Chapter3: AttentioninPerceptionandDisplaySpace Part2:ParallelProcessingandDividedAttention (Wickensetal:Ch.3.3,64-83) (+Ch.2.7,35-40) CriticalAspectsofAttention(Review) SelectiveAttentionlModels • abilitytoselectperceptualchanneltofocusattentionatanyparticularmoment – e.g.visualsampling,targetsearch,etc. • influencesdesignofinstrumentlayouts,procedures,searchtasks,training,etc. DividedAttentionModels • abilitytodivideattention“simultaneously”acrosstwoormoreperceptualchannels – e.g.displaymonitoring,reading,timesharing,etc. • influencesdesignofintegrateddisplays,operatingprocedures,multitasking,etc. FocusedAttentionModels • abilitytofocusonparticularperceptualchannel,andexcludestimulationfromother (adjacent)channels • ≈‘opposite’ofdividedattention – e.g.targetsearch,displaymonitoring,studying,problemsolving,translating,etc. • influencesdesignofworkenvironments,warningsystems,etc. SustainedAttentionTasks • abilityofhumanstomaintainvigilancelevelsoverlongperiodsoftime • influencesdesignofworkenvironments,displayinterfaces,etc. ExampleofLowLevelPre-attentive Processing(Review) “Find Large T” versus "Find Red S” or “Find X or K” • letterS“popsout”automatically...becauseitisred,whileothersaren’t • doesn’tneedtoinvokeconsciousattentionalprocessesèparallelsearch • (searchforXorKinvolvesone(diagonal)feature) Figure-GroundPhenomenon Faces? orVase? SaxophonePlayer? orWoman? YoungWoman? orOldWoman? Somelesswellknownexamples OldMan? orYoungCouple? Whichbuildingisinfront? Wickens’InformationProcessingModel Pre-attentive Cognitive • separationofpre-attentiveandcognitiveperception • notallperceptionisbottom-up(basedonelementsofretinalimages) • semanticcuingcaninfluenceperception,inatop-downmanner HolisticInformationProcessing • Whenwholeisperceiveddirectly,ratherthanasaconsequence ofseparateanalysisofconstituentelements. • Complexityofconstituentelementsdoesnotaffecttimerequired toprocessstimuliholistically...butdoesaffectthetimerequired foranalyticprocessing • Objectsperceivedasintegralwhole,dependingon: – surroundingcontours – correlatedattributes – familiarity AWell-knownExample CanyouseetheDalmatian? DynamicDemosofHolisticProcessing GlobalversusLocalProcessing • Global/HolisticProcessing – Organises(visual)worldintoobjects,groups, patterns,etc. – Automatic/Pre-attentive – ExhibitsGlobalPrecedence versus • Localprocessing – Separateprocessingofindividualobjectsindisplay – Dependsonmorefocusedselectiveattention GlobalvsLocalProcessingExample T T T T T T T T T T T T T TT T T T T T T T T *^&#&+%$$@*# $ * & & $ ^ # % “LargeLetter”=T & ( # 0 “SmallLetter”=T Inthefollowingslide,replytoquestion“WhatistheSmallLetter”or“WhatistheLargeLetter” GlobalvsLocalProcessingExample TTTTTTT T T T T T TTTTT T T T T T T FFFFFFF F F F F F FFFFF F F F F F F TTTTTTT T T T T T TTTTT T T T T T T FFFFFFF F F F F F FFFFF F F F F F F Whatrelativeresultsshouldwehaveobserved(ifexperimentdoneproperly)? WhatistheSMALL Letter? WhatistheSMALL Letter? WhatistheLARGE Letter? WhatistheLARGE Letter? Slowest/ LeastAccurate SecondinSpeed &Accuracy? ThirdinSpeed &Accuracy? Fastest/ MostAccurate GlobalvsLocalProcessingExample Which‘groups’oflettersaremoresimilartoeachother? L L L L L L T L T T L L T T T T T T T T T T T T L L T T T T L L L L L L T T T T T L L L L T T T T T L L T T T T T LessSimilarity T T T T T T T T T GreaterSimilarity LT TT Textons: Fundamental ElementsinPreattentiveVision andPerception ofTextures (Julesz,1983) a)Texturepairdefinedby presenceorabsenceof crossinglinesegments. b)Texturepairsduetogap differenceinL-shaped elementsindistinguishable. c)T-andL-shapedelements indistinguishable.Endofline terminatornotadequatefor