poster print size: open vs. closed shapes: a dimension of ...image quality: you can place digital...

1
Three experiments were performed to test the reliability of the open/closed dimension; the first two from the perspecFve of ahenFonal allocaFon, and the third experiment in the context of mulF-class scaherplot displays. Experiment 1: Flanker Task Flanker tasks examine the influence of distractor symbols on target symbol processing. ParFcipant response Fmes are modulated when distractor and target are the same shape, share features, or are different. Experiment 2: Same-Different Task In same-different tasks, parFcipants judge whether the symbols in the display are all the same or different. Longer response Fmes represent greater difficulty in determining heterogeneity. Experiment 3: Visual AnalyIcs Tasks We varied the visual encoding in three analysis tasks using scaherplots: average value, numerosity, and trend judgments. Response latency and errors reflect the discriminability between categorical variables. Study Design Open vs. Closed Shapes: A Dimension of Perceptual Awareness? David Burlinson, Kalpathi Subramanian Ph.D., Paula Goolkasian Ph.D. The University of North Carolina at Charlohe {dburlins, krs, pagoolka}@uncc.edu David Burlinson The University of North Carolina at Charlohe Email: [email protected] Phone: (843) 465 - 6283 Contact 1. W. S. Cleveland and R. McGill. Graphical percepFon: Theory, experimentaFon, and applicaFon to the development of graphical methods. Journal of the American staFsFcal associaFon, 79(387):531–554, 1984. 2. C¸ . Demiralp, M. S. Bernstein, and J. Heer. Learning perceptual kernels for visualizaFon design. IEEE transacFons on visualizaFon and computer graphics, 20(12):1933–1942, 2014. 3. M. Gleicher, M. Correll, C. Nothelfer, and S. Franconeri. PercepFon of average value in mulFclass scaherplots. IEEE transacFons on visualizaFon and computer graphics, 19(12):2316–2325, 2013. 4. L. Tremmel. The visual separability of ployng symbols in scaherplots. Journal of ComputaFonal and Graphical StaFsFcs, 4(2):101– 112, 1995. 5. S. Lewandowsky and I. Spence. DiscriminaFng strata in scaherplots. Journal of the American StaFsFcal AssociaFon, 84(407):682–688, 1989. 6. J. Li, J. J. van Wijk, and J.-B. Martens. EvaluaFon of symbol contrast in scaherplots. In 2009 IEEE Pacific VisualizaFon Symposium, pp. 97–104. IEEE, 2009. 7. R. A. Rensink and G. Baldridge. The percepFon of correlaFon in scaherplots. In Computer Graphics Forum, vol. 29, pp. 1203–1210. Wiley Online Library, 2010. References IntroducIon Same-feature flankers seemed to cause response inhibiFon due to their similariFes, whether open or closed. Different-feature flankers and irrelevant flankers did not interfere. InteresFngly, same-feature blocks with closed shapes performed significantly faster than same- feature blocks with open shapes. Similarly to in Experiment 1, parFcipants reliably had an easier Fme responding to closed shapes than to the open shapes. In addiFon, parFcipants took longer deciding that the shapes were heterogeneous when the different shapes shared the open or closed feature dimension. Overall, parFcipants were quicker and made fewer errors in trials with one open and one closed shape, but displayed no preference for closed shapes. However, closed targets were impacted far more drasFcally – both facilitaFon by different-feature and interference by same-feature distractors – than open targets. Results Our results provide support for the importance of the open/closed feature category as a dimension to consider when using symbols in visualizaIon displays. The abstract cogniFve tasks in experiment 3 did not show the same closed-shape preference exhibited in the basic tasks from experiments 1 or 2. Task differences in results within experiment 3 suggest more abstract tasks incur more variability in response rates and error proporFons. Our results necessitate further invesFgaFon to understand the complex effects of symbol differences. We plan to explore addiFonal features of staFsFcal charts that may interact with categorical shape encodings. Conclusions EffecFve communicaFon using visualizaFons relies on viable visual encoding strategiesdecisions about how to represent data visually. For scaherplots with mulFple categorical variables, we need to disFnguish different variables’ points. Hue is a more dominant encoding than shape 1,3 . Different shapes work when color is not available, but which shapes should we choose? Two families of shapes polygonal and asterisk-based shapes are common in charFng soHware, and the literature on symbol discriminaFon generally supports their usage 2,3,4,5,6,7 . We propose that these two families exemplify a dimension that influences perceptual awareness: closed shapes bound a region of space and open shapes do not. We are interested in whether differences in this open vs. closed dimension meaningfully impact abstract cogniFve tasks in scaherplot displays. Figure 1. High and Low Load sFmuli in the Flanker task. ParFcipants were primed with two target shapes in each block. One appeared as the target in one of six posiFons. The flanker was placed to the leH or right of the display. Methods Figure 2. Flanker compaFbility (difference between mean incompaFble and compaFble response Fmes, in ms) for each block in low load and high load condiFons. In low load trials, same-feature blocks (1 and 6) had the smallest compaFbility effect, whereas they had the largest (with one excepFon) compaFbility effect in the high load condiFon. Figure 3. 2- and 3-shape variants of the sFmulus displays in the Same-Different task. Figure 4. The impact of the interacFon between feature and condiFon on response Fmes. Same-shape was significantly easiest to respond to, different-shape/different-feature was very close across both features, and different-shape/same-feature trials reliably took the longest to respond to. Figure 5. Average value judgment: Which set of shapes is higher on average? Figure 6. Numerosity judgment: Which set of shapes has more elements? Figure 7. Linear trend judgment: Which set of shapes displays a linear relaFonship? Figure 8. For numerosity and linear trend judgments, same-feature distractors reliably caused longer reacFon Fmes and the effect was modulated by target features.

