlicensorarecreditedandthattheoriginalpublicationinthisjournaliscited,inaccordancewithacceptedacademicpractice.Nouse,distributionorreproductionispermittedwhichdoesnotcomplywiththeseterms. *Correspondence:ErikaBorella,[email protected],[email protected] ThisarticleispartoftheResearchTopic CognitiveandBrainPlasticityInducedbyPhysicalExercise,CognitiveTraining,VideoGamesandCombinedInterventions Viewall 50Articles Peoplealsolookedat Download