Article
Reference
The effect of age and individual differences in attentional control: A sample case using the Hayling test
BORELLA, Erika, et al.
Abstract
Individual differences in working memory (WM) have been shown to reflect the ability to control attention in order to prevent interference. This study examines the role of WM capacity in resisting interference in the Hayling task, in samples of younger and older adults. In each age group, high and low WM span individuals had to complete high-cloze sentences with either expected words (initiation) or words providing no meaning to the sentences (interference). Results showed increased response times and decreased correct responses in interference, as compared to initiation. As interference increased, older adults demonstrated lower accuracy than younger ones. Further, low spans demonstrated higher interference costs than high spans on accuracy, while the reverse pattern was found for response times. Our findings suggest that both age and individual differences in WM capacity need to be considered to account for differences in the ability to resist to interference.
BORELLA, Erika, et al . The effect of age and individual differences in attentional control: A sample case using the Hayling test. Archives of Gerontology and Geriatrics , 2011, vol. 53, no. 1, p. e75–e80
DOI : 10.1016/j.archger.2010.11.005
Available at:
http://archive-ouverte.unige.ch/unige:87070
Disclaimer: layout of this document may differ from the published version.
1 / 1
The effect of age and individual differences in attentional control: A sample case using the Hayling test
E. Borella
a,*, C. Ludwig
b,c, D. Fagot
c, A. De Ribaupierre
c,baFacultyofPsychology,UniversityofPadua,ViaVenezia8,35131Padova,Italy
bCenterforInterdisciplinaryGerontology,UniversityofGeneva,RoutedeDrize7,CH-1227Carouge,Switzerland
cFacultyofPsychologyandEducationalSciences,UniversityofGeneva,BoulevardduPontd’Arve40,CH-1205Geneva,Switzerland
1. Introduction
Mostoftheworkingmemory(WM)modelssharethecommon featureofconsideringasupervisoryattentionalsystemoracentral executive component, responsible for allocating attentional resources and managing controlled, intentional, conscious or effortfulprocesses(Miyakeand Shah,1999). BecauseWMtasks implysimultaneousprocessingandstorage,theyentailcontrolled attention to conduct processing and/or to alternate between processingand storageinordertoachievethetaskgoal.Hence, cognitive/attentionalcontrolappearsasacentralfeatureofWM.
Several authors have further proposed that cognitive control accountsfor eitherindividual (e.g.,Miyakeand Shah, 1999; de Ribaupierre,2000)or/and agedifferences (e.g.,Craik and Byrd, 1982)inbothWMperformanceandcomplexfluidcognition(e.g., Borellaet al.,2006).
With respect to individual differences, Engle (2001) has proposed theconstructof executive,or controlled, attentionto accountforquantitativeandqualitativedifferencesinprocessing across individuals. In this perspective, executive attention is viewedasa generalcontent-freecognitiveresourcethat canbe
reliably capturedby meansof complexspantasks, suchasthe readingspan(seeConwayet al.,2005).
Engle and Kane (2004) recently suggested that executive attentionisimportanttomaintaininformationinactivememory, andtoresolveconflictsresultingfromcompetitionbetweentask- appropriateresponsesandprepotentbutinappropriateones.Engle and collaboratorslargely developed their view by means of an extremegroupmethodology:theyfirstidentifiedhighandlowspan individualsonthebasisoftheirperformanceoncomplexspantasks, and subsequently compared these individuals on various tasks.
These authors reported that in conditions in which automatic responseswerelargelysufficienttoachievethetasks,highandlow spans did notsignificantly differ. However,when attentionwas required, highspans outperformedlow spans;high spans could maintain activea larger amount ofinformationand were more efficienttoresistinterference(EngleandKane,2004;Kaneet al., 2004).Accordingly,individualdifferencesinWMcapacityseemto accountfordifferencesintheabilitytokeeprelevantinformation temporarily available, and mostparticularly, wheninhibition of irrelevant information is also required. More specifically tothe present purpose, Engle’s findings suggest that differences in content-free executive attention account for differences in the abilitytousecontrolledprocessingtoresistinterference.
Interestinglyenough,similarhypothesescanbeadvancedwith respect to age differences. It is commonly reported that, as compared toyoungeradults, olderadults showreducedperfor- ARTICLE INFO
Articlehistory:
Received10July2010
Receivedinrevisedform2November2010 Accepted3November2010
Availableonline7December2010
Keywords:
Workingmemory Interference Individualdifferences Cognitiveaging
ABSTRACT
Individualdifferencesinworkingmemory(WM)have beenshownto reflecttheabilityto control attentioninordertopreventinterference.ThisstudyexaminestheroleofWMcapacityinresisting interferenceintheHaylingtask,insamplesofyoungerandolderadults.Ineachagegroup,highandlow WMspanindividualshadtocompletehigh-clozesentenceswitheitherexpectedwords(initiation)or wordsprovidingnomeaningtothesentences(interference).Resultsshowedincreasedresponsetimes anddecreasedcorrectresponsesininterference,ascomparedtoinitiation.Asinterferenceincreased, olderadultsdemonstratedloweraccuracythanyoungerones.Further,lowspansdemonstratedhigher interferencecoststhanhighspansonaccuracy,whilethereversepatternwasfoundforresponsetimes.