Upload: others

Post on 07-Mar-2021

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Poster Print Size: Open vs. Closed Shapes: A Dimension of ...Image Quality: You can place digital photos or logo art in your poster file by selecFng the Insert, Picture command, or

PosterPrintSize:Thispostertemplateis21”highby36”wideandisprintedat200%fora42”highby72”wideposter.Itcanbeusedtoprintanyposterwitha7:12aspectraFo.

Placeholders:ThevariouselementsincludedinthisposterareonesweoHenseeinmedical,research,andscienFficposters.Feelfreetoedit,move,add,anddeleteitems,orchangethelayouttosuityourneeds.Alwayscheckwithyourconferenceorganizerforspecificrequirements.

ImageQuality:YoucanplacedigitalphotosorlogoartinyourposterfilebyselecFngtheInsert,Picturecommand,orbyusingstandardcopy&paste.Forbestresults,allgraphicelementsshouldbeatleast150-200pixelsperinchintheirfinalprintedsize.Forinstance,a1600x1200pixelphotowillusuallylookfineupto8“-10”wideonyourprintedposter.

Topreviewtheprintqualityofimages,selectamagnificaFonof100%whenpreviewingyourposter.Thiswillgiveyouagoodideaofwhatitwilllooklikeinprint.Ifyouarelayingoutalargeposterandusinghalf-scaledimensions,besuretopreviewyourgraphicsat200%toseethemattheirfinalprintedsize.

Pleasenotethatgraphicsfromwebsites(suchasthelogoonyourhospital'soruniversity'shomepage)willonlybe72dpiandnotsuitableforprinFng.

[Thissidebarareadoesnotprint.]

ChangeColorTheme:Thistemplateisdesignedtousethebuilt-incolorthemesinthenewerversionsofPowerPoint.

Tochangethecolortheme,selecttheDesigntab,thenselecttheColorsdrop-downlist.

Thedefaultcolorthemeforthistemplateis“Office”,soyoucanalwaysreturntothataHertryingsomeofthealternaFves.

PrinFngYourPoster:Onceyourposterfileisready,visitwww.genigraphics.comtoorderahigh-quality,affordableposterprint.EveryorderreceivesafreedesignreviewandwecandeliverasfastasnextbusinessdaywithintheUSandCanada.

Genigraphics®hasbeenproducingoutputfromPowerPoint®longerthananyoneintheindustry;daFngbacktowhenwehelpedMicrosoH®designthePowerPoint®soHware.

USandCanada:1-800-790-4001Email:[email protected]

[Thissidebarareadoesnotprint.]

Threeexperimentswereperformedtotestthereliabilityoftheopen/closeddimension;thefirsttwofromtheperspecFveofahenFonalallocaFon,andthethirdexperimentinthecontextofmulF-classscaherplotdisplays.Experiment1: FlankerTaskFlankertasksexaminetheinfluenceofdistractorsymbolsontargetsymbolprocessing.ParFcipantresponseFmesaremodulatedwhendistractorandtargetarethesameshape,sharefeatures,oraredifferent.Experiment2: Same-DifferentTaskInsame-differenttasks,parFcipantsjudgewhetherthesymbolsinthedisplayareallthesameordifferent.LongerresponseFmesrepresentgreaterdifficultyindeterminingheterogeneity.Experiment3: VisualAnalyIcsTasksWevariedthevisualencodinginthreeanalysistasksusingscaherplots:averagevalue,numerosity,andtrendjudgments.Responselatencyanderrorsreflectthediscriminabilitybetweencategoricalvariables.

StudyDesign

Openvs.ClosedShapes:ADimensionofPerceptualAwareness?DavidBurlinson,KalpathiSubramanianPh.D.,PaulaGoolkasianPh.D.