distinguishingtexture. d)Linecrossingatmid-point orendoflinesinsufficient fordistinguishingtexture. Canweapplytheseprinciplesofpre-attentiveprocessing tothedesignofcomplexhuman-machineinterfaces? Inotherwords,howcandisplaydesigners exploitmodelsofperceptualorganisation andpre-attentiveprocessing? • Goals: – tosupportparallelprocessing(dividedattention) whenappropriate – tosuppressparallelprocessing(andsupportfocussed attention)whenappropriate • PrinciplesofSpatialProximity • PrinciplesofObject-BasedProximity • ProximityCompatibilityPrinciple producing partial masking. Minimal separation is more likely with miniaturized hand-held displays, or with display overlay, as found with head-up displays or database overlays (Kroft & Wickens, 2003; Beck et al., 2010). Numerosity and readout clutter have been referred to as global density and local density clutter, respectively (Tullis, 1988; see also Beck et al., 2010; Wickens, Vincow, et al., 1997). • Disorganizational clutter describes the random location of distractors in search fields that are not “structured.” Examples of structured and unstructured search fields are shown in the left and right panels respectively, of Figure 3.3. Unstructuredlayoutperceivedas(disorganisational)clutter • Heterogeneous clutter refers to the heterogeneity of the non-target background features =>requireslocalprocessing (like color, shape or size) which we saw above was an impediment to visual search. StructuredvsUnstructuredSearchFields • these clutter factors are manifest in the use of maps, whose design we discuss in more detail • All Organised/structuredlayoutfacilitatesglobalprocessing in Chapter 5. Various researchers have quantified either individual factors (Yeh & Wickens, 2001) or developed metrics that combine the factors (Beck, Lohrenz, & Trafton, 2010; Rosenholtz, (a) (b) FIGURE 3.3 (a) Structured Gestalt principles of display organization. (b) Unstructured search field. EmergentFeatures. • • • • Emergentfeaturesarerelatedtoglobalprocessingconcept Relatedtoaggregationofstimuli;notevidentineachitemonitsown Facilitatespre-attentive(rapid)processing Shouldbecompatiblewithtask Chapter 3 • Attention in Perception and Display Space (a) 65 (b) FIGURE 3.4 Local (a) and global (b) perception in aircraft engine dials. Determining whether the pointer indicates a normal position in (a) requires separate examination of each dial. In (b) the dials have been rotated so that the normal state is straight up. This arrangement means that each column of pointers creates an emergent feature (a series of vertical lines). Deviation from the vertical is easy to detect with this arrangement. The arrows indicate the mapping of display column to aircraft engine (discussed further in text). will discuss such spatial display compatibility further in Chapter 4. Banbury, Selcon, and McCrerie (NB:Considerengineparameterdisplayarrangement:{Column1&3forLeftengine;Column2&4forrightengine} (1997) found a four-fold increase in check-reading errors for this type of engine panel arrangeIsthiscompatiblewiththephysicalsystem?) ment, compared to a redesigned panel which grouped both the primary and the secondary left engine dials on the left side, and the right engine dials on the right side. We will touch upon principles relating to the compatibility between display and task requirements later on in this chapter. Howcandisplaydesignersexploitmodelsof perceptualorganisationandpre-attentiveprocessing? • Goals: – tosupportparallelprocessing(dividedattention) whenappropriate – tosuppressparallelprocessing(andsupportfocussed attention)whenappropriate • PrinciplesofSpatialProximity • PrinciplesofObject-BasedProximity • ProximityCompatibilityPrinciple SpatialProximity • Placingvisualinformationsourcesclosetogether(withinUFOV), withexpectationthatparallelprocessingwillbesupported withinforwardfieldofview(FFOV) • AugmentedReality(AR)Displayssuperimposecomputer generatedimagesontoreal-worldscenes,forexample: – WindowmountedHead-upDisplays(HUDs), – HMDs:Head(orspectacle)mounteddisplays(e.g.GoogleGlass) – Hand-heldARdisplays(egPokemonGo) Principle:bysuperimposingtask-related informationonviewofoutsideworld, bothcanbeattendedtoinparallel,due to: • reducedscanningtransitiontimes (recallSEEVmodel) • reducedvisualaccommodationtime andeffort • reducedlight-darkadaptation HUDAugmentedRealityintheAutomobile AudiNightViewAssist head-down configuration. The airplane can be seen, poised to “move out,” in Figure 3.5b. As we saw earlier, placing information sources together does not necessarily gurarantee that both will be processed (remember inattentional blindness and the gorilla and the basketball players). This apparent contradiction hinges on the expectations of the observer. We are likely to see advantages to the HUD format when the observer expects the events in the superimposed background; that is, when they are likely to occur (high expectancy). However, the HUD format will impair Head-upDisplaysandForwardFieldofView NB:Noteprincipleof ConformalSymbology: •Correspondenceof objectsacrossviews •Createsnew“shared object” (a) (b) (c) FIGURE 3.5 (a) Head up display (HUD) used in aviation. (b) Head up display with conformal imagery. Note the airplane on the ground by the runway. (c) Head up display with conformal imagery (runway overlay). Source: (3.5a) Richard Baker/Corbis. However…bewareofinattentionblindness Conformalimagery(good)……however…….potentialinattentiontootheraircraft!! Spatialproximitycancauseproblems(failuresofdividedattention)whenstimuliunexpected. HelicopterHUDExample Howcandisplaydesignersexploitmodelsof perceptualorganisationandpre-attentiveprocessing? • Goals: – tosupportparallelprocessing(dividedattention) whenappropriate – tosuppressparallelprocessing(andsupportfocussed attention)whenappropriate • PrinciplesofSpatialProximity • PrinciplesofObject-BasedProximity • ProximityCompatibilityPrinciple SpatiallyproximalHead-upDisplayssupportparallelprocessing. Whataboutsuppressingparallelprocessing? SomeProblemswithspatialproximity • Absenceof(orinsufficient)conformality • Spatialseparationindepth: – co-locatedstimulimayappearatdifferentdepthplanes (e.g.usingstereoscopicdisplays) • Co-locatedstimulinotimmunetochangeblindness • Perceptualcompetition • Responseconflicts ResponseConflictsduetoSpatialProximity Eriksen&Eriksen(1974) http://cognitivefun.net/test/6 EriksenFlankerTest:Somesamplestimuli “Pressthe<or>keyasquicklyaspossible,accordingtothedirectionofthecentralarrow.” http://cognitivefun.net/test/6 ResponseConflictsduetoSpatialProximity Eriksen&Eriksen(1974) Twotargets,ForH: H(PressRightbutton) (PressLeftButton)F Control: RedundancyGain: H PerceptualCompetition: KHK HHH ResponseConflict: FHF ImplicationsofResponseConflictsduetoSpatialProximity (Eriksen&Eriksen,1974) • PerceptualCompetition(KHK) – Increaseddisplayclutterleadstofailuretofocusbetween relevantandirrelevantèRTincreases • Responseconflict(FHF)versusRedundancygain(HHH) – Proximityallowstwochannelstobeprocessedinparallel… evenifundesired – Whentherearedifferentimplicationsforaction,sourcesof informationcancompeteattheperceptuallevel – Iftheydisagree,canleadtoresponseconflict: • RTincreasesand/orprobabilityoferrorincreases – Iftheyagree,canleadtoredundancygain: • RTdecreasesand/orprobabilityoferrordecreases (RT:ResponseTime) FailuresofSpatialProximity