OurfindingssuggestthatbothageandindividualdifferencesinWMcapacityneedtobeconsideredto accountfordifferencesintheabilitytoresisttointerference.
ß2010ElsevierIrelandLtd.Allrightsreserved.
*Correspondingauthor.Tel.:+390498276911;fax:+390498276600.
E-mailaddress:erika.borella@unipd.it(E.Borella).
ContentslistsavailableatScienceDirect
Archives of Gerontology and Geriatrics
j ou rna l h om e pa ge : w w w. e l s e v i e r. co m/ l oc a t e / a rch ge r
0167-4943/$–seefrontmatterß2010ElsevierIrelandLtd.Allrightsreserved.
doi:10.1016/j.archger.2010.11.005
mancesinWMtaskssuchasthereadingspan(e.g.,Borellaet al., 2008).Ithasalsobeensuggestedthatage-relateddifferencescome fromareductioninprocessingresources(CraikandByrd,1982),or fromadeclineintheabilitytocontrolforinterference(Hasherand Zacks, 1988). Thus evidence tends to support an age-related declineincontrolledattention,oftenmanifestintasksthatrequire activelyinhibiting,suppressing,orclearingirrelevantinformation (McDowdet al.,1995).
However, toourknowledge,no studyhasbeeninterestedin consideringtogetherageandindividualdifferencesincontrolled attention,andininvestigatingwhethertheyexertajointinfluence in explaining performance in tasks requiring resistance to interference.
Thepresentstudyaimedatspecificallyaddressingthisissue.
Usinganextremegroupdesign,itsobjectivewastocomparethe performanceofyoungerandolderadultswithhighandlowWM spansonatestinvolvingactiveinhibitionofirrelevantresponses:
the Hayling test (Burgess and Shallice, 1996). In this task, participants have to complete the final word of series of 15 high-cloze sentences either with a word that is driven by the context(initiation)orwithawordthatissyntacticallycorrect,but gives no meaning to the sentence (interference). The latter condition requires the participant to draw upon controlled attention to resist interference from the word that is highly activated by the sentence. Empirically, inhibition failures are reflected by a decrease in the proportion of correct responses, while the requirement of inhibiting irrelevant competing responsesisreflectedbyanincreaseinresponsetimes.
With respect to the present study, predictions were that individualswithreducedWMcapacitywouldbemorehampered in actively suppressing the competing irrelevant information.
Therefore,lowspanparticipantswouldbemorehamperedthan highspanonesintheinterferencecondition.Similarly,olderadults would,onaverage,bemorehamperedthanyoungeradultsinthe lattercondition.Finally,olderlowspanindividuals,whodemon- stratethe lowest WMcapacity, wereexpected tobe themost hampered when the task requires an active suppression of irrelevantresponses.
2. Subjectsandmethods 2.1. Participants
Theinitialsampleconsistedin241youngand161olderadults, all French speakers. Younger participants aged 18–35, were University of Geneva undergraduates who participated to the experimentaspartofcoursecredit.Olderparticipants,aged60–88, werehealthycommunitydwellingindividuals,recruitedbymeans ofnewspaperadvertisementorfromtheUniversityoftheThird AgeoftheUniversityofGeneva.Allparticipantswereindividually tested.Participantswithlessthan85%ofcorrectresponsesinthe conditionAoftheHaylingtaskwereexcluded,leaving237young and155olderadults.Thiscriterionwaschosentoensurethatthe expectedwordininitiationconditionhadbeenprovided.
Inordertoequatesamplesizesacrossagegroups,155younger adultsweresubsequentlyrandomlyselectedfromthe237ones.In eachoftheremainingsamples,thatis 155younger (meanage:
22.523.27), and 155 older adults (mean age: 69.805.88), participantswithhigherandlowerWMscoreswereidentifiedon thebasisoftheirperformanceinthereadingspantask.Atercilesplit wasappliedwithineachagegroup.Individualswithscoresfallingin theuppertercilewerequalifiedashighspansandindividualswith scoresinthelowertercilewerequalifiedaslowspans.Individuals with scores in the center tercile were discarded. The remaining sampleconsistedin52younghighspans,48younglowspans,52old highspansand52oldlowspans(Table1).ResultsoftheANOVAage
(younger, older)span (high, low) on the WM score, showed a significantmaineffectofage,F(1,200)=213.49,p<0.01,
h
2p¼0:51, of span, F(1,200)=829.23, p<0.01,h
2p¼0:81, and a significant agespaninteraction,F(1,200)=32.53,p<0.01,h
2p¼0:14.Overall, olderadultshadapoorerWMcapacitythanyoungeradults,andlow spansscoredlowerthanhighspans.Resultsofposthocpair-wise comparisonsrevealedthatyoungerhighspansscoredsignificantly higherthanallthreeothergroups.Olderhighspansscoredhigher thanbothyoungerandolderlowspans;however,olderlowspans scoredlowerthanallthreeothergroups(allcomparisonsatp<0.01).Itisworthstressingthatoldhighspansandoldlowspansdidnot differsignificantlyonchronologicalage,asindicatedbytheresultsof at-testforindependentsamples.
