Shape, Pattern, Form
• What is needed for shape (pattern, form)?
• Facts a theory must account for
• Organization
• Four categories of theories
• The Information Processing approach
• Bottom-up and Top-down processing
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
1
What Is Needed For Shape
• Contours
– discontinuity, change, edge, border
• Contours are necessary for shape but do not equal the
shape
– ambiguous figures
– sketches
– illusory or anomalous contours
• shape without the retinal image of the shape
• need information to indicate the boundaries
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
2
Illusory or Anomalous Contours
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
3
Facts To Be Explained
• Retinal image is ambiguous
– size and shape constancy
• the retinal images changes but perception remains the same
– ambiguous figures
• retinal image remains the same but perception changes
• Segmentation of image into objects
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
4
Facts To Be Explained
• Perceptual Learning
– learn new discriminations
• Figural Adaptation
– figural affereffects
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
5
Organization Of The Retinal Image
• Figure/Ground segregation
– separating objects from background
• Gestalt Grouping Principles
– figuring out which parts of the image belong with which objects
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
6
Figure/Ground Segregation
• What part of the image becomes the figure
–
–
–
–
–
bounded (closed) area
smaller area
symmetric area
convex area
brighter area
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
7
Bounded and Smaller
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
8
Symmetric and Convex
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
9
Figure/Ground Qualities
• Figure
–
–
–
–
–
–
• Ground
shape
bounded
near
identified
meaning
remembered
Sensation and Perception - shape.ppt
–
–
–
–
–
–
shapeless
continuous
far
not identified
meaningless
forgotten
© 2001 Dr. Laura Snodgrass, Ph.D.
10
Gestalt Grouping Principles
• Proximity
A
AAA
A
AAA
A
AAA
• Similarity
AAAAABBBBAAAAAAA
AAAAAABBAAAAAAAA
AAAAABBBBAAAAAAA
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
11
Gestalt Grouping
• Good continuation
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
12
Gestalt Grouping
• Closure
– illusory contours are based on good continuation and closure
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
13
Gestalt Grouping
• Common fate
– camouflage is broken by common fate
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
14
Gestalt Grouping
• Pragnanz
– simplicity and regularity
• Set and past experience
– familiarity and expectations
– “Street figures” and ambiguous figures
– perception organized by your expectations
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
15
Four Categories Of Theories
• Theories differ on
– what is encoded
– how the encoded information is stored and used
• Categories
–
–
–
–
analytic feature models
spatial frequency models
template models (prototype models)
cognitive construction models
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
16
Analytic Feature Models
• What is encoded?
– Small local features or elements
– based on Hubel and Wiesel feature detectors
• Theories within this category differ on how the
features are reassembled
• Example: Selfridge’s Pandemonium Model
– hierarchy of feature detectors
– parallel processing (all features detected at once)
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
17
Selfridge’ Pandemonium Model
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
18
Spatial Frequency Model
• Fourier analysis (what is encoded)
– break complex image into component sine waves
– sine waves are a graph of the distribution of brightness over space
(blurry bars)
• The components (blurry bars) can vary in
–
–
–
–
frequency
amplitude
orientation
phase
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
19
Sine Waves
Sine wave patterns
Also called blurry bars or gratings
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
20
Sine Waves
Top pattern is low frequency, high amplitude
The next three increase in frequency, but
decrease in amplitude
The first four patterns start in-phase, but quickly
get out-of-phase (the peaks and valleys do not
line up after the first peak)
As you add the patterns together you get a
square wave (black and white stripes)
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
21
Spatial Frequency Model
• Adding the components together reproduces the
original image
• Theories in this category vary in
– how the components are added together
• Examples of evidence
– contrast sensitivity function
– block portraits
– selective adaptation
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
22
Template Model
• What is encoded
– whole objects (but not whole scene)
– model (template) stored in memory
– Prototype
• Incoming image is compared to stored representation