StroopEffectDemo(ppt) StroopEffectSummary • Whathappenswhendifferentinformationchannelscomprise differentdimensionsofthesamephysicalstimulus? – e.g.colour,shape,location,size,etc. • MostcommonformofStrooptask:“Nameinkcolour” XXXXXXXX GREENRED • Ingeneral: – RTand/orerrorrateincreasewhencolournameandink colourconflict(relativetoneutralcontrol), • i.e.responseconflict – Redundancygaincanalsooccur,wheninkcolourand colournamecoincide(èreducedRTand/orerrorrate) McGurkEffect:AnotherFailureofSpatialProximity McGurkEffectDemo http://www.moillusions.com/2011/05/mcgurk-effect-audio-video-illusion.html (3min) Howcandisplaydesignersexploitmodelsof perceptualorganisationandpre-attentiveprocessing? • Goals: – tosupportparallelprocessing(dividedattention) whenappropriate – tosuppressparallelprocessing(andsupportfocussed attention)whenappropriate • PrinciplesofSpatialProximity • PrinciplesofObject-BasedProximity • ProximityCompatibilityPrinciple Object-BasedProximity • Similaritycanbebasedonmorethanspatialproximity • ObjectFiletheoryofattention: – objectfeaturesperceivedinearlystage,&thencombinedlater – perceptualprocessingisparallelwithinthefeaturesofasingle object,butserialacrossdifferentobjects • Inotherwords,organisingfeatures(includingemergent features)ofmultipledisplayeditemscanresultin perceptionofintegralobjects – e.g.conformalsymbology – e.g.objectdisplays https://en.wikipedia.org/wiki/Feature_integration_theory Canweapplyobjectbasedproximityprincipletodesign displaysforcomplex,multi-degree-of-freedomsystems? 70 Chapter 3 • Attention in Perception and Display Space CORE EXIT 579/422°F 0 SUBCOOL WID RCS PRESS 1265/2235 PSIG 768 MCP Examples/Applicationsof Object-BasedDisplays STARTUP 0 0 DPM PRZR LEV 0/41% CNTMT PRESS 15 PSIG RV LEV 70% RAD CNTMT SEC OTHER WID SG LEV lp 2 39/50% (a) Ventilator Patient Volume Dead space Alveolar space Rate 1 2 3 4 5 7 8 9 10 11 6 (b) 159 PVR 1363 SVR 2394 SVR 159 PVR 100 SaO2 4 CVP 30/10 18 PAP 8 LAP HR 72 SV 70 CO 5.0 120/80 90 MAP 92 SaO2 2 CVP 25/5 12 PAP 9 LAP HR 103 SV 34 CO 3.5 70/40 50 MAP (c) Wickensetal,Fig.3.7 FIGURE 3.7 Examples of object displays. (a) safety parameter display, (b) medical display of oxygen exchange, (c) graphical cardiovascular display for anesthesia. Source: Human Factors: The journal of the Human Factors Society by Human Factors Society; Human Factors Society of America. Reproduced with permission of Human Factors and Ergonomics Society in the format republish in a book/journal via Copyright Clearance Center. The object display concept can be applied to text as well. When a graphic designer places text on a display, there are various techniques for ensuring that its content is associated with the object it identifies (e.g., spatial proximity, arrows, similar colors,). The ultimate method for ensuring association between elements is achieved by making multiple elements part of the same PolygonSafetyParameterDisplay(Westinghouse) • Connectspokestoformpolygon • Assumingparallelprocessing(dividedattention)required, integratesseveralinformationsourcesintoonedisplay • Takesadvantageofobjectbasedproximity • Emergentfeatures–propertyofwholeshape,notseen whenpartsinisolation Example:GraphicalCardiovascularDisplay (Drews&Westenskow,2006) Traditionaldisplayinanaesthesiology: Single-Sensor,Single-Indicator(SSSI) ‘GraphicalPolygonDisplay’ Example:GraphicalCardiovascularDisplay (Drews&Westenskow,2006) (a) - Showingnormalstate,withsymmetry - Incorporatesclinician’smentalmodel (b) - Showingmyocardialischemia - Asymmetriesshowchangesinvolumestatus(a), withhighpreloadandlowafterload - emergentfeatureofcrinkledheart(b)with reductioninsizeofleftheart;reducedcardiac