Developmenlal Psychology 1990. Vol. 26. No. 5. 845-854 Copyright1990by the AmericanPsychological Association, Inc. 0012- l 649190/$00.75 Effectsof Age and Naturally Occurring Experience on SpatialVisualizationPerformance Timothy A. Salthouseand DeboraR. D. Mitchell Georgia Institute of Technology A questionnaire designedto assess experience with activitiespresumedto requirespatialvisualization abilities,and psychometrictestsof theseabilities,wereadministeredto 383 adultsranging from 20 to 83 yearsofage. Although researchparticipantsvariedconsiderably in the amountof self-reported experience, statisticalcontrolofexperienceresultedin relativelymodestattenuations ofthe relationsbetweenageand spatialvisualizationperformance. Thesefindingsseeminconsistentwith a strongdisuseinterpretationofcognitiveagingphenomena andsuggest thatat leastsome age-related differencesin cognitivefunctioningare independentofthe amountofexperiencewith relevantactivities. One o[the most popular hypothesesproposed to account for the age-relateddeclines observedwith certain measuresofcognitive functioning attributes those declines to various forms of disuse or lack of practice. Although seldom articulated as an explicit theory, the following sample of quotations illustrates that this perspectivehas been implicitly acceptedfor more than half a century. Abilitytasksthatarecommonly usedin everyday lifetendto be insensitive to age.(Birren,Cunningham, & Yamamoto, 1983, p.552) morestrongly to theecological niches thatthe IW]hentasksrelate olderpersoninhabits, age-related deficits arelessprominent. (Charness, 1985,p. 226) A decrease in testability amongadultsis probablycausedby the fact that adults,as they grow older,exercisetheir minds lessand lesswith the materialsfound in psychologicaltests.(Sorenson, 1 9 3 3p, . 7 3 6 ) The"losses" arein largemeasure.. . a by-productofdisuse. . . old age actsselectivelyand most decidedlyon those functions which have sufferedfor want of practice. (Sward, 1945, pp. 478-4'79) [ ] n one'sown fieldwhereexperiencehasbeenaccumulatingover a periodof manyyears,thereis little evidencefor anydeclinewith the years,at leastuntil extremeold ageis reached.(Gilbert, 1952, p.130) Thosewho havespenttheir livesworking with their handsand interpretingperceptualdata retain the ability to deal with perceptual and constructionalproblems.(Williams, 1960, pp. 217-218) oftheorganismwithinhisownenviron[S]tudiesofthe functions mentshowrelativelysmall age-related differences or changesand in manycasesadvancingageis correlatedwith improvement. (Fozard& Thomas,1975.p. I l7) [T]he declinesthat are observedin abilitieswhich are usedfrequentlyappearto beginat a laterageandto belessdrasticthanare (Denthe declinesin abilitieswhichareexercisedlessfrequently. n e y , 1 9 8 2p ,. 8 2 4 ) This researchwassupportedby NationalInstituteon Aging Grant AG006858to Timothy A. Salthouse. We would like to thank ReinholdKliegl for valuablecommentson an earlierdraft of this articleand E. Breitmeister, A. Fann,H. Yanous. and M. Yanousfor assistance in data analysis. Correspondence concerningthis articleshouldbe addressedto Timothy A. Salthouse, SchoolofPsychology, GeorgiaInstituteofTechnology,Atlanta,Georgia30332. 845 An importantcategoryofresearchrelevantto the disusehypothesishasinvolvedcomparisonsacrosspeoplepresumedto differ in the natureand extentof their experiences. Research within this categoryhas varied with respectto whetherthe focuson the individual'sexperience and cognitiveperformance has been broad or narrow. Studies with a broad focus have attemptedto relatecharacterizations of the individual'sgeneral activity level(e.g.,Arbuckle,Gold, & Andres,1986;DeCarlo, 1974;Schooler,1984)or his or her self-assessed cognitivedemands(e.g.,Owens,1953;Schwartzman, Gold, Andres,Arbuckle,& Chaikelson,I 987)eitherto a varietyof miscellaneous cognitivemeasuresor to a compositescoreof generalintelligence.Most of thesestudieshavereportedratherweak relationsbetweenexperience and cognitivefunctioning.For example, the semipartialcorrelationbetweena measureof the frequencyof 23 activitiesand a compositemeasureof intelligence in the Schwartzman et al. (1987)studywasonly . 13. Althoughnot without value,studieswith a broadfocussuffer from two problemsassociated with the grossness of the categorization of both the experienceand the cognitionconstructs. One problem is that it is dilficult to rule out the influenceof potentiallyconfoundingthird variables(suchas healthstatus) whenthe evaluations neitherofexperiencenor ofcognitionare very specific.A secondproblem is that the relationsbetween experience andcognitionarelikely to be quiteweakwhenthose constructsareassessed in verygeneralterms.That is,thegreatesteffectsofexperiencewill probablybe evidentbetweenspecific measuresof cognition and particular frequently performedactivities,ratherthan betweenglobalmeasures of cognition or generalintelligenceand grosscategorizations of experience. One meansof achievingcloserlinkagesbetweenexperience and cognition is to rely on samplescomprisingmembersof 846 TIMOTHY A. SALTHOUSE AND DEBORA R. D. MITCHELL particular occupationalgroups and to investigateage-related effectson occupationallyrelevantmeasuresofcognitive performance.Perhapstheearliest,and almostcertainlythe largest,of the occupation-specific studiesrelevantto aginginvolveda battery of perceptualand cognitivetestsadministeredto 544 atrcrewofficers(Glanzer& Glaser,1959;Glanzer,Glaser,& Richlin, 1958).The statedpurposesof this projectwereto "measure the skills requiredfor performanceof aircrewofficers[andto] measure the effects of aging upon skilled performance" (Glanzer& Glaser,1959,p. 89). Unfortunately, the assessment ofage-relatedeffectswas