Face recognition in applied context

Question 1

Why are faces important? (4)

Answer 1

We use faces to help us to establish age, gender, and mood
Identification
Attractiveness
Emotional Expression

Question 2

How is face recognition separate from object recognition? (3)

Answer 2

Faces are dynamic
Different neurological pathways and processes
Prosopagnosia

Question 3

Identification of familiar faces is generally very good under which conditions? (4)

Answer 3

Lighting changes
Disguises
Viewpoints
Expressions

Question 4

What are the problems in unfamiliar face recognition? (5)

Answer 4

Viewpoint – much of the research has focused on frontal views of faces but faces are 3-D and complex

Profile is particularly bad especially when generalising from one profile to another

Viewpoint changes have a particularly large effect on unfamiliar faces

Profiles obscure much of the configural information that seems to be important

Inverting a face also makes recognition difficult

Question 5

Why is 3/4 view best recognised? (2)

Answer 5

Partly because this lies between frontal and profile so any change (learning – test) is relatively small

Question 6

When is profile view particularly bad? (1)

Answer 6

Profile is particularly bad especially when generalising from one profile to another

Question 7

How do negative and lighting effects impact face recognition? (4)

Answer 7

Negatives are very hard to recognise

Representations must encode more information than this from the original image – hence the difficulty

It is the loss of shading that causes difficulty in recognising negatives

Lighting an image from below has a similar effect to negation and also disrupts identification

Question 8

What information can negatives provide? (2)

Answer 8

Position and size of facial feature

Question 9

What is configuration? (1)

Answer 9

The relationship between different components of a face

Question 10

Why does inverting face cause problems with facial recognition? (3)

Answer 10

Makes viewers less sensitive to configures information compared to upright faces
- causes loss of configural information
- Thatcher illusion (Thompson, 1980)
- inversion effect is greatest with faces compared to houses or other objects

Question 11

What is Bartlett and Searcy (1993) do and what did they find? (2)

Answer 11

Made unreasonable configural adjustments
However, ‘grotesqueness’ ratings were much lower when the faces were inverted

Question 12

How are distinctive faces recognised? (3)

Answer 12

Distinctive faces are more likely to be remembered and more quickly recognised if distinctive

Question 13

How are caricatures recognised? (3)

Answer 13

Caricatures are recognised more quickly than ‘true drawings’

Caricaturing improved recognition in typically poor settings

Caricatured photographs may be rated as a better likeness of the face than the actual photograph

Question 14

What influences recognition? (7)

Answer 14

Distinctiveness / Familiarity

Disguises – covering the hair, sunglasses because these features are important in recognition

Carragher & Hancock (2020) reported the surgical masks have a detrimental impacton matching faces

Wells and Olson (2003) noted that light levels should influence recognition andhence EWT accuracy

Length of exposure to culprit although ‘weapon focus effect’

Also if abstract judgements are made then recall of the face is better

Changes over time – facial hair / ageing

Question 15

Are some witnesses better than others? (5)

Answer 15

Gender – small effect, but overall very little difference
Age
Intelligence
Race
Little evidence that personality is a factor though anxious make fewer mistaken ID’s

Question 16

How does gender of witness impact face recognition? (3)

Answer 16

Small effect, but overall very little difference (Shapiro & Penrod, 1986)

Some advantage for females

Haj et al. (2019) reported that single males showed an advantage for recognition of female faces compared to in-relationship males or females (any status)

Question 17

How does age of witness impact face recognition? (2)

Answer 17

Young and old make errors – mainly when the culprit is absent

Brackmann et al. (2019) reported more bystander errors in adolescents than adults and children

Question 18

How does intelligence of witness impact face recognition? (2)

Answer 18

Shakeshaft & Plomin (2015); Gignac et al. (2016) found a correlation, but others found no relationship

Question 19

How does race of witness impact face recognition? (3)

Answer 19

Better memory for faces from the same race - own race bias.

Recent research has shown that this is due to early processing as the effect also exists in visual working memory

Stelter & Degner (2018); Burgund (2021) found evidence of different eye fixations for different race faces

Question 20

What are the problems with traditional measures of retrieval? (2)

Answer 20

Mug shots can cause interference

Photo / in-person line-ups can be biased

Question 21

How are traditional measures of retrieval improved? (5)

Answer 21

Double blind procedure
Witness told that procedure is double blind
Suspect should not stand out
Statement of confidence taken prior to any feedback
Told that culprit may or may not be present

Question 22

What are the problems with using CCTV footage for facial recognition? (3)

Answer 22

They may be:
Very small
Poor quality
Taken from strange angles
Badly lit

Question 23

Describe Henderson, Bruce and Burton, 2001 study (1)

Answer 23

Participants were required to match CCTV images to a face from a choice of 8 ‘mug-shots’

Question 24

What did Henderson, Bruce and Burton, 2001 study find (7)

Answer 24

Pooling first and second choices gives a mean accuracy (for both ‘robbers’) of 28.5%

Better than chance but some faces were chosen more frequently than the actual robber

Matching faces from poor quality CCTV is clearly difficult

Matching photos to photos improves to 76% for Robber 2 (33% for Robber 1)

Still high error rates and other faces are still frequently selected

Matching of unfamiliar faces is difficult even with quality images

Mileva & Burton (2019) confirm that CCTV images remain poor at times

Question 25

What did Megreya, Sandford & Burton (2013) find? (1)

Answer 25

Found increased error rates when images from different days were matched, suggesting underestimates of difficulty

Question 26

What did Wilkinson and Evans (2009) find about facial imagery expert? (3)

Answer 26

Wilkinson and Evans (2009) compared face imagery experts and the public on matching CCTV images to photographs - double the accuracy rate and half the error rates

Suggests that training can be effective and worthwhile

Question 27

What studies were conducted on length of exposure? (3)

Answer 27

Memon, Hope and Bull (2003) examined the effect of the length of exposure to culprits

Memon and Bartlett (2002) reported that older adults (60-80) were more prone to false identification errors

Memon et al. (2003) used 2 (exposure lengths) x 2 (line up type) x 2 (age group) design

Question 28

What did studies of length of exposure find? (5)

Answer 28

Longer exposure did yield higher accuracy

Also higher accuracy when target present than absent

No effect of age

Longer exposure tends to lead to higher confidence ratings

However – remember weapon effect and type of judgements made

Question 29

What did Yuille and Cutshall (1986) do and what did they find? (2)

Answer 29

Assessed accuracy by attempting to reconstruct an event via various sources of information

Found high accuracy and reliability

Question 30

What did Woolnough and MacLeod (2001) do and what did they find? (4)

Answer 30

Compared statements with actual CCTV footage of crimes – thus a true ‘objective assessment’ of eyewitness accuracy

Good recall for violent crimes – attracts and directs attention – possibly only to activity rather than perpetrator

Both victims and bystanders had accurate memory for events

96% accuracy rates – are these memories flashbulb like?

Question 31

What did Tollestrup et al. (1994) find? (1)

Answer 31

Fraud victims act like lab participants with rapid forgetting, little information, poor identification of culprits etc