output CVP=CentralVenousPressure;PAP=PulmonaryArteryPressure;PVR=PulmonaryvascularResistance; LAP=LeftArterialPressure;CO=CardiacOutput;SV=StrokeVolume;MAP=MeanArterialPressure; SVR=SystemicVascularResistance;SaO2=BloodOxygenSaturation;ST=SegmentDepression. Example:GraphicalCardiovascularDisplay (Drews&Westenskow,2006) NormalState Hypovolemia:abnormaldecreasein volumeofbloodcirculatinginbody Bradycardia:abnormaldecreaseheartrate CVP=CentralVenousPressure;PAP=PulmonaryArteryPressure;PVR=PulmonaryvascularResistance; LAP=LeftArterialPressure;CO=CardiacOutput;SV=StrokeVolume;MAP=MeanArterialPressure; SVR=SystemicVascularResistance;SaO2=BloodOxygenSaturation;ST=SegmentDepression. AnotherExampleofIntegratedObjectDisplay Chapter 3 • Attention in Perception and Display Space FIGURE 3.8 Axis Maps. Words as streets: an integrated object display. Source: “Chicago Typographic Map” by We are beginning to see a pattern here: there is a relationship between the choice of a display representation and a particular set of task demands (what we might call a task representation; Smith, 71 Review(fromChapter2): OrthogonalvsCorrelatedDimensionsofSource • OrthogonalDimensions – wheninformationpresentedalonganydimensionis independentofinformationpresentedalongother dimensions – maximisespotential‘efficiency’oftransmission • CorrelatedDimensions – whenlevelofinformationinonedimensionconstrains amountofinformationavailablealongotherdimensions (i.e.redundancy) – maximisesreliabilityoftransmission • BotharePropertiesofInformationSource – (i.e.theydonotcharacteriseinformationtransmission) (FromChapter2) PhysicalFormofPresentation(Stimulus): SeparablevsIntegralDimensions SeparableDimensions • whenlevelofinformationalonganyonedimensioncanbe transmittedwithoutspecifyinglevelsalonganotherdimension • eachdimensionperceivedasindependentofotherdimensions • i.e.it’seasytofocusononedimensionandignoretheother • e.g.colourandfilltexture • e.g.perpendicularvectors (FromChapter2) PhysicalFormofStimuli: SeparablevsIntegralDimensions IntegralDimensions • whenlevelofinformationalongeachdimensioninfluences perceptionoflevelsalongotherdimensions • dimensionsareperceivedasbeingdependent • inotherwords,it’sdifficulttoignoreotherdimensionsandfocuson justonedimension • e.g.colourandbrightnessofanbject • e.g.rectangleheightandwidth SourceInformationvsPhysicalForm PHYSICALFORM(DISPLAY)DIMENSIONS SEPARABLE DIMENSIONS ORTHOGONAL DIMENSIONS SOURCE DIMENSIONS CORRELATED DIMENSIONS INTEGRAL DIMENSIONS western_railway_train_time_table_down.gif (GIF Image, 665×901 pixels)... Correlated Information EncodedSeparably (moreorless) • • • • MumbaiWesternRailwayTimetable Eachentryisseparatedeparturetime Someredundancy:downwards=later Butnocommunicationofspeed, direction,relativetimes,etc. http://binishphilip.files.wordpress.com/2009/02/western_railway RailTimetablewithIntegralDimensions (byE.J.Marey,1880’s,copiedfromEdwardTufte:TheVisualDisplayofQuantitativeInformation) AnotherReal-WorldExample:AnaesthesiaMachine Exercise: • Aretheinformation sourceshereorthogonal orcorrelated? • Arethedisplayshere separableorintegral? DesignImplications • Ifdata(source)dimensionsarecorrelated,itis generallybettertouseintegraldisplaydimensions – (Lateron,we’lltalkaboutconfiguraldisplays) • Ifdata(source)dimensionsareorthogonal,itis generallybettertouseseparabledisplaydimensions – Unidimensionaljudgmentcanbeimpairedbyintegral displaysofseparabledimensions • Isthereunderlyingcorrelationbetweenvariables? – Ifso,integraldimensionsmayhelp – e.g.altitude+temperature – e.g.tidallungvolume+respiratoryrate Example:ObjectDisplayofRespiratoryFunctioning [Breaths/minute]*[volume/breath]=>[volume/minute]=Areaofrectangle RevisitStroopEffect • Whathappenswhendifferentinformationchannelscomprise differentdimensionsofthesamephysicalstimulus? – e.g.colour,shape,location,size,etc. • MostcommonformofStrooptask:“Nameinkcolour” XXXXXXXX GREENRED • Ingeneral: – RTand/orerrorrateincreasewhencolournameandink colourconflict(relativetoneutralcontrol) • Thisisanexampleoforthogonalsourcedimensions presentedasadisplaywithintegraldimensions – Redundancygaincanalsooccur,wheninkcolourand colournamecoincide(èreducedRTand/orerrorrate) • Thisisanexampleofcorrelatedsourcedimensions presentedasadisplaywithintegraldimensions Pairsof(generallyaccepted)Integral&Separable Dimensions(Table2.2) Integral Dimensions Separable Dimensions HeightofRectangle WidthofRectangle VerticalPosition HorizontalPosition Lightness ColourSaturation Size(Area) ColourSaturation Hue ColourSaturation Size(Area) Brightness Pitch Timbre Shape ColourSaturation Pitch Loudness Shape(LetterShape) Colour Timbre:Pitch Location Duration Location FacialFeatures SpacingofFeatures Orientation(angle) Size SpatialLocation TemporalOrder a display and the task requirements. This is illustrated in Figure 3.9. In the next section, e precise nature of this compatibility in the form of a principle for display design. oximity Compatibility Principle (PCP) ProximityCompatibilityPrinciple(PCP): 10, we summarize the one key implication of what we have just discussed. On the ConceptofTaskCompatibility e figure, two (or more) elements on a display (or in the natural environment) can be Display representation Task list. 1._____ 2._____ 3._____ Task representation Taskcompatibility: Thewayataskisdisplayed shouldbecompatiblewith theactualinformationneeds ofthattask. FIGURE 3.9 Task compatibility refers to the relationship between display and task representations. 71 9/8/12 1:09 AM Recallearlierexampleofengine parameterdisplays: Left1-4;Right1-4;Left5-8;Right5-8 TheProximityCompatibilityPrinciple(PCP) • TaskProximity:Extenttowhichdifferentinformationsourcesare actuallyprocessedtogetherwithinthesametask. – (a.k.a.‘processingproximity’or‘mentalproximity’) – dividedattentionismanifestedasmentalinformationintegration(*) • DisplayProximity:How‘close’displayelementsaretoeachother …w.r.t.eitherspatialorobject-baseddistance. • Ifinformationsourceswithinataskrequirehightaskproximity, thentheyshouldhavehighdisplayproximity. – assistsinintegratinginformation • Ifinformationsourceswithinataskrequirelowtaskproximity, thentheyshouldhavelowdisplayproximity. – assistsinseparatinginformationsources – usingintegrateddisplayswhentaskrequireslowtaskproximitycan beadisadvantage;canleadtoresponseconflict (*dualtaskprocessing,includingresponses,tobedealtwithlater) TheProximityCompatibilityPrinciple(PCP) Chapter 3 • Attention in Perception and Display Space Good High task proximity Divided attention (information integration) Performance 72 Low task proximity Bad Focused attention Low (distant) High (close) By space By objectness Display proximity FIGURE 3.10 An illustration of the proximity compatibility principle. The graph at the top of the figure shows performance as a function of display and task proximity. The bottom part of the figure shows two ways to manipulate display proximity: by distance or by objectness. either “distant” from each other or “close,” where closeness can be defined by either spatial proximity (the circles) or belonging to the same object (“objectness,” the rectangles). This distinction forms the x-axis of the graph above. On the y-axis we have plotted performance; the quality of DimensionsofDisplayProximity Chapter 3 • Attention in Perception and Display