not very powerful becauseofa relativelynarrowrangeof ages,with only 14,or lessthan 37o,of the researchparticipantsoverthe ageof40. Despitethis restricted age range, significant negativecorrelations between age and performancewerereportedon 8 of the l4 tests.Furthermore, the largestageeffectswereevidenton a test with the highest facevalidity as a measureof pilot skill. This wasa test titled InstrumentComprehension, in whichthe examineeis required to integrateinformationfrom a compassand an artificialhorizon to indicatethe currentpositionofan airplane.Not only was thesimpleagecorrelationwith this measurestatisticallysignificant (r : -.33), but it wasonly slightlyattenuated(to r = -.24) after statisticallycontrollingthe presumablyrelevantvariable of total numberof hoursof flyingexperience. A similar finding of significantage-related cognitivedifferencesfavoringyoungeradultswithin a sampleof adultsfor whom the relevantabilitiescan be assumedto havebeen in continuoususewas recentlyreportedby Salthouse,Babcock, Skovronek,Mitchell, and Palmon(1990).Most of the participantsin this projectwerearchitects, of cogniand the measures tiveperformance consisted ofscoreson testsofspatialvisualization. The 47 architectsin Study3 ofthat reportrangedfrom 2 l to 7l yearsof age,with a correlationof .97 betweenageand numberof yearsof usingspatialvisualizationabilitiesin one's job. Although it seemsreasonable to assumethat all of these practicingarchitectshad extensive, and nearlycontinuous,use of spatialvisualizationabilities,highly significantage-related of -.69, -.71, and -.47) were declines(i.e.,agecorrelations observedin three measuresofspatial visualizationperformance. projectsjust deThe resultsof the two occupation-specific providing in rather discouragscribedare thereforeconsistent of cognitiveaging.Obing evidencelor the disuseperspective jectionscan be raisedagainsteachof thesestudies,howeveq and it is thus desirableto replicatethe major resultsbefore reachinga definitiveconclusionregardingthe disuseinterprethe strategyof examiningagetrendsin tation. Unfortunately, samplescomprisingmembersof a particularprofessionor occupation is hamperedby the difficulty of recruitingappropriateresearchparticipants.To illustrate,in our recentstudyof architects,more than 1,100lettersweremailed to nearlyall of the membersof the American Instituteof Architectsprofessionalorganizationresidingin a largemetropolitanarea,and approximately400 of theseindividualswerelatertelephonedto makeadditionalappealsfor participation.Ultimately,however, recruitedto particionly about60 individualsweresuccessfully patein the two relevantstudies.Furthermore,it wasimpossible to determine whether the architectswho participated in the project were representativeof the larger population of architects. A different researchstrategywasusedin our presentproject by recruiting participantsfrom the generalpopulation and then administering a questionnaireto evaluatethe extent of eachindividual'sexperiencewith different activitiespresumed to requirespatialvisualizationability. Three types of informain thequestionnaire to assess recentexperition wererequested ence,cumulativeexperience,and subjectiveability.The two categoriesofexperienceweredistinguishedto allow investigation ofage relationswith both the currentfrequencyand the accumulatedfrequencyof activitiespresumedto be relevantto spatial visualization.Informationaboutboth kinds of experience is desirablebecause,althoughproponentsof the disuseperspectivegenerallyargue that increasedage is associatedwith with relevantactivities,the lesseramountsof recentexperience cumulativeexperience of an individualmayactuallybe greater with increasedage.Ratingsof subjectiveability wereincluded becausepeoplewho spendconsiderable time performinga given activity might be assumedto havehigher perceptionsof their levelof ability in that activity than peoplewho devote relativelylittle time to the activity.In this respectthe subjective ability ratingsmayproveusefulin evaluatingthe validityof the experienceinformation. questionnaire, In addition to the experience six cognitive testswerealso administeredto all researchparticipants:two spatialvisualizationability,two designedto designedto assess assess the closelyrelatedcognitiveability of inductivereasonthe presumably ing,andtwo designedto assess unrelatedcognitive ability of perceptualspeed.The purposeof the testsof inductivereasoningand perceptualspeedwasto providea further checkon the validityof the informationobtainedfrom the to the experiexperiencequestionnaire.That is, if responses encequestionnaireare accurateindicationsof the amount of experience eachindividualhashadwith explicitlyspatialactivities,then a gradationin the magnitudeof the correlationsbeand the measures of cognitweenthe questionnaireresponses tive performancewould be expected,with the highestcorrelalowercorrelations tions for the spatialvisualizationmeasures, and the lowestcorrelafor the inductivereasoningmeasures, tions for the perceptualspeedmeasures. The primary questionsinvestigatedin the project were whetherexperiencewith activitiesrequiringspatialvisualization ability either mediatesor moderatesage-relateddifferencesin measuresof spatialvisualizationperformance.The mediationpositionwould be supportedif there is little or no effectassociatedwith ageafter statisticallycontrolling the influenceof variablesreflectingamount of relevantexperience.A somewhatweaker hypothesisis that differentialexperience doesnot mediatethe effectsrelatedto agebut insteadmoderatesthoseeffectssuchthat the age-relatedinfluencesare smallestamongindividualswith the greatestamountof experience. thereThe specificpredictionfrom the moderationperspective, fore, is