Patients with Urbach–Wiethe disease constitute a unique nature experiment as more than half have bilaterally symmetrical damage in the amygdaloid region. Ten such patients were studied neuropsychologically and, nine of them, neuroradiologically with static (CT) and functional imaging techniques [single‐photon emission computed tomography (SPECT) and PET]. Their principal bilateral amygdala damage was confirmed. Neuropsychologically, the patients showed cognitively little deviation from normal subjects, while they differed emotionally. This was evident in their judgement of all emotions in facial expressions, in an odour–figure association test as well as in remembering negative and positive pictures. This suggests that the human amygdala influences both negative and positive emotional processing.
Keywords: emotion; long‐term memory; functional brain imaging; SPECT; PET
Received May 8, 2003. Revised and accepted June 20, 2003
A crucial engagement of the amygdala for processing affective stimuli and emotional long‐term memory is suggested from animal research as well as from functional imaging and clinical studies (Markowitsch, 1998). The amygdala is a critical structure for processing biologically relevant stimuli, in particular those that signal fear (Breitner et al., 1996; Morris, et al., 1996, 1998). Its role in processing stimuli of positive valence is less clear, with some studies (Canli et al., 1998; Lane et al., 1999; Garavan et al., 2000) negating, and others stressing such a role in addition to that for processing negatively valenced stimuli (Garavan et al., 2001; Hamann and Mao, 2002; Hamann et al., 2002). Differences in methodology and in aetiology of the amygdaloid damage may account for the divergent findings (Hamann et al., 2002). Clinical studies also provide inconsistent results. Adolphs et al. (1994, 1995) investigated a patient suffering from Urbach–Wiethe (UW) syndrome, a very rare autosomal recessive disease, which produces bilateral calcifications in the anterior medial temporal lobes, especially of the amygdalae, in 50–75% of cases (Newton et al., 1971; Staut and Naidich, 1998). They found her selectively impaired in recognizing negative emotional expressions in human faces, while her ability to recognize happy faces was totally preserved. In contrast, two herpes simplex encephalitis patients with complete, but non‐selective, bilateral amygdala lesions were unimpaired in the recognition of similarities between facial expressions (Cahill et al., 1996). The question of why only some patients with bilateral amygdala damage are impaired in recognizing affective facial expressions is still unresolved. In an early study of two UW patients, impairments in learning and recalling odour–figure associations were observed (Markowitsch et al., 1994). As amygdaloid nuclei constitute a part of the olfactory system, and as the olfactory sense has among all senses the closest affinity with emotions, an emotional evaluation of stimuli is centrally dependent on the amygdala.
Neuroimaging studies have shown that the amygdala plays a key role for emotional memory processing. PET (Cahill et al., 1996; Hamann et al., 1999) and functional MRI studies (Canli et al., 1999, 2000) have identified a correlation between amygdala activation and episodic memory for highly emotional, but not for neutral stimuli. The findings suggest that the amygdala modulates long‐term memory processing for emotionally intense stimuli. They are consistent with clinical results from three UW patients with circumscribed bilateral amygdala lesions who demonstrated impaired episodic memory for visual and verbal emotionally arousing stimuli (Babinsky et al., 1993; Markowitsch et al., 1994; Cahill et al., 1995; Adolphs et al., 1997). However, the small number of only three UW patients does not allow a reliable conclusion on the necessity of the amygdala for encoding and consolidating highly emotional material.
In the present study, we investigated a group of nine UW patients with confirmed and one with probable bilateral lesions of the amygdala. First, we tested the hypothesis that bilateral amygdaloid damage impairs the recognition of prototypical facial expressions, especially fearful faces, as well as the recognition of blends of emotions shown in a single facial expression. Secondly, we examined whether bilateral amygdaloid lesions compromise the recognition of emotionally arousing material.
