Bruce Et Al: Relative Importance Of External And Internal Features Od Facial Composite

Question 1

Method

Answer 1

3 lab experiments

Question 2

Participants

Answer 2

1- 30 staff and students from Stirling uni

2- 48 undergraduates at Stirling uni

3- 8 staff and student
(4 psych)
(4 computer science)

Question 3

Procedure

Experiment 1

Answer 3

Stimuli were 10 target photographs of 10 celebrities (clean shaven,spectacles avoided)

50 composite images produced by
Efit, ProFit, sketch and EvoFit

3 sets of composit- ‘complete’ ‘internal features ‘ ‘external features’

Tested individually in one of 3 conditions

Asked to place each composite in front of celebrity face , not timed

Question 4

Experiment 2

Answer 4

Used photo array with distracter faces ( makes task more difficult )

Faces and foils made very different (easy) or similar (hard)

Composites presented one at a time along with photo array and participant had to pick out from array the celebrity face that matched composite

Composite was either of external or internal features

Question 5

Experiments 3

Answer 5

Participants had to make 32 composites of 4 computer science and 4 psych staff members

Repeated experiment 1 with 54 volunteers (Glasgow science museum) and 26 psych and computer science from Stirling uni

(Summary: participants made composites of familiar and unfamiliar faces and experimenters investigated if there was a difference in accuracy )

Question 6

Results

Answer 6

1- whole composites and external composites sorted similarly (35% correct) composites of internal features (19.5%)

2- external features identified more easily (42%) internal (24%)
Consistent accross array type (hard or easy)

3- overall accuracy 57.7% 
Familiar faces 50% 
Unfamiliar faces 48.7%
External 53.3%
Internal 32.6%
Complete 61.1%

Question 7

Conclusions

Answer 7

This could indicate that there’s something about the internal features about the face which do not work well when trying to create a reconstruction

This effect obvious when familiar face shown which sheds doubt on previous research

Familiarity makes less difference in accuracy than whether internal or external features are shown

Question 8

Aim

Answer 8

To investigate reactive recognisability