Influences on Face Recognition

O’Toole, Deffenbacher, Valentin, and Abdi (1994) reviewed several factors that influence face recognition. Typicality, one of these factors, refers to how similar a face is to a prototypical face. A prototypical face is a blend of typical or homogeneous features that are “normal.” A prototypical face is a conceptual tool used to describe similarity of features. Closely related faces of an individual’s concept of the prototypical face blend with the features of that prototype, whereas unusual features or non-prototypical faces would tend to “stand out” in memory.

Faces are retained in a manner that enhances the prototypical features of the face. Thus an exemplar is encoded for recognition. Accuracy in recognition occurs where the facial detail differ from the exemplar. Faces that are considered unusual are considered more familiar than typical faces. For example, glasses, facial hair, a facial scar, or tattoo would create a specific memory that could be easily recalled. Lewis and Johnson (1997) found that distinctiveness of a face predicted miss errors, failure to identify a face as seen previously, whereas familiarity of a face predicted false positives, identifying a face as seen previously that was new. As little as the addition or removal of glasses, or the growth of or shaving of a beard can drastically reduce accuracy of face recognition.

Facial attractiveness and memorability are related (Vokey & Read, 1992; O’Toole, et. al). Moreover, faces judged as likeable or neutral are more easily recognized than faces judged unlikable.

Attractiveness is related to typicality or homogeneity of faces. Faces that are viewed as attractive are those closely related to the prototype. A negative correlation exists between facial attractiveness and memorability. Less attractive faces are more easily remembered because they deviate from typical and are more distinct than attractive faces. They are more distinctive. Attractiveness is positively correlated with typicality. Facial distinctiveness may be affected both locally (e.g., scars) and globally (e.g., unusual facial proportions) with regards to the attractiveness of the face.

Additionally, different races find different aspects of the face attractive. Cultural groups have different criteria for judging attractiveness. Social/cosmetic cues such as plucked eyebrows or hairstyles affected sex recognition for Japanese versus Caucasian subjects. Japanese women tend to pluck their eyebrows more heavily than Caucasian women, which is an identifiable cue as to race and gender. O’Toole, et. al also found that males and females use different cues for identification of sex of another face. A same sex bias exists for females but not for males (Vokey & Read, 1992).
Females rate female faces as more typical than male faces. However, the rated familiarity of faces was not related to recognition judgments. This suggests that familiarity judgment out of context is not the source of the effects of typicality on face recognition.

A question in face recognition is whether individual facial features have a role in recognition. Experimentation with cardinal features (e.g., hair, shape, age) has shown that individual features, while important for recognition, are not individually weighted for re-cognition. It appears that multiple features and combinations of features are considered when making a judgment of recognition.

Subjects show a strong preference for eyes and eyebrows, followed closely by the hairline above the temples, the mouth and upper-lid area, and the lateral hairline beside the temples. Although preference was shown for these areas, subsequent recognition testing argued strongly against feature lists as a means of recognition. By changing the spatial location of the eyes, it has been found that subjects’ recognition of the face was impaired, which further emphasizes the holistic face representation model.

The upper-face features of hair and eyes were recognized better by subjects than the lower-face features of nose, chin, and mouth. Using composite pictures (pictures where one or several features have been changed) changes in the upper area of the face affected recognition significantly more than changes in the lower area. Participants in this study were worse at recognizing composite faces or old faces, and judged them as new.

Expressions also have an effect on recognition (Oda, 1997). Recognition of happy expression took less time than recognition of angry expressions of the same face. Confident recognition of facial expressions took 100-250 milliseconds. Face recognition was significantly better than expressional recognition. However, subjects were better at recognition of happy faces than neutral faces. Expression and mood of facial expression appear to be independent of identity. However, both expression and mood can affect delays in recognition.

Stereotypes and preconceived ideas can affect recognition ability. When subjects believed that male targets were defectors in a prisoner’s dilemma game, defectors were more likely to be recognized than cooperators a week later. However, both cooperators and defectors that happened to be female were equally likely to be recognized a week later. Gender evaluations and stereotyping were also found to be a factor in recognition. They found an interaction between recognition of faces and gender stereotyped words presented to subjects. Gender stereotyped words were more easily recognized after seeing the presentation of the same gender face.

Context of the seen face has been shown to impair memory of that face (O’Toole, et. al). People tend to associate a face with other items in the environment, such as hairstyle, clothing, location, or even time of day. When viewed in another context, such as another location, the memory of the face is impaired. Memory is impaired because the viewed face tends to be “chunked” with its surroundings. The face is encoded with the other information in its environment, such that hair, location, or clothing becomes part of the memory of that face. When chunked, the memorability of the face depends on if that face is viewed in the same or similar context.

There is evidence that clothing may have a role in identification of suspects that look similar to a perpetrator. They found that false identifications of a look-alike suspect were greater when the suspect wore the same clothing as the perpetrator. Dissimilar suspects were correctly rejected in target absent lineups, even when dressed in the same clothing as the perpetrator.

You May Also Like