Material and methods
We studied one Austrian and nine Caucasian South African UW patients (four women, six men) between the ages of 17 and 61 years (Table 1). All UW patients had a long‐standing diagnosis based on evidence for the standard symptoms (most prominently a hoarse voice and verrucous nodules). Nine healthy South Africans, matched for gender, school education, ethnic group and age, were investigated as a comparison group; they were friends or healthy family members of the patients, or they were recruited from the employees of the Academic Hospital Pretoria, where the main part of the study was performed.
None of the subjects was taking psychoactive medication and they did not report any psychiatric disease.
The study was approved by The Ethical Committee of the University of Pretoria.
For the purpose of this study, the patients and the comparison subjects were investigated with a broad battery of neuropsychological tests on three consecutive days (Table 2). Two UW patients (A.W. and E.W.) had reduced English language abilities, and therefore they were not investigated with verbal tests. To ensure that A.W. and E.W. understood the test instructions, an Afrikaans translator was employed.
Rating of blends of emotions in facial expressions
Odour association test
Tower of Hanoi
Neurological investigation of the comparison subjects
Neuroradiological investigations of the UW patients
In addition, eight South African UW patients were studied by CT and single‐photon emission computed tomography (SPECT) to detect intracranial calcifications and perfusion or metabolic changes of the amygdaloid complex. Due to occupational reasons, one patient could not take part in the neuroradiological examinations. The Austrian patient was examined by CT and PET. A medical examination of the comparison subjects was carried out to exclude neurological abnormalities.
For a neuropsychological profile, all subjects were examined with an extensive neuropsychological battery (Table 2). The Wechsler Memory Scale‐Revised (WMS‐R) (Wechsler, 1987) was given as a test of verbal and figural short‐ and long‐term memory. The Austrian UW patient completed the adapted German‐language version of the WMS‐R (Härting et al., 2000). The Auditory Verbal Learning Test (AVLT) (Spreen and Strauss, 1998) was used to assess immediate verbal memory span, new learning, susceptibility to interference and recognition memory. The corresponding German‐language test (Helmstaedter et al., 2001) was used for the Austrian patient. The Rey–Osterrieth Complex Figure Test (CFT) was included to test incidental visual memory as well as the visuospatial constructional ability (Spreen and Strauss, 1998). Furthermore, speed of attention and mental flexibility were assessed with the Trail Making Test (TMT) (Spreen and Strauss, 1998), and short‐term concentration ability was tested with the non‐verbal ‘Aufmerksamkeits‐ und Belastungstest (d2)’ test (Brickenkamp, 1995). For an assessment of executive functions, the Tower of Hanoi was conducted in the four discs version.
The Beck Depression Inventory‐Second Edition (BDI‐II) (Beck et al., 1996) and the German version of this questionnaire (Hautzinger et al., 1995) were used for depression screening. Finally, the Benton Facial Recognition Test (BFRT) (Benton et al., 1994) was performed as a standardized procedure for assessing the ability to identify and discriminate photographs of unfamiliar human faces.
To study affective processing, subjects were shown black and white photographs of facial expressions presented in a notebook. We chose 36 emotional and three neutral faces from six different actors (both female and male) (Ekman and Friesen, 1975). Each actor displayed all of the six basic emotions: fear, anger, surprise, disgust, sadness and happiness. In addition, three of them depicted a neutral face. The recognition of prototypical facial expressions was tested by presenting 36 facial expressions of the basic emotions. The 36 faces were divided into six blocks consisting of six prototypical faces of one emotion. Subjects rated the faces according to the given prototypical emotion of the block on a 5‐point scale from 0 to 5 (0 = the face does not express the given emotion at all; 5 = the face expresses the given emotion very much). For the judgements of blends of emotions shown in a single facial expression, subjects received the same 36 faces as in the previous task, as well as three neutral faces. All 39 faces were presented six times in random order. In each trial, the subjects judged each face on the same 5‐point scale with respect to one of the following adjectives: afraid, angry, surprised, disgusted, sad or happy.
The episodic memory for emotional material was tested by presenting 30 pictures (10 each of positive, negative and neutral pictures). The pictures had to be judged with regard to their valence on a rating scale from 1 (very pleasant) to 7 (very unpleasant). After a delay of 1 h, the subjects had to recognize the pictures from a pool of 90 (30 each of positive, negative and neutral pictures). Some of the pictures were selected from the International Affective Picture System (Lang et al., 1995) (Table 3A). The remainder were taken from material that had been collected for previous studies. These pictures were rated by 40 control subjects with respect to their valence (Table 3B).
Codes and valence ratings of the non‐IAPS pictures
Mean of valence rating
Mean of valence rating: –3 = very unpleasant, +3 = very pleasant; IAPS = International Affective Picture System (IAPS).
Furthermore, in an odour association test, six odours (orange, apple, lemon, cinnamon, peppermint and vanilla) had to be associated with six nonsense drawings over a minimum of three and a maximum of six learning trials. The test was finished as soon as the subject made all associations correctly. After 1 h, as well as 1 and 2 days later, the odours were presented again and the subject had to indicate the correct nonsense drawing from a template.
Eight South African patients were studied by CT and SPECT. The SPECT scans were performed with an SP4 HR high resolution gamma camera. The radionucleide 99mTechnetium (99mTc) was infused 30 min before the scans started. The Austrian patient was examined by CT and PET. PET scans were performed on a high resolution scanner (Siemens ECAT EXACT HR) with septa retracted in 3D acquisition mode. A dose of 370 Mbq (10 mCi) of [18F]2‐fluoro‐2‐deoxy‐d‐glucose was administered.
The data were analysed using effect sizes (d; Cohen, 1992), a statistical power analysis quantifying the size of the difference between the groups. The index is interpreted as small for d = 0.20, medium for d = 0.50, and large for d = 0.80. We tested the hypothesis whether patients with bilateral amygdaloid damage were impaired in the recognition of prototypical facial expressions and in the recognition of blends of emotions shown in a single facial expression. For these analyses, an average group rating was calculated for the six prototypical facial expressions of an emotion, and for the second condition the average group rating was computed for the six prototypical facial expressions of an emotion for each of the six trials.
To explore further the ability to remember emotionally arousing and neutral pictures, these were analysed by comparing the mean correct recognitions of negative, positive and neutral pictures with the same statistical methods as described above and with t‐tests for two independent groups. For the analysis of the correct odour–figure associations in the first three learning trials and the three recognition trials, means of correct associations were compared with the same statistical methods as used for the picture test.
Neuropsychological test performance
The results of the neuropsychological test battery are summarized in Table 4. All nine comparison subjects as well as seven of the UW patients were within the normal range or, in part, above average in memory, attention and executive functions. Isolated deviations from the norm were interpreted as transient concentration deficits if the subjects showed normal performance of the same function in a more complex test. Three UW patients had more extensive impairments in the memory domain. Patient E.W. was 1 SD below average in the visual memory index of the WMS‐R. Because of her limited English‐language ability, she was not able to handle verbal tests so that no information about her verbal memory is available. Patient M.A. was impaired in her verbal memory, and patient R.L. was clearly affected in verbal and non‐verbal memory functions and slightly impaired in her visuospatial construction ability. Neither patients nor comparison subjects demonstrated abnormalities in their ability to identify and discriminate photographs of unfamiliar human faces as tested by the BFRT.
The analyses of the recognition of prototypical emotional faces revealed that the patients tended to judge the expression of anger in prototypical angry faces and the expression of sadness in prototypical sad faces as more intensive than the comparison subjects (Table 5). On the other hand, the comparison subjects tended to rate the expression of fear, surprise and disgust in prototypical emotional faces as more intense than the patients. There was no difference between the groups in the assessment of happy faces.
