huamn attention

HumanattntionfiltrsforsinglcolorsPngSuna,b,1,CharlsChubba,CharlsEWrighta,andGorgSprlinga,c,1aDpartmntofCognitiScincs,UnirsityofCalifornia,Irin,CA92697;bDpartmntofPsychology,NwYorkUnirsity,NwYork,NY10003;andcDpartmntofNurobiologyandBhaior,UnirsityofCalifornia,Irin,CA92697ContributdbyGorgSprling,Sptmbr6,2016(sntforriwJun3,2016;riwdbyJulianHochbrgandRichardMShiffrin)ThisualimagsinthyscontainmuchmorinformationthanthbraincanprocssAnimportantslctionmchanismisfaturbasdattntion(FBA)FBAisbstdscribdbyattntionfiltrsthatspcifyprcislythxtnttowhichitmscontainingattnddfatursarslctilyprocssdandthxtnttowhichitmsthatdonotcontainthattnddfatursarattnuatdThcntroidjudgmntparadigmnablsquick,prcismasurmntsofsuchhumanprcptualattntionfiltrs,analogoustotransmissionmasurmntsofphotographiccolorfiltrsSubjctsusamoustolocatthcntroid—thcntrofgraity—ofabriflydisplaydcloudofdotsandrciprcisfdbackAsubstofdotsisdistinguishdbysomcharactristic,suchasadiffrntcolor,andsubjctsjudgthcntroidofonlythdistinguishdsubst(g,dotsofaparticularcolor)Thanalysisfficintlydtrminsthprciswightinthjudgdcntroidofdotsofrycolorinthdisplay(i,thattntionfiltrforthparticularattnddcolorinthatcontxt)Wrport32attntionfiltrsforsinglcolorsAttntionfiltrsthatdiscriminatonsaturatdhufromamongsnothrquiluminantdistractorhusarxtraordinarilyslcti,achiingattnddunattnddwightratios>20:1Attntionfiltrsforslctingacolorthatdiffrsinsaturationorlightnssfromdistractorsarmuchlssslctithanattntionfiltrsforhu(ginqualdiscriminabilityofthcolors),andthirfiltrslctiitisarproportionaltothdiscriminabilitydistancofnighboringcolors,whrasinthsamranghuattntionfiltrslctiityisirtuallyindpndntofdiscriminabiltyfaturbasdattntion|colordiscrimination|slctiattntion|cntroidstimationThisualworldonourplantisincrdiblycomplxThmainmchanismforslctingasubstofthnironmntalisualinformationtoprocssisorintingourhadandbodyandpointingourystowardthmostsignificantlocationsEnthn,achrtinalimagcontainsnormouslymorinformationthanthbraincanprocssTofurthrconstrainthflowofinformation,prcptualattntionprocsssslctlocationsinspac,intralsintim,objctsthatcontainparticularfaturs,andultimatlyparticularcomplxobjctsforfurthrprocssingHr,wfocusonattntiontoaparticularcolor,aninstancoffaturbasdattntion(FBA)ItiswllstablishdthatFBAopratsbroadlyacrossspac,hightningsnsitiitytothattnddfaturnatlocationsthatarirrlanttothtaskathand(1–4)WxploitthisglobalproprtyofFBAinacntroidparadigmtodrihumanattntionfiltrsforcolorJustasthphysicaldscriptionofacolorfiltrdscribsthrlatitransmissionofthfiltrforachwalngthoflight,acolorattntionfiltrdscribsthrlatiffctinsswithwhichachcolorinthrtinalinputultimatlyinfluncsprformancUsingthCntroidParadigmtoDriAttntionFiltrsInacntroidtask,asubjctiwsabriflyprsntdstimuluscontainingacloudofdotsThtaskistothnmousclickthapparntcntrofthdotcloudDrwtal(5)firstusdacntroidtasktostimatattntionfiltrsThprocdurwassubstantiallyrfindbySuntal(6)intothcntroidparadigmusdhrSubjctsattmpttolocatthcntroid(cntrofgraity)ofonlythositmsthathaatobattnddfatur(targts)andattmpttoignorthnontargts(distractors)Inthprsntxprimnts,thtargtsarthrdotsofthtobattnddcolor,andthdistractorsar21dots,thrachofsnothrcolors(Fig1A)Insofarassubjctscanprfctlyattndtotargtsandcompltlyignordistractors,andcanprfctlycomputthcntroidsofthattnddtargts,thjudgdcntroidlocationwouldbthcntroidofthtargtsplusrandomrsponsrror(nois)thatisindpndntofthdistractors’colorsandlocationsInrality,howr,rportdcntroidsarinfluncdtosomxtntbyryitminthstimulus,rflctingthsubjct’sinabilitytoslctilyattndxclusilytotargtsandtocompltlyignordistractorsByanalyzingjudgdcntroidlocationsfrom100orsotrials,thrlatiwightofachcolor’scontributiontothcntroidjudgmnt—thattntionfiltr—canbinfrrdBcausatrialtaks3–4s,anattntionfiltrcanbroughlystimatdin5–7minOfcours,mortrialswouldgigratrprcisionThinfrncprocssisillustratdbythmodlofslctiattntionillustratdinFig1BparticularizdforthprsntxprimntsOnachtrial,astimuluscontainingthrdotsofachofightcolorsisprsntdThattntioninstructionistoassignqualwighttodotsofthtobattnddcolor(thrtargts)andzrowighttothrmaindr(21distractors)LtthightcolorsbrprsntdasCi,i=f1,2,⋯,8g,andltthtargtcolorbCkThn,thtargtattntionfiltrϕkisϕkðiÞ=1ifi=kandϕkðiÞ=0ifi≠k(i,byconntion,thtotalattntionwightis10)Thmodl’srprsntationofahumansubjctattndingtocolorCkhasanattntionfiltrfkðiÞthatcaussthmodl’sprdictdcntroidstobstmatchthobsrdcntroidsThoptimallyprdictimodlfiltrfkðiÞisobtaindbysimpllinarrgrssion(MatrialsandMthods)Thmodl’soptimallyprdictifiltrfkðiÞiscalldthobsrdattntionfiltrIttypicallyisarygoodprdictorofasubjct’sobsrdcntroidjudgmnts†Thrfor,wsayforshortthatfkðiÞisthsubjct’sattntionfiltrforattndingtocolorCkinthatcontxtSignificancThysprsntthbrainmuchmorinformationthanitcouldpossiblyprocssOnimportantwaytoprioritizinformationisbyslctiattntiontofaturs,procssingonlyitmscontainingthattnddfatursandblockingothrs(i,forminganattntionfiltr)Hrwdmonstratanxtrmlyfficintparadigmandapowrfulanalysistoquantitatilymasur,asaccuratlyasonmightmasurphysicalcolorfiltrs,32suchhumanattntionfiltrsforsinglcolorsThsdataaranssntialbasisforathoryofattntiontocolorThcntroidparadigmitslf,bcausitquicklyandquantitatilycharactrizsbasicattntionprocsss,hasnumrousapplicationsAuthorcontributions:PS,CC,CEW,andGSdsigndrsarch,prformdrsarch,analyzddata,andwrotthpaprRiwrs:JH,ColumbiaUnirsity;andRMS,IndianaUnirsityThauthorsdclarnoconflictofintrst1TowhomcorrspondncmaybaddrssdEmail:pngsun@nyuduorsprling@uciduAmorralisticmodlofasubjct’simprfctcntroidcomputationimprosthprdictionofthobsrdcntroidjudgmnts,butwfindthatitdosnotimprothstimationofthsubjct’sattntionfiltr,andsowomitthatcomplication†Sthfirstfootnotwwwpnasorgcgidoi101073pnas1614062113PNASEarlyEdition|1of9PSYCHOLOGICALANDCOGNITIVESCIENCESPNASPLUSOriwWmasurdattntionfiltrsforsinglcolorsThsignificantprcptualattributsofcolorarhu,saturation,andlightnssThattributsofisolatdcolorsonanutralbackgroundarfullydscribdbythrlatistimulationoflong,mdium,andshortwalngthsnsitirtinalcons,hrrprsntdasa3Dspac(Fig2A)Inthisrprsntation,huisrprsntdasacirclinanquiluminantplanwithtwoorthogonalaxs,rd–grn(R–G)andblu–yllow(B–Y);thothr(nonorthogonal)axisisblack–whit(Bl–Wh)Indiffrntblocksoftrials,wtstdthfourstsofightcolorsmbdddinth3DconcolorspacshowninFig2A,whichalsoshowsanxamplstimulusfromachstUsingthcntroidparadigm,attntionfiltrswrdridforachof32diffrntattnddcolors,achinthprsncofthsndistractorcolorsfromitsstFig3Bshowsthrsults:32masurdslctiattntionfiltrsforachoffisubjctsAttntionfiltrsforcolorsonthhucirclarrmarkablyprcis,similarforallsubjcts,andmuchmorslctithanattntionfiltrsalongthsaturationaxs(R–GandB–Y)orlightnssaxis(Bl–Wh)Todtrminwhthrthbttrhuattntionfiltrswrdutostimulionthhucirclbingmordiscriminablthanthstimulionthaxs,Exp2masurdthdiscriminabilitydistancsbtwnthdiffrntstimuliintrmsofthnumbrofjustnoticabldiffrncs(JNDs)sparatingthmFiltrslctiitywascloslyrlatdtoJNDdistancsforstimulialongththraxsbutnotforhustimuliInExp3,colorswrarrangdaroundthhucirclbutwiththJNDspacingofthaxiscolorsThJNDarrangmntalongthhucirclwasnormouslymoradantagousthanthsamJNDspacingalongthBl–Whaxis,confirmingExp2:Inthcolorattntionsystm,attntionfiltrsforaspcifichuarfarmorffctiandfollowdiffrntrulsthanattntionfiltrsthatslctforacolormbdddwithinonofthothrdimnsionsofcolor(saturationandlightnss)GnralProcdursThrwrightattntionconditionsforachstofstimuli,achcorrspondingtoattndingto(i,judgingthcntroidofdotsof)justonthightdiffrnttargtcolorsandignoringthothrsEachattnddcolorwaststdinsparatcountrbalancdblocksoftrialsRandomlychosnstimulifromthsamstwrusdinallightattntionconditionswithinthatstBcausthstimuliwrchosnfromthsamurnforallightdiffrntattntionconditionswithinast,anystatisticallyrliabldiffrncsinrsults(i,diffrncsbtwnthobsrdattntionfiltrs)arduntirlytothdiffrntattntioninstructionsThtrialprocdurisillustratdinFig2C–F(sMatrialsandMthodsfordtails)Exp1:AttntionFiltrsforFourStsofEightSinglColorsFig3Ashowsthattntionfiltrachidbyanaisubjct,S1,foratypicaltargtcolorinachofthfourstimulusstsThillustratdtargtcolorwasarbitrarilychosnasththird(ofight)colorsinachstimulusstForthhust,S1’sattntionfiltrquitcloslyapproximatsthtargtattntionfiltr,butS1’sattntionfiltrsarfarfromthtargtfiltrsforthothrstimulusstsSuntal(6)introducdfiquantitatimasurstoaluatattntionfiltrs,twoofwhicharusdhrTh“slctiityratio”isdfindasthfiltrwightofthtargtcolordiiddbythmanwightofallofthdistractorcolorsAslctiityratioof20forahufiltrmansthatadotofthtargtcolorhas20timsthinflunconthcntroidjudgmntasadistractorcolorddotEfficincyisdtrmindbygiingaprfctcntroidcomputrthsubjct’sachidattntionfiltr,showingthcomputrthsamstimuliasthsubjctxcptthatsomstimulusdotsarrandomlyrmodThfractionofrmainingstimulusdotsthatthcntroidcomputrndstoprformwiththsamaccuracyasthsubjctisthsubjct’s“fficincy”‡Anfficincyof90%forthhuattntionfiltrmansthatwhn2of24dots(83%)arrandomlyrmodfromthstimulusthprfctcntroidcomputrisslightlymoraccuratthanthsubjct,butwhnthrdots(125%)arrmoditdosworsAltrnatily,90%fficincymansthattharagnumbrofdotsusdforthcntroidjudgmntis27(of3)Clarly,subjctS1’s20:1slctiityand90%fficincydscribanxtrmlygoodhuattntionfiltrCompardwithattntionfiltrforstimulus3inthhust,attntionfiltrsforstimulus3inthR–G,B–Y,andBl–WhstimulusstsarsignificantlyworsbothintrmsofslctiityandfficincyCompardwiththnarrowtuningfunction(slctiity=20)ofS1’sattntionfiltrforstimulus3inthhust,thtuningsoffiltrsforS1’sR–G,B–Y,andBl–Whstimulusstsarquitbroad,consistntwiththirrlatilylowrslctiityratios:38,41,and28(Fig3)WhnthprfctcntroidcomputrishandicappdwithS1’sattntionfiltr,itcanmatchthaccuracyofS1’sjudgdcntroidsbyusingonlyslightlymorthanhalfthstimulusdots(i,fficincisof55%,61%,and51%forththraxisstimulusstsAttntionfiltrsforallsubjctsandstimulusstsarshowninFig3BForthcolorsonthhucircl,allofthcolorattntionfiltrsarsharplytundThattntionfiltrsforththrcolorstsonththrcoloraxsarsimilartoachothrandarmuchlssslctiandmuchmorariablthanattntionfiltrsforhuAttntionfiltrsatthtwondsofacoloraxisandthattntionfiltrsforcolorsclossttothgraybackgroundarthmostslctifiltrsFig3Cshowsasummaryofthattntionfiltrs’slctiitis(Fig3C,Top)andfficincis(Fig3C,Bottom),aragdacrossallsubjctsFiltrslctiitisandfficincisforcolorslyingonthhucirclargnrallygratrthanthosforcolorsonthaxsAgain,forcolorattntionfiltrsforcolorsonththraxs(R–G,B–Y,andBl–Wh),slctiitisandfficincisargratrforcolorsatthtwoCntroidof3targtdots(Orang)Cntroidofall8dotsAInputImagFiltrWightsCntroidComputationOutput:x,yRsponsB10Fig1Thcntroidparadigm:procdurandanalysis(A)Examplofasubsctionofadisplaythatshowsthcntroidoforangtargtsamongdistractorsofdiffrntcolors(B)ModlofthattntionwightdcntroidcomputationThinputimagconsistsofdotsofdiffrntcolorsFiltrwightsshowsthfiltrwightsinducdbyattntioninstructions(g,attndingtoorang)whnprformingthtaskThfiltrisdridsuchthatitbstprdictsthsubjct’scntroidjudgmntThcntroidcomputationcomputsthcntrofgraity(x,y)ofthdotlocations,wightingachlocationbythfiltrwightforthcolorofthdotatthatlocationThoutputx,yisthmodl’sprdictionforthatsubjct’sjudgdcntroidforthisparticularinputimag‡SSuntal(6)fordtails2of9|wwwpnasorgcgidoi101073pnas1614062113SuntalndsofththrdiagonalaxsandforcolorsthatarnarthbackgroundgrayApossiblxplanationforthparamtricproprtisofthcolorattntionfiltrsisxplordinthnxttwoxprimntsExp2:ThDpndncofColorAttntionFiltrsonTargt–DistractorDiscriminabilityColorattntionfiltrsforsomcolorsarmorslctithanforothrsIsthissimplybcausthstargtcolorsarfurthrawayfromthdistractorcolorsinsomprcptualspacAnobjctimasurofprcptualdistancbtwntwocolorsisthnumbrofJNDssparatingtwocolorsincolorspacTomasurJNDdistancs,thspacalongth(inisibl)linincolorspacbtwnachpairofnighboringstimuluscolorswassampldatariouspointsAtachsamplpointincolorspactwonarbycolorswrchosnandassigndtotwostimulusdotsWmasurdthlocalJND(i,thminimumdistancincolorspacbtwnthnarbycolorsthatnabld82%discriminationofthcolordiffrncbtwnthtwodots)IntrpolationfromthcollctionoflocalJNDsnablsstimatingthnumbrofJNDsthatsparattwostimuluscolorsJNDdistancswrmasurdforalladjacntcolorpairsforthrofthoriginalfisubjcts—twowrunaailablbcausthyhadgraduatdThsJNDmasurmntswranordrofmagnitudmortimconsumingthanthattntionfiltrmasurmntsThJNDlocalisolationofatargtcolorisdfindas1=ð1=d1+1=d2Þ,whrdiisthJNDdistancbtwnitanditstwoadjacnt(i,mostconfusabl)colorsForbritywomit“local”andrfrsimplytoJNDisolationForthndcolorsonththraxsthJNDisolationissimplythJNDdistancfromthonlyadjacntcolor,conformingwiththabodfinitionBcausthJNDisolationmasursthdiscriminabilityofthtargtcoloragainstitsadjacntcolor(s),tomakcomparisonsplausiblwcomputdanwfiltrslctiitymasurasthtargtwightdiiddbythmanwightofjustnarstdistractorsWcallthisnwmasurlocalslctiityandcomparittoJNDisolationforachcolorFig4Ashowsthcomparisonsbtwncolorattntionfiltrlocalslctiity(targtadjacntdistractorwight)andJNDisolationThmanlocalslctiityandthmanisolationratioaragdforththrsubjctsargininFig4BWhrasFig4AandBshowlocalslctiityandJNDisolationofcolorattntionfiltrsasfunctionsoftargtcolor,Fig4Cshowsdirctlythrlationshipbtwnthtwodpndntariabls,JNDBCDEF1s300msInfInfHuBlWhRGBYSLMHuBlWhBluYllowRdGrnAL+M+S321Fig2AdiagrammaticrprsntationofthfourstsofstimuliusdinExp1,samplstimuli,andthtstingprocdur(A)Thstimulirprsntdina3DspacdfindintrmsofrtinalconsCircls:thighthuststimulithatliquallyspacdonacirclinanquiluminantplanSquars:1,stimuliinthB–Ystthatlionadiamtrdfindbythshortwalngth(S,blucon)snsitiityfunction;2,thR–Gststimulithatlionadiamtrprpndiculartothbluconaxiscorrspondingtothlongwalngth(L,rd)minusmdiumwalngth(M,grncon)snsitiityfunctions;3,thachromaticststimulithatlionalin(L+M+S),commonlycalldthBl–WhaxisNotthatthachromaticlinisnotprpndiculartothquiluminantplan(thprpndicularaxistothquiluminantplanisthL+Maxis,whichisyllowish,notachromatic)StimulialongachradiusarspacdquallyinconspactousthmaximumcolorrangaailablonthmonitorAnxamplstimulusfromachofthfourstimulussts:hu,Bl–Wh,R–G,andB–Y(B)ExamplofaninstructiondisplayshownbforablocksssionstartsSubjctsarinstructdtomousclickthcntroidofththrdotsofthsinglcolorinthtoprowandtoignorthothrdots(sncolorsinthbottomrow)(C–F)Stimulussqunconachtrial(C)Fixationfram,1s(D)Stimulus,300ms(E)Rsponsdisplayshownuntilarsponsismad(F)Fdbackframthatshowsthstimulus,thtargtcntroid,thsubjct’srspons,thtargtcolor(xtrmlft),andthtobignordcolors(xtrmright)ThfdbackframisdisplayduntilthsubjctinitiatsthnxttrialSuntalPNASEarlyEdition|3of9PSYCHOLOGICALANDCOGNITIVESCIENCESPNASPLUSisolationandlocalslctiityFig4CmaksitidntthatwhrasthrisastrongpositicorrlationbtwnlocalslctiityandJNDisolationforaxiscolors,thrisnosuchcorrlationforhucolors,ninthsamrangofJNDisolationsExp3furthrxplorsthisdiffrncbtwncolorattntionfiltrsforhuandcolorattntionfiltrsforthothrdimnsionsofcolor124816326412801050105010501050105Filtrwights00204060810020406081Slctiityratio=20Efficincy=90%Slctiityratio=38Efficincy=55%Slctiityratio=41Efficincy=61%Slctiityratio=28Efficincy=51%FiltrwightsabcdABSlctiityratioEfficincyCHuAchRGYB123456780020406081024681234567812345678StimuluscolorsHuBlWhRGBY0051Fig3Fisubjcts’attntionfiltrs,filtrslctiityratios,andfiltrfficincisfor32attntionconditionsInallpanls,thightcolordcirclsonthabscissarprsntthightcolorsinthparticularst(A)Fourxampltargtandachidattntionfiltrs;achfiltrcorrspondstoaparticulartargtcolorinonofthfourstimulussts:(a)hust,(b)achromatic(Bl–Wh)st,(c)R–Gst,and(d)Y–BstDashdandsolidcursartargtandthsubjct’sachidfiltrsFiltrcursarpaintdinthcorrspondingtargtcolorsErrorbarsthroughoutFig3rprsnt95%confidncintrals(B)Attntionfiltrsoffisubjcts(sparatdinrows)forfourstsofstimuli(sparatdincolumns)Columns(fromlfttoright)arhust,Bl–Whst,R–Gst,andY–BstWithinachpanl,achcolordcurcorrspondstothachidattntionfiltrforthatparticulartargtcolorFiltrcursforallofthcolorswithinastimulusstarshowninonpanlalthoughthyardridfromdatacollctdinsparatblocksoftrials(C)Manfiltrslctiitis(targtdistractorfiltrwights)andfficincis(fractionoftargtsusd)aragdacrossfisubjctsThcolorsofpointsarthcorrspondingtargtcolors;thlincolorsararbitrary4of9|wwwpnasorgcgidoi101073pnas1614062113SuntalExp3:ColorAttntionFiltrsforSpacingAdjustdHuColorsOnpossiblxplanationforthadantagofcolorattntionfiltrsforthsaturatdhucirclcolorsrsusthR–G,B–Y,andBl–WhaxiscolorsisthathucolorsarmorwidlysparatdinprcptualcolorspacInExp3,forachofththrsubjctsightnwquiluminantcolorswrchosnonthhucirclsothatthJNDdistancsbtwnthmwrxactlythsamasthJNDdistancsbtwnthightstimulionthBl–WhaxisSlctiityratiosofthcolorattntionfiltrsforthsightspacingadjustdcolorsarshowninFig5ThnormousattntionslctiityadantagofhucolorfiltrsorBl–WhattntionfiltrsisstillobsrabldspitaJNDspacingofthighthucolorsthatisxactlyquialnttothspacingofthightBl–WhcontrastsAsinExp2,thslctiityofattntionfiltrsforspcifichussmstobirtuallyindpndntofthdiscriminabilitydistancofthnarstnighbors,quitdiffrntfromattntionfiltrsforspcificBl–WhcontrastsThspcialproprtisofattntionfiltrsforspcifichusmostliklyrflctthatgratrcologicalalidityofhu(rsussaturationorlightnss)asaclutothtrunaturofasurfacthatdmandsattntionDiscussionThbhaioralffctsofFBAhamostoftnbncharactrizdintrmsofithrarducdractiontim,whnsarchingforattnddfaturs,orarducdthrsholdfordtctinganddiscriminatingattnddfaturs(7,8;srf9forariw)ThsmasursralhowmuchmorffctiFBAmaksthattnddfaturcompardwithwhnthsamfaturisnotattnddHowr,othrimportantaspctsofFBAhabnlftlarglyunaddrssdForxampl,towhatdgrdosFBAinfluncthffctinssofirrlantisualfatursThsaspctsofFBAarliklytobspciallyimportantinnaturalsttingsinwhichirrlantfatursaboundParadigmaticEfficincyToprobfaturattntionquantitatilyandfficintlywdlopdthcntroidparadigm(5,6)ThcntroidparadigmnablsthsimultanousmasurmntofthrlatiwightoftargtitmsinthprsncofothritmsinanutralattntionconditionandinanattntionconditionwhnaparticulartargttypisattnddandthothritmtypsardistractorsInbothsituations,thcntroidparadigmproidsthrlatiwightsofallofthothritmtypsrlatitothtargttyp(s),annormousimpromntinboththamountofinformationgaindandinthparadigmaticfficincyofgainingthisinformationcompardwithpriousFBAparadigmsInprincipl,ahumanslctiattntionfiltrforahucanbdtrmindasaccuratlyasthtypicalphysicalspcificationofanopticalcolorfiltrforacamraMasuring32colorattntionfiltrswouldhabnimpossiblwithoutsuchahighlyfficintparadigmSomintrstingproprtisofFBAmrgdasarsultExtrmSlctiity,thSpcialStatusofHuThprincipalrsultofthxprimntsisthatattntiontohu,asmasurdinthcntroidparadigm,isxtrmlyslctiInExp1,slctiity(thwightratioofthattnddhudotsdiiddbytharagwightofthsndistractorhudots)xcdd20:1inmostcassInExp2,thlocalslctiity(wightratioofthattnddhutothmostsimilardistractorhu)xcdd10:1formostr=092L=1372r=096L=721r=091L=608r=080L=329JNDlocalisolation0481216Localslctiityratio1416JNDlocalisolation10203001020300JNDlocalisolationLocalslctiityratio0481216116644Localslctiityratio216328428322832283264ACBAgS1S4S5048121604812161416141632323212345678Stimuluscolors123456781234567812345678SELJNDFig4Exp2:targtattntionfiltrlocalslctiityratiosandtargtJNDisolationfor32targtsandthrsubjcts(A)ColorattntionfiltrlocalslctiityandcolorJNDisolation(rightandlftordinats)asafunctionoftargtcolor(abscissa)forthrsubjcts(rows)Inachpanl,thblucur(withdatapointsindicatdby+)plotsJNDisolation(stxtfordfinition)Magnta,black,rd,andyllowcursconnctslctiityratiosforhucolors,contrasts,RGandBYcolors,rspctilyFromlfttoright,subfigursarforhu,Bl–Wh,RG,andYBcolorsts(B)Arag(orththrsubjcts)slctiityratiosandJNDisolationsThcorrlationsrbtwnthstwoaragmasursargininachpanlListharagoralllngthinJNDsofthtrajctoryformdbyachstofcolors(C)ScattrplotsillustratingthrlationshipbtwntargtJNDisolation(abscissas)andtargtlocalslctiityratio(ordinats)forththrsubjctsandtharagThcolorofachpointindicatsthtargtcolorDatapointsfromthsamstimulusstargroupdbyllipsscolordasfollows:hu,magnta;Bl–Wh,black;RG,rd;andBY,yllowSuntalPNASEarlyEdition|5of9PSYCHOLOGICALANDCOGNITIVESCIENCESPNASPLUSattnddhusWhnquiluminanthusarmatchdindiscriminabilitydistanctoBl–Whstimulithatdiffrincontrast,tharagattntionslctiityforhuis96timsgratr(aragdforththrtstdsubjcts)Asnotdabo,thgratattntionadantagofslctiityforahursusslctiityforaaluwithinaBl–Wh,R–G,orB–Yaxisprobablyiscological:Bcausofrflctancdiffrncscausdbyshadows,angloflighting,andanglofiw,huisamorrliablcutomatrialidntitythanithrlightnssorsaturationAscondsignificantspcialproprtyofattntiontoahuisthatattntionslctiityforahuisalmostindpndntofthJNDdiscriminabilitydistancofnighboringdistractorhus,whrasinthsamdiscriminabilityrangofthBl–Wh,R–G,andB–YstimulusstsattntionalslctiityandJNDdistractordistancarhighlycorrlatdWhnthJNDdistancsbtwnhuswrrducdinExp3,attntionslctiitystartdtoshownhancddpndncisonJNDdiscriminabilityfortwooutofthrsubjctsItisliklythatJNDdiscriminabilitydtrminsattntionalslctiityforrycloslyspacdcolors,butJNDdiscriminabilityquicklybcomsmuchlssrlantwhnhusarmordiscriminablfromachothrWhatdtrminsattntionalslctiityforhusisanimportantqustionbutisbyondthscopofthcurrntstudyNuralEnhancmntofanAttnddFaturAdominantthoryofFBAstmsfromthnuralcorrlatsofFBA(10)Thrsponsmagnitud(impulssprscond)ofnuronsinthmiddltmporalcorticalaraofmacaqumonkyswasmasurdforstimulimoinginattndd,nutral,andunattnddmotiondirctionsWhnamonkyattnddtothprfrrdmotiondirctionforanuron,thnuron’srsponsmagnitudincrasdbyafactorofabout115rsusnutralattntion;attndingtothantiprfrrdmotiondirctionrducdnuralrsponssby094ItwasproposdthatFBAhadamultiplicati(amplifying)ffctonnuralrsponsmagnitudThrarsimilarfindingsofsuchsmall(rlatitothprsntxprimnts)butstatisticallysignificantffctsofattntioninthamplitudofrsponssinsinglnurons(11),EEGwaformamplituds(12,13),andonfMRIBOLDrspons(1)DiffrntProcssingLls:AttntiontoColorPrcdsAttntiontoMotionThmultiplicatiffctofattntiononnuronsthatrportdirctionofmotionopratsatadiffrntbrainprocssingllthanthffctofattntiontocolorNuronsthatarinfluncddirctlybyattntiontocolorrprsntthinputstomotioncomputingnuronsBlasrtal(14)masurdattntiontoacolor(rdrsusgrn)inanambiguousmotionparadigmandalsofoundarlatilysmallamplificationfactor(typically13,rang12–22)ThnormouslysmallrffctofattntiontohuinanambiguousmotionparadigmrsusacntroidtasksuggststhatattntionslctiityarisnormouslybtwntasksthatinoldiffrntbrainprocsssAcriticaldiffrncbtwnthcntroidtaskandthambiguousmotiontaskofBlasrtal(14)isthatgroupingofitmsmaybryimportantinthcntroidtask(15,16)butisirrlantinthambiguousmotiontaskThatis,FBAsmstohaanspciallybiginfluncongroupingprocsssDiffrntProcssingLls(MultiplObjctTrackingVrsusStatisticalSummaryRprsntation)DpndingonNumbrofTargtItmsTypically,acntroidjudgmntisassumdtobastatisticalsummaryrprsntation(SSR),thatis,astatisticthataccuratlydscribsaproprtyofagroupofitmsnwhnthrarsomanyitmsthatthsubjcthasaccuratinformationonlyaboutfw,ifany,ofthindiidualitms(17)Forxampl,injudgingthcntroidof16itms,subjctsachifficincisofabout08(6),whichmansthatanidaldtctorwouldhatoknowthprcislocationof08×16=128itmstomatchthsubjcts’prformancsThisisthrorfourtimsthnumbrofitmsthatsubjctscanidntifythlocationsofHowr,whnthnumbrofitmsissmalladiffrntprocsscomsintoplayForxampl,inasarchxprimnt,thrtypicallyisonlyontargtitmandthsubjcttypicallyknowsthlocationandothrproprtisofthsingltargtInrsotal(18)foundthatsubjctsarasilyandaccuratlyabltofindthcntroidoftworticallinsrandomlyplacdamongtwohorizontaldistractorsand,similarly,thcntroidoftwohorizontallinsamongtworticaldistractorsVrysurprisingly,subjctswrunabltofindthcntroidoffour,six,orightrticalsamonganqualnumbrofdistractors,althoughthysuccddasilywithblackrsuswhitattntionalslctionwiththsamstimulusstsThpointhrisnotthpointofInrsotal(18),thatthrissomthinguniqulydifficultaboutabsolutrticalrsushorizontalattntionalslction,butrathrthatthrisadiffrntjudgmntprocsswhnthnumbroftargtsislssthanfourInariousxtnsilystudidtrackingprocdursinwhichanumbroftargtsmorandomlyamongrandomlymoingdistractors,subjctstypicallycantrackaboutfourtargts(19,20);thytrackfwrtargtswhndisplaysarmorcomplx(21)ThnumbroftargtsinallofthprsntattntionxprimntswasthrThrfor,winfrthatthsubjcts’judgmntsprobablydidnotxclusilyrlyonSSRs—ifatall—butalsoinoldthirabilitytokptrackofthlocationsofthrtargtsThisagrswithoursubjcts’introspctirportsThosquridsaidthywrawarofthactuallocationsofthtargtdotsThisdosnotchangthsignificancofthcolorattntionfiltrsthatwrobsrd;rathr,itinformsthprocsssbywhichthcntroidswrcomputdWhthrthsamcolorattntionfiltrswouldbobsrdinparadigmsthatinoldlargrnumbrsofdotsandrquirdSSRsisamattrforfuturinstigation841216412164812JNDlocalisolationS1S4S51166441166441416S1S4S5rlum=090rcol=042rlum=093rcol=058rlum=094rcol=0091020030182416124818216324JNDlocalisolationDISCRIMINABILITYATTENTIONLocalslctiityratio643216328StimuluscolorsFig5Exp3:slctiityratioandJNDisolationdridfromattntionfiltrsforaBl–WhcolorstandanquialntJNDspacingadjustdhucolorst;dataforthrsubjcts(Lft)BlackcursarJNDisolationforthBl–WhstimuliandalsothJNDisolationforhusspacdalongthcolorcircltohaxactlythsamJNDisolationasafunctionoftargtcolorBlucursarthslctiityratioforBl–Whtargts,whichishighlycorrlatdwithJNDisolationMagntacursarthslctiityratioforthhutargts(Right)SlctiityratiorsusJNDisolationforthtwoquialntlyspacdstsofcolortargtsThgrayllsandcolorsofthpointsindicatthcontrastsandcolorsofthtargtstimuliSlctiitycorrlatsstronglywithJNDisolationforBl–WhtargtsbutnotforidnticallyJNDspacdhutargts6of9|wwwpnasorgcgidoi101073pnas1614062113SuntalSimilarProcssingofDarkandLightTargtItmsItiswlldocumntd(22)thatforthsamcontrastratiomagnitudofWbrcontrastsdarkrthanbackgroundstimuli(“dark”)armordiscriminablthanlightrthanbackgroundstimuli(“light”)Figs4Cand5showthmasurdJNDsparationsbtwnbothadjacntcontrastlightandadjacntcontrastdarkspotsand,forthsampairsofadjacntdotcontrasts,thlocalattntionslctiityincntroidjudgmntsConsidrcorrspondingpairsofdarkandlightspotsthatarquallydistantfromthnutralbackground(g,−14,+14;−12+12;tc)Althoughitisdifficulttosinrducdsizfigurs,innarlyallcassthdarkpointlistothrightofthcorrspondinglightpoint,indicatingagratrJNDsparationforadjacntdarkrsuslightcontrasts,andinnarlyallcassthdarkpointinthgraphlisabothcorrspondingwhitpoint,indicatinggratrattntionslctiityfordarkrsuslightThrfor,inFigs4Cand5,withinmasurmntrror,boththdarkandspotslionthsamdiagonallinThatmansthattntionslctiityadantag(gratrhightinFigs4Cand5)ofdarkrsuslightspotsisntirlydutothirgratrJNDsparation,notdutodiffrntparamtrsforattntiontodarkrsusattntiontolightExamininglightanddarkspotssparatlyshowsagainthatattntionslctiityforlight–darkcontrastsisgratlydpndntonthirJNDsparationandismuchworsthanattntionslctiityforhu,whichisbttrandirtuallyindpndntofthtargt–distractorJNDsparationforprcislythsamsparations(Fig5)AThrsholdforAttntionFiltringAnintrstingaspctofthlight–darkdataisthatthxintrcptofthlogslctiityrsusJNDisolationgraphsforBl–WhstofspotsisgratrthanzroAgratrthanzroxaxisintrcptmansthat,blowathrsholdnumbrofJNDs(thintrcpt),thattntionsystmisinopratiInFig4C,thattntionthrsholdsmstoblowrforthR–G(aboutfiJNDs)andY–B(aboutthrJNDs)stimulusststhanforthBl–Whstimulusst(attntionthrsholdaboutsnJNDs)FBAActsonLocationRathrthandirctlyassssingthprcptionoffaturs,thmasurdffctofattntiontocolorinthcntroidparadigmaswllasattntiontocolorinmotionparadigmsisonthlocationsofthfatursThatis,bothcntroidandmotioncomputationsdpndonlocationcomputations;fatursntrintothscomputationsonlyindirctlyinsofarasthychangthwightsassigndtodiffrntlocationsSimilarly,ShihandSprling(23)inarapidstramofisualdisplayssarchxprimntconcluddthat“slctiattntiontoaparticularsizorcolordosnotcausprcptualxclusionoradmissionofitmscontainingthatfatur;itactsbyguidingsarchprocssstospatiallocationsthatcontainthtobattnddfaturs(rf23,p758)”Prioritizinglocationsis,infact,thguidingprinciplincontmporarycomputationalthorisofisualsarch(g,rfs24–26)Toconclud:Althoughcntroid,motion,andsarchtasksarusdtomasurFBA,countrtointuition,thstasksdonotmasuradirctffctofFBAonthprcptionofthattnddfaturs;thymasuranindirctactionofFBAinslctilywightingthlocationsoffatursNonlocationMasurmntsofFBAInparadigmsothrthancntroid,ambiguousmotion,andsarch,thattnddfaturdosnotncssarilyinollocationAmongthparadigmsthatmasurffctsofFBAothrthanitsffctonlocation,thmostprominntisadaptationofthattnddfatur,mostcommonlyadaptationtoaparticularmotiondirctionasraldbythmotionaftrffct(g,rf27)AdaptationtogratingorintationalsoarisdpndingonFBA(28)AusfuldmonstrationofFBAoccursinrptitiondtctionofdigitsthatsharcolor,spatialfrquncy,orintation,siz,orcombinationsofthsfaturs(29)Thxtnttowhichandhowthasytunmasurd,implicitattntionfiltrsactiatdbythsmanyFBAparadigmsrlattoattntionfiltrsdridfromcntroidjudgmntsisacriticalunanswrdqustionMatrialsandMthodsApparatusThxprimntwasconductdonaniMacintlcomputrrunningMATLABwithaPsychotoolboxpackag(30)Thbuiltin23in,60Hzfrshrat,LEDmonitorwith1,920×1,080rsolutionwasusdtodisplaythstimuliThluminanclookuptablcontaind256grayllsgnratdbyastandardcalibrationprocdurThmanluminancofthmonitorwas521cdm2Stimuliwriwdatafixddistancof60cmGnratingTstColorsThightquiluminanthucolorswrdridusingaminimummotionparadigm(31–33)Subjctsadjustdthlightnssofacolorsothatthcontrastbtwnthcolorandthmonitor’smanluminancproducdnomotion,onlyflickrThatis,allcolorswrmadquiluminanttothmanluminancofthmonitoraccordingtothluminancsnsitiityofmotionsystmThndpointsofthR–G,Y–B,andBl–WhaxswrthlargstaailablcontrastsaailablonthmonitorInadditiontobinginthquiluminantplan(asdtrmindforachparticipantindiidually),thgamutoflightsalongthR–GaxisischosntoproducinariantactiationofthScons[asdfindbythStockman–Sharp(34),2°fundamntal],andthY–Baxisisorthogonal(inconactiationspac)tothR–GaxisThsconstraintsxhaustththrdgrsoffrdomofthcolordisplaymonitorThcoordinatsofthmonitorscrnwrspcifidinthconspacThightBl–Whcontrastswr−1,−075,−05,−025,025,05,075,and1,whr−1isblackand+1isthmostintnswhit,andthmonitorbackgroundis0ThothrhalfaxsR,G,Y,andBalsowrquallydiiddinphysicalconspacThighthuswrchosntoliinacirclinth3Dconspacthatfllonanquiluminantplan(29)ThCntroidExtractionParadigmSubjctsFournaisubjcts(S1–S4)wrundrgraduatpsychologystudntswhowrunawarofthpurposofthstudyThothrsubjct,S5,wasPSAllsubjctsgainformdconsnttoparticipatinthstudyAllmthodsusdwrapprodbythUnirsityofCalifornia,IrininstitutionalriwboardStimuliThstimulusdisplaywas512by512pixls(isualangl121°)cntrdinadisplayof1,920×1,080pixlsDotswrninpixlwidsquars(isualangl021°)Inanyoncondition,onlyonofthfourstsofightcolorswaststdErystimuluscontaindthrdotsofachofthightcolorsTargtanddistractordotlocationswrdrawnfromtwobiariatGaussiandistributionswithdiffrntmansbutthsamSDsof3°Thmansofthtwodotdistributionsrodindpndntly,bothwithSDof07°Ifasamplddotlocationoccurrdwithin10pixlsofanothrdotthlocationwasrsampldsothattwodotsnrorlappdThroingariationimprosthstatisticalpowrofthparadigmandisimprcptiblProcdurBforthmainxprimnts,subjctswrtraindtojudgthcntroidof1,3,6,12,or24dots,rspctilyAftr50trainingtrialsonondotandabout200trainingtrialsonachofthothrdotnumbrs,allwithfdback,subjctswrabltolocatthcntroidryaccuratlyandwrnolongrimproingEachofthfourconditions(hu,Bl–Wh,RG,andBY)waststdinablockddsignThtargtcolorwasfixdwithinablockEachblockcontaind100tsttrialsintrladwith20controltrialsinwhichonlytargtswrshownWiththidntificationnumbrgintoachcolorinFig3Ccolorswrtstdoncinthordr7,8,2,3,1,5,4,6andthnoncagaininthrrsordr—anABBAdsignEstimatingFiltrWightsandFiltrEfficincisLtCibthcolorofthithdotinadisplayandgðCiÞbthwightassigndtoit[implyinggðCiÞisdpndntonCionly]Thmodlstimatsthmanlocationofth24dotarray(itscntroid)withachdotwightdbyitscorrspondinggðCiÞThnthmodl’sx,yrsponsRx,RyisginbyRx=Pi=24i=1gðCiÞxiPi=24i=1gðCiÞ+Qx and Ry=Pi=24i=1gðCiÞyiPi=24i=1gðCiÞ+Qy ,[1]whrQxandQyrprsntnoistrmsLtXk,YkbthsumofthlocationsofallthrdotsofcolorCk;thn,Eq1canbrwrittnasRx=Xk=8k=1fðCkÞXk+Qx and Ry=Xk=8k=1fðCkÞYk+Qy,[2]whrfðCkÞisqualtogðCkÞ=DforDthdnominatorinachofthtwopartsofEq1Eq2dscribsasimpllinarmodlinwhichfðCkÞcanbstimatdbymultipllinarrgrssionSuntalPNASEarlyEdition|7of9PSYCHOLOGICALANDCOGNITIVESCIENCESPNASPLUSFindingthNumbrofJNDsBtwnTargtColorsThJNDisamasurofthcolordiscriminabilitybtwnnarbycolors(hr,colorslyingononofththrcoloraxsoronthhucircl)(Fig2)TomasurasinglJND,twodotsthatwrconfigurdxactlylikthstimulusdotsinthattntionxprimntswrshownforthsamdurationasinthattntionxprimnts(300ms)sparatdby093°cntrdonthscrnOndotisthstandardcoloraroundwhichthJNDismasurd,andthothrisaariablcolorThstandarddothadafixdcoloronasquncoftrials,andthariabldottookonof31(or62)narbycolorsonthcoloraxisofthatstAstaircasprocdurwasusdtofindthtwodotcolordiscriminationthrshold(JND)atachofthightcolorsinacolorstJNDswrmasurdforthfourstsofightstandardtstingcolorsNotthatxcptforthndpointsthrartwoJNDsatachstandardcolor,oninachdirctiononthaxisorhucirclJNDswrmasurdsparatlyforachcolorstForachadjacntcolorpair,thcontinuumofth33colors(twoadjacntcolorsand31colorsbtwnthm)wrshowntothsubjctbforthstaircasprocdurbganInthtstphas,subjctswrtoindicatwhichofthtwodotsinapairwasclosrincolortoadsignatdndofthcolorcontinuumForxampl,tostimatthJNDthrsholdforthcolororangtowardthcoloryllow,subjctswraskdtoindicatwhichdotwasmoryllowThrwasfdbackaftrachtrialForachofththrdiagonalaxs,onadditionalJNDthrsholdwasmasurdaroundthbackgroundgrayInthiscas,onofthtwodotshadacolordrawnfrombtwnonofthunsaturatdaxiscolors(g,unsaturatdrd)andthbackgroundgray,andthothrdothadacolordrawnfrombtwnthothrunsaturatdcolor(g,unsaturatdgrn)andthbackgroundgrayThsubjctsindicatdwhichofthdotswasclosrtoonofthunsaturatdcolor(g,whichonwasmorrddishorgrnish)Twothrdown,onupstaircasswrintrladwithinablocktomasurthJNDthrsholdsforacolordiffrncAstaircastrminatdaftr16rrsalshadoccurrdThinitial10trialsinachblockwrdiscarddThstaircasdatawrfitwithaWibullfunction;thcntringparamtrwasusdasththrshold(i,82%corrctthrshold)ThnumbrofJNDs(i,prcptualdistanc)btwntwocolorswasstimatdusingthmthoddscribdblowOthrwaysofcalculatingJNDisolationsandstimatingprcptualdistancswralsotridbutdidnotaltrthconclusionsDfinitionofthStimulusSpacThstimuliinthprsntxprimntsardscribdindimnsionsthatardfindintrmsofthhumanconspac,thatis,intrmsofthactionspctraofthblu,grn,andrdsnsitirtinalconsOr,asthyarcalldbyisionscintists,thshort(S),mdium(M),andlong(L)walngthconsTodscribstimuliintrmsofconspac,itwasfirstncssarytomasurthspctralcharactristicsofthmonitorthatprsntsthstimuli;thiswasdonbystandardmthodsWdfinthconspacintrmsofthStockman–Sharp(34)2°short,mdium,andlongwalngthconfundamntalsEachofS,M,andLisafunctionthatmapsthwalngths390,395,400,⋯,830nmintoRdscribingacon’srlatiwalngthsnsitiityWrprsntS,M,andLascolumnctorsoflngth89Thn,ltBbththrcolumnmatrixinwhichthfirstcolumnB1isqualtoSrscaldtohanorm1,thscondcolumnB2isqualtoM+LorthogonalizdrlatitoB1andrscaldtohanorm1,andththirdcolumnisL−MorthogonalizdrlatitoB1andB2andrscaldtohanorm1Bdscribsthstimulusspac,throrthogonaldimnsions(B–Y,R–G,andluminanc)dfindbyconsnsitiitis(ThSandL−MdimnsionsarillustratdinFig2A)AstimulusofcolorCwithawalngthspctrumQ(acolumnctoroflngth89)isrprsntdinthstimulusspacbyth3DctorCQ:BTQ=24q1q2q335=CQ,[3]whichisrfrrdtoinformallyinthtxtasa“color”ThprcptproducdbyQdpndsntirlyonCQAnylightsQ1andQ2forwhichCQ1=CQ2willapparidnticaltohumanision(suchlightsarcalldmtamrs)andthprcptualdiscriminabilitytwocolorsCQ1andCQ2isasmoothlyincrasingfunctionofthEuclidandistancbtwnthm;thiswillbimportantinmasuringJNDsCalculatingthNumbrofJNDsSparatingTwoStimuliTocalculatthnumbrofJNDsbtwntwoadjacntstimuluscolorswithspctraQ1andQ2,wmasuronJNDatachndofthlinsgmntinR3btwnCQ1andCQ2andthn,asxplaindblow,infrhowmanyJNDsmustbbtwnthmtofilldistanccordbythatlinsgmntLtabththrsholdEuclidandistancfromCQ1towardCQ2;ltcbththrsholdfromCQ2towardCQ1;andltD=a+b+cbthtotalEuclidandistancbtwnCQ1andCQ2Inallcassinthcurrntstudy,b>0,anda≠cWanticipatthathumanisionmaybmorsnsititoariationsinsomrgionsdirctionsofcolorspacthaninothrrgionsdirctionsHowr,JNDswrmasurdonlybtwnadjacntcolors,sowcanassumthatsuchariationsinsnsitiityargradualnoughthatdiscriminationthrsholdsbtwnthpairsofadjacntcolorsinthcurrntstudyundrgoasmoothtransitionfromatoc,withachsuccssiJNDqualtoafixdscalarαtimsthprcdingonThisimplisthatfornqualtothnumbrofJNDsfromCQ1toCQ2,c=aαn−1,[4]andinthcasinwhichnisanintgr,D=aXn−1k=0 