Face recognition is an area of psychological research concerned with pattern recognition, which investigates how we process and recognise faces. Recognising faces is very important for human social functioning. It can help us to form relationships, recognise a friend in a crowd and provide some non-verbal cues about what a person is thinking and feeling.
One of the most active current research areas in pattern recognition concerns the perception faces. All faces have eyes, a nose and a mouth placed in the same relative locations and yet we are able to identify thousands of different faces. Furthermore, we are able to recognise a particular face, such as the face of a best friend, regardless of the expression – angry, sad, amused etc.
Feature analysis versus holistic processing
Some psychologists have suggested that we might recognise faces using a feature using a feature-detection approach, i.e. we might use information about individual features, such as eye colour or mouth shape, to build up a representation of the face. This bottom- up approach is the basis of identikit – a system used by some police forces to build up the face of a criminal on a feature by feature basis.
Bradshaw and Wallace (1971) used identikit to construct pairs of faces that shared a number of features, but differed in others. Participants were asked to decide as quickly as possible whether pairs of faces were the same or different. The more differences there were between the two faces, the faster the judgments were made. They concluded that participants were processing facial features independently and in serial fashion. Using sequential comparison of features, participants would encounter differences between the two faces sooner if there were several features that differed. However, the data can be interpreted differently.
Sergent (1984) pointed out that the faces that differed in several features were also very different in terms of their overall configuration, and that it may have been this factor which allowed participants to make speedy judgements. Sergent conducted her own study based on Identikit pictures and found that the faces were being processed in a holistic form, rather than as a set of independent features. Holistic processing refers to recognition based on overall shape and structure, rather than on individual elements.
In another study, Young et.al (1987) supported the idea that we recognise faces by processing information about the overall configuration of the face, rather than by analysing individual features. For example, we analyse the spatial relationships between individual features and the way the shape of one feature might interact with another (e.g. between the nose and mouth). They combined photos of the top half of one celebrity face with the bottom half of another. When the two halves were closely aligned to create a single face, participants experienced great difficulty in identifying the top half. This task became much easier when the two halves were misaligned, or when the top half was presented in isolation. This seems to suggest that the close alignment in…