Ratings of the intensity of the emotion shown in prototypical facial expressions
Prototypical facial expressions
Max = maximum; M = average; d = effect size; appraisal: 0 = the face does not express the given emotion at all; 5 = the face expresses the given emotion very much; *α = 5%.
The ratings for blends of several emotions in a single facial expression differed between groups. Patients showed a tendency to judge similarities between emotional facial expressions higher than the comparison subjects (Table 6). For example, the patients assessed fear in prototypical sad facial expressions (Table 6) as well as the expression of happiness in prototypical surprised faces as more intense than the comparison subjects. The stronger appraisal of the combination of several emotions shown in a single facial expression occurred for nearly all emotional combinations and was not unique for one emotion.
Effect size (d) and average rating (M) of emotions in facial expressions from patients/comparison subjects of judgements to a given adjective
Emotional facial expressions
d = 0.25
d = 0.58
d = 0.99
d = 0.69
d = 0.89
d = 0.33
M = 3.50/3.69
M = 1.93/1.20
M = 1.95/0.93
M = 2.52/1.65
M = 0.25/0.04
M = 1.20/0.90
d = 0.35
d = 0.56
d = 0.20
d = 0.75
d = 0.90
d = 0.19
M = 1.38/0.98
M = 4.22/3.98
M = 0.75/0.63
M = 3.05/2.00
M = 0.15/0.02
M = 1.08/0.90
d = 0.75
d = 0.35
d = 0.35
d = 0.80
d = 0.90
d = 0.52
M = 1.65/0.89
M = 0.90/0.54
M = 3.85/3.46
M = 1.45/0.83
M = 1.60/0.89
M = 0.48/0.24
d = 0.33
d = 0.61
d = 0.52
d = 0.28
d = 0.62
d = 0.18
M = 0.78/0.44
M = 2.85/1.96
M = 0.38/0.17
M = 3.88/3.54
M = 0.18/0.06
M = 0.63/0.83
d = 0.51
d = 0.41
d = 0.52
d = 0.59
d = 0.89
d = 0.06
M = 0.30/0.00
M = 0.22/0.02
M = 2.02/1.19
M = 0.22/0.02
M = 4.51/3.83
M = 0.05/0.06
d = 1.28
d = 0.54
d = 0.78
d = 0.01
d = 0.99
d = 0.40
M = 1.53/0.41
M = 1.27/0.72
M = 0.38/0.09
M = 1.30/1.20
M = 0.38/0.04
M = 3.73/3.37
d = 0.58
d = 0.76
d = 0.85
d = 0.29
d = 0.79
d = 0.67
M = 0.63/0.19
M = 0.60/0.11
M = 1.00/0.22
M = 0.57/0.37
M = 1.93/1.22
M = 1.97/1.26
M = average; appraisal: 0 = the face does not express the given emotion at all, 5 = the face expresses the given emotion very much.
The ability to recognize emotionally arousing and neutral pictures from a set of 90 pictures after a delay of 1 h differed between groups. Patients compared with comparison subjects remembered fewer pictures for each of the three categories (Table 7). Differences between groups were stronger for negative (d = 1.07) and, in particular, for positive pictures (d = 1.29) than for neutral ones (d = 0.73). t‐tests for two independent groups also resulted in significant differences for negative (P = 0.038) and especially for positive (P = 0.013), but not for neutral pictures between patients and controls (Table 7). Additionally, the patients rated negative (d = 0.35) and positive pictures (d = 0.21) somewhat more pleasant, and neutral pictures (d = 0.36) somewhat more unpleasant than comparison subjects (Table 8).
M = average; d = effect size; appraisal: 0 = very pleasant, 7 = very unpleasant.
