aburgoynehoninguvanl dbountouridisjmhvanbalenuunl ABSTRACT Although there has been some empirical research on ear worms songs that become caught and replayed in ones memory over and over again there has been surprisingly little empirical research on ID: 40649
Download Pdf The PPT/PDF document "HOOKED A GAME FOR DISCOVERING WHAT MAKES..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
aswellasthetheoriesunderlyingitsdesignandtheresultsofapilotstudyweundertooktocheckitsscientiÞcvalidity.WefoundsigniÞcantdi!erencesintimetorecallpiecesofmusicacrossdi!erentsegments,identiÞedparametersfor content[].Thisareaofstudy,incontrast,isbroaderthanourareaofinquiry.Althoughcatchinessiscertainlycorrelatedwithpopularity,manypopularsongsarequiteforgettable,andwearemostinterestedinmusicthatre- blurryÕ[ memory;onestudyhasshownthatafteronly400ms,listen-erscanidentifyfamiliarmusicwithasigniÞcantlygreaterfrequencythanonewouldexpectfromchance[12].Wehavedesignedanexperimentthatwebelievewillhelptoquantifythee inwhatwefeltwasthemostentertainingfashionpossible.Inthisway,wehopetobeablerecruitthelargestnumberofsubjectspossiblewithoutsacriÞcingscientiÞcquality.2.1RecognitionTaskTherecognitiontaskistheheartofthegame.ItstemsfromtheideathatthedeÞningaspectofcatchinessitse!ectonlong-termmemory.Inparticular,theeasierafragmentofmusicistorecallafteralongperiodoftime,thecatchieritshouldbe.Thus,aÔdrop-the-needleÕstylequiz,wherebyapieceofmusicstartsplayingfromapointinthemiddleandplayersareaskedtorecogniseit,seemedtobeappropri-ate.Asnotedabove,thereisaconsensusinthetheoreticalliteraturethatthehookshouldstartatthebeginningofanewstructuralsection(possiblyincludingthebeginningofthepieceitself),andweextendedthisideatolimitthenumberofstartingpointstoastatisticallytractablesubset:Musicwillalwaysstartplayingfromthebeginningofastructuralsection.Thentheamountoftimeittakesaplayertorecognisethepieceisaproxyforhoweasythatsectionistorecall,orinshort,howcatchyitis.Figure1aillustratestherecognitiongameasimplemen-tedinourcurrentiOSprototype.Apieceofmusicstartsplayingfromthestartofastructuralsection,andplayershaveseveralsecondstodecidewhethertheyknowit.Whileplayersarelistening,pointsarecountingdown;thefasterplayersrecognisesthepiece,themorepointstheycanwin.2.2VeriÞcationTaskInacontrolledlaboratoryenvironment,itmightbejusti-Þabletotrustsubjectstobehonestinclaimingtohave thenameoftheperformingartist;moreover,evenforthoseuserswhodoknowsuchtrivia,theextracognitiveloadinrecallingitinadditiontorecognisingthemusicitselfwould beabletofollowalongwiththesongintheirheadsforsometimeevenafterthemusicstopsplaying.Moreover,thereisevidencethatabsolutetempoispartofmusicalmemory,al-thoughtheerrordistributionissomewhatskewedinfavourofoverlyfasttempi[14].InHooked,assoonasplayersclaimtoknowasong,playbackmutesforafewseconds.Duringthemute,playersareaskedtoimaginementallyorsingalongactivelyforafewseconds(Figure1b).Whenthesoundreturns,halfthetimethemusicreturnsthecorrectplace(i.e.,themutewasgenuinelyonlyamute)andhalfthetimetheplaybackiso catchymembersofeachpairandsendingthosepredic-tionstothoseopponentsforarecognitiontask;wewould Section Start Time (s)Mean Response Time (s)0246810Figure2:Meanresponsetimesfromtherecognitiontaskondi!erentsectionsofAdeleÕsÔRumourHasItÕ.Errorbarsreßectstandarderror.Controllingformultiplecomparisons,therearesigniÞcantdi!erences(p05)inresponsetimesbetweenthebridge(133.9s)ortheverseat91.3s,andtheinitialentryofthevocals(12.6s)orthepre-chorusat38.9s.percentthrough.Nonetheless,becausewerandomisedthepresentationofsectionsandparametersettingsforeachsubject,wehavenoreasontobelievethatthemissingdatashouldexhibitanysystematicbias.Fortherecognitiontask,theBox-Coxproceduresug-gestsalogtransformonresponsetime,andweassumethatresponsetimesarelog-normallydistributed.Regressingthusacrossallsongsections,theaverageresponsetimeforsuccessfullyveriÞedclaimstoknowasongis5.2s. areashighas4s.InordertotunetheveriÞcationtask,weneededtode-terminethebestvaluestousefor tioncorrectly.AkaikeÕsInformationCriterion toplayinghonestlyÐagain,a64percentimprovementintheoddsofansweringtheveriÞcationcorrectlyÐsuggestthatwemayfeelcomfortablethatplayershaveanincentivetoplayhonestlyregardlessoftheparametersettingsandthuscanfocusonmakingthegameaspleasantforhonestplayersaspossible.Assuch,weintendtoallow15sforrecognitionandusethe"15-sdistractoro!set.Weweresurprisedthatthedistractoro!sethadsuchastronge!ectontheplayersÕaccuracy,andtheideathatdis-tractorsfromthepastareeasiertoidentifyasincorrectthandistractorsfromthefutureisespeciallyintriguingfromacognitiveperspective:Isiteasiertorewindmusicalmemorythanitistofast-forward?Anotherpossibility,perhapssim-pler,isthattheforwarddistractorismorelikelytobeinthesamestructuralsectionastheoriginalfragment,whereasbecausewehavechosenourfragmentsalwaystostartatthebeginningsofnewsections,thebackwarddistractorwillal-waysbeinadi!erentone.Assumingthatstructuralsectionsmaintainsomedegreeoftimbralconsistency,thebackwarddistractormaymoreofteno!ertimbralcluestotheplayerthatsomethingisnotrightwhenthesoundreturns.Thedataforthepredictiontaskdonotlendstrongsup- second,butoverall,playersdonotappeartobeconsistentwiththeirrecognitionbehaviourwhenmakingpredictions.theirpersonalfeelingsaboutcatchiness;infact,justthereverseseemstobetrue.AkintoWilliamsonandMllen-siefenÕsworkonearwormsandBurnsÕsmorespeculativework,[4,19],aswerollHookedouttolargeraudienceandtherebygeneratealargerdatabase,weplantoÞndsetsaudiofeaturesthatcorrelatewithrecognitionandpredic-tionperformance.Thedi!erencebetweenthesetwosetswillhelpclarifythisdivergencebetweenlistenersÕactuallong-termmusicalmemoriesandtheirexpectationsofthem.6.REFERENCES[1]L.vanAhnandL.Dabbish:ÔDesigningGameswithaPurposeÕ,Communicationsofthe!"#,Vol.51,No.8,