Aseriesof22ANOVAscomparingage(younger,older)and span(high,low)oneducationallevel andvocabularyscorewas alsoconducted(seeTable1fordescriptivestatistics).
With respect to the number of years of education, results revealedthatthemaineffectofage,F(1,200)=45.22,p<0.001,
h
2p¼0:19, of span,F(1,200)=4.62,p<0.05,h
2p¼0:02, and the interactionagespan,F(1,200)=4.62,p<0.05,h
2p¼0:02,were significant.Plannedcomparisonsindicatedthatthetwogroupsof youngadultsunderwentmoreyearsofeducationthanhighand lowspanolderadults(allcomparisonsatp<0.01).Furthermore, whereasyounghighspansdidnotdifferfromyounglowspans,old highspanshadasignificantlyhigherlevelofeducationthantheold lowspans(p<0.01).Asconcernsthevocabulary,assessedby theMillHill, partB (Deltour,1998),themaineffectofage,F(1,200)=28.18,p<0.001,
h
2p¼0:12, of span, F(1,200)=8.40, p<0.01,h
2p¼0:04, and the interactionagespan,F(1,200)=4.02, p<0.05,h
2p¼0:02,were significant. Planned comparisons revealed that old high spans presenteda significantlyhigherperformancethan botholdlow spans andthetwo groupsofyounger adults(allcomparison at p<0.01). In contrast, young high and low spans were not significantlydifferentfromeachother.2.2. Tasks
WMcapacitywasassessedbymeansofacomputerizedFrench adaptationofthereadingspantask(Delaloyeetal.,2008;Robertet al., 2009). Participants were visually presented with series of individualsentencesandhadtoprovideasemanticjudgmentupon thecontentofeachsentence,whiletemporarilymaintainingthe finalwordofeachsentence.Seriescontainedfrom2to5sentences, witheachsetdefininganitem.Atotalof16itemswereused,4at eachsetlength,foratotalof56sentences.Beforethetestphase, practiceitemswerepresented.Attheendofeachset,participants wererequiredtoorallyrecall thewordsin theoriginalorderof appearance. WMcapacity was scored by the mean number of words correctly recalled independently of the recall order computedasthetotalnumberofcorrectwordscorrectlyrecalled
Table1
Characteristicsoftheyoungandolderadults,classifiedashigh-andlow-span participants,meanS.D.
Youngadults Olderadults
High(N=52) Low(N=48) High(N=52) Low(N=52) Agea 21.812.41 23.043.50 68.735.80 70.625.02 Educ.b 16.00.0 16.000.0 14.383.71 12.873.01 Vocab.c 37.032.73 36.642.74 40.312.57 38.124.37 RSpan.d 3.370.06 2.680.31 3.110.16 2.070.54
aAgeinyears.
b Numberofyearsofformaleducation.
cVocabularyperformanceassessedbytheMillHillPartB.
d Performance at the readingspan test (mean number of words correctly recalled).
E.Borellaetal./ArchivesofGerontologyandGeriatrics53(2011)e75–e80 e76
(max=56)dividedbythenumberofitems(16).Theuseofthis scorewasmotivatedbypreviousfindingsdemonstratingthatnot considering theoriginal recall order yields better distributions (e.g.,Borellaetal.,2006).
ResistancetointerferencewasassessedbymeansofaFrench computerized adaptation of the Burgess and Shallice (1996) Haylingtask(seeBorellaetal.,2009).Participantswereprovided withlistsofhigh-clozesentencestobecompletedeitherwithan expectedword(initiationcondition,A)orwithawordproviding no meaning to the sentence, but grammatically correct with respecttocaseandgender(interferencecondition,A0).Froma largerpoolofsentences,30wereselectedbasedonapilotstudy conductedona large sampleof young adults. Sentenceswere chosen on the basis of the probability of completion by the expectedword(between0.97and0.99).Thenumberofsyllables, thestartingphonemesandthelengthofthetargetwordwere furthertakenintoaccount.Intheinterferenceconditionconsid- ered,thesentencesprovidedatinitiationwerepresentedasecond time.The presentadaptation ofthe Haylingtask entailedtwo interference conditions: one in which new sentences were presented(asintheoriginalBurgessandShallice,1996,taskB), andoneinwhichthesentencesusedatinitiationweredisplayeda second time (A0). Because preliminary analyses showed no significantdifferencesbetweenthesetwoconditions,weretained one,theA0,forsubsequentanalyses.Thechoicewasdrivenbythe fact that using the same sentences at interference than at initiationensuresthatthewordstobeinhibitedwereactually previouslyactivated(seeBorellaetal.,2009).
