... Trying to pay attention to a cell phone conversatio

1
Visual Perception and Attention

• Lecture 3

2
Why Attention?

• Limited Capacities
• We receive more sensory information than we can
  possibly process.
• Example You probably arent aware of the feeling
  of the socks on your feet because you have better
  things to attend to.

3
Limits of Attention

• Selective Attention Errors
• Focused Attention Errors
• Divided Attention Errors

4
Limits of Attention

• Selective Attention Errors
• Errors are caused by selecting the wrong
  information
• Example paying attention to the radio rather
  than the road during poor weather conditions.
• Focused Attention
• Divided Attention

5
Limits of Attention

• Selective Attention Errors
• Errors are caused by selecting the wrong
  information
• Focused Attention
• Problems excluding irrelevant information
• Example neighbor plays music too loudly while
  you are trying to study.
• Divided Attention

6
Limits of Attention

• Selective Attention Errors
• Errors are caused by selecting the wrong
  information
• Focused Attention
• Problems excluding irrelevant information
• Divided Attention
• Problems dividing attention between two sources
  of information at once.
• Example Trying to pay attention to a cell phone
  conversation while also paying attention to the
  road.

7
Visual Limits

• Visual Angle - measures the size of the image
  falling on the retina.

8
Visual Limits

• If the object moves closer, the visual angle
  increases

9
Visual Limits

• If the object moves closer, the visual angle
  increases

10
Structure of the Eye
Lens
Retina
11
Retina

• 2 types of photocells
• Rods
• sensitive to shades of gray
• Cones
• maximally sensitive to red, green, or blue

12
Ganglion Cells
13
Spatial Frequency

• used to describe the degree of spatial detail.

High Frequency fine detail
Low Frequency course detail
14
Spatial Frequency
low frequencies only
original
15
Spatial Frequency
high frequencies only (ignore lack of color
saturation in this example)
original
16
Fovea

• Fovea - located at the center of the retina

17
Fovea

• visual angle is roughly the size of your
  thumbnail at arms length

18
distribution of rods and cones

• Cones are concentrated in the fovea.
• Rods occur only in the periphery

19
Normal Image
20
Adjusted for retinal acuity
21
Low Light Conditions

• Cones relatively insensitive in low light
  conditions
• Q As the light level drops, what types of
  photoreceptors do we start to rely on more-and
  more?
• A rods!

22
Low Light Conditions

• If we are seeing primarily with our rods, what
  does this do to our perception?
• Black White
• Course detail/low frequency/low resolution
• Little or no foveal vision

23
Simulated Low light
low frequencies, no color
original
24
Visual Systems
25
Cortical Magnification

• A large chunk of the visual cortex is devoted to
  the fovea.
• Fovea is 0.2 of the retina, but is represented
  by 25 of the visual cortex.
• 1000x as much cortical tissue is devoted to
  processing information from the fovea compared to
  an equal sized patch from the periphery.

26
(No Transcript)
27
(No Transcript)
28
Why Cortical Magnification?

• If the brain had the same amount of tissue as
  devoted to the fovea devoted to the rest of the
  visual field, then our visual cortex would have
  to be 100x bigger (i.e. wed have huge heads).
• Instead, we can make eye movements to move the
  fovea around.

29
Eyemovement terminology

• Fixations
• When the eyes are stopped on an object.
• The eyes are not completely steady during a
  fixation, but show small, involuntary tremors.
• Saccades
• Smooth Pursuit
• Dwell Times

30
Eyemovement terminology

• Fixations
• Saccades
• The ballistic reorientation of the eyes from
  point of interest to another.
• Saccades connect fixations together.
• Visual processing is suppressed during saccades
• Smooth Pursuit
• Dwell Times

31
Eyemovement terminology

• Fixations
• Saccades
• Smooth Pursuit
• Occurs when the eyes track a moving object.
• Not considered a saccade.
• Dwell Times

32
Eye Movements

• Fixations
• Saccades
• Smooth Pursuit
• Dwell Times
• How long the eyes dwell on an item of interest.
• May consist of several fixations.
• Time affected by difficulty of information
  extraction.

