Research on memory for faces

This webpage makes available some research on memory for faces which I conducted in the 1970s, mostly at the Department of Psychology, University College London. It tackles research issues which were current at the time and may still have some relevance today. Since that time, face recognition has attracted a huge amount of research effort. Some idea of this can be gained from visiting the Face Recognition website.

Richard Phillips


Why are faces hard to recognize in photographic negative?

R J Phillips, Perception & Psychophysics, 1972, 12(5), 425-426.

Faces may be difficult to recognize in photographic negative simply because they contain a large range of grays, while printed words and geometric shapes, which contain no grays, are easy to recognize in negative. This explanation was partly tested in an experiment where Ss had to recognize positive and negative pictures of well-known people, both using normal monochrome photographs and using lith photographs in which all areas of gray were removed. Lith photographs were harder to recognize than normal photographs, but the difference between positive and negative was the same for lith pictures as for normal ones. This does not rule out an explanation in terms of grays, but it does put a major constraint on it.

Full paper as pdf file (483KB)


Recognition, recall and imagery of faces

Phillips, R.J. (1978) Recognition, recall and imagery of faces. In: Gruneberg, M.M., Sykes, R.N. and Morris, P.E. (Eds.), Practical aspects of memory pp.270-7, London: Academic Press.

Two exploratory studies are reported on the relationship between recognition, recall and imagery of faces. The first investigated undergraduates' memory of a single well known face. The second was a correlational study testing faces in general. Both studies suggest that subjects' reported imagery ratings for faces reflect performance on face recognition tasks more strongly than performance on recall tasks. It is argued that our difficulty in recalling faces arises principally from problems of recoding rather than retrieval.

Full paper as pdf file (4650KB)


Some exploratory experiments on memory for photographs of faces

Richard J. Phillips, Acta Psychologica 43 (1979) 39-56.

Yin (1970) has revived Bodamer's (1947) contention that faces are processed separately from other visual objects. This was investigated in a series of exploratory experiments in which memory for photographs of the center of the face (eyes, nose and mouth) was compared with photographs showing only the periphery (hair, ears and chin). Because facial expression emanates from the center, it is argued that any specific mechanism may affect the center but not the periphery. Although the two types of stimuli were equally difficult to recognise, the center of the face was considerably harder when paired-associates were learned. In another experiment, whole faces were compared with geometric shapes, and the faces were superior on a recognition memory test but the shapes were superior on an associate learning test. The difficulty in learning paired-associates with faces may be linked to a difficulty in constructing good verbal descriptions of faces: compared to other visual stimuli, faces appear to have low association values. A face specific mechanism would predict this, but there are more parsimonious explanations, for example, in terms of the way the brain stores visual information.

Full paper as pdf file (4304KB)


Recognition of upright and inverted faces: a correlational study

Richard J Phillips & Richard E Rawles, Perception, 1979, volume 8, pages 577-583.

An investigation of ninety-five university admission candidates failed to replicate the finding by Yin of a negative correlation between the ability to recognise upright and inverted faces. A zero correlation was obtained when unknown faces were both learned and recognised upside down, but when well-known faces were presented normally and upside down for identification, a significant positive correlation appeared. Rock has suggested that inverted faces are difficult to recognise because they overtax a mechanism for correcting disoriented stimuli. This explanation satisfactorily accounts for the data with the proviso that, when inverted faces are to be remembered, the best strategy is not to attempt to correct their orientation, but to learn isolated features of the face. This describes the data more parsimoniously than Yin's face-specific mechanism.

Full paper as pdf file (645KB)


Undergraduates' immediate memory span for faces is three

Richard J Phillips & Christina M Zahra, Perceptual and Motor Skills, 1979, 48, 1098.

A research note based on Christina Zahra's undergraduate project. It challenged a claim in circulation at the time that we have an immediate memory span of only one face.

View as pdf file (85KB)


The uniqueness of faces

Short extract from R J Phillips, A comparison of memory for faces and other visual stimuli, University of London PhD thesis, 1977.

This discussion attempts to estimate the quantity of information stored about a familiar face.

View as pdf file (244KB)


Here is Richard Phillips's homepage.


Updated 20 November 2006.