PreAttentive Processing

1
Certain aspects of visual processing seem to be
accomplished simultaneously (that is, for the
entire field at once) and automatically (that is,
without attention being focused on any one part
of the visual field). Other aspects of visual
processing seem to depend on focused attention
and are done serially, or one at a time, as if a
mental spotlight were being moved from one
location to another.
Ann Treisman, 1986
2
Overview

• Ware
• Fundamental visual processing done
  pre-attentively
• List
• Samples
• Multiple feature elements can be combined

• Additional Research
• Feature Integration Theory
• Automatic Processes
• Hemispheric Processing
• Prior Knowledge
• Inhibition of Return

3
P.A.P. Features (Ware)

• Ware provides an excellent list
• Skipped closure specifically lack of closure
• Possible to combine multiple features in one
  interface, but requires skill

4
Conjunctive Processing (Ware)

• Results when multiple features are present and
  target shares same features as distractors (see
  example)
• Possible to maintain P.A.P. if objects are
  separated so they appear as distinct groups (see
  example)
• Also possible to use prior-knowledge to assist in
  processing

5
Competing Features

• How do we choose when multiple features present?
• Feature Integration Theory (Treisman, 1986)
• Early vision creates feature difference maps
• Each map relates to a primitive feature
• The stack of maps is navigated via a master map
  (see illustration)

6
Feature Difference Maps

• Feature differences are added together
• Area with the highest score wins attention
• Areas are then examined in descending order
• Maps can be short-circuited if a single primitive
  is used it creates a pop-out effect
• This explains why Wares examples were so weak
  no clear winner of attention

7
Automatic Processes

• Automatic Processes (habits) appear to be very
  similar to P.A.P
• Both function independently from attention
• Habits developed through practice and training
• P.A.P. innate or acquired early in childhood
• Functionally different
• Habits support skilled behavior, little
  transferability
• P.A.P. support low-level perception in all areas

8
Hemispheric Processing Styles

• Right hemisphere of brain is holistic, uses
  template matching when performing a task
• Process music and visuo-spatial information
• Ability to integrate multiple pieces of
  information
• Left hemisphere recognizes well-learned
  individual units, then serially integrates them
• Counting, processing unfamiliar words, phonetics
• Expression via language

9
Hemispheric P.A.P.

• Janiszewski (1988) conducted experiments testing
  hemispheric P.A.P.
• Experiment
• Fake newspapers with target ads
• Subjects tested for conscious recall very few
• Subjects tested for ad preference
• Preconscious preference formation is
  hemispherically based

10
Hemispheric P.A.P. (cont)

• Ads can be processed pre-attentively
• Different ads were used to test hemispheric
  preference
• Based on results
• Graphic ads are more effective in left peripheral
  vision
• Text based ads are more effective in right
  peripheral vision

11
Hemispheric vs. Exposure

• Janiszewskis study was challenged four years
  later
• Shapiro conducted experiments with no significant
  preference to hemisphere, just preference by
  exposure
• Both studies indicated they were not conclusive
  and more research needed to be done.

12
Prior Knowledge in P.A.P.

• Treisman conducted experiment to test conjunctive
  processing and prior-knowledge
• Experiment
• A number flanked by two images shown for 200 ms
• Followed by a checkerboard to flush iconic buffer
• Asked users what numbers and images appeared (see
  sample)

13
Prior Knowledge (cont)

• Images described as shapes (orange triangle,
  black ring, and blue ellipse) and as objects
  (carrot, tire, and lake)
• Occasionally colors were changed to test for
  shape AND color perception
• Shapes resulted in 29 false results
• Objects resulted in 5 false results

14
Prior Knowledge (cont)

• Based on results
• Prior knowledge does help attention use
  pre-attentive processing effectively in conjoined
  searches
• The P.A.P. still works the same way, but serial
  processing is assisted

15
Inhibition of Return (IOR)

• The basic claim underlying inhibition of return
  (IOR) is that after attention is reflexively
  shifted to a location in space, there is a
  delayed responding to stimuli displayed at that
  location (Theeuwes, 2001)
• Theeuwes experimented to test this inhibition

16
IOR Experiment

• Placed eight objects in a ring
• One was a distracting color/shape see example
• Targets (small gray dots) were placed inside,
  near, and far from distractor
• After viewing for 1300 ms subject had to
  determine if a gray dot was turned off

17
IOR Results

• Target inside distractor slowest response
• Target near distractor slow response
• Target far from distractor fastest response
• It seems reflexive selection of an object
  inhibits processing of neighboring objects
• Possibly the distractor pulls resources away from
  near-by elements

18
Design Implications

• Feature Integration Theory
• Greater differences get processed first
• Make important areas strongly contrast
  surroundings
• Too many differences cancel each other out
• Single feature primitive can create pop-out
• Automatic Responses (habits)
• Can be used like P.A.P. for trained audience
• Should not be used with general public

19
Design Implications (cont)

• Hemispheric Processing
• Efficacy still being debated
• Put graphics on left side of screen
• Put text on right side of screen
• Prior Knowledge
• People respond more accurately to known objects
• Try to keep design based on existing knowledge

20
Design Implications (cont)

• Inhibition of Return
• If possible, strongest visual element should be
  most crucial element
• If strongest visual element is not most crucial,
  separate two by lots of space

21
Questions?
End
22
Ware P.A.P. Primitive Features

• Line orientation, length, width,
  collinearity,size
• Curvature, grouping, added marks, numerosity
• Color hue, intensity
• Motion flicker, direction
• Spatial Position 2D, stereoscopic,
  convex/concave

Return
23
Ware P.A.P. Samples
Return
24
Ware Skipped Feature - Closure
Closure

• Lack of Closure

Return
25
Ware - Conjunctive Search
Return
26
Ware Conjunctive PAP Search
Return
27
Treisman F.I.T. Maps
Return
28
Poor Feature Difference Maps
Return
29
Treisman Prior-Knowledge Test
Return
30
Theeuwes IOR Experiment
Return