33
(No Transcript)
34

• The amount of information we can usefully extract
  from the environment is not only determined by
  retinal acuity and cortical magnification, but is
  also determined by attention.
• Useful Field of View is our attentional window
  on the world.
• The UFOV is centered in the middle of our visual
  field.
• Its width is determined by the ease of image
  extraction and other factors such as the effects
  of aging.

35
Eye movements UFOV

• Saccades are used to connect attentional windows
• What happens when items are difficult to
  discriminate?
• UFOV shrinks -- fewer items examined in each
  fixation
• Smaller UFOV leads to more eye movements

36
Eye movements Attention

• Covert attention can move around while the eyes
  are still.
• Example checking out someone out of the corner
  of your eye while pretending to read a
  newspaper.

37
Eye movements Attention

• Covert attention can move around while the eyes
  are still.
• Overt attention eye movements or any physical
  orienting of the body.

38
Eye Movements Covert Attention

• When the eyes are still, covert attention is free
  to move independently.
• Programming an eye movement always leads to a
  shift of covert attention.
• That is, covert attention gets to the intended
  target before the eyes begin to move.

39

• If we ask you to move your eyes to a specific
  location (e.g. left), and
• flash a target right before your eyes move
• you will be more likely to identify the target
  when it appears where your eyes will land than
  when it appears somewhere else.

40
0 msec 50 msec 100 msec 150 msec 200 msec
Left
T
T
41
Attentional Guidance

• Top-Down
• Environmental Expectancies
• Where
• should I look on the ceiling for my car keys?
• Attentional Set
• What
• is it smaller than a bread box?

42
Attentional Set

• Singletons (Folk et al.)
• items that are identifiable by a unique (compared
  to other visible items) feature
• e.g. the only red car in the parking lot.
• Search for feature singletons is extremely
  efficient.

43
Attentional Guidance

• Bottom-up (attentional capture)
• Onsets -- items that appear in a display.
• Must include a luminance change
• Color changes without a luminance change do not
  count as onsets.

suddenly
44
Im not an onset
I am
So am I
45
Attention Switching

• Bottom-up (exogenous, involuntary)
• Fast, automatic, unintentional
• Top-down (endogenous, voluntary)
• slow, intentional

46
voluntary arrow cue (75 valid)
involuntary onset cue (25 valid -- chance)
47
(No Transcript)
48

• 2 or 5

49
(No Transcript)
50
(No Transcript)
51
(No Transcript)
52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
Onsets

• Why are onsets special?
• Detectable by magnocellular system
• Magno is sensitive to motion luminance change
• Magno is also faster conducting than parvo
• If magno, then where (spatially) should onsets be
  most effective?
• in the Periphery
• no rods in fovea

56
Of course, web advertisers have noted than, and
have used this knowledge to make web browsing
extremely annoying
57
(No Transcript)
58
(No Transcript)
59
Search

• Targets the items we are looking for
• Distractors the stuff were not looking for

60
Serial Self-Terminating Search (SST)

• Assumes that items are examined one at a time.
• Search terminates when the target is found or all
  items have been examined.
• Each item takes roughly the same amount of time
  to examine.

61
SST

• If the target is present, on average, how many
  items need to be examined?
• roughly half
• (n1)/2
• where n size of the search set
• If absent, all items will be examined.

62
SST (set size 3)

• 1
• 2
• 3
• 6 ? 3 2 items

63
Search Task

• Is there an L present in the display?

64

• T T
• T
• T T
• T L
• T T T
• T
• T T T T
• T L

65

• According to SST, on average, you should have
  searched half the items.
• So, why was search so easy?

66

• T L
• T
• T T
• T T
• T T T
• T
• T T T T
• T L

67
Guided Search (Wolfe)

• Attention is attracted to the most salient item
  in the display.

68
Guided Search (Wolfe)

• Attention is attracted to the most salient item
  in the display.
• Saliency determined by
• Attention is attracted to items that are highly
  dissimilar from their neighbors (local
  differences).

