DOI10.1068p3134 The effect of name category and di

1IntroductionColouriswellestablishedasamethodofenhancinginformationdisplays(egChrist1975).Anenhancementoftenreportedisinimprovingthesearchforinformation.Typicallyresearchinthisareahasconcentratedonthecolour-differencerequirementsforobtainingparallelsearchwithincoloursets,andtheadequacyofstandardcolour-differencemetrics(egDEuv)inprovidingusefulmeasurestopredictparallelsearch(egCarterandCarter1982;Luriaetal1986;SanchezandNagy1989).Lessresearchhasbeendirectedtowardstheroleofnameability(easeofnamingofcolours)inalteringthesearchcharacteristicsofcoloursets.Thisisperhapssurprisingsincemuchresearchhasdeterminedwithinandbetween-cultureagreementinwhatthefundamentalnames(`basiccolourterms'orBCTs)are,andwhatsensationstheydenote(BerlinandKay1969;Rosch-Heider1972;Kayetal1997).Mostresearchhasdeterminedamaximumofelevensuchtermsöwhite,black,red,green,blue,yellow,pink,purple,orange,brown,greyöandithasbeenproposedthatthesederivefromafundamentalstructureofthevisualsystemsharedbetweenallhumans(Ratliff1976;Boynton1997).Suchuniversalssuggestthatanyresultsfromvisual-searchexperimentsinvolvingcolournamingcouldbestable,andwidelyusable.Giventheabovedata,thequestionarisesastowhyonlylimitedefforthasbeendirectedtowardstheuseofnameablecoloursinsearch(andother)tasks.Inpartthisisbecausedefining`easytoname'isnotitselfeasy,unlikedefiningperceivedcolourdifference(egDEuv).TherearenoanalogoussystemstoallowaDEequivalentforcolour`nameability'.NoristhereanamingframeworkwhichallowsstatementsTheeffectofnamecategoryanddiscriminabilityonthesearchcharacteristicsofcoloursetsPerception,2002,volume31,pages445^461SteveGuestDepartmentofExperimentalPsychology,OxfordUniversity,SouthParksRoad,OxfordOX13UD,UK;e-mail:[e-mail]tofPsychology,UniversityofPortsmouth,KingHenryIStreet,PortsmouthPO12DY,UK;e-mail:[e-mail]18September2000,inrevisedform14September2001Abstract.Within(andbetween)cultures,peopletendtoagreeonwhichpartsofcolourspaceareeasiesttonameandwhatthenamesfortheseregionsare.Thereforeitislikelythatthemanipu-lationofeaseofnaming(nameability)ofcoloursshouldchangeperformanceintaskswherecategorisationbycolournameisimportant.Morespecifically,highly`nameable'coloursetsshouldleadtobetterperformancethanmetricallyequivalentbutlesscategoricallydistinctsets,whenthetaskrequirescategorisation.Thishypothesiswasinvestigatedbytestingobserversonaname-basedtask,thenamingandsubsequentidentificationbynameofcoloursetswithuptosixteenmembers.Thesesetsweredesignedtobeeasytoname(nameable),maximallydiscrim-inable,ormatcheddiscriminable.Thefirstwerederivedfrompreviouslygenerateddata,thesecondbyastandardalgorithmtospacecolourswidelyincolourspace,andthelatterbycloselymatchingtheirmetriccharacteristicstothoseofaneasy-to-namecolourset.Thisfinalconditionwasmetrically(butnotcategorically)equivalenttothenameableset.Itwasfoundthatsetsdesignedtobenameabledidindeedleadtosuperiorperformanceasmeasuredbyresponsetimes,confidenceratings,andresponseaccuracy.Perceptualcoloursimilarity,measuredbyaDEmetric,didnotpredicterrors.Nameabilitymaythusbeavalid,manipulable,aspectofsetsofcolours,andonewhichisnototherwiseduplicatedinthemetriccharacteristicsofsuchsets.DOI:10.1068/p3134suchas`thisgreenisabettergreenthanthatcolourisared',or`thesetwogreensaremorelikelytobeconfusedtogetherthanthosetworeds',tobemade.Anotherreasonwhylittleresearchhasbeendirectedattheroleofcolournamesinsearchtasksisbecause,formanysearchtasks,categorisationmaybeunnecessary.Forinstance,nocolour-spacingalgorithmsinclude,orseemtorequire,categoricaldifferenceforoptimalsearchperformance(egCarterandCarter1982;DeCorte1990).However,describingacolourbynamerequiresacategoricaljudgment,suchaswhendirectinganobserver'sattentiontothe`purpleobject'onacomputerscreen,orwhencueingatargetgrouporclassbyacolourname.Thislattertaskis,atleastinitially,acategoricalone,althoughoncethetargetsubgrouphasbeenidentifiedfromthenamethesubsequentsearchmightnolongerbecategorical.Itisthesesortsoftaskswherenameabilitycouldbeimportant.Theprimaryresearchintoeaseofnamingofcolours(orcolour`nameability')asavariableinvisualsearchhasbeenthatofSmallmanandBoynton(1990,1993).Inthefirststudytheyinvestigatedtheusefulnessofbasiccolourcodinginavisual-searchtask,withbasiccoloursdefinedviacolour-namingdatafromBoyntonandOlson(1987).Inthesestudies,adesiredtargetgroupwascodedina(basic)colour,andthisgroupwassubsequentlycuedbyanexampleoracolourname.Itwasfoundthatbasiccodesdidindeedleadtoexcellentsearchperformance.However,thisperformancewasnobetterthanwhensimilarlymetricallyspacednonbasicexemplarswereused.Theconclusionwasthatanysearchefficacyofbasiccoloursetswasonlybecausethebasicexemplarswerewidelyseparatedincolourspace,thoughitmustbenotedthatforthenonbasicpartsoftheexperiment,noname-basedcueingwasused.ThelaterSmallmanandBoyntonstudyusedamoreinteractivedesign,withobserversnavigatingcolourspacetosetbasicand`personal'(ienonbasic)exemplarsforuseinasubsequentsearchtask.Resultswereessentiallyinlinewiththe1990study,althoughtherewasasug