4. Perceiving & Recognizing Objects

Question 1

extrastriate cortex

Answer 1

• region of the cortex bordering V1
• contains multiple areas involved in visual processing (VS, V3, V4, inferotemporal cortex, etc.)

Question 2

how are extrastriate cells different?

Answer 2

• can recognize “border ownership”; for a given border, which side is part of the object and which side is the background?
• receptive fields more sophisticated

Question 3

2 pathways for processing of object information in extrastriate cortex

Answer 3

where and what

Question 4

where pathway

Answer 4

• concerned with locations of objects, not names or functions
• also important in deployment of attention
• goes to parietal lobe

Question 5

what pathway

Answer 5

• concerned with names and functions of objects, not location

Question 6

does the pathway principle always hold?

Answer 6

no, some basic object info can be represented simultaneously in both pathways

Question 7

inferotemporal (IT) cortex

Answer 7

• part of the cerebral cortex in the lower portion of the temporal love
• part of the “what” pathway
• important for object recognition
• close connections with regions in temporal lobe (ex: hippocampus)

Question 8

receptive field properties of IT neurons

Answer 8

• very large (some cover half of visual field)
• don’t respond well to spot or lines
• respond well to objects holistically instead of specific features (ex: hands, faces, objects)

Question 9

grandmother cell

Answer 9

• specialized neuron that fires exclusively in response to a specific stimulus

Question 10

object recognition speed

Answer 10

150ms

Question 11

feed-forward process

Answer 11

a process that carries out a comutation (ex: object recognition) one neural step after another, without the need for feedback from a later stage to an earlier stage

Question 12

reverse-hierarchy theory

Answer 12

• proposition that feed-forward processes give initial, crude information about the world by activating high-level parts of visual cortex
• more detailed information becomes available when activation flows back down the hierarchy (from cortex) to lower visual areas
• probably because you need memory to understand the meaning and details

Question 13

3 levels to conceptualize how the visual system works

Answer 13

1. low-level vision: extraction of basic features from the image (color, light, etc)
2. mid-level vision: perception of shapes and surfaces; piecing together the individual features from low-level
   3- high-level vision: object recognition and scene understanding

Question 14

what does mid-level vision involve?

Answer 14

analysis and synthesis

Question 15

analysis

Answer 15

using reflected and transmitted light to infer about the shape and composition of objects

Question 16

synthesis

Answer 16

ability to put local bits of information together into recognizable objects

Question 17

illusory contour

Answer 17

a contour that is perceived even though there are actually no physical edges present

Question 18

gestalt psychology

Answer 18

the whole is greater than the sum of its parts

Question 19

gestalt grouping rules

Answer 19

• a set of rules that describe when elements in an image will appear to group together
• visual system uses these rules to try to make sense of the vast, ambiguous and noisy inputs from the early stage of visual processing

Question 20

rules of evidence

Answer 20

Gestalt grouping principles
1- good continuation
2- pragnanz
3- similarity
4- proximity
5- common fate
Not Gestalt
6- texture segmentation
7- parallelism
8- symmetry
9- common region
10- connectedness

Question 21

good continuation

Answer 21

two elements will tend to group together if they seem to lie on the same contour

Question 22

pragnanz

Answer 22

• every stimulus pattern is seen in such a way that the resulting structure is as simple as possible
• aka principle of good figure or principle of simplicity

Question 23

similarity

Answer 23

similar looking items tend to group

Question 24

proximity

Answer 24

items that are near each other tend to group

Question 25

common fate

Answer 25

things that are moving in the same direction tend to be grouped together

Question 26

texture segmentation

Answer 26

- carving an image into regions of common texture properties - depends on the statistics of textures in one region vs another - 2 of the strongest principles of texture segmentation: similarity and proximity - not a gestalt grouping principle

Question 27

parallelism

Answer 27

parallel contours are likely to belong to the same group

Question 28

symmetry

Answer 28

symmetrical regions are more likely to be seen as a group

Question 29

common region

Answer 29

items will group if they appear to be part of the same larger section

Question 30

connectedness

Answer 30

items will tend to group if they are connected

Question 31

2 of the strongest principles of texture segmentation

Answer 31

similarity and proximity

Question 32

commitees

Answer 32

"committees" must integrate conflicting opinions and reach a consensus when there are perceptual ambiguities - many different and sometimes competing principles are involved in perception

Question 33

ambiguous figure

Answer 33

a visual stimulus that gives rise to 2 or more interpretations of its identity of structure - perceptual committees tend to obey the laws of physics

Question 34

accidental viewpoint

Answer 34

a viewing position that produces some regularity in the visual image that is not present in the world - perceptual committees assume viewpoints are not accidental - but sometimes the visual system can be fooled by an accidental viewpoint

Question 35

figure-ground assignment

Answer 35

the process of determining that some regions of an image belong to the foregroup object (figure) and other regions are part of the background (ground)

Question 36

principles at work in the assignment of regions to figure or ground:

Answer 36

- surroundedness: if one region is entirely surrounded by another = figure - size: smallest region = figure - symmetry: symmetrical region = figure - parallelism: regions with parallel contours = figure - relative motion: if one region moves in front of the other, the closer region = figure

Question 37

relatability

Answer 37

- degree to which two line segments appear to be part of the same contour - in the real world objects are often hidden by other objects (occlusion)

Question 38

nonaccidental features

Answer 38

- a feature of an object that is not dependent on the exact (or accidental) viewing position of the observer - T, Y and arrow junctions indicate specific input

Question 39

global superiority effect

Answer 39

the properties of the whole object take precedence over the properties of parts of the object

Question 40

5 principles of middle vision

Answer 40

1. bring together that which should be brought together (ex: Gestalt grouping rules and relatability) 2. Split asunder that which should be split asunder (ex: edge finding processes, figure-ground mechanisms, texture segmentation) 3. Use what you know (ex: T, Y, arrow junctions) 4. Avoid accidents (ex: accidental viewpoints) 5. Seek consensus and avoid ambiguity (ex: visual syst. needs to resolve any ambiguity and deliver a single solution to the perceptual problem at hand)

Question 41

what does object recognition involve?

Answer 41

- moving from V1 to IT in the "what" pathway, neurons respond to more and more complex stimuli - By area V4, cells are interested in stimuli such as fans, spirals, and pinwheels - it's difficult to know exactly what V4 neurons like, but it's something more complicated than spots or bars of light