69
find the vertical line (easy search)
70
find the vertical line (hard search)
71
Guided Search (Wolfe)

• Attention is attracted to items that are highly
  dissimilar from their neighbors.
• Local differences are along a feature dimension
• Features include
• color
• edge/line orientation
• stereopsis
• direction of movement
• size (spatial frequency)
• luminance
• more

72
Guided Search (Wolfe)

• Saliency determined by
• Attention is attracted to items that are highly
  dissimilar from their neighbors (local
  differences).
• Top-down attentional set.
• If you are looking for your car keys, you are
  looking for
• Something silver and shiny
• Something small and key shaped
• you are using you top-down set to weight the
  feature qualities (including local differences)

73

• T T
• T
• T T
• T T
• T T T
• T
• T L T T
• T T

74
Guided Search (Wolfe)

• You probably looked at the red T first, because
  previous experience had you looking for a red
  item.

75
Guided Search (Wolfe)

• Saliency determined by
• Attention is attracted to items that are highly
  dissimilar from their neighbors (local
  differences).
• Top-down attentional set.
• Whether an item has been examined already
• When an item is examined, its saliency is
  suppressed.
• This keeps attention from being stuck on the
  same item.
• i.e. you have a memory for items that you have
  already examined

76
Guided Search (Wolfe)

• When is search fast?
• when an item is very different from its
  neighbors.
• when the item youre looking for matches your
  attentional set.

77
Guided Search (Wolfe)

• When is search slow?
• When the target is similar to its neighbors
• Example camouflage
• When the item you are looking for does not match
  your attentional set.
• hmmmhe was thinner and had a lot more hair when
  I last saw him 10 years ago

78
Parallel Search

• Search in some sense is always parallel.
• The question is how many items can you examine
  at once?
• Q When might it be easy to examine several items
  in parallel?
• A when the items are highly dissimilar to each
  other.

79
Parallel Search

• When items are highly dissimilar, the UFOV is
  large and many items can be examined in a single
  fixation.
• When items are highly similar, the UFOV becomes
  smaller, perhaps encompassing only a single item.
• Example Wheres Waldo?

80
UFOV Parallel Search

• When the UFOV is
• Small a greater number of fixations will be
  required to search the environment.
• Large fewer fixations (perhaps only a single
  glance) will be required.

81
Efficiency During Search

• As more and more items are examined, the
  likelihood of finding the target increases
• i.e. you have a memory for which items you have
  examined and eliminate them from your list of
  candidates.

82
Efficiency During Search

• As more and more items are examined, the
  likelihood of finding the target increases.
• However, as time goes on, your probability of
  finding the target decreases at a diminishing
  rate.

83
Efficiency During Search

• As more and more items are examined, the
  likelihood of finding the target increases.
• However, as time goes on, your probability of
  finding the target decreases at a diminishing
  rate.
• Sometimes items are poorly examined the first
  time and are later reexamined.
• Strategies might lead you to concentrate to much
  on a single area.

84
General Principles of Search

• Target-distractor similarity affects search speed
• Pop-out - when target is a singleton
• Memory set/ of targets
• Am I looking for Bob, or anyone from the office?
• Automaticity - highly overlearned items
• Can you spot the green-crested stickleback?

85
Structured Search

• Examples
• Scanning Airline departure times.
• Scanning a pull-down menu.

86
Structured Search

• Search isnt free
• search is structured to be performed generally in
  a linear fashion.
• there often are no distinguishing features that
  can be processed in the periphery -- items must
  be examined individually.

87
Structured Search

• Given this knowledge, what can we do to make
  search more efficient?
• Place the most likely targets at the start of the
  search (usually the top).
• Add structure
• Alphabetize items
• Group items by similarity

88
items grouped by similarity of function
inactive items are deemphasized
89
Parallel processing / Divided attention

• What is parallel processing?
• items are processed all at once.
• Example letters in a word are processed all at
  once, not one-at-a-time.

90
Early vision / perceptual organization

• Early in the visual system, information is
  extracted and organized in parallel.
• much of this is effortless, automatic, and does
  not require attention.
• extracts features and organizes them into objects.

