The Biology of Visual Perception

1
The Biology of Visual Perception

• Eric Postma
• Computer Science

Course 3.3 Perception, Imagery, and Brain
2
Block homepage
http//www.cs.unimaas.nl/postma/imag.htm
3
Introduction

• Perceptual Phenomena
• perceptions (veridical and illusory)
• behavior (e.g., eye-movements)
• Biological Underpinnings
• neural structures (e.g., anatomy)
• neural processes (e.g., spiking)

4
A Grand Tour of the Visual System

• The Eye
• Primary Visual Cortex
• preprocessing the retinal image
• extracting features
• Higher Visual Areas
• extracting information from the (preprocessed)
  retinal image
• Recognizing Visual Objects
• invariant representations

5
The Eye(s)

• Vision is active
• The eyes act as the cameras of the brain
• The retinal network of the eye is a brain in
  itself

6
Yarbuss (1967) eye movement studies
7
Eye movements (overt attention)

• saccades
• sudden movements 50-100 ms
• preparation takes 200 ms
• fixations
• collecting visual information (and preparing a
  saccade)
• lasts for 200 ms
• about 4 fixation-saccade cycles per second

8
Nystagmus demonstration
9
Stopped images

• when images are stabilized on the retina they
  dissappear (Pritchard, 1961)

10
Demonstration stabilized images

• explanation
• the sharp disk remains visible due to the edges
• the blurred disk fades while the edges are not
  detected

11
Functional significance of involuntary eye
movements

• Eye movements are required for vision
• They need not to be compensated for
• Eyes are not like cameras (steady cams)
• Recent results stabilized images disappear
  within a fraction of a second
• Involuntary movements help maintaining the image

12
Patient with stable eyes
From Gilchrist (1999)
13
The retinal sampling lattice

• Retina rods and cones
• Rods for brightness (120 million)
• distributed over the entire retina
• Cones for color (7 million)
• most of them in the fovea
• Compression 1.5 million ganglion cells collect
  the signals from the rods and the cones

14
Distribution of rods and cones
15
Sunflower distribution

• The distribution of the scopes of the sampling
  elements is like a sunflower

16
Visual Cortex two major pathways

• The identification pathway (WHAT)
• striate cortex to inferotemporal cortex
• The location pathway (WHERE for action)
• striate cortex to parietal cortex

17
Cortical processing anatomy
18
(No Transcript)
19
Multiple feature analysis
V1
V2
20
Five types of visual processing

• Feature analysis detection of motion, color, and
  form (V1 and V2)
• Identification of color (V4)
• Identification of motion (V5/MT)
• Identification of form by shading
• Object recognition (IT)

21
Feature analysis in V1

• orientation selectivity
• color selectivity
• direction of motion selectivity
• size (spatial-frequency) selectivity
• texture selectivity

22
Spatial distribution of orientation preference

• All orientations are detected for each
  (sampling) point of the visual scene
• Top view of V1feature map

23
Why are all these features detected at V1?

• A theory from psychology...

24
Feature Integration Theory

• Treisman

visual pop-out
25
Assumption 1 Features

• Visual scenes are decomposed into elementary
  features such as
• oriented edges
• colors
• shapes
• sizes
• explains pop-out phenomena and visual search
  curves

26
Assumption 2 Objects

• Features are bound to form objects (binding
  problem)
• The process of binding may require attention

27
Two stages

• Stage 1 feature extraction from image
• preattentive
• Stage 2 identification of objects
• attentive

28
What are the features?

• Biology (of course)
• Adaptation experiments
• color adaptation (color aftereffect)
• motion adaptation (waterfall illusion)
• spatial-frequency adaptation
• orientation adaptation (tilt aftereffect)

29
Demonstration of the tilt and spatial-frequency
aftereffects
30
Further processing

• Retinal signals proceed via LGN (Lateral
  Geniculate Nucleus) to V1 (primary visual cortex
  or striate cortex)
• The left and right visual field are sampled by
  the left and right eye
• The left striate cortex processes the right
  visual field of both eyes, the right cortex the
  left field of both eyes.
• Integration?
• Retinotopy...

31
Texture segregation
no segregation
segregation
32
Therefore,

• Line orientations are important in segregation or
  grouping
• Arrangements of lines are not

33
Boundary defined by shape
OOOVV OOOVV OOOVV
34
Boundary defined by color
OVVOO VOOOV OVVVO
red
yellow
35
Boundary defined by a conjunction of shape and
color
OOVOV VVOVO VVOOV
36
Preattentive vs attentive vision

• Preattentive vision deals with individual
  features, not with feature conjunctions
• Attentive vision deals with combinations of
  features and glues them together to form objects
• Note what should be glued with what?
  (combinatorial explosion)

37
Illusory conjunctions 1
S
S
S
S
S
brief presentation
perception
38
Illusory conjunctions 2
brief presentation
perception
39
Illusory conjunctions 3
brief presentation
perception
40
Illusory conjunctions 4
brief presentation
perception
41
Search curves for features

• Reaction times do not increase with the number of
  items (distractors) in the display
• Feature search
• no target present or target present
• RT more or less constant
• conclusion parallel search

42
(No Transcript)
43
Conclusions on FIT

• Features are processed in parallel
• Conjunctions have to be glued together and are
  therefore processed in a serial fashion
• Note the link to biologymultiple processing
  streams

44
Retinotopic mapping

• There is a rough mapping of the retinal lattice
  onto V1
• Adjacent cells respond to adjacent positions of
  the visual field (as represented on the retina)
• What does this mean?

INTERMEZZO
45
A picture at V1...
INTERMEZZO
46
No! (Homunculus problem)
INTERMEZZO
47
Do not confound structure with function!!!

• A common fallacy among psychologists and
  biologists
• There is no picture in the head, because there
  is no viewer in the head.
• Perceptions mediate actions

INTERMEZZO
48
Example Light follower
Do they behave differently?
INTERMEZZO
49
Thats it