that the ageand experiencevariableswill haveinteractive effectson measuresof spatial visualizationperformance. Method Subjects A total of383 adultsbetween20 and 83 yearsofage contributed valid datato the project.The data from 5 additionalindividualswere AGING AND SPATIAL EXPERIENCE consideredinvalidand werediscardedprior to analyses becausethese participantshad difficulty understandingthe test materialsor be_ causethey arrived at the testingsessionin an obviouslyinebriated state.All participantswererecruitedfrom newspaper advertisements and weretestedin smallgroups.Participantsconsistedofl g6 menand 197women,wilh 20-47 individualsin eachdecade-sex groupingfrom the 20sto 70+. Eachindividualwaspaid $10for his or her parricipa_ tion in the 90-minsession Procedure The testingsessionbeganwith the researchparticipantscompleting a questionnaireintendedto assess the amountof recentand cumula_ trveexperiencethe individualhad with activitiespresumedto require spatialvisualizationabilitiesandto obtaina self-appraisalofhis or her levelofability in eachactivity.For eachofl0 activities(listedin Table 2), the individualwasaskedto (a)rate his or her ability on a 5-point scalerangingfrom | (muchaboveaverage)to 5 (muchbelowaverage),(b) estimatethe average numberofhours per monthdevotedto that activity overthe last6 months,and(c)estimatethe numberof yearsin which an average ofat leastI 5 hr per month had beendevotedto that activity. The remainderofthe testsessionwasdevotedto the performance of six cognitivetests.The tests,in theorderin whichtheywerepresented, werethe NumberComparisonTest,the paperFoldingTest,the Letter SetsTest,the AbstractionTest,theSurlaceDevelopment Test,and the Finding As Test.All but the AbstractionTest were from the Kit oi .|c.rl.r Factor-Re/brenced Cognitive (Ekstrom, French,Harman. & Dermen, 1976).The AbstractionTest was from the ShipleyInstituteof LivingScale(Shipley, 1986). The PaperFoldingand SurfaceDevelopmenttestswereintendedto assess spatialvisualizationability.The taskin the paperFoldingTestis to determinewhich patternof holeswould resultif a pieceol'paper werefoldedin the mannerillustratedand a holepunchedin the specified location.The individualis allowed3 min to completeas manyof the l0 five-alternative multiple-choiceitemsas possible.ltems in the SurfaceDevelopmentTest consistof an unfoldedand an assembled drawingof a three-dimensional object,for which the examineeis required to determinethe correspondence betweenedgesin the two drawings.Individualsareallowed6 min to completeasmanyof the 30 itemsaspossible. The LetterSetsand Abstractiontestsweredesignedto assess inductive reasoningability.The task in the LetterSetsTest is to determine whichof fivesetsof lettersis differentin somewayfrom the remaining setsof letters.The examineeis allowed7 min to performthe l5 problemsin the test.The AbstractionTest is a seriescompletrontestcontaining sequences of numbers,letters,or words that are to be completed by supplyingthe item that most naturally continuesthe sequence.The individualis permitted5 min to solvethe 20 itemson the test. The Number Comparisonand Finding As testsweredesignedto perceptualspeed.The task in the NumberComparisonTestis assess to decideas rapidlyas possiblewhethertwo numbersare the sameor different.A time limit of 90 s is providedlor examinees to completeas manyof the 48 itemsaspossible.The task in the FindingAs Testis to locateall of the wordscontainingthe letter"a" in fivecolumnsof 4l wordseach.Theexamineeis allowed2 min to detectasmanvofthe 100 targetsas possible. Results Cognitive PerfbrmanceMeasures Formostof theanalyses, performance in eachtestwassummarizedby thenumberof itemsanswered correctly rninusthe numberof itemsanswered incorrectly. Thisscoringmethodhas 847 the dual advantage ofproviding a correctionfor guessing while also increasingthe range of possiblescores.The correlation matrix illustratingthe relationsamongthesecognitiveperfbrmancemeasures and the variablesofage, sex,education,and self-reported healthstatusis displayedin lbble l. Becausethe cognitivetestswereselecteda priori to represent threedistinctabilities,and becausethe largestcorrelationwith each meastrrewas generallywith the other measurehypothesizedto representthe sameability (Tablel), compositeability scoreswerecreatedby averagingthe z scoresfrom the two relevant measures. That is, a spatialvisualizationcompositewas createdby averagingthe individual's z scoresfrom the paper Foldingand SurfaceDevelopmenttests,an inductivereasoning compositewas createdby averagingz scoresfrom the Letter Setsand Abstractiontests,and a perceptualspeedcomposite wascreatedby averagingz scoresfrom the Number Comparisonand FindingAs tests.Correlationsof thesecompositemeasureswith chronological agewere-.37 for spatialvisualization, -.27 for inductivereasoning, and -.28 for perceptualspeed(all significantat p <.0 l). Althoughconceptually distinct,thecompositemeasures werenot independent,becausethe intercorrelations were .69 betweenspatial visualizationand inductive reasoning,.33 betweenspatialvisualizationand perceptual speed,and .5 I betweeninductivereasoningand perceptual speed. Questionnaire Responses Meansand standarddeviationsof the responses to the individualquestionnaireitemsarepresented in Table2. Responses weremissingon oneor moreitemsin 40 of the questionnaires; hence,all subsequent analyses are basedon datafrom the 343 individualswith completerecords.In all cases,highernumbers reflectgreaterquantitids,with the valuesin the recentexperiencecolumn representinghours per month over the last 6 monthsand thosein the cumulativeexperiencecolumn representingyears,with an average of at leastl5 hr per month.Most of the distributionsof