Spatial properties of flicker adaptation - PDF document

Spatialpropertiesof”ickeradaptationAlanE.Robinson,VirginiaR.deSaDepartmentofCognitiveScience,UniversityofCalifornia,SanDiego,USADepartmentofPsychology,UniversityofCalifornia,SanDiego,USAarticleinfoArticlehistory:Received26August2011Receivedinrevisedform21July2012Availableonline8August2012FlickeradaptationSpatial“ltering Correspondingauthorat:DepartmentofPsychology,UniversityofCalifornia,SanDiego,9500GilmanDrive,LaJolla,CA92093-0109,USA.E-mailaddress:(A.E.Robinson).Adaptationisfoundformanyothertypesofvisualstimuli,andthepurposeofadaptationisstilldebated.Webster(2011)providesanexcellentreviewforthose VisionResearch70(2012)2…6 ContentslistsavailableatSciVerseScienceDirectVisionResearchjournalhomepage:www.else 2.Experiment1:adaptingtouniform”ickerThereareatleasttwodifferentkindsofvisualmechanismsthatcoulddetectluminance”icker,differingintheirspatialproperties.First,aspatialcontrastmechanismcoulddetecttheincreaseorde-creaseinlocalcontrastthatoccurswith”icker.Itwouldrespondwhenitsreceptive“eldintersectswiththeborderofthe”ickeringregion,butnotwhenstimulatedbyuniform”icker.Thesecondtypeofmechanismissensitivetotheabsolutechangeinlumi-nancewithinitsreceptive“eld.Itrespondswhenitintersectsa”ickeringedge,butrespondsevenmorestronglywhenitsrecep-tive“eldisentirelywithinthe”ickeringregion.Asthusdescribed,thesetwomechanismsmapontothesustainedandtransientreti-nalmechanismsidenti“edbyTyler,buttoremainagnosticastotheactualrelationshipwewillrefertothesetwokindsofdetectorsasedgeanduniform”ickerdetectors,respectively.Inthisexperimentwemeasurethestrengthof”ickeradapta-tionfortwoconditions:whenthetestdiskisexactlythesamesizeandlocationastheadaptedregion(,andwhenthetestdiskissigni“cantlyrelativetotheedgesoftheadaptingdiskFig.1).Thealignedconditionhasthepotentialtoadaptbothedgeanduniformmechanisms.Intheinsetconditiontheedgesoftheadapterandthetestdonotoverlap,buttherecouldstillbeasen-sitivityreductionduetoadaptationintheuniformmechanism.2.1.Methods2.1.1.SubjectsTwopsychophysicallyexperiencedsubjectsparticipated,oneanauthor,andonenaïvetothepurposeoftheexperiment.2.1.2.ApparatusStimuliwerepresentedona22HM204DTACRTdrivenbyanNVIDIAGeForce8600GTvideocardatarefreshrateof75Hz,inamoderatelylitroom.DisplayluminancewaslinearizedusingaCambridgeResearchSystemsBits++adapterwitha14-bitcolorlookuptable.AUDTphotometerwasusedtoselecttheappropriatelookuptablevaluesforgammacompensation.Achinrestwasusedtomaintainaviewingdistanceof42cm.StimuliweregeneratedanddisplayedusingMatlabrunningthePsychophysicsToolbox,version3(Brainard,1997;Kleiner,Brainard,&Pelli,2007;Pelli,)onaWindowsXPcomputer.Thesameapparatuswasusedinallexperiments.2.1.3.StimuliandprocedureTheadaptingandteststimuliconsistedofatemporally”icker-inguniformdiskcenteredonagraybackgroundwitha“xationdotatthecenter.Conditionsvariedalongtwodimensions:Thediam-eterofthetestdiskandwhetherthetestdiskwasalignedorinsetrelativetotheadaptingdisk.Weusethefollowingshorthandradiusofadapterradiusoftesttosummarizethedegreesofvi-sualangleeachstimulussubtended(seeFig.1).Thealignedcondi-tionswere2/2and10/10,andtheinsetwere4/2,10/2,12/10,and15/10.Duringadaptationtheluminanceofthediskswasmodu-latedbya6.25Hztemporalsquare-wave(80msperframe)be-tweenblack(2.3cd/m)andwhite(130cd/m),onagraybackground(65cd/m).Theblackvaluewasrelativelyhighbe-causetheexperimentwasconductedwiththeroomlightson.Whilethisissomewhatunusualinvisionexperiments,wefoundittobequitebene“cialinthiscase,becauseotherwiseadaptingtolarge”ickeringdiskscausedsigni“cantvisualstrainandfatigue,changingvisualthresholdsoverthecourseofarun.OurparadigmisoutlinedinFig.2.Contrastthresholdsweremeasuredintwodifferentkindsofsessions:Intheadaptsessionsubjects“rstview9.1sof”icker(notshownFig.2).This”ickerismeanttobuildupadaptationbe-foreanymeasurementsaremade.Subjectsareinstructedto“xatethedotatthecenterofthescreenduringthis,andallsubsequentpartsoftheexperiment.Afterthepre-trial”ickersthemeasure-menttrialsstart.Ameasurementtrialinanadaptsessionstartswith2.08sofadapting”icker,whichservestomaintainaconstantstateofadap-tation.Awhite“xationdotisshown,whichbrie”y”ashestoblackonthe2ndtolastcycleof”ickertosignalthatthetestisabouttostart.Nextagrayscreenisshownfor66ms,followedbyagreen“xationdotandthetest”ickers.Ina2AFCtask,subjectsdetectif4cyclesof”ickerareshownorjust1.When1cycleisshownitisatthesametemporallocationasthe3rdcycleinthe4-cyclese-quence,tohelpequatevisibility.Thenextmeasurementtrialstartsimmediately;thesubjecthasupto2.08storespondtothe2AFCtaskfromtheprevioustrialwhiletheyadapt.109trialswerecol-lectedeachsessionandeachsessiontook5.5min.Tomeasureifadaptationreducedsensitivitywecomparedcon-trastthresholdsfortheadaptsessionstosessions.Intheno-adaptsessionsnopre-trial”ickersareshown.Insteadof2.08sofadapting”ickerbetweentests,1cycleof”ickerwasshownjustbefore,andonejustafterthetest”ickers(total=0.032s).Wein-cludedthese”ickerstomakesurethatanymaskingeffects,suchasmeta-contrastmasking,werepresentinboththeadaptandno-adaptsessions.Datawerecollectedovermultipledays.Topreventordereffectsbothsubjectsweretrainedonthetaskuntilthresholdswerecon-sistentacrossmultipledays,andtrainingdatawerenotusedinanalysis.Tofurtherprotectagainstordereffects,subjectscom-pletedthesessionsinrandomorder,exceptwetriedtoincreaseourabilitytodetectdifferencesbetweenadaptandno-adaptses-sions.Whenaconditionwasselectedthathadbothadaptandno-adaptsessions,bothofthosesessionswerecollectedbacktoback(whichwas“rstwasselectedrandomlyoneachday,thoughpilottestsshowedthatthresholdsinno-adaptsessionwerenotshiftedwhenprecededbyanadaptsession).Anaverageoffoursessionspersubjectwerecollectedforeachcombinationofcondi-tionandsessiontype.Tomeasurecontrastthresholdsweadjustthemagnitudeofthetest”ickersusingavariablestepsizestaircase.Thestaircasewasinitializedat30%Webercontrast,withastepsizeof10%Webercontrast.Aftercorrecttrialsthe”ickerstrengthwasreducedby 4/2 10/2 Test disk Adapter (or mask) (a)(b)(c) Fig.1.Examplespatialpro“lesoftheadaptingdiskandtestdisk(testdisklocationdenotedbydashedlines),withthecorrespondingstimulishorthandnamesprintedatleft.(a)Analignedcondition.