Brain (1995), 118, 15-24 Face processing impairments after amygdalotomy Andrew W. Young,1 John P. Aggleton,5 Deborah J. Hellawell,6 Michael Johnson,2 Paul Broks3 and J. Richard Hanley4 lMRC Applied Psychology Unit, Cambridge, the 2Department of Neurology, St James's University Hospital, Leeds, the lNeuropsychology Unit, Royal Hallamshire Hospital, Sheffield, the ^Department of Psychology, University of Liverpool, the 5School of Psychology, University of Wales College of Cardiff and the ^Department of Clinical Neurosciences, Western General Hospital, Edinburgh, UK Correspondence to: Dr Andy Young, MRC Applied Psychology Unit, 15 Chaucer Road, Cambridge CB2 2EF, UK Summary We report an investigation of face processing impairments in D.R., a 51-year-old woman with a partial bilateral amygdalotomy. D.R. was able to recognize pre-operatively familiar faces, but she showed generalized problems of name retrieval and a more circumscribed deficit affecting the recognition of faces learnt post-operatively. In contrast to her poor memory for new faces, D.R. 's ability to match simultaneously presented photographs of unfamiliar faces was unimpaired. However, D.R. also experienced deficits in expression processing which compromised the recognition of emotion from people's faces she was poor both at matching and at identifying photographs of emotional facial expressions. In addition, her interpretation of eye gaze direction was defective, showing a more general problem in reading social signals from the face. The presence of impairments affecting the learning of new faces and the comprehension of gaze direction and facial expressions of emotion is consistent with the hypothesis of a role for the amygdala in learning and social behaviour. Keywords: amygdala face recognition facial expression gaze direction Introduction Selective damage to the human amygdala is unusual, and its consequences have not been definitively established (Aggleton, 1992, 1993). In most cases, there is no evidence of impairment on overall measures of intelligence, and no evidence of global memory impairment, but specific memory deficits have been noted to affect the recognition of nonverbal visual stimuli, and especially faces (Andersen, 1978 Jacobson, 1986 Tranel and Hyman, 1990 Aggleton, 1992). This evidence of an impairment of memory for faces forms one of the few common features that has emerged from studies of the effects of amygdala damage, being noted in a number of cases with a variety of aetiologies (Aggleton, 1992). Findings of face processing impairments after amygdalo- tomy are of particular interest because neurophysiological studies have identified cells which respond selectively to faces in the amygdala for primates (Rolls, 1984 Leonard et al., 1985 Nakamura et al., 1992) and humans (Seeck et ai, 1993). In addition, studies of face processing impairments �� Oxford University Press 1995 following damage to either human or primate cerebral cortex have shown that these can take dissociable forms, selectively affecting different aspects of face processing (Heywood and Cowey, 1992 Young, 1992 Young et al., 1993). A question which therefore arises concerns whether face processing impairments due to amygdalotomy will also show some degree of selectivity. To date, however, most investigators have only looked at recognition memory for faces, and there has only been one detailed case study of the consequences of amygdalotomy across a range of face processing tasks (Jacobson, 1986). Jacobson found that his patient showed poor learning of new faces, borderline abnormalities in the matching of unfamiliar faces, and impaired recognition of familiar faces, with particularly marked problems in naming them. We report here a further amygdalotomy case showing face processing impairments, D.R. The presence and nature of the face processing impairments was established with some of the same tests as Jacobson (1986), allowing direct comparisons

16 A. W. Young et al. between the two cases. In addition, we included a number of further tests of face processing and of visual recognition and visual memory, to identify more precisely D.R.'s preserved and impaired abilities. Case description D.R. is a 51-year-old right-handed woman who first suffered from epilepsy during her second pregnancy at the age of 28 years. Prior to this, she had an uneventful medical history, with no evidence of behavioural or psychiatric disorder and no family history of epilepsy. Her first two seizures were generalized tonic clonic seizures and these continued at very infrequent intervals, but 2 weeks after the onset of epilepsy she began to suffer complex partial seizures which occurred two or three times each day. Since then, D.R. has continued to have three types of seizure. There are tonic clonic seizures about once each month. There are absences several times each day, sometimes associated with some movement of the head from side to side. Thirdly, there are complex partial attacks almost every day, lasting for ~2 min and followed by a period of