TheUniversityofNorthCarolinaatCharlohe{dburlins,krs,pagoolka}@uncc.edu

DavidBurlinsonTheUniversityofNorthCarolinaatCharloheEmail:[email protected]:(843)465-6283

Contact1.  W.S.ClevelandandR.McGill.GraphicalpercepFon:Theory,experimentaFon,andapplicaFontothedevelopmentofgraphicalmethods.JournaloftheAmericanstaFsFcalassociaFon,79(387):531–554,1984.2.  C¸.Demiralp,M.S.Bernstein,andJ.Heer.LearningperceptualkernelsforvisualizaFondesign.IEEEtransacFonsonvisualizaFonandcomputergraphics,20(12):1933–1942,2014.3.  M.Gleicher,M.Correll,C.Nothelfer,andS.Franconeri.PercepFonofaveragevalueinmulFclassscaherplots.IEEEtransacFonsonvisualizaFonandcomputergraphics,19(12):2316–2325,2013.4.  L.Tremmel.Thevisualseparabilityofployngsymbolsinscaherplots.JournalofComputaFonalandGraphicalStaFsFcs,4(2):101–112,1995.5.  S.LewandowskyandI.Spence.DiscriminaFngstratainscaherplots.JournaloftheAmericanStaFsFcalAssociaFon,84(407):682–688,1989.6.  J.Li,J.J.vanWijk,andJ.-B.Martens.EvaluaFonofsymbolcontrastinscaherplots.In2009IEEEPacificVisualizaFonSymposium,pp.97–104.IEEE,2009.7.  R.A.RensinkandG.Baldridge.ThepercepFonofcorrelaFoninscaherplots.InComputerGraphicsForum,vol.29,pp.1203–1210.WileyOnlineLibrary,2010.

References

IntroducIonSame-featureflankersseemedtocauseresponseinhibiFonduetotheirsimilariFes,whetheropenorclosed.Different-featureflankersandirrelevantflankersdidnotinterfere.InteresFngly,same-featureblockswithclosedshapesperformedsignificantlyfasterthansame-featureblockswithopenshapes.SimilarlytoinExperiment1,parFcipantsreliablyhadaneasierFmerespondingtoclosedshapesthantotheopenshapes.InaddiFon,parFcipantstooklongerdecidingthattheshapeswereheterogeneouswhenthedifferentshapessharedtheopenorclosedfeaturedimension.Overall,parFcipantswerequickerandmadefewererrorsintrialswithoneopenandoneclosedshape,butdisplayednopreferenceforclosedshapes.However,closedtargetswereimpactedfarmoredrasFcally–bothfacilitaFonbydifferent-featureandinterferencebysame-featuredistractors–thanopentargets.

Results

Ourresultsprovidesupportfortheimportanceoftheopen/closedfeaturecategoryasadimensiontoconsiderwhenusingsymbolsinvisualizaIondisplays.TheabstractcogniFvetasksinexperiment3didnotshowthesameclosed-shapepreferenceexhibitedinthebasictasksfromexperiments1or2.Taskdifferencesinresultswithinexperiment3suggestmoreabstracttasksincurmorevariabilityinresponseratesanderrorproporFons.OurresultsnecessitatefurtherinvesFgaFontounderstandthecomplexeffectsofsymboldifferences.WeplantoexploreaddiFonalfeaturesofstaFsFcalchartsthatmayinteractwithcategoricalshapeencodings.

Conclusions

EffecFvecommunicaFonusingvisualizaFonsreliesonviablevisualencodingstrategies–decisionsabouthowtorepresentdatavisually.ForscaherplotswithmulFplecategoricalvariables,weneedtodisFnguishdifferentvariables’points.Hueisamoredominantencodingthanshape1,3.Differentshapesworkwhencolorisnotavailable,butwhichshapesshouldwechoose?Twofamiliesofshapes–polygonalandasterisk-basedshapes–arecommonincharFngsoHware,andtheliteratureonsymboldiscriminaFongenerallysupportstheirusage2,3,4,5,6,7.Weproposethatthesetwofamiliesexemplifyadimensionthatinfluencesperceptualawareness:closedshapesboundaregionofspaceandopenshapesdonot.Weareinterestedinwhetherdifferencesinthisopenvs.closeddimensionmeaningfullyimpactabstractcogniFvetasksinscaherplotdisplays.

Figure1.HighandLowLoadsFmuliintheFlankertask.ParFcipantswereprimedwithtwotargetshapesineachblock.OneappearedasthetargetinoneofsixposiFons.TheflankerwasplacedtotheleHorrightofthedisplay.

Methods

Figure2.FlankercompaFbility(differencebetweenmeanincompaFbleandcompaFbleresponseFmes,inms)foreachblockinlowloadandhighloadcondiFons.Inlowloadtrials,same-featureblocks(1and6)hadthesmallestcompaFbilityeffect,whereastheyhadthelargest(withoneexcepFon)compaFbilityeffectinthehighloadcondiFon.

Figure3.2-and3-shapevariantsofthesFmulusdisplaysintheSame-Differenttask.

Figure4.TheimpactoftheinteracFonbetweenfeatureandcondiFononresponseFmes.Same-shapewassignificantlyeasiesttorespondto,different-shape/different-featurewasverycloseacrossbothfeatures,anddifferent-shape/same-featuretrialsreliablytookthelongesttorespondto.

Figure5.Averagevaluejudgment:Whichsetofshapesishigheronaverage?

Figure6.Numerosityjudgment:Whichsetofshapeshasmoreelements?

Figure7.Lineartrendjudgment:WhichsetofshapesdisplaysalinearrelaFonship?

Figure8.Fornumerosityandlineartrendjudgments,same-featuredistractorsreliablycausedlongerreacFonFmesandtheeffectwasmodulatedbytargetfeatures.