(b…c)Insetconditions.A.E.Robinson,V.R.deSa/VisionResearch70(2012)2…6 onestepsize;afterincorrecttrialsitwasincreasedby2.7timesstepsize,andthestepsizewasreducedby20%.Ifthestaircasereacheszerocontrastthenhalfofthelasttestedcontrastisusedinsteadandthestepsizeisalsoreducedby25%.Aftereveryfourincorrectresponsesasingleeasytrialisintroducedtokeepthesubjectsattentionandmotivationfromdecreasingduetofrustra-tion.Toensurethatthresholdsweremeasuredatsteadystatewediscardedthe“rst20trialsofeachsession.2.1.4.AnalysisWe“taWeibullcurvetothedataforeachconditiontoestimatethecontrastnecessarytoelicit75%correctdetection.Abootstrapanalysiswasusedtoestimatethe95%con“denceintervalsofthisthreshold,usingtheBCAmethodasimplementedinMatlabR2010A.Weallowedalapserateofupto10%toaddressthecurve“ttingissuesraisedbyWichmannandHill(2001)2.2.ResultsContrastthresholdsforthetwosubjectsareshowninFig.3Bothshowthesamerelativepatternofresults.Contrastthresholdsweremuchhigherinthe2/2(aligned)conditionthaninthe4/2(inset)condition.Thresholdswerequitesimilarforthe4/2and10/2conditions,suggestingthatadaptationdropsoffsteeplyasafunctionofdistancebetweenadapterandtestedges,reachingasymptoticvaluesbyaround2ofseparationforsmalltest“elds.Largertest“eldsalsoshowedasigni“cantdifferencebetweenthealigned(10/10)andinset(12/10;15/10)conditions.Interest-ingly,thetwoinsetconditionswerequitesimilartoeachother,suggestingthatagain,2ofseparationwassuf“cienttoeliminatemostofthesensitivitychange,evenforlargetest“elds.Thus,itdoesnotappearthattheratioofthesizeofadaptertotestdiskpre-dictsthechangeinthreshold.Instead,itistheabsolutechangeinsizethatmatters,andonlyacoupledegreesdifferenceisnecessary.Anyedge-selectivemechanismstunedtoscaleslargerthan2shouldhavebeenadaptedeveninthe12/10condition.Thus,detectioninthe12/10conditionmustbedrivenbyverysmall-scalespatialcontrastmechanismsrespondingtothe”ickerimme-diatelyalongtheedgeofthetestdisk.Thisiswhythresholdsinthe15/10conditionarenobetter:bothconditionssparethesmall-scaleedgedetectorsequally.Wealsofoundevidenceofadaptationinourinsetconditions,thoughmuchweaker.Ourno-adaptsessionsweredesignedtoduplicateanyshort-termmaskingeffectscausedintheadaptses-sions,whileinducingminimal”ickeradaptation.Weconsistentlyfoundlowerthresholdsthaninthecorrespondinginsetadaptses-sions,andthoughthedifferencesweresmall,theyexceededthe95%con“denceintervalsinallcases(exceptforthe4/2conditionforsubjectAR).Thus,mechanismsthataresensitivetouniform”ickeralsoadapt.Subjectsmaintained“xationthroughouttheexperiment.Sincewedidnotuseaneyetracker,wedonotknowthefrequencyandmagnitudeofeyemovements,thoughbothsubjectswereexperi-encedpsychophysicalobserverssoitisfairtoexpecttheseerrorstobesmall.Furthermore,anyeffectswouldhavebeenconstrainedtothesmalleradaptingandteststimuli,whereeyemovementscouldhavereducedadaptationbyreducingalignmentinthe2/2condition,orincreasedadaptationbyreducingthedistancebe-tweenedgesintheunaligned4/2condition.Inthe10/2condition,onlyimplausiblylargeeyemovementswouldhavecausedsigni“-cantalignmentbetweenadaptandtestedges,andyetthethresh-oldsareveryclosetothe4/2conditions,suggestingthateyemovementsdidnotin”uenceourresults.3.Experiment2:adaptingtoedge”ickerExperiment1showedthatsmall(2)changesinscale(andthusalignment)betweenadapterandtestdrasticallychangethelevelofadaptation.Thusitispossiblethattheedges(thatis,thespatialcontrastcomponent)oftheadaptercausemostofthereductioninsensitivity.Totestthis,inexperiment2weuseanadapterthattargetsedges,whileleavingtheinterioruntouched,toseeifthis 010002000300040005000 Fig.2.Diagramoftheparadigmusedinexperiment1(and2),showingtwoexampletrialsandtwodifferentsessiontypes.