(note that this approach is focusing on the image, while we are focusing on behavior)
(2) The representation is formed through modular analysis.
multiple
representations of the retinal image - what are they doing?
a) parallel pathways
b) hierarchical stages of analysis
a)
parallel pathways
component representations - each carries information about a particular
aspect of the image
(helping in grouping, preceeding segmentation)
independent
analysis of visual attributes: luminance,
texture, color, disparity,
movement,
orientation
evidence is coming from:
(1) physiology and anatomy of monkey visual cortex
single cell technique:
receptive field location, size, structure
receptive field structure: orientation
in V1,
motion
in MT, color in V4, objects
in IT....
topographic maps - economic connectivity
across space & features
(2) neuropsychology: cortical color- and motion-blindness, agnosias
(1) and (2) leading to the concept of 2 parallel cortical pathways:
V1-V2-V4-IT(Br21): object identification
V1-V2&V3/VP-V5-P(Br7): movement and position
(3) psychology: e.g., visual search:
color, orientation, size, curvature
b)
hierarchical levels
- LGN to V1 to all other areas
P/M to ventral/dorsal
- within V1, hierarchy of complexity in RF structure
- road map of cortical organization based
on anantomy of
ascending projections: upper layers to layer 4
descending projections: lower layers to other than 4
Here is a great site to study again:
http://www.med.uwo.ca/physiology/courses/sensesweb/
| Retina | PARASOL
large RF transient response rapid conduction broad-band |
MIDGET
small RF sustained slower conduction color-opponent |
| LGN | 2 MAGNO layers
low resolution high contrast sensitivity fast color-blind |
4 PARVO layers
high resolution low contrast sensitivity slow color selective |
| V1 | 4 C alpha
| 4 B ------------------> | | | |
4 C beta
| blobs in layer 2/3<--|--> interblob in layer 2/3 (color) (orientation, disparity) | | | | |
| Extrastriate | MT & V2 thick
(to MT) (movement) |
V2 thin (to V4)
V2 pale stripes
(to V4) |
click on the figure to get the full roadmap by Ungerleider, 1995
David Hubel
demonstrating color & luminance contrast
| ACTION | PERCEPTION |
| size
shape direction distance |
shape
color texture shading transparency |
| veridical | constancies (size, color, shape, brightness) |
| on-line
fast, robust |
memory requirement is large |
| ego-centric | allocentric |
Can you come up with some more distinctions between the two hypothetical
systems?
How could these be tested experimentally?
Can you list some experimental findings or observations that might
support the theory?