recentexperienceresponses werepositivelyskewed.To illustrate,all of the medians(50thpercentile values)were3 or less,whereasthe valuesat the 95thpercentile for ItemsI throughl0 were37.5,30,30,20, 20,20,40,5,10,and 4 hr per month,respectively. For all exceptthe lastthreeactivities,therefore,a considerable amountofrecentexperiencewas reportedby at leastsomeofthe researchparticipants. To reducethe numberofquestionnairevariablesfor subsequentanalyses, a principal-components analysiswasconducted on the datafrom all 30 itemsin the questionnaire. (Verysimilar resultswereobtainedwith oblique-rotation profactor-analysis cedures;hence,the structural configurationof scoresis not specilicto this particularmethodof analysis)Loadingsof the itemsin excessof .3 on the eightcomponentswith eigenvalues greaterthan l, afterorthogonalrotation,aredisplayedin Table 3. Correlationsbetweenthe componentscoresand the age,sex, spatialvisualization,inductivereasoning, and perceptualspeed variablesare displayedin Table 4. None of the correlations betweenthe componentscoresand the educationor self-reportedhealthvariableswassigniflcant(i.e.,p > .05);thus,they are not reported in Table4. The first three componentscan be interpretedas representing relativelybroador nonspecificcumulativeexperience, sub- 848 TIMOTHY A. SALTHOUSE AND DEBORA R. D. MITCHELL Table I CorrelationMatrix for PerformanceMeasuresand DemographicVariables(N : 353) Measure l. Age (years) 2. Sex" 3. Formaleducation(years) 4. Self-reportedhealthb 5. PaperFolding 6. SurfaceDevelopment 7. Abstraction 8. LetterSets 9. FindingAs 10. NumberComparison M .13 -.0t -.05 .10 .06 -.01 *.38* -.14* .26* .02 -.30* -.22* .23* -.06 .67* -.25* -.07 .30f .00 .60* 1. *.26* .01 .27* -.05 .59* .54* .69t -.17; . 13 .t4* .00 .21* .24+ .33* .44+ -.30r .17* .13 -.08 .29* .30r .36* .47, .46t 45.99 . 51 1 5 1. 4 1.99 1.69 7.23 I l.l0 7.52 27.33 22.20 SD 16.80 .50 2.52 r.09 4.30 I 1.43 5.37 4.'79 10.33 6.16 "Male= 0. female=l. ol= excellent.5= poor. *p<.01. jectiveability,and recentexperiencecomponents,becauseall of the loadingsfor eachcomponentderivefrom the sametype items.That is,ComponentI is basedexclusively of response on questions.Component responses to the cumulativeexperience 2 on responses to the self-ratedability questions,and Component3 on responses to the recentexperiencequestions. As might be expected,Table4 indicatesthat scoreson the cumulativeexperiencecomponentincreasewith age,whereas with age. thoseon the recentexperiencecomponentdecrease ability componentarenegatively correScoreson the subjective latedwith agebut positivelycorrelatedwith both spatialvisual- ization performanceand inductive reasoningperformance. These latter resultssuggestthat the overall self-appraisalsof ability havesomevalidity in that peoplewith higherself-ratings perform better than people with lower self-ratingson testsof spatialvisualizationand, to a lesserextent,alsoon testsof the closelyrelatedinductive reasoningability In contrastwith ComponentsI through3, the patternof loadingsfor the remainingcomponentsis morespecificto the particular activity being describedrather than to the type of Thesecomponentscan thereresponseinformationrequested. fore be inferred to representexperiencewith specific spatial Table 2 Meansand StandardDeviationsof Responses to Spatial ExperienceQuestionnaire(N : 343) Ability rating Item l. lmaginingdifferent arrangementsof furniture or other objects 2. Consideringhow an objeclor building would look from a differentviewingposition 3. Devisingefficientwaysof packingor loadinga box or car trunk 4. Following instructionsfor the assemblyof furniture, toys, models,and so on 5. Visualizingtraveldirections from a verbal description 6. Designingor making clothes accordingto patterns 7. Producingor interpreting technicaldrawings(e.g., blueprints)of three-dimensional objects 8. Perlormingpaper-folding activitiessuchasorigami games 9. Solvingpiece-assembly suchasjigsawpuzzles 10. Workingon spatial-manipulation puzzleslike Rubik s Cube M ,SD 3.5 0.9 Recent experience M SD Cumulative experlence M SD 36.6 7.2 21.0 4.3 8.4 3.8 0.9 6.8 17.0 5.0 8.7 3.5 l.l 5.8 t7.2 4.9 8.6 3.6 1.0 5.9 I1.6 6.2 9.4 2.7 1.3 3.6 t4.5 4.0 10.0 3.0 1.3 5.'l 18.9 3.4 '1 .'7 2.7 l.l 0.8 2.2 1.2 3.'7 3.4 0.9 1.9 4.0 3.5 7.2 2.5 0.9 0.9 3.4 t.2 4.1 AGINGAND SPATIALEXPERIENCE 849 Table3 ComponentLoadings.fiom Principal-Components Analysisof Questionnaire Responses After VarimaxOrthogonalRotation(N : 343) Item Abilr RExl CumExl Abil2 REx2 CumEx2 Abil3 REx3 CumEx3 Abi14 REx4 CumEx4 Abil5 REx5 CumEx5 Abil6 REx6 CumEx6 AbilT RExT CumExT AbilS RExS CumExS Abil9 REx9 CumEx9 Abill0 RExl0 CumExl0 Eigenvalue C2 CI C3 C4 C5 C6 c'7 C8 n- .61 .58 .72 .72 .60 .69 .77 .80 .65 .37 .43 .6t 78 . t l .52 .87 .51 52 .47 4l .'70 .56 .or+ .J-' .o-, .36 .65 .80 .87 .59 .68 .59 .s3 .36 .4t .35 .69 .79 .64 36 57 .ot) . 71 .16 .46 .55 6 .l 3 3.44 2.64 t.96 |.79 t.32 t.2t .52 .68 .71 .64 .57 .63 .44 .62 .68 .66 .68 .'79 .75 .58 .70 .73 .75 .'77 .54 .56 .69 .57 .60 .60 1.03 Not? Cl: Non-specificcumulativeexperience;C2: Non-specificsubjectiveability; C3: Non-specilic recentexperience;C4: Perspectivel C5: Clothes;C6: Puzzlesl; C7: Puzzles2l C8: Directions.Abit subjectiveability; REx = recentexperience; CumEx : cumulativeexDerience. activities. Based on the loading patterns. Components 4 through 8 have been labeled Perspective,Clothes, Puzzles I, Puzzles2, and Directions, respectively.Examination of Table 4 revealsthal, among the specilic components, only Component 4 (Perspective)and Component 8 (Directions) have significant correlations with the composite measure of spatial visualization performance. One meansof examiningthe validity of the experienceassessments is to comparethe