Printer Friendly

Normal and impaired face perception: implications for the optometrist.

Course code: C-34648 | Deadline: February 28, 2014

Learning objectives

To understand the difficulties experienced by patients with an impairment of face processing (Group 2.2.1)

To understand the neurological implications which relate to impairments of face perception (Group 6.1.14)

Learning objectives

To understand the difficulties experienced by patients with an impairment of face processing (Group 2.2.1)


Faces are among the most important objects that humans visualise and they can contain a wealth of information. A brief glimpse of a face is sufficient to allow an accurate judgement about someone's age, gender or ethnic background and whether that face is familiar or not. Humans can easily discriminate between hundreds or even thousands of different faces. Faces also play an important role in communication; humans broadcast cues about their feelings and emotions through facial expressions and, for example, we can infer the focus of someone's attention just from the direction of their eye gaze.

That we are so good at discriminating between faces is in many ways remarkable. After all, faces are based upon the same basic configuration; two eyes, above a nose, above a mouth and are, hence, broadly similar. Nevertheless, the visual system is very sensitive to subtle differences in the position and shape of facial features. As a result, humans are extremely accurate at recognising faces that are familiar to them. From an evolutionary perspective, this ability is critical for any species to interact with others of that species and form social bonds. Our ability to recognise faces allows for changes in lighting, viewing angle, facial expression and cosmetic details (for example make-up and facial hair) which all significantly transform the retinal image. (1) Consider the case of a woman returning home from the hairdresser with an expensive new hairstyle which her partner fails to notice. It could be argued, perhaps, that this omission is not (entirely) due to a lack of attention on his part but rather the result of a very sophisticated visual system that has set its priority at recognising a face as familiar rather than noticing other changes.

Humans are so good at recognising faces that, despite years of research, computer scientists have been unable to produce an algorithm to match the ability of the human visual system. However, this ability is limited to familiar faces --when we look at unfamiliar faces, our ability to differentiate between them is considerably more prone to error. (2) The consequences of this can be serious; it has been estimated that mistaken eye-witness identification is responsible for approximately 75% of wrongful convictions in the USA. (3)


Early experiences

Faces are important right from the start; newborn infants have an innate preference to fixate on a face rather than other visual patterns. (4) There is evidence that new-born babies learn to recognise their mother's face soon after birth (5) and show an early preference for attractive compared to unattractive faces. (6) Despite this innate face processing ability, it takes years to become expert at recognising faces. Recent estimates suggest that the face processing system may continue to mature over the first 30 years of life. (7) Normal face-viewing experience appears to be a requirement for the normal development of face perception. Adults deprived of visual stimulation in early life by congenital cataracts demonstrate marked face recognition deficits. (8) On the other hand, similar levels of visual deprivation do not impair recognition of non-face objects such as houses or even animal faces. (9) These findings suggest that human face recognition is particularly reliant on visual experience.

As visual experience is so important for recognising faces, the processing system becomes tailored by experience with particular face categories. This is best demonstrated by the own-race bias (sometimes referred to as the cross-race effect). People are substantially worse at discriminating between faces of an unfamiliar race than faces that belong to their own race. (10) This own-race bias is not fixed and can be recalibrated by a change in experience. For example, Korean children adopted into French families when they were less than 10 years old were shown to perform more accurately on a face recognition test with Caucasian faces than with Korean faces. (11)


Illusions and effects

Given our remarkable sensitivity to faces, it has long been speculated that the visual system may process faces differently to other objects. Studies with upside-down faces provide support for this assumption. Although recognition accuracy for most objects decreases slightly if they are viewed inverted, face recognition accuracy is disproportionately reduced. (12) The classic explanation for this effect is that inversion impairs the extraction of the fine differences in feature spacing and position which give a person's face its unique identity. This is perhaps best illustrated by the Thatcher illusion, first described by Peter Thompson in 1980 (see Figure I). (13) The two upside-down faces in Figure 1 appear similar. However, if you rotate the page so that you view them in the typical, upright orientation, it is clear that they are markedly different. The face on the right appears grotesque because the mouth and eyes have been inverted. This was first (and famously) demonstrated with a picture of Margaret Thatcher. The Thatcher illusion demonstrates that the visual system is curiously insensitive to differences in facial features when faces are viewed upside-down.

Verbal descriptions of faces are often based on prominent individual facial features, for example, large ears for Prince Charles, or full lips for Angelina Jolie. This is paradoxical because it is well established that face recognition relies heavily upon the pooling of facial features across the whole face and combining these into a single, holistic face, rather than the recognition of isolated features. This can be illustrated by the composite face effect as shown in Figure 2.

The two composite faces in the figure look rather different from each other. The top halves of both composite faces, however, are identical (check this by covering the bottom face halves) but combining identical top halves with different bottom halves creates the impression that the faces are completely different. (14) The visual system appears to automatically merge the top and bottom halves together, creating the perception of a new face identity. Note that the two composite faces (top row) bear very little resemblance to any of the three original faces (bottom row) which were used to create the composites. As a result, the presence of the identical upper part in these images goes unnoticed and requires careful inspection to verify.

