LCS 11: Cognitive Science 1. Neural retina 2. Lateral geniculate - - PowerPoint PPT Presentation

Pomona College

LCS 11: Cognitive Science

Vision 2

Jesse A. Harris April 22, 2013

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 1

Agenda

֠ Visual pathway

• 1. Neural retina
• 2. Lateral geniculate nucleus
• 3. Primary visual cortex

֠ Construction illusory surfaces ֠ Reading for Wednesday: Hoffman, 1998: ch 4 ֠ Group presentations May 1, 6, and 8

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 2

Visual pathway Eye

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 4

Neural retina

◮ Image focuses on neural retina ◮ Neural retina composed of discrete photoreceptor cells

which are attuned to particular stimuli Rods Responsive to movement Cones Responsive to color and detail

◮ Early selectivity within eye itself

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 6

Rods and cones Retinal ganglion cells

Retinal ganglion cells

Respond to small receptive field, consist of cells with ON and OFF regions. ON Excitation OFF Inhibition Some cells have ON center, others have OFF center, approx equal cells that compete with one another in terms of firing rates.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 8

Retinal ganglion cells Nystagmus

Fading images

Many cells in retina are detectors of fading; attuned to changes in stimulus (transient cells).

◮ Constant slight tremor (nystagmus) to reactivate

stimulus on photoreceptors

◮ If eye is stabilized by paralyzing eye muscles with drugs

• r by moving images to account for movement, world

becomes a dull grey.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 10

Troxler fading

◮ Fixate on upper cross for a while: Blurry grey lines

disappear.

◮ Then fixate on lower cross: complementary after images

appear.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 12

Troxler fading

◮ Fixate on central dot and doughnut will disappear,

replaced by orange background.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 14

◮ Left visual field inputs right V1 ◮ Right visual field inputs left V1 ◮ Crossing at the optic chiasm

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 16

Lateral geniculate nucleus

◮ Cells similar to retinal ganglion cells, with small receptive

fields.

◮ No discrimination between more complex features of

image, e.g., orientation or movement.

Retinotopic maps

Nearby parts of image are represented in the nearby cells of the retina, LGN, and V1 areas of the brain. Physiological

• rganization selectively mirrors organization of object

perceived.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 17

Primary visual cortex (V1)

Columnal organization

Cells are ordered via columns (approx 1mm strips) that are sensitive to particular features. Each cell has a preferred

• rientation which responds to

input from a particular receptive field. But columns are not evenly distributed across V1!

Distortion in retinotopic maps

Cortical magnification effect

Cells most dense around fovea - far greater amount of V1 dedicated to processing information from with the fovea. Results in a distorted map, where there is about a 2000-fold magnification for area within 1-2 degrees of visual angle.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 20

Compression system

High resolution in important areas, with low resolution elsewhere. ֠ If the brain were required to deal with all information from the complete visual field with the same degree of acuity as the fovea, it would weight 10 tons.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 22

Simple cells

◮ Cells in LGN don’t response to orientation ◮ Cells in V1 do - if bar covers ON center, than minimal

inhibition, if it crosses inhibitory area enough, then the inhibition cancels out firing.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 23 Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 24

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 25

Complex and hyper complex cells

Complex cells

Larger receptive fields, without clear division between excitatory and inhibitory areas.

Hyper complex cells

Responsive to orientation, as well as length.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 26

The miracle of vision!

The visual system transforms a fragmented 2D image into a coherent 3D construct.

◮ Physiological organization should feed into rules for

constructing percepts

◮ We construct visual perception of everything we see,

according to basic constraints or principles.

◮ Illustrate this best with subjective contours and surfaces,

in which we create the perception of a surface which is not present in the image.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 27

Rules

Rule of generic view

Construct only those visual worlds for which the image is stable (i.e., generic) view.

Rule 1

Always interpret a straight line in an image as a straight line in 3D.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 28

Rules

Rule of generic view

Construct only those visual worlds for which the image is stable (i.e., generic) view.

Rule 2

If the tips of two lines coincide in an image, then always interpret them as coinciding in 3D.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 29

Rules

Rule of generic view

Construct only those visual worlds for which the image is stable (i.e., generic) view.

Rule 2

If the tips of two lines coincide in an image, then always interpret them as coinciding in 3D.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 30

Kanizsa triangle

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 31

Non-accidental features (Rule 12)

If two visual structures have a non-accidental relation, group them and assign them to a common origin.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 32

Non-accidental features

Non-accidental features (Rule 12)

If two visual structures have a non-accidental relation, group them and assign them to a common origin.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 33

Convex curves

Convex cusps (Rule 11)

Construct subjective figures that occlude only if there are convex cusps.

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 34

Blivet figure, aka the devil’s fork

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 35

Blivet figure, aka the devil’s fork

Jesse A. Harris: LCS 11: Cognitive Science, Vision 2 36