responses of membersof occupationsassumedto requirespatialvisualizationabilitieswith the responses of the entiresample.For this purpose,the datafrom I I participantswho reportedtheir occupationsas architects, civil engineers,or interior decoratorsweregroupedtogetheq and their scoreson eachof the componentswerecomputed. Table 4 CorrelationsWith Principal Components(N : 343) Componenl Cl : Cumulativeexperience C2: Subjective ability C3: Recentexperience C4: Perspective C5:Clothes C6: PuzzlesI C7: Puzzles2 C8: Directions * p <.ol Age Sex .26* -.lg+ .06 .u .20* -.l3 .02 -.06 .01 ./)' .38* -.07 .05 -.23* Spatial visualization .08 .36+ -.04 .17* .lI -.01 .ut .17* Inductive reasoning -.tz .21" .00 .02 -.02 -.09 .l 6 * Perceptual speed -.09 .10 .00 -.03 .02 .03 -.02 .08 850 TIMOTHY A. SALIHOUSE AND DEBORA R. D. MITCHELL The mean valuesfor the individualsin this subsamplewere within onestandarddeviationof the samplemeanfor all componentsexceptComponent4, for whichtheir meancomponent scorewas2.29,with a rangeof L04 to 3.78.Component4 is the Perspective component,with primary loadingson Activities2, 4,and7 (seeTable3). Comparisonsof the estimatednumberof hours per month devotedto theseactivities revealedthat the subsample estimatesaveraged 50.9hoursper month "considering how an objector building would look from a differentposition" (comparedwith 7.2 hoursfor the entiresample),27.2 hours per month "following instructionsfor the assemblyof furniture, toys, models,etc." (comparedwith 5.8 hours per month for the entiresample),and 50.7hoursper month "producingor interpretingtechnicaldrawingsof three-dimensional objects"(comparedwith 5.7 hours for the entiresample).The findingthat the estimatesfrom peopleexpectedto havegreater with certainspatialvisualizationactivitiesweresubexperience stantiallyhigherthan thosefrom the entire sampleenhances the credibilityofthe experience ratings. Simultaneous Analysis of Age and Experience A seriesof multiple regression analyseswereconductedon the spatialvisualization,inductivereasoning,and perceptual speedcompositevariables.The first analysiswith eachvariable wasa stepwiseregression to determinewhichof the eightcomponentsftom the questionnairedata had significant(p <.01) effectson the compositemeasuresof cognitiveperformance. (2,4, and 8) weresignificantwith the spatial Threecomponents visualizationvariable,one (Component2) wassignificantwith the inductivereasoningvariable,and nonewassignificantwith the perceptualspeedvariable.The regressionanalyseswere then repeatedwith only the significantcomponentsand ageas predictorsand werethen repeatedagainfor the spatialvisualizationand inductivereasoningvariableswith perceptualspeed as an additionalpredictor.Identicalanalyseswereconducted on compositesbasedon the numberof corrector right responses and on the numberof incorrector wrongresponses, in additionto the primary analysisthat wasbasedon the number of right responses minus the numberof wrong responses. Resultsfrom theseanalyses are summarizedin Table5. The first point to note about Table 5 is that the absolute valueso[ the varianceestimatesare largerwith the composite basedon numberofright responses than with that basedon the right-wrong scores.This is probably a reflection of the lower reliability of differencescoresbecauseall of the agecorrelations with the numberof right scoreswerenegativeand all of thosewith the numberof wrongscoreswerepositive.(Notethat this is inconsistentwith whatone would expectif therewerea greateremphasison accuracythan on speedwith increased age) A secondpoint concerningthe data in Table 5 is that althoughsimilar patternsare evidentin the right-wrongand right scores,very few systematiceffectswere evident in the analyses basedon the numberofwrong responses. It is evidentin Table5 that the age-relatedeffectsin both the spatialvisualizationand inductivereasoningvariableswereattenuatedby statisticalcontrolofthe questionnaire components and ofperceptualspeed.The proportion ofage-associated variance for the spatialvisualizationvariablewas reducedfrom .139to .083after controllingthe significantcomponentsfrom the questionnaire,to .085 after controlling perceptualspeed, and to .048 after controllingboth the questionnairecomponentsand perceptualspeed.Expressedin percentages, the age (. I 39 - .083)l I 391after control of effectwasreducedby 40.3Vo the questionnairecomponents,by 38.87o(. 139 - .085)11391 after control of perceptualspeed,and by 65.5Vo [(.I 39 - .048)/ .1391after control of both. The age effectson the inductive reasoningvariable were reducedlessby controlling the questionnaire componentsand more by controlling perceptual speed.That is, the age effectswere reduced21.67,[(.088.069)1088 I aftercontrolof Component2, 67.}Vo(.088- .029)/ .0881after control of perceptualspeed,and 77.3Vo K.088.020)10881 after simultaneouscontrol of both Component2 and perceptualspeed. As notedearlier,Component2 reflectsthe individualSestimatesof his or her levelof ability acrossall activitiesand, becauseof the methodusedto identifycomponents,is independentof the amountof cumulativeor recentexperience with any of the activities.A moreappropriateevaluationof thecontribution of relevantexperienceto the age effectson measuresof spatial visualization should therefore be restricted to effects with Components4 and 8. The total varianceacassociated countedfor by age,Component4, and Component8 was.176, with.l20 of that uniquelyassociated with age.The reductionof age-associated effectswas thereforel3.7Eo[(139- .120)1139]. After controlofperceptualspeed,the proportionofage-related variancewas .085; this was reducedby l5.3%o, to .072, after control of Components4 and 8. Verysimilar estimatesof the contributionsof