Asinthe originalversionof thetask(BurgessandShallice, 1996) each condition entailed 15 sentences individually pre- sented on a computer screen. Within each condition, the sentencesappearedinarandomorderforeachparticipant.The presentation order of each conditionwas fixed, andinitiation alwaysprecededinterference.Apracticephase(5sentences)was presentedbeforeeachtestcondition.Participantswereinstructed tosilentlyreadeachsentenceandtocompletethesentencewith the appropriate, expected word that fitted the sentences (initiation, e.g., Les abeilles produisent du ...MIEL - Bees produce...HONEY),orbysyntacticallycorrect,butsemantically non-plausibleword (interference, e.g.: Lesabeilles produisent
du...SABLE-Beesproduce...SAND).Presentationwasself-paced
andeachresponsetriggeredthefollowingtrial.Responseswere providedorally.Theexperimentermanuallyrecordedaccuracy, whereas the computer recorded the response time latencies, whichcorrespondedtotheintervalbetweenthesentencedisplay andtheparticipant’soralresponseinthevoicekey.
2.3. Statisticalanalyses
Thedependent variablesconsideredweretheproportionof correct completions (expected words for the phase A and unexpected words for the phaseA0,each dividedby thetotal numberofsentences),andtheaverageresponse times(inms) acrosstrials,computedonresponsetimesforcorrectresponses only.Ininterference, correctwordscorresponded toanyword thatfulfilled the case andgender agreement provided by the context,butneitherwassemanticallyrelatedorrelevanttothe sentence,normadeany referencetothesentence.
Additionally,foreachvariable,aninterferencecostindexwas computedas thedifferencein performance betweeninitiation and interference, weighted by performance at initiation. For correctresponses,thescorewascomputedas[(AA0)/A],andas [(A0A)/A] for response times.Inboth cases,a positive score indicates a cognitive cost associated with interference. These scoresallowedcontrollingforindividualand/orage-differencesin baselineperformance(Borellaet al.,2009).
3. Results
3.1. Proportionofcorrectresponses
FromtheentriesinTable2,andwithrespecttotheproportion ofcorrectresponses,onecanseethatforallgroups,performance was at ceiling in initiation condition. This effect wasexpected becausehigh-clozesentenceswerepurposivelybuilttoholdavery highprobabilityofcorrectcompletion.
Further,forallgroups,performancedecreasedininterference condition. Given the ceiling effect in the initiation condition, inferential statistics were conducted only on the cost index computedas[(AA0)/A]forcorrectresponses,andas[(A0A)/A]
forresponsetimes.
A22ANOVAswithage(younger,older)andspan(high,low) asbetweensubjectfactorswasconducted.Allanalyseswerealso conductedusingeithereducationallevelorMillHillperformance asacovariate,buttheresultswerenotaffected.Resultsrevealeda significantmaineffectofage,F(1,200)=18.58,p<0.01,
h
2p¼0:09, and a significant main effect of span, F(1,200)=7.46, p<0.01,h
2p¼0:04.Theseresultssuggestthatolderadultsdemonstratea highercost(0.200.14)thanyoungerones(0.130.12)inresisting tointerference,andthatlowspans(0.190.14)aremorehampered than highspans (0.140.13).Theagespaninteraction wasnot significant,F(1,200)=0.20,p=0.65.Itisalsoworthmentioninghere thatthemajorityofparticipantsshowedacostassociatedwiththe interferencescondition:75%oftheyoungerhighspans,80%ofthe youngerlowspans,83%oftheolderhighspans,and93%oftheolder lowspans.Theremainingoftheparticipantswasnotaffectedbythe condition (interference score equal to 0), and no participant demonstrated a gain associated with the interference condition (negativescore).3.2. Responsetimes
Turningtoresponsetimesrestrictedtocorrectresponses,the datadisplayedinTable2suggestthattheinterferenceconditionis associated with larger response times than the initiation one.
Inferentialstatisticswereconductedonthisvariableusingamixed design22ANOVAwithage(younger,older)andspan(high,low) asbetweensubjectfactors,andcondition(initiation,interference) asrepeatedmeasure.
Results revealed a significant main effect of condition, F(1,200)=1532.30, p<0.01,
h
2p¼0:88, indicating that response timeswerelonger in interferencecondition (3675738ms),as compared to initiation (1891362ms, mean difference=M- diff.=1780ms).ResultsoftheANOVAfurtherrevealedasignificant spanconditioninteraction,F(1,200)=6.67,p<0.05,h
2p¼0:03,and a significant agespan interaction, F(1,200)=5.33, p<0.05,h
2p¼0:03.Themaineffectsofage,F(1,200)=1.54,p=0.11,andof span,F(1,200)=1.42,p=0.23,werenotsignificant.Neitherwerethe agecondition interaction, F(1,200)=0.10, p=0.75, nor the agespancondition interaction, F(1,200)=0.02, p=0.89. TheTable2
ResultsoftheHaylingtestbyageandspan,meanS.D.