(a)Intensitypro“leoftheadaptingdiskduringnoadapttrials(includedtoequatemasking;minimaladaptationisexpected).(b)Intensitypro“leusedforthesamediskintheadaptsessions,wheremaximaladaptationisdesired.(c)Exampleintensitypro“leofthetestdisknearthebeginningoftheexperimentwhereitissuper-threshold.Ano-adaptsessionwouldconsistof(a)and(c),whileanadaptsessionwouldconsistof(b)and(c).A.E.Robinson,V.R.deSa/VisionResearch70(2012)2…6 causessigni“cantadaptationforuniformdiskteststhatstimulateboththeadaptededgesandtheunadaptedinterior.Allmethods,apparatus,andproceduresarethesameasinexperiment1.3.1.StimuliThenewadapter(polarcheckerboard,Fig.4wasdesignedtomodulatelocalcontrastacrossthetestdisksedge,whileleavingglobalcontrastconstant.Itconsistedofacheckerboardwithacir-cularborderbetweentheinnerandouterchecks.Thisborderwasalignedwiththeouteredgeofthetestdisk.Oneachsuccessiveframetheidentityoftheblackandwhitecheckswasswitched,cre-atinglocal”ickeralongtheborder,whileholdinggloballuminanceandcontrastconstant.Thecheckradiuswasvariedacrosscondi-tionstovarythescaleofthecontrastmodulationacrossthetestedge.Theconditionsaresummarizedwiththefollowingshorthandcheckradius,measuredfromtheadaptingedgeradiusofthetestdisk.Thussummarized,theconditionswere4/10(6holeinthecenter),8/10,and10/10(nohole).3.2.ResultsForbothsubjectsthe4/10adaptsessionshadsigni“cantlyhigh-erthresholdsthanintheno-adaptsessions(Fig.5),buttheeffectwasquitesmall.The4/10conditionshouldhavereducedsensitiv-ityofanyedgemechanismswithreceptive“eldsof4inradiusorless,includingthesmall-scaledetectorsthatappearedtodominatethresholdsinexperiment1.Thus,itappearsthatwhenthesemechanismsareadapted,othermechanisms(suchasuniformmechanisms)canservetodetect”icker,resultinginlittlechangeincontrastthresholds.Goingfrom4/10to8/10(reducingthesizeoftheholeintheadapter),andthento10/10(eliminatingthehole)minimallyin-creasedthresholds,suggestingthatadaptinganylargerscalecon-trastmechanisms(shouldtheyexist)didnotreducesensitivity.4.GeneraldiscussionExperiment1showsthatadaptinguniformmechanismsbutnotedgemechanismsonlyslightlyreduces”ickersensitivity.Experi-ment2showsthatthereversealsoholds:adaptingedgemecha-nismsbutnotuniformmechanismsalsoreducessensitivityonlyslightly.Thusbothmechanismsplayanimportantrolein”ickeradaptation.Largethresholdchangeswereonlyseeninthealignedconditionsinexperiment1,whichwouldhaveadaptedbothedgeanduniformmechanisms,suggestingthatbothmechanismsmustbeadaptedtosigni“cantlyreduce”ickersensitivity.ThismatchesTylerssuggestionthat”ickersensitivitythresholdsaredeter-minedbywhichevermechanismismostsensitivetothat”icker.Thefactthat”ickerdetectionremainswhenonlyonemecha-nismisadaptedindicatesthat”ickerdetectiondoesnotdependontheproperinterplayofedgeanduniformmechanisms…eachoneinisolationservesquitewell.Thismayhaverelevancetothe-oriesofsurfaceperceptionthatdependonthe“lling-inofuniformregionsfromedgesignals(e.g.GrossbergandTodorovicnamebutoneprominentexample).Onesubject(AR)observedinthesecondexperimentthatthetest”ickerperceptinthe4/10conditionhadoneoftwodistinctforms:athighcontrasttheentiretestdiskwasseento”icker,withaclearlyde“nededge,whileatlowcontrast,asomewhatill-de-“nedregioninthecenter6radiusofthetestdisk”ickered,withnoclearboundary.Thiscorrespondedtotheregionthatwasnotmodulatedbytheadaptingcheckerboard.Thisimpliesthatuni-formregionscanbeperceivedwithoutedgesignalstocontainthem.Whilejustspeculation,itdoessuggestthattheparadigmandstimuliusedinthisexperimentmaybeusefultotestandeval-uatemodelsof“llingin.OurresultsarecompatiblewithbutextendthepredictionsofGreen(1981).Hereportedthatfull-“eld”ickerdidnotreducesen-sitivitytomovinggratingsthatwere4cycles/degreeandhigher.Whilemotionisnotthesameasstationary”icker,onemightpre-dictfromthisabsolutelynoadaptationintheinsetconditionsinexperiment1,sinceourstimuluscouldhavebeendetectedfromhighspatialfrequenciesalone(whichGreensuggestsarenotadaptedby”icker).Note,however,a4cycle/degreegratinghasverylittleinthewayofuniformregions(7.5arcminuteswhichis 0.005 0.01 0.015 0.02 0.025 15 / 10 12 / 10 10 / 10 10 / 2 4 / 22 / 2adapt (2.08s)Session typeno adapt (masking only, 0.032s)adapt / test radius (degrees)Experiment 1: Uniform disk adapters and testsSubject AB 0 0.02 0.03 0.04 15 / 10 12 / 10 10 / 10 10 / 2 4 / 22 / 2Subject AR Fig.3.Resultsfortwosubjectsinexperiment1.Errorbarsdenote95%con“denceintervals. 4° Polar checkerboard (4/10) Fig.4.Polarcheckerboard(4/10condition).A.E.Robinson,V.R.deSa/VisionResearch70(2012)2…6 comparabletotherangeofspatialsummationfoundinSinai,Es-sock,andMcCarley(1999).Gratingsofhigherfrequencyhaveevensmalleruniformregionsthatmayresultinuniformspatialsumma-tionmechanismscontributingminimallytogratingdetectionabove4cycles/degree.Eveninour10/2condition,however,thecentraltestdiskwas4indiameterandthuswillactivateuniformmechanismsthatcouldbeadaptedbyuniform”icker.Greensre-sultspredictthatifourtestdiskwasreducedto1/8thofadegreeindiameter,uniform”ickermightnolongercauseanyadaptation.Greendid“ndsigni“cantadaptationforgratingsbelow4cy-cles/degree.Weexplainthisbypositingthatthosegratingswerewideenoughtostimulateasuf“cientnumberofuniformmecha-nismstoprovideausefuldetectionsignal(priortoadaptation).Fortheselargergratings(lowerspatialfrequency)Greenfoundmoreadaptationthanwedidintheinsetconditionsofexperiment1,probablybecauseheusedsinewavegratings,whichwouldhaveonlystimulatededgemechanismstunedtoanarrowrangeofspa-tialscales.Thus,uniformmechanismswouldhavecontributedrel-ativelymoreofthesignalinhisparadigm.Ourworkalsosuggestsanotherinterestinglineofresearch.Gi-ventheevidencethat”ickeradaptationisatleastpartiallycortical,onecouldextendourparadigmtotestifbothmechanismshavecorticalcomponentsbytestingforinteroculartransferofadapta-DAntona,Kremers,andShevell(2011)havefoundevidencethattheperceptionof”ickermagnitudehasamonocularandbin-ocularcomponent,whichcanbedistinguishedbyaminimalbinoc-ularcontributionathigh(12.5Hz)”ickerrates.Ifthesearethesamemechanismswetested,thenourparadigmshouldalsoshowsomedependenceon”ickerratewhentestingfortransferbetweeneyes.Suchworkwouldhelpconstraintheneuralunderpinningsofthemechanismsstudiedhere.AcknowledgmentsWewouldliketothankDonMacLeodforhishelpdesigningtheseexperiments,andPaulHammon,StuartAnstis,andKarenDobkinsforhelpfuldiscussionsaboutthisworkingeneral.ThisworkwassupportedbytheNationalScienceFoundationunderNSFCareerGrantNo.0133996toV.R.deSa.A.E.Robinsonwassup-portedbyNSFIGERTGrant#DGE-0333451toG.W.Cottrell/V.R.deSa,andNSFGrant#SBE-0542013toG.W.Cottrell.Brainard,D.H.