Perception

1
Perception
2
Question of the Day

• Why is recognizing an object so easy for humans,
  but so difficult for computers?

3
Points of Confusion
http//i.ivillage.com/E/325/Celebrities/FaceReader
/FaceReaderIntro_325.jpg
http//www.apogeephoto.com/mag1-6/mag2-5mf1.jpg
4

• Perception
• attaching meaning to incoming sensory information

What is this?
5
Figure 2-1 Distal Stimuli, Proximal Stimuli,
and Percepts.
6
Gestalt ApproachFigureGround Figure 2-2
7
Gestalt ApproachSubjectiveContours Figure
2-3
8
Gestalt principles of Perceptual Organization

• Proximity

• http//www.aber.ac.uk

9
Gestalt principles of Perceptual Organization

• Similarity

• http//www.aber.ac.uk

10
Gestalt principles of Perceptual Organization

• Good continuation

• http//www.aber.ac.uk

11
Gestalt principles of Perceptual Organization

• Closure

http//daphne.palomar.edu
12
Gestalt principles of Perceptual Organization

• Common fate

13
Figure 2-5
Gestalt principles of Perceptual Organization
14
The number 4 from the check is compared to a
list of stored templates.
Bottom-Up Processes
Template matching
15
Bottom-Up Processes

• Problems with Template Matching
• Large number of stored templates needed
• How are new templates made?
• An object can be more or less like the template
• We can recognize many variations of a template

16
Figure 2-8
Bottom-Up Processes
17
Bottom-Up Processes

• Featural Analysis
• features (parts) of a stimulus are recognized
  by feature detectors and added together to help
  us perceive an object
• Lines or edges
• Geons
• Phonemes
• Parts of a face (eyes, nose)

18

• Featural Analysis
• Geons

19
Figure 2-14 A depiction of Selfridges (1959)
Pandemonium model.
Featural Analysis Letter detection
20
Bottom-Up Processes

• Featural Analysis
• Feature Properties
• Detectors can respond at different intensities
• Connections between detectors can have different
  strengths
• It is possible to change what a detector will
  respond to

21
Bottom-Up Processes
Prototype Matching
22
Bottom-Up Processes
Prototype Matching
http//www.palm.com
23
Figure 2-19An example of context effects in
perception.
Top-Down Processes
24
Top-Down Processes

• Perceptual Learning

25
Top-Down Processes

• Change Blindness
• http//viscog.beckman.uiuc.edu/grafs/demos/10.html
  

26
Top-Down Processes

• Word Superiority Effect

K
K D
WORK
OWRK
27
Bottom-Up and Top-Down Processing working together
Flying Animal
_at

• Word Perception
• Connectionist Model

Words
Bat
Letters
B
Fox
Bat
Fat
Features (lines)
I
Rat
Cat
28
Figure 3-26 (p. 91)Example of stimuli used in
the PET scan study of processing words.

• Word Perception
• Neuropsychological Perspective

29
Direct Perception vs. Constructivist Approach

• Biological motion
• http//www.psico.univ.trieste.it/labs/acn-lab/eng_
  p/e051c1m1_curr.html

30
Optic flow
31
Direct Perception

• Affordances
• Information from the stimulus that specifies how
  it can be used

32
Visual Agnosia
http//scien.stanford.edu/class/psych221/projects/
06/cukur/intro_files/image021.jpg
33
Visual Agnosia

• Associative Visual Agnosia
• Can copy, but unaware what it is cannot assign
  meaning to object
• Difficulty in transferring visual info into words
• Apperceptive Visual Agnosia
• Cannot recognize by shape
• Cannot copy drawings
• Often involves prosopagnosia

http//scien.stanford.edu/class/psych221/projects/
06/cukur/intro_files/image021.jpg
34
Agnosia
35
Identification of Faces and Members of
CategoriesProsopagnosia
The Fusiform Face Area
http//www.psy.vanderbilt.edu/faculty/gauthier/pic
ts/mona_lisa.jpg