confusion or automatic behaviour. Over a period of 25 years, drug treatment seems to have had little effect on the number or type of attacks, although there has been some variation in frequency unrelated to treatment. Many different anti-convulsants have been tried at one time or another, but during the period when the tests to be reported here were begun, D.R. was taking Carbamazepine 700 mg daily and Phenytoin 300 mg daily, with Phenytoin serum level 11.1 mg/1 and Carbamazepine level 5.9 mg/1. Later in the testing period, D.R. started taking Lamotrigine, but this was discontinued after 6 months and the dose of Phenytoin was increased to 350 mg a day and Carbamazepine to 800 mg a day, with Carbamazepine level 7.6 mg/1 and Phenytoin 11.9 mg/1. However, during this time D.R.'s performance of the psychological tests to be reported was consistent across different testing sessions, and did not seem in any way affected by changes in medication. Failure of anti-convulsants to control D.R.'s seizures led to a series of electroencephalogram investigations in the 1970s, including three with sphenoidal leads. These established a seizure source in the left anterior temporal lobe and, since 1974, there have been records with an autonomous discharge in the right temporal lobe. Electrocorticography at the time of proposed temporal lobe surgery in 1975 showed widespread discharges involving frontal, parietal and even occipital areas on the left side without evidence of clear derivation from the seizure source located by depth electrodes in the left temporal lobe. Pre-operative CT scans, a ventriculogram, an air encephalogram and a left carotid angiogram did not identify any pathology. Between 1978 and 1981 D.R. underwent a series of stereotaxic procedures targetted initially at the left amygdala and, later, at the left and right amygdala. In all, D.R. had four cryoprobe lesions and one electrocoagulation lesion on the left side, and she had two cryoprobe lesions on the right, following one of which a CT scan showed a haematoma in the region of the right caudate nucleus. These procedures may have had some success in reducing the frequency of seizures, but at present D.R. still has six or seven attacks per day in which she becomes absent and makes fidgeting movements. In some attacks she falls, and some start with shaking of the left arm and leg. Following the seizure she is sometimes confused and disinhibited, but usually she regains normal orientation in time and place quite quickly. MRI scans were performed in 1991 and 1992. T2 axial sequence 2200/80 7 mm thick sections followed by T, coronal sequence 600/15 7 mm sections were acquired. Later, T| coronal sequence 500/15 2.2 mm thick sections were imaged, throughout the region of the basal ganglia and amygdala. Gradient echo 1.5 mm sections through the same region were also acquired. These MRIs showed an extensive lesion of the left medial amygdala, which destroyed much of the basal nuclei but largely spared the lateral nucleus. The area of damage extended throughout the rostro-caudal limits of the left amygdala, just reaching the anterior horn of the left hippocampus. Associated damage extended dorsally beyond the amygdala to involve part of the anterior commissure, lateral putamen and external capsule. In the right hemisphere, there was a small posteriorly placed lesion at the caudal limit of the amygdala, and a second small lesion in the right anterior amygdaloid area. Planimetric measurements of the MRIs (using a Placom KP-90N planimeter) indicated that the amygdalar lesions involved at least 25% of the total volume of the left amygdala and more than 4% of the right amygdala. Figure 1 shows tracings of lesions in the region of the amygdala derived from six coronal MRI sections at 2.2 mm intervals. There was also some additional extra-amygdalar sub- cortical damage in the right hemisphere, probably as a result of the bleeding noted after one surgery this involved a discrete lesion in the pallidal region at the level of the anterior commissure, extending more dorsally within the striatum at a level rostral to the anterior commissure, with possible damage to adjacent parts of the internal capsule and caudate nucleus. In addition, very small areas of cortical abnormality (high signal on T2 and low signal on Tj) were noted in the left occipito-parietal region adjacent to the falx and in the anterior right frontal lobe these were also considered to result from surgery. D.R. no longer has any sense of smell. She readily engages in conversation in a lively manner, but she can have problems in finding the exact word needed and therefore resorts to circumlocutions fairly frequently. These word finding difficulties are the only obvious problem with language, and they do not present any hindrance to communication because D.R. uses effective circumlocutions. In all other respects, her conversation is usually appropriate and grammatical. With respect to emotion, D.R. can readily give examples of occasions when she says that she has experienced feelings