The mental processes responsible for combining features into a single face identity can be disrupted by introducing a horizontal offset between the top and bottom parts, as shown in Figure 3. As a result, the visual system no longer treats the two halves as a single unit and the similarity of the top halves becomes apparent while the difference between the bottom halves remains.

A compelling demonstration of the involuntary and automatic nature of the processing that combines facial features into contextual units is provided by the part-whole effect. (15) In this classic experiment, observers are asked to recognise a particular facial feature (for example the nose) taken from a face with which they had been familiarised (see Figure 4, page 52). When the target nose is shown alongside noses from other individuals, the task is difficult (Figure 4-top). When the same noses are put inside an otherwise identical head, the task is much easier (Figure 4 -bottom). Simplified faces, akin to police e-fits, have been used here to increase the clarity of the part-whole effect.

The part-whole effect is the opposite of what one would expect. Typically, isolated targets are easier to identify than when extra information is added to a scene (see Figure 5). This does not apply to faces; adding extra face parts makes the task easier. The part-whole effect provides strong evidence that facial features are not recognised individually but are combined into a unified representation. Neither the composite nor the part-whole effect is found with non-face objects, for example, cars, dogs or houses. (15) These results suggest that the visual system treats faces as a special class of object and processes them in a unique, holistic way.

Impairments of face perception

Given that recognising familiar faces is an important social skill, it is unsurprising that patients who are unable to recognise faces report anxiety in social situations, with feelings of isolation and depression. (16) Generally, face recognition ability is diminished by reduced visual acuity and contrast sensitivity. (17) For example, patients with age-related macular degeneration (AMD) perform face recognition tasks less accurately than people of a similar age with healthy eyes, (18) with one study reporting that patients with AMD needed to be eight times closer to a face image than age-matched controls to correctly recognise the face. (19) Recently, it has been demonstrated that similar impairments of face perception are also found in patients with advanced glaucoma. (20) These results suggest that face recognition impairment may be an important, but perhaps overlooked, difficulty experienced by patients with ocular disease. Patients with AMD regularly highlight problems with face recognition as a determinant of their quality of life. (21)

Impairments of face perception are also an established characteristic of autistic spectrum disorders (ASD). Compared to typically-developing children, those with ASD spend significantly less time looking at faces when they are embedded in a complex visual scene. (22) When they do look at faces, patients with ASD often avoid looking at the eye region of a face and focus on the mouth. (23) This suggests that the face-viewing experience of patients with ASD during development is markedly different to that of typically-developing individuals. While patients with ASD have difficulty recognising familiar faces, differentiating between unfamiliar faces and interpreting facial expressions, (24) their ability to recognise non-face objects (for example cars, houses) is unimpaired. (25) This suggests that face perception represents a particular problem for patients with ASD. Impairments of face perception have also been associated with a range of other prevalent conditions including Alzheimer's disease, Parkinson's disease and schizophrenia. (26,27,28)


The best-studied and most dramatic impairment of face perception is the rare neurological condition prosopagnosia (sometimes referred to as face blindness). The term prosopagnosia was first coined by Bodamer in 1947 to describe a specific impairment of familiar face recognition in three patients. (29) The loss of face recognition ability in prosopagnosia can be profound; some patients become unable to recognise their spouse or even their own reflection in a mirror. (30) Prosopagnosia is debilitating. Patients often report that they are unable to follow film plots because they cannot recognise central characters. In extreme cases, the condition can necessitate a change in the patient's career and can lead to the development of social anxiety disorders or depression. Patients with prosopagnosia frequently have normal language, memory and intellectual functioning. Moreover, the patient's low-level vision (visual acuity and contrast sensitivity) remains largely intact. Thus, despite being able to view faces clearly, patients with the condition often complain that all faces seem indistinguishable. Indeed, one patient described faces as appearing:

"Strangely flat, white, with emphatic dark eyes, as if made from a flat surface, like white, oval plates, all alike." (31)


The aetiology of prosopagnosia involves damage to the specific brain region linked to facial processing, the fusiform gyrus in the temporal lobes. One of Bodamer's original patients was a young male who developed acquired prosopagnosia after sustaining a bullet wound to the head. The patient had no history of difficulty with face recognition prior to suffering brain damage.

The most common cause of acquired prosopagnosia, however, is stroke. Less frequent causes include carbon monoxide poisoning, head trauma and surgical treatment for intractable epilepsy. (32)

Depending on the type and extent of brain damage, the resulting impairment can be unique to faces. Patients with acquired prosopagnosia may perform very poorly on face recognition tests yet be able to recognise other complex objects including fruit, birds, furniture, cars and even animal faces, as accurately as neurologically-normal people. (33) For example, one patient with acquired prosopagnosia was reported to be able to identify individual members of his flock of sheep, despite being unable to recognise familiar faces. (34) Furthermore, although patients with the condition are unable to recognise faces, some can make accurate judgments about the gender, attractiveness and age of individual faces. Interestingly, patients with prosopagnosia do not exhibit any of the hallmarks of face perception, demonstrated by the three effects described earlier (the face inversion, composite or part-whole effects). (33) This is in line with the view that the condition represents a specific impairment of the face recognition system.