Youngeradults Olderadults
Highspan Lowspan Highspan Lowspan Proportionofcorrectresponses
Initiation(A) 0.980.04 0.970.05 0.980.04 0.970.04 Interference(A0) 0.870.11 0.830.13 0.800.14 0.740.14 Cost 0.110.09 0.150.13 0.160.14 0.230.14 Responsetimes(inms)
Initiation(A) 1827380 1866330 1760324 2112317 Interference(A0) 3704759 3521587 3679868 3783653 Cost 1.070.46 0.930.39 1.110.44 0.810.31
twosignificantinteractionswerefurtherdecomposedusingplanned comparisons.Withrespecttothespanconditioninteraction,the resultsshowedthattheconditioneffectwassignificantforboththe highspans (p<0.01)andthelowspans(p<0.01).High andlow spanssignificantlydifferedintheinitiation(p<0.01,Mdiff.=201ms infavor ofthe highspans),but not inthe interferencecondition (p=0.70, Mdiff.=34ms). Thus, overall, high spans were faster (1793353ms)thanlowspans(1994345ms)atprovidingthe appropriatewordatinitiation(Fig. 1).Concerningtheagespan interaction,theresultsdemonstratedthathighandlowspansdidnot significantlydiffer(p=0.97,Mdiff.=73ms)intheyoungersample.In the older group, the difference was significant (p=0.05, Mdiff.=229ms),infavorofthehighspans.Youngerandolderhigh spansdidnotsignificantlydiffer(p=0.89,Mdiff.=47ms),whilefor the low spans, the difference between younger and older was marginally significant in favor of the younger (p=0.08, Mdiff.=255ms).
Asfortheproportionofcorrectresponses,inferentialstatistics were conducted on the interference cost index computed for responses times by means of an agespan between-subject ANOVA. Results showed only a significant main effect of span, F(1,200)=14.83, p<0.01,
h
2p¼0:07,demonstrating that overall, low spans (0.870.35) were less affected than high spans (1.090.452, Mdiff.=0.22). Neither the main effect of age, F(1,200)=0.46, p=0.50, nor the agespan interaction, F(1,200)=1.84,p=0.18,weresignificant.4. Discussion
The purpose of the present studywas toexamine the joint accountofageandindividualdifferencesinWMcapacityonthe abilitytoinhibitirrelevantresponses.Highandlowspanyounger and older adults were compared on the Hayling task, a task requiring completing high-cloze sentences either with a word semantically prompted by the sentence (initiation) or by a syntacticallycorrectwordthatmadenomeaningtothesentence (interference).Thereby,intheinterferencecondition,prepotentor predominantresponseactivationmustbeactivelysuppressedin orderto givea correct response. Itis worth pointing out that, contrarytotheclassicalversionofthetask,thesamesentences werepresentedbothatinitiationandatinterference.Thispointis
quitecrucialbecauseitensuresthattheto-be-inhibitedwordsat interferencehavebeenactuallypreviouslyactivatedatinitiation (Borella et al., 2009). Hence, only thesentences for which the expectedwordwasprovidedatinitiation,andtherebyactivated, werefurtherconsideredforanalysisatinterference.
Resultsfirstrevealedasignificantconditioneffectonresponse times,reflectinganincreaseinresponselatenciesatinterference,as comparedtoinitiation.Thisincreaseisattributedtotheadditional requirementofactivelyinhibitingtheprepotentresponsesemanti- callytriggeredbythesentence.Thus,asexpected,andinlinewith previousfindings(Bellevilleetal.,2006),performancewasalteredas a function of the task manipulation. Second, span effects were reported on the cost scores both for the proportion of correct responsesandfortheresponsetimes,althoughdifferentpatterns wereobserved.Forthecorrectresponses,resultsrevealedthatlow spans demonstrated higher interferencescores than high spans therebydemonstratingalargersusceptibilitytointerference,and lessefficientinhibitorymechanisms.Thisfindingreplicatesprevi- ousresultsshowingthatlowspansaremorehamperedthanhigh spanswhenirrelevantresponsesmustbeinhibited(KaneandEngle, 2003).
