L5 - Perceptual organisation, gestalt psychology and face perception

Question 1

What is low-level vision?

Answer 1

• Extracts local information about lines, bars and edges
• Our visual world consists of structured wholes and objects (meaningful)

Question 2

What is mid-level vision?

Answer 2

• Joins isolated features into larger groups
• Forming the basis for object recognition in high-level vision
• Takes info from V1 and makes them more meaningful units = hard as it is ambiguous about which pieces of info to group together

Question 3

What are the visual streams for processing?

Answer 3

• Ventral: what - object identification: the long parts of the brain
• If you lesion what stream in animals - hard to distinguish objects but can place the objects e.g which object is closer to them (spatial relationships)
• Dorsal: where: visuo-spatial information processing: very split up

Question 4

Is there separation of the visual processing streams?

Answer 4

• No strict separation
• There is a tendency for the streams to perform how they do but they have complex connectivity between them
• No strict anatomical or functional separation of what/where streams

Question 5

Describe the ‘what’ stream?

Answer 5

• Receptive field size increases from V1 to v4 (mid-level vision) = take more info from visual info and combine to create more meaningful info
• Tuning complexity becomes higher = specific stimuli they respond to becomes more complex e.g neurons that specifically respond to faces

Question 6

How does visual system overcome ambiguity?

Answer 6

Need to use prior knowledge/assumptions as constraining principles to rid ambiguity

Question 7

What are gestalt principles of perceptual organisation?

Answer 7

• Set of laws of perceptual organisation
• Relationship between elements are critical for perception
• Whole is greater/different than the sum of its parts

Question 8

What is the principle of proximity?

Answer 8

• How do you bind these circles that are identical e.g can see them in 5 different columns BUT if you have to group them by proximity, you can say they are in 5 different rows - how close are these elements
• Pieces of info that are close together in space are bound together

Question 9

What is the principle of similarity?

Answer 9

• Pieces of info that are similar to each other are bound together e.g black and white rows

Question 10

What is the principle of common fate?

Answer 10

• Things that move together are bound together
• As soon as this stops, it is almost impossible to see it

Question 11

What is the principle of good continuation?

Answer 11

• Visual system assumes that orientations have smooth contours, as most natural objects have smooth changes in orientation

Question 12

What is Figure-ground assignment?

Answer 12

• How do you know what is foreground/background and how do you separate this
• Other principles of gestalt psychology

Question 13

Gestalt principles of perceptual organisation now?

Answer 13

• Important for perception research to move forward and to consider the role of relationships
• BUT largely descriptive
• Little to no experimental evidence
• Recent work in visual neuroscience and psychophysics has established the why/how of some of these principles

Question 14

Why is face perception important?

Answer 14

• Faces are most important stimuli
• Small changes in faces can tell you things about someone’s identity, gender, age, ethnicity (stable features)
• Tells us about people’s characteristics that change very quickly e.g. facial expressions, gaze direction = guide our social interactions
• Compared to other visual categories faces are processed very efficiently

Question 15

What brain areas are dedicated to processing faces?

Answer 15

• Superior temporal sulcus
• Inferior occipital gyrus: sits on occipital lobe and dedicated to processing facial info
• Lateral fusiform gyrus

Question 16

Is face processing special?

Answer 16

• Domain specificity: mechanisms operate independently of general object perception
• Expertise: mechanism derive from general object perception but become finely tuned due to extensive experience = exposed to faces frequently

Question 17

What is holistic processing?

Answer 17

• Representing features and their relationship as one unit, we don’t use this for other objects other than faces
  1) Part-whole effect: features are easier to identify when presented as part of a face
• Have different eyes, nose and mouth and you can combine these features to construct a number of identities, and ask ppt to memorise identities, test how good they are at identifying the different features in isolation e.g just eyes, or in the context of a face = easier to process whole face than individual features, seen in other stimuli too
  2) Face inversion effect: inversion disrupts processing of fine details and relationship between features
• Show ppts inverse or normal faces, ppts find it harder to identify individuals if you present them upside down, as neurons are only triggered if you see a face in the correct orientation = process the fine details and can combine but when upside down, you do not trigger holistic processing and cannot combine
• Use general processing when faces are upside down, but facial processing when faces are right way up
• People who were experts at facial expressions also show these effects but hardly any replication of these studies

Question 18

What is prosopagnosia?

Answer 18

• Failure to identify or distinguish between faces, despite otherwise normal visual and cognitive ability
• Special case of visual agnosia
• Familiarity but no identification: develop weird delusions: Capgras syndrome - don’t feel familiar
• Highlights modularity of some processes in visual perception

Question 19

What is norm-based code?

Answer 19

• Facial features are represented as deviations from the average face (the norm)
• Makes explicit what is distinctive about a face
• Emphasises subtle variations that define individuals
• Further the face is from the norm = easier to recognise them
• If you change the norm, all faces must be different

Question 20

What are aftereffects in high-level vision? (Study)

Answer 20

• With barrack example: when you stare at the picture = you change the norm
• Neurons that code for facial features adapt to specific characteristics of adaptor face
• Perception of subsequent face is biased away from adaptor characteristics
• Code is based off the norm: e.g two populations of neurones, one responding to Bush and the other Obama, the middle point = mix of Bush and Obama(also known as norm), move in either direction of tuning = reduced sensitivity, so you think the norm is Obama if you adapt to Bush (as norm shifts to Obama)
• Adaptation is like a re-calibration of the norm
• When you adapt to bush, neurons reduce sensitivity

Question 21

What are consequences to norm-based code?

Answer 21

• When you adapt to the norm, the two populations of neurones both decrease, not changing the norm