αk[5]Notthatifbwrqualto0,thnEqs4and5wouldimplythatα=c=aand(asdsird)n=2FromEq4,itfollowsthatca=αn−1[6]FromEq5itfollowsthatDa=αn−1α−1[7]Eq7gnralizsEq5tononintgralusofnSubstitutingEq6intoEq7yildsDa=�ca�α−1α−1,[8]implyingthatα=D−aD−c=b+cb+a[9]SubstitutingEq9intoEq6yildsca=�b+cb+a�n−1,[10]fromwhichitfollowsthatthnumbrofJNDsbtwnCQ1andCQ2isginbyn=1+ln c−ln alnðb+cÞ−lnðb+aÞ[11]Toritrat:aisthEuclidandistancofththrsholdfromCQ1inthdirctionofCQ2,cisthEuclidandistancofththrsholdfromCQ2inthdirctionofCQ1,DisthEuclidandistancbtwnCQ1andCQ2,andb=D−a−c[12]ACKNOWLEDGMENTSThisworkwassupportdinpartbyNationalScincFoundationGrantBCS08438971SrncsJT,BoyntonGM(2007)FaturbasdattntionalmodulationsinthabsncofdirctisualstimulationNuron55(2):301–3122ZhangW,LuckSJ(2009)FaturbasdattntionmodulatsfdforwardisualprocssingNatNurosci12(1):24–253LiuT,MancI(2011)ConstantspradoffaturbasdattntionacrossthisualfildVisionRs51(1):26–334WhitAL,CarrascoM(2011)FaturbasdattntioninoluntarilyandsimultanouslyimprosisualprformancacrosslocationsJVision11(6):155DrwSA,ChubbCF,SprlingG(2010)PrcisattntionfiltrsforWbrcontrastdridfromcntroidstimationJVision10(10):206SunP,ChubbCF,WrightC,SprlingG(2016)Thcntroidparadigm:QuantifyingfaturbasdattntionintrmsofattntionfiltrsAttnPrcptPsychophys78(2):474–5158of9|wwwpnasorgcgidoi101073pnas1614062113Suntal7WangY,MillrJ,LiuT(2015)SupprssionffctsinfaturbasdattntionJVis15(5):158StörmrVS,AlarzGA(2014)FaturbasdattntionlicitssurroundsupprssioninfaturspacCurrBiol24(17):1985–19889CarrascoM(2011)Visualattntion:Thpast25yarsVisionRs51(13):1484–152510TruS,MartínzTrujilloJC(1999)FaturbasdattntioninfluncsmotionprocssinggaininmacaquisualcortxNatur399(6736):575–57911BichotNP,RossiAF,DsimonR(2005)ParalllandsrialnuralmchanismsforisualsarchinmacaquaraV4Scinc308(5721):529–53412MüllrMM,tal(2006)FaturslctiattntionnhancscolorsignalsinarlyisualarasofthhumanbrainProcNatlAcadSciUSA103(38):14250–1425413AndrsnSK,MüllrMM(2010)BhaioralprformancfollowsthtimcoursofnuralfacilitationandsupprssionduringcudshiftsoffaturslctiattntionProcNatlAcadSciUSA107(31):13878–1388214BlasrE,SprlingG,LuZL(1999)MasuringthamplificationofattntionProcNatlAcadSciUSA96(20):11681–1168615SchollBJ(2001)Objctsandattntion:ThstatofthartCognition80(1–2):1–4616HalbrdaJ,SirsSF,FignsonL(2006)MultiplspatiallyorlappingstscanbnumratdinparalllPsycholSci17(7):572–57617ArilyD(2001)Singsts:rprsntationbystatisticalproprtisPsycholSci12(2):157–16218InrsoM,SunP,ChubbC,WrightCE,SprlingG(2016)EidncagainstglobalattntionfiltrsslctiforabsolutbarorintationinhumanisionAttnPrcptPsychophys78(1):293–30819PylyshynZW,StormRW(1988)Trackingmultiplindpndnttargts:EidncforaparallltrackingmchanismSpatVis3(3):179–19720CaanaghP,AlarzGA(2005)TrackingmultipltargtswithmultifocalattntionTrndsCognSci9(7):349–35421PylyshynZW(2004)Sompuzzlingfindingsinmultiplobjcttracking:ITrackingwithoutkpingtrackofobjctidntitisVisionCogn11(7):801–82222LuZL,SprlingG(2012)BlackWhitasymmtryinisualprcptionJVision12(10):823ShihSI,SprlingG(2002)MasuringandmodlingthtrajctoryofisualspatialattntionPsycholR109(2):260–30524WolfJM(1994)GuiddSarch20ArisdmodlofisualsarchPsychonBullR1(2):202–23825ElazaryL,IttiL(2010)ABaysianmodlforfficintisualsarchandrcognitionVisionRs50(14):1338–135226GislrWS,CormackLK(2011)ModlsofortattntionOxfordHandbookofEyMomnts,dsLirsdgSP,GilchristID,ErlingS(OxfordUniPrss,NwYork),pp439–45427BoyntonGM,CiaramitaroVM,ArmanAC(2006)EffctsoffaturbasdattntiononthmotionaftrffctatrmotlocationsVisionRs46(18):2968–297628LiuT,LarssonJ,CarrascoM(2007)FaturbasdattntionmodulatsorintationslctirsponssinhumanisualcortxNuron55(2):313–32329SprlingG,WurstSA,LuZL(1992)UsingrptitiondtctiontodfinandlocalizthprocsssofslctiattntionAttntionandPrformancXIV,dsMyrDE,KornblumS(MITPrss,Cambridg,MA),pp265–29830BrainardDH(1997)ThPsychophysicstoolboxSpatVis10(4):433–43631AnstisS,CaanaghP(1983)AminimummotiontchniquforjudgingquiluminancColorVision:PsychophysicsandPhysiology,dsMollonJD,SharpLT(Acadmic,London),pp155–16632LuZL,SprlingG(2001)SnsiticalibrationandmasurmntprocdursbasdonthamplificationprinciplinmotionprcptionVisionRs41(18):2355–237433HrrraC,tal(2013)HowdothS,MandLconscontributtomotionluminancassssdusingminimummotionJVision9:102134StockmanA,SharpLT(2000)ThspctralsnsitiitisofthmiddlandlongwalngthsnsiticonsdridfrommasurmntsinobsrrsofknowngnotypVisionRs40(13):1711–1737SuntalPNASEarlyEdition|9of9PSYCHOLOGICALANDCOGNITIVESCIENCESPNASPLUS