Withrespecttoresponselatencies,alargerinterferencecostwas foundforthehighspans,ascomparedtothelowspans.Thispattern ofresultcanbeexplainedbylookingdirectlyatthemedianresponse times:whilehighspans werefasterthanlowspans atcorrectly completing the sentence at initiation, both groups did not significantlydifferatinterference.Hence,thedifferenceinresponse latencieswasrelativelylargerforhighspansthanforlowspans, whichisalsoreflectedbyalargerinterferencecost.Thedifferencein response times at initiation in favor of the high spans may be associatedwiththefactthathighspanshavemorepotentialwords available inWMthanlow spans(Cantorand Engle,1993). High spansmayeitherretrievemorequicklytheexpectedword,activate itmoreefficientlyand/ormayhavealargerspreadofactivation across multiple exemplars semantically cued by the context (Buntinget al.,2004).Highspansmayalsocallpreferentiallyupon controlled attention while low spans mainly rely on automatic processes(RosenandEngle,1997).Consequently,atinitiation,high spanscouldmorerapidlyfocusontheappropriatewordthanlow spans.Inanycase,higheractivationatinitiationcouldfurtherresult in a relatively larger slowdown in processing at interference.
Comparedtolowspans,highspansmightneedrelativelymoretime toachievetheactivesuppressionoftheno-longerrelevantwords, preciselybecausethesewordswerepreviouslyinahigheractivation state.Further,andprovidedthatlowspansrelymoreonautomatic processing,timedperformancesoftheseparticipantsmaybeless affectedby thetask manipulationthanthe performanceofhigh spans(e.g.,RosenandEngle,1997).Asacounterpart,however,low spansshoulddemonstratealowerproportionofcorrectresponsesat interference,ascomparedtohighspans.Thiswasactuallythecasein thepresentstudy,asreflectedbyasignificantspaneffectonthe interferencecost.
Turningtotheageeffects,resultsrevealedthatoverallolder adultsweremorehamperedthanyoungeronesatprovidingatask appropriateresponseatinterference.Indeed,theinterferencecost associatedwiththetaskrequirementsfortheproportionofcorrect responseswaslargerforolderadultsthanforyoungerones.This findingisinlinewithpreviousstudiesdemonstratingthatolder adultshavedifficultiesintasksrequiringinhibitingautomaticbut inadequateandnolongerappropriateresponses(e.g.,Hasherand Zacks, 1988). Our results, though, are in contrast with those reportedbyBellevilleet al.(2006)whosurprisinglyfailedtofind age-relateddifferencesonerrorscores.However,andcontraryto ourexpectations,olderadultswerenotsignificantlyslowerthan youngeradultstoprovideacorrectresponse,eitheratinitiation,or atinterference.Thisfindingappearscontradictorywiththewell Fig.1.Responsetimesbyspanandcondition(aggregatedmeansandstandard
errors)intheHaylingtest.
E.Borellaetal./ArchivesofGerontologyandGeriatrics53(2011)e75–e80 e78
establishedage-relatedreductionin processingspeed (seeSalt- house,1996,forareview).However,agedifferencesonresponse timesareusuallythemostprominentonsmalltimeintervalsand tendtodecreaseatlargertimeintervals,probablybecauseofthe associated increase in response time variability (Bherer and Belleville,2004).Inthepresentstudy,responsetimeswerefairly largeanddemonstratedlargeinter-individualvariability,particu- larlyintheinterferencecondition.Thismaybeduetothefactthat, contrarytootherstudies(Burgess andShallice,1996; Belleville etal.,2006),readingtimeswereincludedinthetimemeasures, andnotimelimitswereimposedtoprovidearesponse.Indeed, whenreported,agedifferencesinresponsetimesweremeasured from the offset of the sentence, and time constraints of 30s (Bellevilleetal.,2006)or60s(BurgessandShallice,1996)were imposedtocompletethesentences.Suchproceduresmainlyallow focusingmore directlyon theprocessesofinterest byavoiding confounding sources of variability as reading times could be.
Probably associated with this response time issue is also the absenceofasignificantageconditioninteractiononrawtime measures,andofasignificantageeffectontheinterferencecost score.Furtherworkisthusneededtoinvestigatetheembedded effectsofreadingandnamingtimesindifferentagepopulations.
Finally,contrarytoourinitialhypothesis,olderlowspanswere not more hampered than all three other groups. Indeed, the agespancondition interaction was not significant on raw response times, and neitherwere the agespaninteraction on costindicesforbothvariables.Hence,aboveandbeyondageand spaneffectsontheproportionofcorrectresponses,andspaneffects ontheresponsetimes,ourfindingsdonotpointtoamultiplicative accountofage andindividualdifferencesinWMcapacityinthe abilitytoresistinterferenceinthepresentversionoftheHayling task.Notwithstanding,thisissueremainstobefurtherinvestigated.
Firstbecauseresultsdemonstratedthatontheproportionofcorrect responses, the frequency of participants that showed a cost associated with the interference condition (i.e., score>0) was descriptivelyhigherforolderlow spansthan forallthreeother groups.Second,theanalysisoftherawresponsetimesrevealeda significantagespaninteraction,suggestingthattherewasatrend fortheolderlowspanstorespondslowerthanallthreeothergroups.