91
Gestalt grouping principles

• Proximity

92
Gestalt grouping principles

• Similarity

93
Gestalt grouping principles

• Common fate

94
Gestalt grouping principles

• Good Continuation

95
Gestalt grouping principles

• Closure

96
Global Precedence

• Items are examined at several different scales
• Large is examined before small
• EEEEEEEE
• E E
• E E
• E E
• EEEEEEE
• E E
• E E
• E E
• E E

97
Emergent Features

• A global property of a set of stimuli that is not
  apparent by examining the stimuli in isolation

98
Emergent Feature
99
Simons movie
100
Overlapping views HUDs

• Logic its easier to extract information from
  two items when they are close together than when
  they are farther apart.

101
Problems

• Allows, but does not guarantee parallel
  processing.
• The two displays might be treated as 2 different
  sources of information
• Pay too much attention to the HUD (cognitive
  tunneling).

102
Costs Benefits

• Bad Can clutter the display, making items less
  easy to detect or process.
• Good Works particularly well when the observer
  expects the stimulus
• conformal symbology is an extreme example of
  this.

103
Conformal Symbology
104
Conformal Symbology
105
Focused Attention

• Tight spacing makes it easier for us to share our
  attention between two items.
• However, if items are too close, it becomes
  difficult to focus on a single item.
• How tightly can we focus our attention???

106
Focused Attention

• Flanker Task
• respond to middle item (L or R), ignore the other
  items (flankers).
• 3 spaces, 3 flanker/target compatibilities.
• compatible incompatible neutral
• LLL LRL XRX
• R R R R L R X L X
• L L L L R R X R X

107
Focused attention
Response Competition
Redundancy Gain
108
Object based attention

• When you attend to part of an object, you attend
  to all of the object.
• Demo Name this word

109

• Red

110
Stroop Effect

• The reason that the Stroop task is so difficult
  (at least when the names and colors mismatch), is
  because you automatically process all parts of
  the object (color, form).

111
Proximity-Compatibility Principle

• Items that have close processing proximity should
  have close display proximity.
• Display proximity - how close are two components?
• distance
• other properties, e.g. same color
• Gestalt grouping or part of the same object

112
Proximity-Compatibility Principle

• processing proximity -
• the extent to which 2 sources of information are
  used in the same task.
• example altimeter rate of climb gauge

113
(No Transcript)
114
Proximity-Compatibility Principle

• Moving items close together increases the
  likelihood that they will be processed together,
  as does making them part of the same object

115
Proximity-Compatibility Principle

• Moving items close together increases the
  likelihood that they will be processed together,
  as does making them part of the same object
• Close proximity can cause emergent features to
  occur. If the emergent feature is not part of
  the task or is unintended, this can hurt
  performance.

116
Proximity-Compatibility Principle

• Moving items close together increases the
  likelihood that they will be processed together,
  as does making them part of the same object
• Close proximity can cause emergent features to
  occur. If the emergent feature is not part of
  the task or is unintended, this can hurt
  performance.
• Close proximity can create clutter or response
  conflict. If the task is likely to call for
  focused attention on only a single item at a
  time, then avoid close proximity

117
Color coding

• For the 97 of the population who are not color
  blind, color coding can be of great benefit
• highlighting - if the color is significantly
  different from the rest of the display.
• stereotypical meanings
• red stop red hot
• yellow caution blue cold
• green go

118
Color coding

• For the 97 of the population who are not color
  blind, color coding can be of great benefit
• highlighting - if the color is significantly
  different from the rest of the display.
• stereotypical meanings
• used to tie together spatially separate objects
• mixing board
• Redundant coding
• a traffic light uses color and location

119
Color coding

• Limitations of color coding
• Sensory
• limit the number of colors to 5 or 6
• color is not perceptible under low-light
  conditions

120
Color coding

• Limitations of color coding
• Sensory
• limit the number of colors to 5 or 6
• color is not perceptible under low-light
  conditions
• Cognitive
• Color does not imply an ordered continuum
• Is red more than green?

121
Color coding

• Limitations of color coding
• Stereotypes
• red-yellow-green stoplight pattern might not have
  the same meaning in different parts of the world

122
Color coding

• Limitations of color coding
• Stereotypes
• red-yellow-green stoplight pattern might not have
  the same meaning in different parts of the world
• Coding
• color coding should be relevant and consistent.

123
(No Transcript)