Space 73 Display proximity Sensory/perceptual differences Howcanwemanipulate DisplayProximity? Proximity in: 1. Space o o (vs. o 2. Color o o (vs. o) Separation o) Common object 3. Connections o o 4. Abutment 5. Heterogeneous feature 6. Homogeneous feature (size,brightness,shape) or y 7. Homogenous feature Emergent features 8. Homogeneous feature (again) x H W Area Shape 9. Polygon display FIGURE 3.11 Dimensions of display proximity. perceptual similarities; common object; and emergent features. We describe these below, identified by row number in the figure. Thisonedoesn’t! DimensionsofDisplayProximity Chapter 3 • Attention in Perception and Display Space Display proximity Thislabelrelatestothisfigure. Sensory/perceptual differences Proximity in: 1. Space o o (vs. o 2. Color o o (vs. o) Separation o) Common object 3. Connections o o 4. Abutment 5. Heterogeneous feature 6. Homogeneous feature or y 7. Homogenous feature Emergent features 8. Homogeneous feature (again) 9. Polygon display x H W Area Shape 73 Sensory/perceptual differences Proximity in: DimensionsofDisplayProximity Separation o) 1. Space o o (vs. o 2. Color o o (vs. o) Common object 3. Connections o o 4. Abutment 5. Heterogeneous feature 6. Homogeneous feature or y position 7. Homogenous feature Emergent features 8. Homogeneous feature (again) x H W Area Shape 9. Polygon display FIGURE 3.11 Dimensions of display proximity. al similarities; common object; and emergent features. We describe these below, identiw number in the figure. • Heterogeneousfeaturestypicallyusedforseparabledisplays • Homogeneousfeaturescanbeseparableorintegral Close proximity in space (or spatial contiguity; Ginn, 2006). As we have discussed, spacebased proximity is related strongly to the effort required to move attention (and particu- NSORY/PERCEPTUAL SIMILARITIES 4. Abutment 5. Heterogeneous feature or y DimensionsofDisplayProximity 6. Homogeneous feature 7. Homogenous feature Emergent features 8. Homogeneous feature (again) x H W Area Shape 9. Polygon display (a)symmetry FIGURE 3.11 Dimensions of display proximity. larities; common object; and emergent features. We describeversus: these below, identimber in the figure. /PERCEPTUAL SIMILARITIES proximity in space (or spatial contiguity; Ginn, 2006). As we have discussed, spaceSomedisplayscanbeconfiguredtocreate emergentfeatures proximity is related strongly to the effort required to move attention (and particuhe eyes) from one location to another. One example is the book designer’s goal to SampleGraphfromRecentThesis Graphinfirstdraftofthesis. Todistinguishbetweenconditions, usedonly: • colour(sensorydifferences) • heterogeneousfeatures(shape ofsymbols) Problem:requiredsignificanteffort tomentallyassociatepointsand inferpotentialtrendsand/or relationshipsamongconditions. Samedatainfinalthesis. Todistinguishbetweenconditions, used: • colour(sensorydifferences) • heterogeneousfeatures(shapeof symbols) • connections(linesjoiningdatafrom samecondition) • redundantcolour(lines&points) Example:ClusterAnalysis Napoleon’sMarchThroughRussia,1812-13 (SeeTufte) Summary:Focused/DividedAttention PositiveAspectsofParallelProcessing • ifinformationsourceshavethesameorsimilar meaning(i.e.redundancygain) – e.g.colour+shapecoding – e.g.objectdisplays(e.g.temperatureandpressure) • also…ifinformationsourceshaveindependent implications,butarenon-interfering – e.g.computercursor – e.g.flightdirectors Summary:Focused/DividedAttention NegativeAspectsofParallelProcessing • wheninformationsourcesincloseproximity haveincompatibleresponseimplications – e.g.Stroopeffect • resourcecompetition:wheninformation sourcesincloseproximitycompeteforthe