Lecture 6 Face Recognition Flashcards

1
Q

What is the inverted face phenomenon? Describe a study by Yin (1969) that proved this ability to be unique in faces.

A

The inverted face phenomenon is the idea that faces presented upside down are hard to recognise. Yin (1969) argued that all things presented upside are hard to recognise. To test this he presented faces, houses, planes, men in motion to ps. He then performed a memory recognition task where pairs of the stimuli were presented either upright or inverted. One of the stimuli was seen before the other has not. Ps had to indicate which one they recognised. Found that faces were disproportionately harder to recognise invertedly then the other objects. Suggested that this was evidence for the unique processing of
Faces.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is the Thatcher Illusion?

A

The Thatcher illusion is the idea that spatial relations crucial to perception of upright faces (configurable information) are not apparent when presented upside down.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is configuration and feature information when it comes to faces? What does the Thatcher illusion suggest faces are recognised by?

A

Configuration information is information about the arrangement of features on the face. Feature information is the information about individual features on the face. Because only configuration information is distorted when faces are presented invertedly then it has been suggested that faces are detected using configuration information

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is prosopagnosia? What did McNeil and Warrington (1993) find about patient WJ and his sheep?

A

Prosopagnosia is the inability to recognise familiar faces. McNeil and Warrington (1993) looked at the sheep farmer WJ who had prosopagnosia to see if he can distinguish between human faces as well as his sheep, which he was well accustomed too. They found that WJ could not tell about human faces but could tell apart his sheep. Suggested that this again was evidence that faces are uniquely processed and it did not have to do with familiar stimuli.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is the super expertise argument for face recognition not being unique,

A

The super expertise argument suggests that we process all stimuli of which we are experts in using the same mechanisms. Therefore since we are well accustomed to seeing faces each day we are experts in distinguishing them.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What did Diamond and Carey (1986) do to see if the super expertise account for facial recognition could be supported?

A

Diamond and Carey (1986) presented dog expert and novices with pictures of dogs and faces that were either inverted or not. They used the same procedure as Yin. Found that dog experts suffered in their performance of viewing dogs upside down considerably more than novices. Results suggested that the inversion effect is not unique to faces but to other stimuli where expertise is apparent. So perhaps facial recognition is not unique

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What did Valentine and Bruce (1986) find when they took this expertise idea and applied it to same and different race face photos? What did Bruce argue could be the reason for these results?

A

Valentine and Bruce suggested that because humans are better at discriminating faces within their own race as opposed to other races that the face inversion effect would be more apparent for same race faces if the inversion effect is due to familiarity and expertise. However they found that the opposite was true, the inversion effect was larger for different race faces. Bruce suggested that this is because both stimuli are faces and we do not treat different race faces as seperate objects. Therefore processed similarly.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Describe the Greeble study by Gauither et al (1997) and what it was trying to determine.

A

Gautier et al (1997) wanted to know if this inversion effect was due to expertise/familiarity or if it was due to faces being unique. To do this they trained ps to be experiments in identifying greebles and then presented them in an inverted and upright memory recognition task. Found that the Greeble experts experienced an inversion effect. Providing support for the support expertise account.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What did Gautier et al (1997) find in how Greeble experts and novices processed the greebles? How does this map onto face processing?

A

Found that experts used more configural information when identifying the greeble however the novices looked more at the feature information. As previously found, faces are processed using configural information

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What did Bauer and Verfaellie (1988) discover that may suggest that prosopagnosics still have an implicit knowledge of faces?

A

When ps were asked to match faces to names they could not. However when names and faces did match their electrodermal responses increased significantly.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What two testable predictions did Bruce and Youngs Model of facial recognition (1986) produce?

A

That familiar faces are processed separately from unfamiliar faces.
And that people should be unable to access the name of someone without knowing other information about the person.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What did Malone et al (1982) find that supports that familiar faces are processed differently from unfamiliar faces?

A

Malone and colleagues found a patient who could match together unfamiliar faces well but could not recognise familiar faces. They also found another patient where the opposite was true.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

How did Young et al (1985) support the second testable prediction of Bruce and youngs facial recognition model that people can’t access a name without knowing other information about the person?

A

Had people keep a diary of their everyday facial recognition and the problems they experienced with this. One out of 1008 problems reported being able to recall a name but no other information about the person. However 190/1008 problems involved not being able to name a person but knew information about them.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What are some limitations of Bruce and Youngs face recognition model?

A
The cognitive system is vaguely defined. 
We don't usually know everything about a person before we can recall their name.
Despite the evidence, a seperate module for proper names is evolutionary uncompelling (ie. useless)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What are some alternative accounts for the name specific module first suggested by Bruce and young ?

A

Cohen- that names are meaningless and therefore harder to recall then important semantic information.
Burton and Bruce- concepts are more likely to remember if primed, names don’t get primed but semantic info does.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is the McGurk effect?

A

When you see a person saying ga and the audio track over top says ba so you hear the person saying da. What you hear is affected by what you see.

17
Q

How can feature and configural information processing be related to Farahs 2 process model of object recognition?

A
Feature = analysis by parts
Holistic = configuration
18
Q

Why aren’t things like PhotoKit or identikit good for eyewitness testimony?

A

They both rely on the features of the face but research has shown that faces are processed through their configurations.

19
Q

Why is Evofit a better program for eyewitness testimony? What does it involve?

A

Evofit uses configural information for facial recognition. Doesn’t involve the witness describing the face. It blurs external, irrelevant features and it uses holistic scales to adjust things like weight, age, health. It asks people to select the most familiar 6 faces in an array and then generates another 72 faces based on those 6. Ps are asked to select another 6 and so on.

20
Q

What is the current recognition success rate for Evofit ?

A

60%

21
Q

Name two ideas that support the idea that face recognition is unique.

A

The inverted face phenomenon and prosopagnosia