experiencewerederivedfrom the measureof the numberofright responses, asthe reductionsin age-related variancewereI 3.I 7owithoutconsideringperceptualspeedand l4.l%ofor the speed-adjusted measures. Multipleregression werealsoconductedwith AgeX analyses Component cross-productinteraction terms enteredafter age and the eightquestionnairecomponents.None of the interactions wassignificant(i.e.,all p > .20)for eitherthe right-wrong or the right scoresfor the inductivereasoningor perceptual speedvariables.Noneofthe interactionsreachedthe.0I significancelevelwith the right-wrong scoresfor the spatialvisualization variable,but the interactionsof agewith Components2, 4, (i.e.,p < .l0) with oneor both of and 8 approached significance the right-wrong or the right scores.Another analysiswasthereforeconductedasa furthercheckon the possibilitythat experiencemay havemoderatedage-relatedeffectson spatial visualization. For this purpose,individuals were categorizedinto threegroupson the basisoftheir scoreson Component2 (Subjective Ability), Component4 (Perspective), and Component8 (Directions).Regression equationsrelatingageto thecomposite measureof spatialvisualizationperformancewerethen computed for the individualsin eachof thesethree groups.The resultingregressionlinesareillustratedin FigureI (Component 2), Figure2 (Component4), and Figure3 (Component8). Confidence intervals around the regressioncoefficientsrevealed that only the medium Component2 and low Component2 regressionequationshad significantly(p < .01) different slopes. The important point to note in Figures1,2, and 3 is that althoughthe regressionlines for the individualswith higher I I I l AGING AND SPATIAL EXPERIENCE 851 Table5 ProportionofvarianceAccountedfor in HierarchicalRegressionAnalyses(N : 343) Right-Wrong I I Variable R2 Cum.R2 Risht R2 Wrong Cum. R2 R2 C u m .R 2 Spatialvisualization Age .t39* .r39 .206* .206 .023 .023 Component2 Component4 Component8 Age .I29* .029+ .027* .093* .t29 .t57 .184 .267 .150* .043* .035* .129* .150 .1 9 3 .228 .351 . 0 51 * .004 .007 .01I .051 .055 .062 .073 PerceptualSpeed Age .104* .095* .104 .189 .085* .149* .085 .234 .074* .006 .0-/4 .080 PerceptualSpeed Component2 Component4 Component8 Age .104* .l07t .03li .020* .048* .104 .2tl .242 .262 . 3l 0 .085* .I29* .0464 .028* .093* .085 .214 .260 .288 .381 .074* .039* .006 .004 .00r . ll 9 .123 -124 Age .0gg* .088 . 15 7 * . t5 7 .005 .005 Component2 Age .044* .069* .044 .l13 .052* .l3l* .052 .183 . 0l 6 .002 .016 . 0 18 PerceptualSpeed Age .227+ .0294 .227 .lJtJ .2634 .069+ .263 .332 .087+ .000 .087 .087 PerceptualSpeed Component2 Age .2274 .021* .020* .227 .254 .274 .263* .032* .056* .263 .295 . 3 5l .087* .009 .001 .087 .096 .097 .08l .006 .074 . l l J Inductive reasoning Perceptualspeed Age . 0 81 * .081 .081* Mrlr,. Cum. R2 : cumulative squared multiple correlation. * < .01 (florR' only). I valueson the componentsare elevatedrelativeto thosewith lower values(reflectingthe significantmain effectsof these components), the slopesof the lines,and particularlythoseof the extremegroups,are nearlyparallel.This suggests that the ageeffectsaresimilar throughoutthe rangeof componentval_ uesand impliesthat it is not the casethat the magnitudeof the ageeffectsis attenuatedamong individualswith the sreatest amountof experience or self-assessed abilitv. Discussion I I Beforeconsideringthe implicationsofour results,it is impor_ tant to notethat the relationbetweenageand spatialvisualiza_ tion performanceevidentin this study is consistentwith that foundin numerousearlierstudies.Forexample,the correlation of -.37 betweenageand the compositemeasureof spatialvisu_ alizationin this studyis nearlyidenticalto the medianof -.39 for l8 correlationsbetweenage and spatialability measures summarizedin TableI 2. I of Salthouse (1985).This is notewor_ thy becausethe currentsampleis relativelyselect,with an aver_ ageofoverI 5 yearsofeducationanda -.0 I correlationbetween ageand amountof education(Table1). The age-related effectson both the spatialvisualizationand inductivereasoningvariablesweresubstantiallyreducedafter controlof perceptualspeedand self-ratedability.The findings with perceptualspeedreplicatethoseof earlier studies(e.g., Hertzog,1989;Salthouse, Kausler, &Saults,l ggg;Schaie, l9g9) and are consistentwith suggestions that at leastsomeof the adultagedifferences in cognitivefunctioningareattributableto age-related reductionsin the rateof processinginformation. The effectsassociatedwith Componenl2 are nor easyro evaluatebecauseit is not clear how the self-ratingsofability shouldbe interpreted.In particulaqit is difficult to determine the extentto whichtheseratingsreflectpersonalitycharacteris_ tics such as self-confidence or feelingsof self-efficacy, as opposedto actuallevelsofcognitiveability.Ifthe self-ratings are merelyalternativeindicatorsof generalcognitiveability,then they are of limited interestas potentialmediatorsor modera_ torsofage-related differences in cognitivefunctioning.Unfortunately,it wasnot possibleto distinguishbetweentheseinterpre_ tationsof the self:ratingmeasures in our study. The major conclusionimplied from our findingsis that many of the age-relatedeffectson spatial visualizationobservedin this study,and presumablyotherstudies,seemto be 852 TIMOTHY A. SALTHOUSE AND DEBORA R. D. MITCHELL 1 o o o 1 o "'--.. x- 0.5 HighC2 a I MediumC2 N o o) (u o <r. o o o (t 0 0.5 MediumC8 I N o o) (u "'* AT 0 LorvG8 ----: L -0.5 (t) --t---:::$ LowC2 -1 -0.5 -1 20 30 40 50 60 70 Age Chronological 80 20 30 40 50 60 70 Age Ghronological 80 l'igurc I. Regressionlines relatingcompositespatial visualization scoreto agefbr individualsin the top. middle,and bottom thirdsof the distributionof scoreson