For the odour association test, all subjects reported that they were able to differentiate the odours from each other, even though the discrimination between the odours lemon and orange was difficult for some subjects. The UW patients learned clearly fewer odour–figure associations and achieved fewer recognitions at all three delayed recalls (Fig. 1). The differences between the patients and the comparison group were substantial for the first three learning trials (d = 0.97), the first recall after 1 h (d = 0.84) and the third recall after 2 days (d = 1.06), and medium for the second recall after 1 day (d = 0.56). Tested with the t‐test for two independent groups, the number of correct associations in the first three learning trials (P = 0.05) and in the third recall (P = 0.039) were significantly higher (Fig. 1) for controls. Differences in the first recall (P = 0.087) and in the second recall (P = 0.235) remained insignificant.
Fig. 1 Number of correct odour–figure associations at three different times. *Significant differences at P < 0.05.
Neuroradiological data revealed that the dominant brain damage was a bilateral calcification of the amygdaloid complex in six patients. CT and SPECT results provided this uniform outcome and demonstrated the full‐blown degeneration in all portions of the amygdaloid complex (Fig. 2). In three further patients (J.K., A.vdW. and R.L.) the existence of calcifications could not be diagnosed by CT with certainty, but the SPECT results confirmed a bilateral decreased perfusion in the temporal lobes. The calcifications tended to expand into the uncinate and parahippocampal gyri. One patient had small bilateral calcifications in the lateral parietal cortex. The lentiform nucleus was bilaterally affected in one, and the right basal ganglia in another patient. Furthermore, one of the patients had minor calcifications in the left semioval centre, and four patients showed calcifications of the pineal gland.
Fig. 2 CT scans of eight UW patients. Note the consistent pattern of bilateral calcifications of the amygdalae in all of them.
This study provided the opportunity to investigate a comparatively large number of patients with bilateral amygdala damage (n = 10) by neuroradiological and neuropsychological methods, and to compare their performance with that of a matched comparison group (n = 9).
In principle, our analyses with structural and perfusion imaging methods confirmed previous findings that the principal brain degeneration in patients with UW disease lies within the amygdaloid region and seems to develop over time. Brain damage within the amygdaloid region is usually extensive and bilaterally rather symmetrical (cf. Fig. 2). Contrary to previous findings, our anatomical data indicate that patients with UW disease may have more widespread brain damage and may have brain damage which is not uniform across subjects. In particular, surrounding and more posterior temporal lobe regions may manifest degenerations; but also more remote structures may be affected in single patients.
Interestingly, and in line with other data in which structural and perfusion/functional imaging methods were employed, structural imaging methods lead, compared with functional methods, to a more restricted and conservative estimation of affected brain tissue (though the SPECT analyses do not have the spatial sensitivity of PET or functional MRI).
On the basis of previous studies, we had hypothesized that UW patients with bilateral amygdaloid damage would be impaired in the recognition of prototypical facial expressions, in particular fearful faces, and in the recognition of combinations of emotions shown in a single facial expression. Moreover, we expected that the patients were impaired in their memory for emotional stimuli, especially for negatively valenced pictures. Because of the direct participation of amygdaloid nuclei in the olfactory system, we hypothesized that the patients would learn and remember fewer odour–figure associations than the comparison subjects.
The results from the facial recognition test indicate differences between patients and comparison subjects for numerous emotional judgements. In contrast to our first hypothesis and to previous findings (Adolphs et al., 1994, 1995), the patients differed only slightly in their ratings of prototypical fearful faces and they perceived anger in angry faces to be more intense than the comparison subjects did. On the other hand, the patients’ judgements of the facial emotions of surprise and disgust were less intense than those of the controls. Our results indicate that, at least for UW patients, amygdaloid lesions do not necessarily produce impairments in the recognition of basic emotions such as fear and anger. It is possible that the patients (four of them had high‐school education and two of them had a university degree) generated cognitive strategies to compensate their deficits in recognizing facial prototypical expressions. Alternatively, they were able to perform that task of emotional processing by using intact structures apart from the amygdala.