(1997).Thepsychophysicstoolbox.SpatialVision,10(4),433…436.DAntona,A.D.,Kremers,J.,&Shevell,S.K.(2011).Changesinperceivedtemporalvariationduetocontext:Contributionsfromtwodistinctneuralmechanisms.VisionResearch,51(16),1853…1860.Green,M.(1981).Psychophysicalrelationshipsamongmechanismssensitivetopattern,motionand”icker.VisionResearch,21,971…1084.Grossberg,S.,&Todorovic,D.(1988).Neuraldynamicsof1-Dand2-Dbrightnessperception:Auni“edmodelofclassicalandrecentphenomena.Perception&Psychophysics,43(3),241…277.Kelly,D.H.(1969).Flickeringpatternsandlateralinhibition.JournaloftheOpticalSocietyofAmerica,59(10),1361…1370.Kleiner,M.,Brainard,D.,&Pelli,D.(2007).WhatsnewinPsychtoolbox-3.Perception,36,14.Legge,G.E.(1978).Sustainedandtransientmechanismsinhumanvision:Temporalandspatialproperties.VisionResearch,18(1),69…81.MacLeod,D.I.(1978).Visualsensitivity.AnnualReviewofPsychology,29,613…645.Moulden,B.,Renshaw,J.,&Mather,G.(1984).Twochannelsfor”ickerinthehumanvisualsystem.Perception,13(4),387…400.Nilsson,T.H.,Richmond,C.F.,&Nelson,T.M.(1975).Flickeradaptationshowsevidenceofmanyvisualchannelsselectivelysensitivetotemporalfrequency.VisionResearch,15(5),621…624.Pantle,A.(1971).Flickeradaptation.I.Effectonvisualsensitivitytotemporal”uctuationsoflightintensity.VisionResearch,11(9),943…952.Pelli,D.G.(1997).TheVideoToolboxsoftwareforvisualpsychophysics:Transformingnumbersintomovies.SpatialVision,10(4),437…442.Schieting,S.,&Spillmann,L.(1987).Flickeradaptationintheperipheralretina.VisionResearch,27(2),277…284.Shady,S.,MacLeod,D.I.A.,&Fisher,H.S.(2004).Adaptationfrominvisible”icker.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,(14),5170…5173.Sinai,M.J.,Essock,E.A.,&McCarley,J.S.(1999).Spatialsensitizationofincrementsanddecrements:Aborder-contrastprocessandanet-excitationprocess.Research,39(10),1847…1860.Tyler,C.W.(1975).Analysisofvisualmodulationsensitivity:Twocomponentsin”ickerperception.VisionResearch,15(7),843…848.Watson,A.B.(1986).Temporalsensitivity.InK.Boff&J.Thomas(Eds.),Handbookofperceptionandhumanperformance(pp.6.1…6.43).NewYork:Wiley.Webster,M.A.(2011).Adaptationandvisualcoding.JournalofVision,11(5),3,1…23.Wichmann,F.A.,&Hill,N.J.(2001).Thepsychometricfunction:I.Fitting,samplingandgoodness-of-“t.PerceptionandPsychophysics,63,1293…1313.Wu,S.,&Burns,S.A.(1996).Analysisofretinallightadaptationwiththe”ickerJournaloftheOpticalSocietyofAmericaA,Optics,ImageScience,andVision,13(3),649…657. 0.005 0.01 0.015 0.02 10 / 10 8 / 104 / 10 % contrast required for 75% correct 0 0.02 0.03 10 / 10 8 / 10 % contrast required for 75% correctadapt / test radius (degrees)Experiment 2: polar checkerboard adapter, uniform testsubject ABsubject ARadapt (2.08s)Session typeno adapt (masking only, 0.032s) Fig.5.Resultsfortwosubjectsinexperiment2.Errorbarsdenote95%con“denceintervals.A.E.Robinson,V.R.deSa/VisionResearch70(2012)2…6