Third, because the methodology employed deserves further consideration despite of partially satisfying results. To our knowledge, only few studies in the cognitive aging literature appliedamethodologybasedonanextremegroupdesign,andnone ofthemidentifiedthehighandlowspanyoungerandolderadultsin exactlythesamewaythaninthepresentstudy.Forinstance,some studiesusedamediansplit,ratherthanatercilesplit,toidentify individualswithhighandlowWMspan(KemtesandKemper,1997, 1999).Othersusedaprocedureinwhichhighandlowspanyounger individualswereselectedbymatchingtheirWMscoreswiththe onesofhighandlowolderadults(Kemperet al.,2004,Experiment 2).Finally,otherresearchersinterestedinthisissue,onlypartially consideredjointly age and individual differences in WM, either comparing high and low span younger adults to older adults (Oberauer,2005),orthereverse,thatishighandlowspanolder adultscomparedtoyoungerones(Smileret al.,2003).Inbothcase, andconsideringthemethodologicalrecommendationsformulated byConwayet al.(2005),methodologicalprecautionswereoften only partially taken to reliably identify high and low spans. It remainshowever,that qualitativedifferences inprocessing may existacrossyoungerandolderadultsintheWMtaskitself(e.g.,Li andLindenberger,2002).
5. Conclusion
In summary, the results of the present study highlight the crucial role of controlled attentionin tasks in whichirrelevant
information needs to be suppressed (Engle and Kane, 2004).
IndividualdifferencesinWMcapacitywereindeedassociatedwith differences in the abilityto resist interference. Together,these findings support the idea that individual differences in WM capacity mayberelatedtoindividualdifferencesinthetype of processes engaged preferentially in the task, with high spans relying mainly upon controlled processes, and low spans upon automatic ones. Further, our findings demonstrated that, with respecttoresponseaccuracy,olderadultsshowedalowerability toresistinterferencethanyoungerones,whichwasinlinewiththe hypothesis of an age-related decline in inhibitory processes (HasherandZacks,1988;Borellaetal.,2008,2009).However,no agedifferenceswerereportedwithrespecttoresponsetimes,and contrary to our expectations, older low spans were not more hamperedthanallothergroups.
Although the present study yielded unexpected findings, it provided interesting highlights for considering jointly age and individual differencesinWMcapacity whenaccounting forthe abilitytoresistinterference.Ithaspointedtoamethodologythat hasprovenveryconstructiveinthestudyofindividualdifferences inWMcapacityinyoungeradults,butwasrarelyextendedinthe fieldofcognitiveaging.Futureworkusingsuchamethodologywill undoubtedly help disentangling both sources of differences in cognition,andprovideaninterestingmeantobetterunderstand individual differencesin cognitive aging,not only in inhibitory tasks, butin all tasksreported todemonstrateanaverage age- relateddecline.
Conflictofintereststatement None.
Acknowledgements
Thestudyreportedinthepresentpaperwaspartofalarger study financially supported by two successivegrants from the SwissNationalResearchFoundationtoAnikdeRibaupierre(grants 1213-065020and1114-052565).
References
Belleville,S.,Rouleau,N.,vanderLinden,M.,2006.UseoftheHaylingtaskto measureinhibitionofprepotentresponsesinnormalagingandAlzheimer’s disease.BrainCogn.62,113–199.
Bherer,L.,Belleville,S.,2004.Age-relateddifferencesinresponsepreparation:the roleoftimeuncertainty.J.Gerontol.B:Psychol.Sci.Soc.Sci.59,66–74.
Borella,E.,Carretti,B.,Mammarella,I.C.,2006.Doworkingmemoryandsuscepti- bilitytointerferencepredictindividualdifferencesinfluidintelligence?Eur.J.
Cogn.Psychol.18,51–69.
Borella,E.,Carretti,B.,DeBeni,R.,2008.Workingmemoryandinhibitionacrossthe adultlife-span.ActaPsychol.(Amst.)128,33–44.
Borella,E.,Delaloye,C.,Lecerf,T.,Renaud,O.,deRibaupierre,A.,2009.Doage differencesbetweenyoungandolderadultsininhibitorytasksdependonthe degreeofactivationofinformation? Eur.J.Cogn.Psychol.21,445–472.
Bunting,M.F.,Conway,A.R.A.,Heitz,R.P.,2004.Individualdifferencesinthefan effectandworkingmemorycapacity.J.Mem.Lang.51,604–622.
Burgess,P.W.,Shallice,T.,1996.Responsesuppression,initiationandstrategyuse followingfrontallobelesions.Neuropsychologia34,263–273.
Cantor,J.,Engle,R.W.,1993.Workingmemorycapacityaslong-termmemory activation:anindividualdifferencesapproach.J.Exp.Psychol.:Learn.Mem.
Cogn.19,1101–1114.
Conway,A.R.A.,Kane,M.J.,Bunting,M.F.,Hambrick,D.Z.,Wilhelm,O.,Engle,R.W., 2005.Workingmemoryspantasks:amethodologicalreviewanduser’sguide.
Psychonom.Bull.Rev.12,769–786.