The analysis of the similarity judgements indicates that patients and comparison subjects rated prototypical emotional faces adequately. Both groups showed extremely concordant rankings with respect to emotional expressions of faces to be most similar or dissimilar. For instance, when all kinds of emotional faces had to be judged with respect to whether they looked angry, angry faces received the highest ratings, followed by disgusted, then fearful, sad, surprised, neutral and, finally, happy faces. In contrast, patients and controls differed in their judgements concerning the intensity of perceived similarities between emotional facial expressions. Compared with the comparison subjects, the patients gave higher similarity ratings for nearly all facial expressions of negative and positive emotions. The clearest discrepancies between the groups were found for the ratings of happiness. The findings indicate that the patients overestimated the blends of negative and positive facial expressions.
In comparison with the result of the first task, it can be presumed that the recognition of emotional blends in faces is more difficult than the recognition of prototypical emotions. Moreover, the present results indicate that the amygdala is a critical structure not only for differentiating between emotional facial expressions related to threat and danger, but also for other negative and even for happy faces. The interpretation is in agreement with neuroimaging studies that found an activated amygdala during the presentation of negative and positive emotional visual stimuli (Garavan et al., 2001; Hamann et al., 2002; Yang et al., 2002).
Memorizing emotionally arousing material was highly impaired, as measured by the picture recognition test. The patients remembered emotional and neutral pictures much more poorly than the controls. Negative and even more so positive pictures showed the most pronounced differences. On the other hand, valence judgements of negative, positive and neutral pictures were rather similar in comparison subjects and UW patients.
Our results suggest that the amygdala is a bottleneck structure for emotional memory processing, but that it is not crucially activated in processing stimuli of differing emotional valence. The results are consistent with earlier studies, pointing to arousal but not valence of emotional stimuli as the critical component for encoding and remembering (Bradley et al., 1992; Cahill and McGaugh, 1998; Phelps et al., 1998).
The finding of a limited number of correct odour–figure associations supports the idea that the amygdala is strongly involved in emotional memory processes. The results are in agreement with a recent imaging study that found a bilateral increased regional cerebral blood flow induced by emotionally valenced but not neutral olfactory stimuli (Royet et al., 2000).
In summary, our findings suggest that a normal recognition of prototypical emotional faces can occur independently of the amygdala, while the recognition of similarities between facial expressions (depicting different, including positive, emotions) depends on intact amygdalae. Additionally, the amygdala appears to be necessary for moderating the encoding not only of arousing negative, but also of arousing positive visual stimuli. The results support the theory that the amygdala is involved in processing biologically relevant stimuli, independently of their valence (Yang et al., 2002).
With respect to patients with UW disease, our results from a larger sample of such patients strongly indicate that even patients with considerable bilateral damage of the amygdala can perform fairly normally in most everyday situations and may have more subtle memory impairments than initial studies of patients with this aetiology had suggested (Markowitsch et al., 1994; Adolphs et al., 1997). Furthermore, it seems that the most likely slowly progressing degenerative process in the brains of patients with UW disease seems to initiate compensatory strategies which make these patients less impaired than those with amygdala damage due to other aetiologies (Adolphs et al., 2000; Boucsein et al., 2001; Fine et al., 2001). Recently, Papps et al. (2003) similarly demonstrated that a patient who had had repeated stereotactic operations targeted at the left amygdala had preserved long‐term memory for emotional material.
We wish to thank all patients and their relatives, Rena Steenkamp for great help during the stay of H.J.M. and M.S. in Pretoria, Dr Friedrich Wörmann from the MRT Unit of Mara Hospital of the von Bodelschwinghsche Clinics in Bielefeld‐Bethel (Germany) for his careful additional blind reviewing of the CTs, and Drs Coetzee, Lampbrecht, Greeff and Vennote of the Eugene Marais Hospital. This study was supported by the German Research Council (DFG) through grant Ma 795/31.
Royet J‐P, Zald D, Versace R, Costes N, Lavenne F, Koenig O, et al. Emotional responses to pleasant and unpleasant olfactory, visual, and auditory stimuli: a positron emission tomography study. J Neurosci2000; 20: 7752–59.