More recently, a different form of prosopagnosia has been described.

With manifestations similar to the acquired form, developmental prosopagnosia is a lifelong impairment of face recognition, where patients have no structural brain abnormalities (as confirmed by MRI and CT scans) and no impairment of intelligence or memory. Low-level visual function (visual acuity, colour vision and visual fields), as in the acquired form, is typically normal. The precise aetiology of developmental prosopagnosia has not been determined; although patients with the condition are often related to other people who experience difficulty with face perception suggesting a genetic basis. One report identified significant impairments of face perception within 10 members of the same family. (35) Although cases of acquired prosopagnosia are relatively rare, it does seem that the developmental form of the condition may be considerably more common.

In children, the symptoms of developmental prosopagnosia can lead to an erroneous diagnosis of ASD or the child simply being labelled as antisocial and unfriendly. Adults with prosopagnosia often report feeling guilty and embarrassed about their inability to recognise people from previous social interactions. For many patients, the diagnosis and an explanation about the nature of the condition provides reassurance. When asked about their experiences of living with developmental prosopagnosia before and after diagnosis, one patient commented: "I didn't want to let anybody know how weak I was--whereas now I don't put it down to weakness, I put it down to just the way I'm wired, and it's not a problem in terms of telling people." (16)

There is, as yet, no formal treatment for the condition. Although preliminary evidence suggests that intensive training may improve the face recognition ability of patients with developmental prosopagnosia. (37) Others have suggested that improvements can result from inhalation of oxytocin, a hormone that has a key role in social bonding. (38)

It has been estimated that the prevalence of the developmental form is 2-3%.36 The condition may be under-diagnosed because patients may be unaware that their face- perception ability was abnormal, with deficits camouflaged by coping strategies, such as relying on distinctive voices, gait, clothing, spectacles or hairstyles to recognise individuals. These strategies, however, are considerably less robust than face recognition; patients with prosopagnosia experience difficulty if an individual changes their appearance (for example a new hairstyle, switching from spectacles to contact lenses). When meeting new people, such as visiting an optometry practice, patients with prosopagnosia find name badges particularly helpful. Patients who experience difficulties with face recognition also prefer to interact with only small numbers of new people. It may be helpful, then, to ensure that, on visiting an optometry practice, patients with known prosopagnosia are consistently assisted by the same members of staff.



Faces engage mechanisms which are distinct from those used to recognise other objects. The normal visual system is adept at recognising minute differences between individual faces, which are important for normal social functioning. Impairments of face perception can result from both ocular disease and neurological disorders. Optometrists are, therefore, likely to encounter patients who present with difficulties in face recognition. It may not, however, be immediately obvious during a typical optometric consultation that problems with face perception exist. In the case of ocular disease, the optometrist needs to be aware of perceptual problems with faces as a consequence of the impaired retinal and/or optic nerve functioning. In the case of neurological disorders, neither the ocular health nor the visual acuity may point towards face perception deficits. Where face deficits are identified, the optometrist should keep in mind the considerable impact these can have on a patient's quality of life and provide appropriate advice and support.


The authors would like to thank Frank Munro for generously allowing us to use his face photograph. The authors would also like to thank Lesley Miller and Eilidh Martin for comments on an earlier version of this article.

Andrew J. Logan BSc (Hons), MCOptom, Dr Gael E. Gordon PhD, Dip. Optom and Dr Gunter Loffler PhD


References Visit, click on the article title and then on 'references' to download.

Exam questions Under the new enhanced CET rules of the GOC, MCQs for this exam appear online at Please complete online by midnight on February 28, 2014. You will be unable to submit exams after this date. Answers will be published on and CET points will be uploaded to the GOC every two weeks. You will then need to log into your CET portfolio by clicking on 'MyGOC' on the GOC website ( to confirm your points.

Reflective learning Having completed this CET exam, consider whether you feel more confident in your clinical skills--how will you change the way you practice? How will you use this information to improve your work for patient benefit?

Andrew Logan is an optometrist studying for a PhD in vision science at Glasgow Caledonian University (GCU) under the supervision of the co-authors, Dr Gael E. Gordon, optometrist and lecturer in vision science at GCU and Dr Gunter Loffler, optometrist and reader in vision science at GCU. The group have published widely on the subject of face perception and presented their work at both national and international conferences.
COPYRIGHT 2014 Ten Alps Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:1 CET POINT
Author:Logan, Andrew J.; Gordon, Gael E.; Loffler, Gunter
Publication:Optometry Today
Geographic Code:4EUUK
Date:Jan 31, 2014
Previous Article:Colour vision deficiency part 3--occupational standards.
Next Article:Diary dates.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters