1. Are faces special?

2. Prosopagnosia
Brain damage can produce problems in face recognition - even own reflection (Bodamer, 1947)
Prosopagnosia usually results from localized brain damage

3. Prosopagnosia
Farah (1990) reported that 94% of prosopagnosia patients had damage in the right hemisphere
Damage usually occurs in the ventral occipital or temporal lobes.
This led to the idea of a face-specific module that resides in the right ventral occipitotemporal regions of the brain

4. The Thatcher Illusion

5. The Thatcher Illusion

6. The Thatcher Illusion Features analysed independently
Each feature coded relative to gravity

7. The inversion effect Humans are most attuned to upright faces
Yin (1970) studied patients with Right hemisphere damage
Recognition test with pictures of faces and houses
Items presented upright and upside down
R hemisphere patients did as well as normal subjects in recognising inverted faces, but were worse at recognising upright faces

8. The inversion effect
Yin’s interpretation = when faces are upright they are processed by special mechanism in the right hemisphere
Faces presented upside down do not stimulate this mechanism and are treated like objects

9. Face detection cells
First face detection cells were discovered by Gross et al (1972)
Monkey temporal cortex
Cells did not respond to simple stimuli nor to other complex objects

10. Brain activity - Faces vs. objects
FFA – fusiform face area
From FMRI in awake humans

11. Brain Asymmetry Brain pathways Visual and motor pathways crossed
Sensory input more direct from contra-lateral side
Motor output stronger to contra-lateral limbs

12. Brain Asymmetry Brain pathways In other words…
The left hemisphere processes information from and controls the right side of the body
The right hemisphere processes information from and controls the left side of the body

14. Face expression Face expression
More prevalent on left side of the face (80%)
Explanation
Emotional/motor control bias to right hemisphere
right
left

15. Who is More Masculine?

16. Same faces – different halves

17. Why is Face Perception Biased?
right
left

18. Brain asymmetry in face processing
Perceptual bias
Judgments of face identity, sex, age, attractiveness are biased to left side of face
Explanation
Left ½ face projects to observer's right hemisphere
Facial identification is a right hemisphere specialization

19. Facial recognition

20. Visual Cues for Face Recognition
Features
Internal features (eyes, mouth, nose, etc)
External features (hair)

21. Visual Cues for Face Recognition
Features
Internal features (eyes, mouth, nose, etc)
External features (hair)
Configuration (features have to be in the right place)
Familiar and unfamiliar faces
External features more important for unfamiliar faces
Internal features more important for famous faces

22. Making Facial Averages
Warp + +

23. Caricature exaggerating difference from average 100%
Which looks most like Jack Nicholson? (The Joker in ‘Batman’, ‘The Shining’)
Accurate drawing
Caricature exaggerating difference from average 100%

24. Hugh Grant?

25. Recognising Familiar Faces
4 stages:
Form description of seen face
Match to stored representations of known faces
Access semantic information
Recall person’s name

26. Face Recognition Image of Face Description of face
Stored Descriptions of faces (Face Recognition Units)
Other routes to identity (e.g. Voice, name)
Person Identities
Names

27. Neuropsychological Evidence
Prosopagnosic patients: fail to recognise familiar people from their faces (PH, de Haan et al., 1987)
Semantic impairments: faces seem familiar but unable to recall semantic info or name (ME, de Haan et al., 1991)
Anomia: Faces familiar, semantic info accessed but unable to recall name (EST, Flude et al., 1989)

28. Experimental Evidence Latencies of Decisions to Faces
Young, McWeeny, Ellis & Hay (1986a):
Familiarity decisions (774 ms) made faster than semantic decisions (929 ms).
Young et al., (1986b):
Semantic decisions (934 ms) made faster than naming (1290 ms)

29. Experimental Evidence Repetition Priming
Bruce & Valentine (1985)
Face familiarity decision to a face is faster if it was presented earlier
Long lasting
Within-domain: names don’t prime faces
Suggests existence of representations of faces to which names don’t have access (Face Recognition Units)

30. Face Recognition Image of Face Description of face
Stored Descriptions of faces (Face Recognition Units)
Other routes to identity (e.g. Voice, name)
Repetition Priming
Person Identities
Names

31. Experimental Evidence Interference Effects
Young et al. (1986)
Names interfere with face naming but not face classification
Bill Clinton
faces have indirect access to names (via semantics)

32. Face Recognition Image of Face Description of face
Stored Descriptions of faces (Face Recognition Units)
Other routes to identity (e.g. Voice)
Person Identities
Name Generation

33. Other Meanings
We can derive several types of socially important meanings from faces
Identity
age
gender
mood
speech
gaze & attention
                    

34. Bruce & Young (1986) EXPRESSION ANALYSIS EXPRESSION ANALYSIS
STRUCTURAL ENCODING
FACIAL
SPEECH
ANALYSIS
FACIAL
SPEECH
ANALYSIS
FACE
RECOGNITION UNITS
DIRECTED
VISUAL
PROCESSING
PERSON IDENTITY NODES
COGNITIVE
SYSTEM
NAME GENERATION

35. Expressions and Identity
Expression tasks unaffected by familiarity (e.g. Bruce, 1986; Young et al., 1986)
Double dissociation in neurological patients (e.g. Young et al., 1993) )
Prosopagnosics can often do expressions
Opposite pattern also found
Different cells in primate temporal lobe (e.g. Perrett et al., 1984, Hasselmo et al, 1989)
Different brain regions in humans (Sergent et al, 1994)

36. Summary
Sequence of stages proposed for identification stood up against evidence
B & Y (1986) model proposes independent routes for the processing of expression, facial speech and identities

37. Facial attractiveness

38. Peahens prefer peacocks with more elaborate trains
Improved growth and survival of offspring of peacocks with more elaborate trains
Petrie et al. 1991, Anim. Behav.
Petrie 1994, Nature

39. Sexual Dimorphism in Human Faces
Masculine facial traits (large jaws, prominent brows) in males are testosterone dependent
Testosterone suppresses the immune system and handicaps the individual producing it
Only high quality individuals can afford this handicap
So masculine face traits should be attractive in males

40. Sexual Dimorphism in Human Faces
Feminine facial traits in females are oestrogen dependent
Oestrogen is related to health and fertility
So feminine face traits should be attractive in females

42. Preferred in UK and Japan
Perrett et al. 1998, Nature
Feminised
Masculinised

43. Preferred in UK and Japan
Perrett et al. 1998, Nature
Feminised
Masculinised

44. Masculine Vs. Feminine Male Faces
Masculine male:
Dominant
Low warmth
Bad parent
Feminine male:
Honest
Warm
Good parent
Perrett et al. 1998, Nature

45. Non-Human Species Preferences for Symmetry
Female barn swallows prefer males with symmetrical tail feathers
Female zebra finches prefer symmetrical males
Møller, 1994, Anim Behav
Large literature showing in a wide variety of animals show preferences for symmetry.
Swaddle & Cuthill, 1994, Nature

46. Symmetry
Most features are supposed to be symmetrical: any deviation from symmetry is a reflection of imperfect development
Only high quality individuals can maintain symmetric development under environmental and genetic stress
Symmetry thus serves as a measure of quality
Large literature showing in a wide variety of animals show preferences for symmetry.

47. Symmetry
Across many species males with low asymmetry have more mating success
Symmetric human males report more sexual partners than more asymmetric men
Møller & Thornhill 1998
Thornhill & Gangestad 1994
Large literature showing in a wide variety of animals show preferences for symmetry.

48. Measuring Symmetry
Attractiveness ratings of women correlate with measured symmetry
rs= .28, p = 0.02, n=66
Penton-Voak, et al. 2001, Proc Royal Soc

49. Computer Graphics Which Face is Most Attractive?
15 face pairs –
% symmetry preference = number of symmetrical faces selected out of 15

50. Symmetry Manipulation
The lower images are made using the left/right average of the 224 feature points (only 4 are Marked here)

51. Original
Symmetric
Symmetry is found attractive in both male and female faces Perrett et al. 1999, Evo Human Behav; Little et al. 2001, Proc Royal Soc

52. Averageness
How similar a face is to a population mean influences its attractiveness
Average appearance may indicate diverse mixture of genes
With composite images, the more faces that go into an image the more attractive the face becomes
Langlois et al., 1991, Psych Science

53. Averageness
3 images
6 images
12 images

54. Averageness
3 faces faces faces
Little & Hancock, 2002,

55. Averageness Average texture Original Average shape
Little & Hancock, 2002

56. Averageness

57. General Preferences
Femininity, symmetry and averageness appear to be universally attractive traits

58. What you should know…
The evidence that face processing may be a specialised module in the brain
Bruce & Young’s model of face recognition
The biological basis of some attractive features: femininity/symmetry/averageness

59. Friday 3pm – visual deficits & illusions
Next lecture…
Friday 3pm – visual deficits & illusions

60. Reading Recognition Are faces special? Cheaters
Kanwisher (2000) Domain specificity in face perception, Nature Neurosci (see website to download)
Cheaters
Gaulin:
Place in brain- prospoagnosia
Gaulin: , Kanwisher (2000)
Attractiveness
Masculinity & averageness
Introduction of Little & Hancock (2002) Brit J of Psychology (see website)
Symmetry
Gaulin: p