sameperceptualanalysers–inotherwords, “displayclutter” – e.g.followingtwoauditorymessagesatonce – e.g.monitoringtwoadjacent,similardisplays – e.g.Brooks’memorytask(tobestudiedlater) GuidelinesforDesigningTasksInvolving Divided Attention • Useintegraldisplays(highproximity),asappropriate – (butthesearenotnecessarilyalwaysappropriate!) • Maintainconformality,asappropriate • Minimisenumberofpotentialsourcesofinformation** • Iftimesharingisnecessary,indicate,wherepossible,the relativeprioritiesoftasks/stimuli • Whenappropriate,optimise(ratherthanjustminimise) loadlevelofsecondarytasks. • Minimisedifficultyleveloftasks,wherepossible **Distracteddrivingisdangerous; don’tusemobilephone(talkingortexting)whiledriving! GuidelinesforDesigningTasksInvolving FocusedAttention • Separatecompetingchannelsphysically,asfarapartas possible • Makecompetingchannelsasdistinctfromeachother aspossible • Minimisethenumberofcompetingchannels • Maximisesalienceofthemostimportantchannel(s), usingcolour,brightness,loudness,centrallocation,etc. AttentionintheAuditoryModality Comparedtovisualmodality: • Auditorymodalityisomnidirectional – incontrasttovisualattention;no“earball” – (butsomeabilitytoattendtolocalisedsoundsources) • Auditorymodality(almost)always“on” – i.e.indark,evenwhensleeping • Mostauditoryinputsaretransient – heldinshorttermsensorystore,andthenvanishes – (auditorySTSSlongerthanvisualSTSS) • Auditoryinputshavegoodchanceofbeingnoticed – Advantageforauditorywarningsignals(vsvisual) – Disadvantageifneedtoignoreirrelevantsounds AttentionintheAuditoryModality: TheCocktailPartyEffect 1 5 2 3 4 6 WallStreetJournal,23/4,2012 ClassicalResearchonAuditoryAttention: DichoticListeningTask • Subjectrequiredto‘shadow’inputfromoneear,whileignoringinputfromotherear • Experimenterkeepstrackoferrors…andwhethercontentfrom‘ignoredear’appearsinoutput www.simplypsychology.org/attention-models.html ClassicalResearchonAuditoryAttention: Broadbent’sFilterModelofSelectiveAttention • attendingtooneparticulareardependsondistinguishablecharacteristics • e.g.whichear,typeofvoice,gender,etc. • filteringblocksoffonesource,infavourofanother • semanticprocessinghappensafterthefilter • takestimetoswitch(~500ms) www.simplypsychology.org/attention-models.html ClassicalResearchonAuditoryAttention: Treisman’sAttenuationModel • tocounterweaknessinBroadbentmodel • e.g.CocktailPartyeffect:‘unattended’informationneverthelessperceived • filterattenuates,ratherthaneliminates,unattendedmaterial • allowsperceptionofownname,plusotherimportantmaterial www.simplypsychology.org/attention-models.html AuditoryDividedAttention • dividedattentionachievedbyrapidswitchingbetweensources • reliesonretentionofinformation(forafewseconds)inSTSS • easiertoswitchoverifunattendedinformationiscontextually pertinent(e.g.ownname) • possibletodesignlessintrusiveauditorywarningsbyrelyingon personalisedalerts,expertexperience,etc. • possibletocreate‘auditoryobjectdisplays’ – e.g.words+melodyofsongs,meaning+voiceinflections – e.g.auditoryalerts,thatcombinepitch,loudnessandtimbre – e.g.sonificationdisplays,inmedicine,music,seismology,etc. • demo:http://www.youtube.com/watch?v=zoagpsLd8iM AuditoryFocusedAttention • cocktailpartyeffectexemplifiesabilitytofocusononesource – manifestsasauditorystreamsofstableanddistinctauditoryobjects – e.g.differentvoices,differentinstrumentsinorchestra,etc. – mediatedbypitch,timbre,spatiallocation,timing,etc. • easiertofocuswithdichoticlistening,relativetomonaural • despiteomni-directionality,abletolocalisesoundinspace • canusespatiallylocalisedsoundforseparateauditorychannels – e.g.headphones,speakerarrays,etc. – e.g.cuingofwarningsignals