Component2 (C2:SubjectiveAbility). spatialvisualization linesrelatingcomposite Figure3. Regression andbottomthirdsof the in thetop,middle, scoreto ageforindividuals 8 (C8;Directions). on Component of scores distribution relatively independent ofthe amount ofrelevant experiencethe individuals havereceived.That is, experiential factorsappear to be responsiblefor only about I 57oofthe total age-relatedvarizince observed in measuresof spatial visualization. However, acceptance of this conclusion is contingent on a number of assumptions that can each be challenged. It is therefore useful to consider arguments that can be raised in defense of three questionnaireprovidea valid indicationof the actualexperiencesof the individuals.Evaluatingthe validity of self-report informationof this type is alwaysdifficult,but thereareseveral reasonsto haveconfidencein our questionnaireresults.F'irst, to the questionnaireitemsapthe distributionsof responses near the middle of the pear plausible,with averageresponses rangefor the subjectiveability ratingsand relativelysmall for mostactivities(seeTable2). amountsof reportedexperience analysisresultedin a coherSecond,the principal-components to ent patternofboth generalcomponents(reflectingresponses each type of scale)and specificcomponents(representing meaningfulconfigurationsof self-ratedability, recentexperience,and cumulativeexperiencefor specificabilities).Third, membersof occupationsin which one would expectfrequent usageof spatialvisualizationabilitieshad exceptionallyhigh scoreson the componentconcernedwith spatialperspective. A secondassumptionimplicit in our approachthat could be challengedis that the rangeofexperiencewassufficientto reAlveal the expectedinfluencesof differentialexperience. aboutthis thoughit maybe impossibleto dispelall reservations assumption,it is important to point out that a considerable wasreportedacrossparticipantsin rangeofrelevantexperience this study.To illustrate,the individualsin the top third ofthe of distributionof valueson Component4 reportedan average 40 hr per month lor performingthe threeconstituentactivities (i.e.,2,4,and 7), whereasthosein the bottomthird of the distribution reportedan averageof only 6.5 hr per month for these activities.Despitethis substantialdifferencein the amountof time spentperformingwhat appearto be relevantactivities,the data in Figure2 indicatethat the agetrends in measuresof spatialvisualizationperformancefor the two subgroupswere nearlyidentical.It is clearlypossiblethat individualswith more extremelevelsof experiencemight be found and that differential age trendsmight be evidentwithin that sample,but the critical assumptions. One assumption of our approach is that the responsesto the 1 co *...."tsn ""--.. o. o 0.5 o () A-_-_"1 U' I N o o) (u o 0 LowC4 -0.s Medium C4 z -----:! ---------- ""'* ,"" -!a -1 20 30 40 50 60 70 Age Chronological 80 I'igure 2. RegressionIines relatingcompositespatial visualization scoreto agefor individualsin the top. middle,andbottom thirdsofthe distributionof scoreson Component4 (C4; Perspective). AGING AND SPATIAL EXPERIENCE rangeofnaturally occurringexperiencewith spatialvisualization activities in our sampledoes appear unrepresentativeof that expectedin the generalpopulation. A third assumptionimplicit in the approachused in this study is that the activitiesmentionedin the questionnaireare amongthe most relevantfor spatialvisualizationabilities.An objectioncould be raisedthat the requirements of theseactivitiesarenot sumcientlysimilarto whatthe examineemustdo in the PaperFoldingand SurfaceDevelopmenttestsusedin the assessment ofspatial visualizationability to expectsubstantial relationsbetweenexperienceand spatial visualization performance.This is a plausibleconcern,but we havebeenunableto identify relativelycommon activitiesthat appearto havegreater relevanceto the spatialvisualizationconstruct.Moreover,it is interestingto considerthe implicationsfor the disusehypothesisof the difficultyof finding activitiesrelevantto the abitities observedto decrease with increasedage:Ifthere are no activitiesthat provideappropriateexperience, then neitherthe conceptsof usenor of disusemay be very meaningfulwith respect to the maintenanceor declineof spatialabilitiesacrossthe life span. Additional objectionsto the current procedurescould undoubtedlybe raised,but it is noteworthythat resultssimilar to thosefound in this study havepreviouslybeen reportedwith quitedifferentmethodologies. For example,threestudieshave useda strategyof examiningagetrendsin molecular,or basic, processes afterequatingindividualsofdifferentagesin the proficiencyof a molar targetactivity.Charnessexaminedunexpectedrecallof bridgehandsamongbridgeplayers(Charness, 1979)and of chessconligurationsamongchessplayers(Charness,l98l ), and Salthouse examinedmeasures of perceptualmotorspeedamongtranscription typists(Salthouse, 1984).In each case,significantage-related declineswere found in the measureof the molecularprocesses despitewhat can be assumedto be moderateto high amountsof relevantexperience lor most researchparticipants. Two studiesexaminingthe joint effectsof ageand reading habitson recallof prosematerialarealsopertinentto the disuseperspective if it is assumedthat experiencewith readingis relevantto the taskofrecallingprosematerial.In a I 986study,a questionnaire wasadministeredto assess the numberof hours per weekdevotedto readingdifferenttypesof materialand the individual'spreferences for variouskinds of reading(Rice & Meyer,1986).Althoughseveralof the summaryscoresderived from a principal-components analysisweresignificantlyrelated to both ageand total recall performance,therewasno evidence that the ageeffectsvariedasa functionofthe amountofreading experience. A laterstudy by the sameinvestigators (Rice, Meyer,& Miller, 1988)involvedthe simultaneous examination of proserecall performanceand degreeof readingactivity,as determinedfrom analyses of diaries.Unflortunately, Riceet al. did not report the extent to which the age-relatedeffectsin recall were attenuatedby controlling for amount of reading experience,but they did indicate that the effectsof age and educationallevelweremuchgreaterthan thoseassociated with readinghabits. In summary,the findingsof our study,in conjunctionwith the resultsofthe studiesjust reviewedand the occupation-specific studiesdescribedearlier,appearinconsistentwith the dis- 853 use perspective.Not only are the age-relatedeffectsgenerally similar acrossdifferent levelsof presumablyrelevant experience(thisstudyand probablythe studiesof Rice& Meyeq1986, and Rice et al., 1988),but they appearto be substantialeven amongsamplesselectedto be equivalentwith respectto occupation(e.g., Glanzer& Glaser,1959;Glanzer,Glaseq& Richlin, 1958;Salthouseet al., 1990)or to level of molar ability (e.g., Charness,1979,l98l; Salthouse,1984).The seeminglyinescapable conclusionfrom this bodyof evidenceis that manyof the age-relatedeffectson measuresof relativelybasic abilities are largelyindependentof the amountof relevantexperience. We hastento point out, however,that this conclusiondoes not imply that thereare not positivebenefitsof experienceor that increasedage in adulthoodis inevitablyassociatedwith declininglevelsof competence. More extensive experiencefrequently resultsin greaterknowledge(both declarativeand procedural),betterdiscriminationbetweenrelevantand irrelevant information,more successful executionof complexactivities, and perhapsmoreeffectivemonitoringand deploymentof basic abilities.What remainsto be resolvedis the dynamicrelation betweenthe efficiencyof basicabilitiesand the operation ofthesehigherorderprocesses, and whether,and ifso how this relationshipchangeswith age. References Arbuckle, T.Y.,Gold,D.,& Andres, D (1986). Cognitive functioningof olderpeoplein relationto socialand personalityvariables.psy,g7r1ogvand Aging,l.55-62. Birren,J. E.,Cunningham,W R.,& Yamamoto,K. (1933).psychology ofadult developmentand aging.Annual Revieu,olpsvchologu34, 543-575. Charness,N. (1979).Componentsofskill in bridge.CanadianJournal tulP.sychology, JJ, l-16. Charness,N. (1981).Aging and skilled problemsolving.Journalrt' E.yperimentalPsychologv:General,I I 0, 2|-38. Charness,N. (1985).Aging and problemsolvingperformance.In N. Charness(Ed), Aging and human perlbrmance(pp. 225-259\. Chichester,England:Wiley. DeCarlo,T. I. (1974).Recreationparticipationpatternsandsuccessful aging.Journal ol Gerontology,29, 4 | 6-422. Denney,N. W (1982).Aging and cognitivechanges.In B. B. Wolman (Ed,),Ilandbook ofdevelopmentalpsvt'hology(pp. 807-827). Englewood Cliffs. NJ: Prentice-Hall. Ekstrom,R. 8.,French,J.W, Harman,H. H.,& Dermen.D.0976l. Kit of .fitctor-reference utgnitive tesls.Princeton,NJ: EducationalTesting Service. Fozard,J. L., & Thomas,J. C. (1975).Psychology ofaging: Basichndingsand somepsychiatricapplications.ln J.G. Howells,Eil. Modern perspectives in the p sychiatry of oId age(pp. I 07- I 69). New york: Brunner/Mazel. Gilbert, J. G. (1952).Understandingold age.New york: Ronald press. Glanzer,M., & Glaser,R. (1959).Cross-sectional and longitudinalresultsin a studyofage-relatedchanges.Educationaland psychological Mea.surcment. 19, 89- l 0 l . Glanzer,M., Glaser,R., & Richlin, M. (1958).Development ol'a test battery.for study oJ'age-related changesin intellectualand percepttnl abilities.RandolphAir ForceBase,TX: Schoolof Aviation Medicine.#56-I 38. Hertzog,C. (1989).Influencesof cognitiveslowingon agedifi'erences in intelligence.DevelopmentalPsychology, 25, 636-65l. Owens,W A. (1953).Age and mentalabilities:A longitudinalstudy. GeneticPsychologvMonographs,48, 3-54. 854 TIMOTHY A. SALIHOUSE AND DEBORA R. D. MITCHELL effectsof complexenvironments SchooleqC. (1984).Psychological Rice, G. E., & Meyer,B. J. F (1986).Proserecall: Effectsof aging, 4 I , 469during the life span:A reviewand theory.Intelligence,8, 259-281. verbalability, and readingbehavior.JournalofGerontology, Schwartzman,A. 8., Gold, D., Andres, D., Arbuckle, T. Y, & Chaikel480. son, J. (1987).Stability of intelligence:A 4O-yearfollow-up. CanaRice,G. 8., Meyer,B. J. F, & Miller, D. C. (1988).Relationof everyday 41,244-256. dian Journalof Psychology, activitiesof adultsto their proserecall perfbrmance.Educational Shipley,V{ C. (1986).Shipley Institute of Living Scale.Los Angeles: y, I 4, 147-1 58. Gerontolog WesternPsychology Servies. Salthouse,T. A. (1984).Effectsofage and skill in typing. Journalrd' Sorenson,H. (l 933).Ability overa wide rangeofadultages.Journalof E"tperimentalPsychoktgy:General,I 13,345-37l. AppliedPsychology,I 7, 729-'141. aging.Amsterdam:NorthT. A. (1985).A theoryol cognitive Salthouse, Sward,K. (1945).Age and mentalability in superiormen.American Holland. 58, 443-479. Journal oJPsychology, Salthouse,T. A., Babcock,R. L.. Skovonek,E., Mitchell, D. R. D., & Williams, M. (1960).The effectof past experienceon mentalperforPalmon,R. (1990).Age and experienceeffectsin spatialvisualiza33, 215mancein the elderly.British Journalof MedicalPsychology, 26, 128-136. rcn. DevelopmentPs)'chology, 22t. Salthouse,T. A.. Kausler,D. H.. & Saults,J. S. (1988).Utilization of path-analyticproceduresto investigatethe role of processingreReceivedJuly 7, 1989 and Aging,J, I 58-166. sourcesin cognitiveaging.Psvchologv RevisionreceivedDecember22, 1989 Schaie,K. W (1989).Perceptualspeedin adulthood:Cross-sectional AcceptedJanuary4,1990 t and Asing,4, 443-453. and longitudinalstudies.Psvt'hologv