Craik,F.I.M.,Byrd,M.,1982.Agingandcognitivedeficits:theroleofattentional resources.In:Craik,F.I.M.,Trehub,S.(Eds.),AgingandCognitiveProcesses.
PlenumPress,NewYork,NY,USA, pp.191–211.
deRibaupierre,A.,2000.Workingmemoryandattentionalcontrol.In:Perrig, W.J., Grob,A. (Eds.), ControlofHuman Behavior,MentalProcesses,and Consciousness:EssaysinHonorofthe60thBirthdayofAugustFlammer.
LawrenceErlbaumAssociates,Mahwah,NJ,USA, pp.147–164.
Deltour,J.J.,1998.EchelledeVocabulaireMillHilldeJ.C.Raven.Paris:Editionset ApplicationsPsychologiques.
Delaloye,C.,Ludwig,C.,Borella,E.,Chicherio,C.,deRibaupierre,A.,2008.L’Empan delecturecommee´preuvemesurantlacapacite´ deme´moiredetravail:normes base´essurunepopulationfrancophonede775adultesjeunesetaˆge´s.Eur.Rev.
Appl.Psychol.58,89–103(inFrench).
Engle,R.W.,2001.Whatisworkingmemorycapacity? In:Roediger,H.L.I.,Nairne, J.S.,Neath,I.,Surprenant,A.M.(Eds.),TheNatureofRemembering:Essaysin HonorofRobertG.Crowder.AmericanPsychologicalAssociation,Washington, DC,USA, pp.297–314.
Engle,R.W.,Kane,M.J.,2004.Executiveattention,workingmemorycapacity,and atwo-factortheoryofcognitivecontrol.In:Ross,B.H.(Ed.), ThePsychology ofLearningandMotivation,vol.44.Elsevier,NewYork,NY,USA,pp.145–199.
Hasher,L.,Zacks,R.T.,1988.Workingmemory,comprehensionandaging:areview anda newview. In:Bower,G.H. (Ed.), The Psychologyof Learningand Motivation,vol.22.AcademicPress,NewYork,pp.193–225.
Kane,M.J.,Engle,R.W.,2003.Working-memorycapacityandthecontrolofatten- tion:thecontributionsofgoalneglect,responsecompetition,andtasksetto Stroopinterference.J.Exp.Psychol.Gen.132,47–70.
Kane,M.J.,Hambrick,D.Z.,Tuholski,S.W.,Wilhelm,O.,Payne,T.W.,Engle,R.W., 2004.Thegeneralityofworkingmemorycapacity:alatent-variableapproachto verbalandvisuospatialmemoryspanandreasoning.J.Exp.Psychol.Gen.133, 189–217.
Kemper,S.,Crow,A.,Kemtes,K.A.,2004.Eye-fixationpatternsofhigh-andlow-span youngandolderadults:downthegardenpathandbackagain.Psychol.Aging 19,157–170.
Kemtes,K.A.,Kemper,S.,1997.Youngerandolderadults’on-lineprocessingof syntacticallyambiguoussentences.Psychol.Aging12,362–371.
Kemtes,K.A.,Kemper,S.,1999.Agingandresolutionofquantifierscopeeffects.J.
Gerontol.B:Psychol.Sci.Soc.Sci.54B,P350–P360.
Li,K.Z.H.,Lindenberger,U.,2002.Relationsbetweenagingsensory/sensorimotor andcognitivefunctions.Neurosci.Biobehav.lRev.26,777–783.
McDowd,J.M.,Oseas-Kreger,D.M.,Filion,D.L.,1995.Inhibitoryprocessesincogni- tionandaging.In:Dempster,F.N.,Brainerd,C.J.(Eds.),InterferenceandInhibi- tioninCognition.AcademicPress,SanDiego,CA,USA, pp.363–400.
Miyake,A.,Shah,P.,1999. ModelsofWorkingMemory:MechanismsofActive MaintenanceandExecutiveControl.CambridgeUniversityPress,NewYork, NY,USA.
Oberauer,K.,2005.Bindingandinhibitioninworkingmemory:individualandage differencesinshort-termrecognition.J.Exp.Psychol.Gen.134,368–387.
Robert,C.,Borella,E.,Fagot,D.,Lecerf,T.,deRibaupierre,A.,2009.Workingmemory andinhibitorycontrolacrossthelifespan:intrusionerrorsinthereadingspan test.Mem.Cogn.37,336–345.
Rosen,V.M.,Engle,R.W.,1997.Theroleofworkingmemorycapacityinretrieval.J.
Exp.Psychol.Gen.126,211–227.
Salthouse,T.A.,1996.Theprocessing-speedtheoryofadultagedifferences in cognition.Psychol.Rev.103,403–428.
Smiler,A.P.,Gagne,D.D.,Stine-Morrow,E.A.L., 2003.Aging,memoryload,and resourceallocationduringreading.Psychol.Aging18,203–209.
E.Borellaetal./ArchivesofGerontologyandGeriatrics53(2011)e75–e80 e80