pa 4

Question 1

How is the face broken down and remembered?

What evidence is there for this

Answer 1

Faces are not broken down into their constituent parts. Remember faces as an overall pattern.

photofit system in court.
Not a good way of forming representations fo the face (choosing eyes, mouth etc from a book)
- people have difficulty reproducing likeness of even familiar faces

Question 2

Faces are represented configurally, meaning

Answer 2

faces are processed holistically.

spatial relationships between features are as important as the features themselves.

the face features interact eith one another

Question 3

Features vs configurations. Larry experiment

Answer 3

Tanaka and Farah.

Learned Larry’s face either as normal or scrambled.

Then asked, which is Larry’s nose, which is Larry from options.

FOUND
when learnt normal face, better at recognising in learning context

When learned as scrambled face, you’re better at recognising the feature alone

Concluded that the representation of whole faces is based on a holistic representation

Question 4

When are single features of a face processed

Question 5

what does the thatcher illusion suggest

Answer 5

we process faces a whole when upright, but as individual features when upside down

Question 6

Importance of surface properties for faces

3 pieces of evidence

Answer 6

pigmentation.

e.g. laser scans remove pigmentation, and found that recognition decreases.

e.g. only edge info reduced recognition. only shadow and edges raised

e.g. Negation seems to detriment due to reversing pigmentation/shading levels

Question 7

Distinctive vs typical face in familiarity vs Categorisation tasks

Answer 7

distinctive faces were processed quicker in The familiarity task, BUT typical faces were quicker in the Categorisation task

Question 8

Valentines face space

Answer 8

almost like a normal distribution of face configurations. Where distinctive faces are placed in a space that is not very populated and so don’t get confused with others and are easier to identify

Question 9

Bruce and Young model of face recognition

4 stages of recognising familiar faces

Answer 9

1. Structural encoding (info on contrast, pigmentation etc)
2. Face recognition units (one for each person u know)
3. Person identity nodes (info ab person, where u know them)
4. Name generation

Question 10

Bruce & Young model

Features of face recogntion of identity

Answer 10

• Expression anlysis. happy or sad.
• facial speech analysis. oo or ee
• Directed visual processing (other info like old or young)
• Cognitive system

Going on at same time but independantly of facial recognition system

Question 11

Independence as a problem for Bruce and Youngs model

Answer 11

maybe the systems have some overlap. e.g.
it’s easier to make expression decisions for familiar faces than unfamiliar faces

Question 12

Covert recognition as a problem for Bruce and Youngs model

Answer 12

prosopagnosic patients with no OVERT recognition of famous faces, but showed COVERT faces.

suggests either some cog “leakage” from systems
OR
two face processing routes, conscious and unconscious

e.g. after biking accident, can’t recognise faces, better at matching faces together when familliar than unfamilliar.
Better at trying to learn name of familiar person compared to wrong name

Question 13

Computer model of face recognition

Answer 13

similar to Bruce young model.

Pools of units - feature units / FRUs (face recogniton units) / NRUs (name recognition units)/ PINs (person identity nodes) / SIU s (semantic information units) connected by bidirectional excitatory links.

Face link, Name link.

SIU info comes from either face or name, and then spreads back and activates the model for all semantic info like priming (e.g. activates all royals)