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
23
Cognitive Construction Models
• What is encoded:
– depends upon expectations
– can encode an entire scene at a time if given proper schema
– can encode local elements if no schema is available or if incoming
information conflicts with schema
• Schemas
– sets of expectations
• e.g. a living room
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
24
Cognitive Construction Models
• Image is processed in steps
– make hypothesis based on schema (expectations)
– if incoming information matches the hypothesis there is rapid
identification
– if incoming information does not make, must make new hypothesis and
try again
– cycle through matching procedure until get a match
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
25
Cognitive Construction Models
• Examples of evidence
–
–
–
–
ambiguous pictures
“Street” figures
scene perception research
research on the effects of “set” and context
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
26
Set and Context
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
27
Information Processing Approach
• Stages of processing of form through the memory
and recognition system
• Not a separate theory of form perception
• The models of shape perception
– refer to processing in LTM that results in recognition
– propose different mechanisms that shape may be stored in LTM
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
28
Information Processing Approach
• Distal Stimulus
– object in the environment
• Stage I
Proximal Stimulus
– transduction produces the retinal image
• Stage II Sensory Register
– iconic storage for 1/4 sec
• Stage III Short Term Memory
– factors that influence further processing
• Stage IV Long Term Memory
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
29
Sensory Register
• Sperling’s Partial Report Technique
– can report 3 -5 letters before image fades
– proves that entire image is available but fades fast
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
30
Short Term Memory
• Limited capacity
– 7 +/- 2 items
• Many factors influence either what gets further
processing or the speed of further processing
• Any information that does not get further processing
is lost
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
31
Factors That Influence Processing
• Task requirements
– may not need to do a full analysis
– global discrimination
• in which of the following would it be easier to find the letter Q?
Why?
– EKLXAVFTZ
– ODCBGPRU
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
32
Factors that influence processing
• Instructional Set
– process only the features you are instructed to attend to, such as color,
location, form, or size
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
33
Factors That Influence Processing
• Familiarity
– familiar things are processed faster
• word superiority
– “m” presented alone
– “aemg” m presented as part of a letter string
– “game”
m presented as part of a meaningful
word
• the letter is detected fastest when part of a familiar word, even
faster than if presented alone
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
34
Factors That Influence Processing
• Interference
– slows down processing
– features you can’t ignore
– e.g. Stroop Task - color words presented in incongruous ink colors.
• The automaticity of reading interferes with color naming.
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
35
Stroop
•
•
•
•
•
•
•
•
•
Blue
Green
Red
Blue
Green
Yellow
Yellow
Red
Blue
Sensation and Perception - shape.ppt
•
•
•
•
•
•
•
•
•
Yellow
Green
Red
Orange
Green
Red
Blue
Yellow
Orange
© 2001 Dr. Laura Snodgrass, Ph.D.
36
Factors That Influence Processing
• Schemas and expectations
– you are faster at processing things you expect to see
– you are faster at processing things that fit your schemas
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
37
Long Term Memory
• Incoming sensory information is compared with the
representation stored in LTM
• Each of the theories of form perception makes
different predictions about:
– what is stored in LTM
– the mechanism of the comparison process
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
38
Direction of Processing
• Bottom-up (data driven)
• Top-down (conceptually driven)
– the features encoded rom the
environment determine the
final representation
– usually fast and automatic,
relatively fixed
– e.g. most feature models and
spatial frequency models
Sensation and Perception - shape.ppt
– the representations already
stored in LTM determine what
gets encoded from the
environment
– slower, more flexible
– experience, expectations,
schemas, and set influence
encoding
– e.g. cognitive construction
models
© 2001 Dr. Laura Snodgrass, Ph.D.
39
Direction of Processing
• Both direction of processing are necessary
– may use one or the other in different situations
– sometimes both are needed
• e.g. ambiguous figures
– abstract art
Sensation and Perception - shape.ppt
© 2001 Dr. Laura Snodgrass, Ph.D.
40