Object and Facial Recognition

Question 1

What are the Gestalt Laws?

Answer 1

Proximity, similarity, closure, good continuation, common fate

Question 2

What is the Law of Pragnanz?

Answer 2

You should see the simplest interpretation of the scene

Question 3

What are the 3 components of Marr’s level of analysis?

Answer 3

Why (the problem)
What (strategy)
How (network of neurons)

Question 4

What are the 4 steps of recognition?

Answer 4

1- Input image
2- Primal sketch
3- 2 1/2 D sketch
4- 3D model rep

Question 5

What are the 2 theories of object recognition?

Answer 5

1-Image-based models. Specific views are ‘stored’ and recognition performance is based on these. Works on efficiency.

2- Structural Description Model- Info about structures of an object is extracted from a single view. Basic units called geons represent objects

Question 6

What are Geons?

Answer 6

A 3D shape. Variations in a small number of basic parameters called Non-Accidental Properties NAPs

Question 7

Which theory of object recognition is supported by Filipe et al?

Answer 7

Supports image-based models. Found better accuracy at learnt than interpreted viewpoints. Also found 3D viewing condition was easier to perform (For structural description mode)

Question 8

What is the most recognisable view?

Answer 8

Cannonical

Question 9

What is the frequency hypothesis and the maximal information hypothesis?

Answer 9

Proposes cannonical viewpoint is recognised quicker as it is the view we mostly see in our lives.

Maximal information- the viewpoint contains the most informative information about the object

Question 10

What did Johnson and Morton find about facial recognition?

Answer 10

New-born babies will preferentially view faces from day one.

Question 11

What is:

a) the Featural Hypothesis
b) Configurational Hypothesis
c) Holistic Hypothesis

Answer 11

Featural- faces are primarily remembered due to their features

Configurational- emphasises relationship amongst facial features

Holistic- takes face as a whole.

Question 12

What did Yin find about faces?

Answer 12

We are better at recognising upright faces than objects, but worse at inverted faces

Question 13

What does the Thatcher illusion illustrate?

Answer 13

Shows we perceive faces in a global configuration and faces are processed holistically.

Question 14

What does the Functional Face Recognition Model suggest?

Answer 14

Faces are structurally encoded and activated FRUs.

If encoding and FRUs match, then semantic information can be accessed

Question 15

During an fRMI, what 2 areas of the brain were more active when

a) viewing faces
b) viewing objects

Answer 15

faces- FFA

objects- Lateral occipital complex

Question 16

What is the inversion effect?

Answer 16

Healthy PPs are better at recognising upright faces but prospagnosics are better are recognising inverted faces

Question 17

What did Qulroga et al do/find?

Answer 17

Single-cell recording. PPs shown 100s of photos. Neurons in Medical Temporal Lobe responded only when seeing photos of Jennifer Aniston alone, not with Brad Pitt

Question 18

What did Gauthier and Tarr find?

Answer 18

It was possible to train PPs to be greeable experts in hours. When experitised, brain activity shifted from object to face area- suggesting face rec may only reflect expertise

Question 19

What are the 5 properties of NAPs?

Answer 19

Curvilinearity
Parallelism
Contermination
Symmetry
Collinearity