Chapter 3 - Object Recognition

Question 1

What is the goal of vision?

Answer 1

To identify an object within a scene (separates object from background)

Question 2

How do neurons begin to represent objects?

Answer 2

• a neuron responsible to detect Jennifer Aniston (example) receives multiple inputs from neurons responsible for detecting individual features of the face
• reduces input signals until you can’t reduce it any more (up to a photon of light stimulating a single photoreceptor)
• a neuron that detects a whole receives projections from neurons that detect some of its parts and makes inferences about the rest)

Question 3

How is Gestalt psychology related to object representation?

Answer 3

The perceptual whole is greater than the sum of its parts

Question 4

What are some factors that influence object recognition?

Answer 4

1. object familiarity
   - we can recognize objects as being objects regardless of if we know what that object is used for
2. image clutter
   - the visual system has to identify objects among cluttered scenes (objects are often covered by other objects)
3. object variety
   - exemplars: variety within a category
   - states: identical objects in various states/views
4. memory for objects
   - we cannot store all possible examples/states/objects

Question 5

Describe perceptual organization

Answer 5

1. represent edges - abrupt changes in brightness and colour
2. represent uniform regions bounded by edges
3. divide these regions into figure and ground - assign border ownership
4. group together regions that have similar properties
5. fill in missing edges and surfaces

Question 6

How does object recognition work?

Answer 6

It uses higher level processing to represent objects fully enough to recognize them
- then it matches current object representation to representation in memory

Question 7

How does the visual system disentangle what is figure (surface) from what is ground (background)?

Answer 7

through edge extraction, border ownership and perceptual grouping

Question 8

What is edge extraction?

Answer 8

the process by which the visual system determines the locations, orientation and curvature of edges in the retinal image

Question 9

How are edges identified?

Answer 9

by sudden changes in illumination

Question 10

Why are other systems important besides edge extraction when recognizing an object?

Answer 10

Shadows can create edges

Question 11

What needs to happen to clearly identify borders?

Answer 11

They must be assigned to the appropriate part of the image (borders of objects vs. borders of the ground)

Question 12

What are the principles of figure ground organization?

Answer 12

1. depth
2. surroundedness
3. symmetry
4. convexity
5. meaningfulness

Question 13

What is the law of Pragnaz (law of simplicity)?

Answer 13

Every stimulus pattern is seen in a way that renders its structure in the simplest way possible

Question 14

What are the Gestalt grouping heuristics?

Answer 14

1. proximity
2. similarity
3. common motion
4. symmetry and parallelism
5. good continuation

Question 15

What happens to the retinal image when an object is occluded?

Answer 15

the retinal image abruptly ends

Question 16

What are the two mechanisms involved in perceptual interpolation?

Answer 16

1. edge completion: filling in the occluded edges within a scene
2. surface completion: filling in particularly hidden surfaces

Question 17

What is object permanence?

Answer 17

our ability to understand that objects can be occluded, but continue to exist

Question 18

What does the neural mechanism of object representation seem to suggest?

Answer 18

the principles of object recognition occur with the neurons in V2

Question 19

Describe the neurons in V2

Answer 19

• slightly larger receptive fields than neurons in V1
• receives inputs from multiple V1 cells (tend to all have the same orientation preference)
• tend to be interested more in global pattern less so in underlying components
• have a location preference

Question 20

What do specialized networks between V1 and V2 allow neurons to do?

Answer 20

Know where an object is located

Question 21

What do neurons in V2 respond to?

Answer 21

Perceptual grouping

- can also base their responses on info outside of their receptive field

Question 22

In what way is organized perception not perfect?

Answer 22

Some of the guiding principles we use do not reflect the truth of what we are seeing. ex:

• bistable images (which change unpredictably how we perceive them)
• accidental viewpoints (like a bird’s eye view)