Lecture 2 Flashcards
(23 cards)
1
Q
Why study face perception?
A
- Gives information about relatively stable features like identity, gender or age
- Can gain information from changing states such as emotion and intention
- Long history of people interested in faces due to peculiarities with faces
2
Q
What was a study looking at infant face perception?
A
- Newborns in the first hour after birth look at stimuli with a face-like configuration = humans directed at faces
- Used a stimulus (paddle) with a normal face, scrambled face and no stimuli
- Looking at how much newborns rotate their heads = rotate more with face stimuli
3
Q
What is adult attention like?
A
- Our attentional selection prioritises faces and social information
- Others’ gaze triggers automatic shifts of attention
- Can study this with a cueing paradigm, but instead of using arrows to influence gaze direction, we use a picture of eyes looking left/right = target will either be congruent or not and measuring RT
- Participants know that faces do not indicate where the cue is
- People are faster with congruent conditioning = gaze following is automatically triggered
4
Q
What was a study about exposure to faces?
A
- Glasses worn during day and capture visual input during walking time = 30 frames/sec
- Found 22.5% of frames contained a face = every 15 mins in an hour you see stimuli of face
- Lots of exposure to faces
5
Q
What is holistic processing?
A
- Mechanisms used to engage holistic processing are triggered when you see faces in the right side up = processing things as one unit = looking at general relationship between features
- When upside down = not using specialised face processing so harder to see the differences between the two faces
6
Q
What is the genetic contribution?
A
- Large-scale study with mono/di twins and controls
- Face recognition ability is substantially heritable (61%)
- Heritability is mostly specific to face recognition rather than shared with general object recognition or general intelligence
7
Q
What are face specific disorders?
A
- Developmental Prosopagnosia = never knowing who they are = hard to recognise people from face
- Acquired prosopagnosia (accident/trauma) = husband of 20y parked car and put on a cap and wife could not recognise her husband
- People with prosopagnosia use other coping things like hair
- Individuals have largely normal visual perception but have an inability to recognise people they have known for many years - cannot create face discriminations that are distinguishable enough to separate people
8
Q
What are the models of face perception? Bruce & Young
A
- Faces are inputs to processing system
- Starts with structural encoding = image based encoding = recognise them as 2d images
- Faces familiar are send to face recognition units = construct 3d face = link that to other info about that person = fed into cognitive system
- Early image based processing fed into direct visual processing = for unfamiliar faces = image based representation = do not make a 3d light/pose independent representation
- Processing of identity is separate from process of expressions
- Shared early processing
9
Q
What is the Haxby Model?
A
- Have core and extended system
- Inferior occipital gyri: early perception of feature
- Identity is processed by lateral fusiform gyrus - perception of unique identity
- Superior temporal sulcus: used for changeable aspects of faces - perception of eye gaze
- Extended system: Further processing in concert with neural systems
- e.g Intraparietal sulcus, auditory cortex, amygdala, insula, limbic system, anterior temporal
10
Q
What was a study looking at face-selective areas in the human brain?
A
- Using fMRI and ppts were in scanner, saw a number of different faces, rest, and then they saw objects = brain responds more to faces than objects
- In 12/15 subjects showed face selective activations in fusiform gyrus
- Showed this is a functional localiser = cannot anatomically detect where things are activated in FFG to identify which areas specifically respond to faces than objects
- Ran experiments comparing faces to scrambled features faces = area reacts more to normal than scrambled
- Compared different exemplars of same category = no difference
- Face selectivity is not due to difference in low-level properties
- Face selectivity is not due to faces being animate
- Face-selectivity is not due to attention allocation
11
Q
What are three keys areas where neurons respond better to faces?
A
- Superior temporal sulcus = STS = changeable features like expressions
- Inferior occipital gyrus = OFA = early shared processing
- Lateral fusiform gyrus = FFA = Identity
- All face specific areas, all part of Haxby model and part of the core system
- Right hemisphere
12
Q
What is hierarchical encoding of faces? (M&M)
A
- Researchers used face morphs = using two identities fused together and some features look like marilyn and others look like maggie
- Pictures start with marilyn, and are perceived as her, then features change, slowly encompassing features of maggie but perception is still marilyn, until there is a sudden change in identity when too many features are maggie = categorical perception
- Used pairs of images where one image was always the same (70% maggie) = change can occur either within category where physical is slightly different but identity is the same, or between category where both are different
- Within: change in image but within category boundary = ppts say they are not two different individuals
- Between: same change (%) but across boundary = ppts sat they are different individuals
13
Q
What was a follow up study to maggie and marilyn?
A
- Used these in the scanner to locate OFA and FFA via functional localiser
- Used fMRI-adaption to see if OFA and FFA differ in how they processes these pictures
- If you present same stimulus back to back = activity for second stimulus is lower than first stimulus activity = can work out if stimuli looks the same to the brain (if brain activation is same = brain thinks it’s different)
- Used the pairs again with within/between
- Found that amount of activation in OFA in response to second face in the within AND between category, you see more activity than the identical stimuli presented again
- In FFA, activation is non-significant in within category changes but is significant in between category changes = cares about identity
- FFA does not care about image-based changes = only cares if images are perceived as coming from two different identities, OFA cares about physical changes in the stimulus
14
Q
What was a study looking at part-based and holistic representations?
A
- Stimuli where either change features or configuration of features
- How do two different areas process features/config that are face relevant
- In both OFA and FFA = higher activation when features are present, relative to when they are absent = both care about face-specific features
- When looking at configurations regardless of features, OFA does not distinguish between these conditions but FFA does
- Identity specificity shows FFA cares about features and configuration of features
- Relationship between features and configurations is slight = slight overlap = same neurons responding to face relevant features also respond to configuration
15
Q
What was another study on part-based/holistic?
A
- Used RTMS (inject noise into certain parts of brain in order to figure out what performance does = should see a drop in performance if the area is important for processing the stimuli)
- Showed two faces back to back: one matched and one that either changed configuration (distance between eyes) or changed features (eyes/nose/mouth)
- Are they the same or not? Changed config/features
- Applied TMS when second face was shown = to point of brain where OFA sits
- If you compare the config and then TMS OFA, you see a drop = important for features not for config in 60-100ms = suggesting OFA starts processing features during this time
16
Q
What is the temporal relationship of encoding?
A
- Looked at FFA, STS, OFA localised through fMRI and EEG (for when)
- Looked at specific signatures in EEG signals - p8 measured looked at contrast between faces and other objects
- Looked at correlation between size of faces vs other objects and the size of p8 at 2 different timepoints: one early after stimulus presentation and one late
- OFA shows clear relationship with EEG signal later but not early
- FFA: early on no relationship but later there is = similar to STS
- Info processed in OFA and then fed to FFA and STS that processes further
17
Q
What is the summary of hierarchical encoding of faces?
A
- OFA representation is encoding earlier than FFA representation
- OFA encodes physical, low-level properties of faces in a part-based fashion
- FFA contributes to representation of identity - its response is less dependent on physical properties and more holistic
18
Q
What is the relationship between neural and perceptual processes?
A
- FFA represents faces holistically
- Facial identity is represented holistically
19
Q
What is the face inversion effect?
A
- When two faces are presented upright, you can identify them
- Becomes harder when they are inverted
20
Q
What are the neural substrates of face inversion effect?
A
- Localise areas: OFA, FFA, LOC
- Little difference if face is presented inverted or not in FFA
- But upright and inverted is seen much differently in OFA
- Relate behavioural face inversion effect to neuronal effect (behaviour/perception)
- OFA has no relationship between these 2 variables but FFA does
- FFA is important for perception that we use to make responses = area that contributes to creating realistic faces
21
Q
What is the composite face effect?
A
- IF you take two celebrities and cut the top off one and cut the bottom off the other and combine them
- When you ask people to focus only on the top half of the picture = they cannot recognise it even though there is no identity change
- Because we have a holistic representation to one representation = we see a new identity emerging
- Disrupted when faces are not perfectly aligned
22
Q
How was the composite face effect measured experimentally?
A
- Take top half of one face and combine with another bottom half
- Show ppt two diff faces where you can change bottom of face and ask ppt if face changed or not when they focus on the top and when aligned or not
- Ppts more likely to be correct if top part has changed if faces are aligned = composite face effect
23
Q
What are the neural substrates of composite face effect?
A
- Experimentally, you present aligned/not faces, and identical/bottom changes faces
- In FFA, dramatic difference between these two conditions but less difference in right OFA
- Most pattern tells you that holistic processing is done in FFA and OFA does not contribute to this
