Topic 5b: Spatial vision

Question 1

What type of cells in V1 detect orientated edges and bars?

Answer 1

Simple cells
Complex cells
Hypercomplex cells

Question 2

How do simple cells process?

Answer 2

• Have receptive fields
• Structured on and off regions
• Orientation selective
• Some position selective
• Some size selective
• Length summation (strong response to long bar / little response to short bar)

Question 3

What do simple cells in V1 respond to?

Answer 3

Edges at particular locations and orientations within the visual field

Question 4

How do complex cells process?

Answer 4

• NO separate on and off regions
• Orientation selective
• Some size selective
• Not position selective
• Length summation
• High resting output = no stimulus = more activity

Question 5

How do hypercomplex cells process?

Answer 5

• Like complex cells.
• But, sensitive to line length and orientation.
• Cell turned off by stimulus outside its ‘classical’ receptive field.
• Hyper complex cells have a peak sensitivity for line length AND line orientation.

Question 6

How might inhibition of hypercomplex cells work?

Answer 6

Inhibitory input from receptive fields at either end of the excitatory receptive field could signal length (end-stop signal).
Making a hyper-complex cell from complex cells to provide length sensitivity.

Question 7

What is texture segmentation theory?

Answer 7

• Textures are defined by statistical properties rather than absolute values
• If two textures share the same statistical properties (as far as a human is concerned) they are the same.
• If two textures have different statistical properties (as far as a human is concerned) they are different and will be segmentable / separable.
• This enables us to see edges of objects define by changes in their surface markings or makeup.

Question 8

What is Julesz’ Texton theory?

Answer 8

Texton (texture atom) the simplest defining properties of textures (based on properties of V1 cells).
Textures will segment if they differ in terms of the density of their textons.
Textons are:
Elongated blobs (lines, rectangles, ellipses) with orientation and length.
Line endings.
Line crossings.
Pattern elements will form segmentable textures (i.e., look different) if they have different numbers of textons.
Broken L’s and broken T’s each have 6 textons (2 lines and 4 line endings).
Crosses have seven textons (2 lines, 4 line endings and a line crossing).

Question 9

Why is Julesz texton theory wrong?

Answer 9

Segmentation based on line crossings and line endings is easily disrupted.

Question 10

What is Nothdurft’s orientation contrast model theory?

Answer 10

Texture segmentation based on line crossings and line endings is more affected by certain visual manipulations, than is segmentation by orientation.
For example, random variations in size, position and luminance (grey value) disrupts segmentation based on crossings and endings (but not segmentation based on orientation).
So orientation differences and size or compactness (i.e., local contrast) may be the only real ‘textons’.

Question 11

What does Northdruft’s theory on texture segmentation mean?

Answer 11

Texture segmentation based on line crossings and line endings is more affected by certain visual manipulations, than is segmentation by orientation.
For example, random variations in size, position and luminance (grey value) disrupts segmentation based on crossings and endings (but not segmentation based on orientation).
So orientation differences and size or compactness (i.e., local contrast) may be the only real ‘textons’.

Question 12

Texture segmentation by orientation contrast model means what?

Answer 12

Segmentation by orientation contrast
The two regions have the same mean orientation
They segment on the basis of local orientation contrast or differences.
It is also possible to segment objects based on local size differences

Question 13

How do centre surround cells work for orientation? Single and double opponent cells

Answer 13

Single Opponent Cell
(Orientation Sensitive)- Responds to the local contrast of a line of a certain orientation difference

Double Opponent Cell
(Orientation Insensitive)- Responds to the local contrast of a line of any orientation difference

Question 14

What causes the simultaneous tilt illusion?

Answer 14

Centre-surround for orientation: Centre-surround cells for orientation in V1 act a bit like the centre-surround cells in the retina, but are sensitive to orientation NOT luminance.
Tilt-illusion works like simultaneous contrast illusion, but with orientation replacing luminance.
Tilt-illusion is an example of the neural code being changed by inhibition over space.

Question 15

What causes Titchener circles / Ebbinghaus illusion?

Answer 15

Inhibition over space simultaneous contrast

Question 16

What causes the tilt after effect?

Answer 16

inhibition over time in area V1