A Facial Composite - A visual likeness of the perpetrator of a crime that is constructed based on an eyewitness' description. These are then used by the police to catch the offender.
Composite System - A method used to construct the facial composite
How are the composite systems used?
A witness gives a verbal description of the perpetrator, facial features that match the description are selected by someone trained to use the composite system and then the composite is adjusted until the witness agrees that an appropriate likeness has been produced.
- Contains 855 features, printed on jigsaw-like pieces that slot into a template
- Features can be enhanced; e.g. by ageing with wrinkles
- When placed with distractor faces, Photo-FIT composites were only identified 12.5% of the time
- Feature demarcation lines seemed to interfere with identification
- Insufficient features were available
- Mac-A-Mug Pro is a computer programme that attempts to deal with some of the problems encountered on the Photo-FIT system; it contains more features and has eliminated feature boundary lines
- This is a computerised version of Photo-FIT
- The composites are more realistic because photographs of features are used
- Davies et al. - Participants exposed to target faces for one minute then constructed composites using either E-FIT or Photo-FIT. No significant differences were found between the two systems
- In the Davies study, artistic elaboration was not allowed. Such enhancement is used in police work, which means that the findings from the Davies study cannot be generalised outside the laboratory situation
- Complete faces are generated using a statistical technique known as principal components analysis
- 6 faces are selected that most resemble the suspect are chosen and become the 'parents'
- The components of the seperate faces are then mixed together and the 'offspring' are selected. These offspring are then bred together until a good facial likeness is created
- The EvoFIT system has been produced to fit in with the holistic model of face recognition
- The EvoFIT system allows a composite to evolve from complete faces (i.e. the composite is not constructed from individual features)
Frowd et al.
Aim: To compare and evaluate the use of five composite systems
Method: Composites from unfamiliar targets were constructed from memory after a four-hour delay, using experienced operators for each of the composite systems: E-FIT, PROfit, Sketch, Photo-FIT and EvoFIT.
Results: The average naming rate was 10%. PROfit performance was superior except on the more distinctive targets, when E-FIT was superior. Photo-FIT and EvoFIT were the least effective techniques.
Conclusion: The computerised composite systems (PROfit and E-FIT) appear superior.
Tanaka and Farah
Aim: To investigate the ability to recognise single facial features in isolation
Method: Participants were trained until they could name a series of composite faces correctly. After initial training, participants were presented with pairs of faces: one, the original 'Larry', and one that differed from the original only by a single feature like the nose. In a further condition, participants were required to look just at the two noses and identify which one belonged to Larry.
Results: In the whole-face condition, where the two faces only differed by one feature, participants could still tell the two faces apart and could identify Larrys original nose. In the isolated features condition, participants had great difficulty identifying Larrys nose.
Conclusion: Isolated facial features are very difficult to recognise, which supports the holistic theory of face recognition and perhaps explains why composite systems, that require witnesses to select from features in isolation, are of limited use.