Abstract

The present study attempts to identify cortical lesion sites associated with deficits in the decoding of emotional facial expressions and with changes of emotional states. A consecutive sample of 68 subjects with lesions caused by microsurgical tumor resection or by intracerebral hemorrhage were compared with 16 clinical and 15 healthy control subjects. A set of neuropsychological test measures – two experimental tasks requiring the sorting and rating of emotional facial expressions, and an adjective checklist as a measure of mood state – were administered. An analysis according to the functional properties of the areas with lesions revealed that subjects with combined lesions of heteromodal and limbic/ paralimbic cortices of the right hemisphere showed the strongest deficits in the decoding of emotional facial expressions. In contrast, subjects with lesions of limbic/paralimbic cortices alone, or subjects with lesions of unimodal or primary motor or sensory cortices displayed only minor deficits. Subjects with combined lesions of heteromodal and limbic/paralimbic cortices of the left hemisphere showed the most negative mood states. Our results suggest that heteromodal and limbic/paralimbic, i.e. transmodal cortices may be viewed as ‘bottleneck’ structures for the decoding of emotional facial expressions and for the modulation of mood states. Lesions of transmodal cortices deprive limbic structures of one of their main sources of input and are therefore likely to produce deficits in various forms of complex human emotional-cognitive behaviors and emotional states.

Introduction

The recognition of facial affect seems to be an important prerequisite for the social communication and judgement of human beings. While the contribution of the amygdala to human emotion recognition has been broadly investigated (Aggleton, 2000), studies focusing on the role of the cerebral cortex have tended to rely on the hemispheric side of lesions (Blonder et al., 1991), and only vaguely specified lesion sites within a given hemisphere.

The specification of cerebral cortical areas involved in human emotion recognition may be done on the basis of structural (i.e. architectonical) or functional features. However, the structural approach has the disadvantage that parcellation of the mammalian cerebral cortex into a great number of anatomically distinct areas hinders the understanding of complex and integrative functions that underlie human cognition and emotion. It is a fundamental tenet of modern neuroscience that integrated actions of interconnected cortical regions are crucial for complex human behavior (Geschwind, 1965). Subdividing the cerebral cortex based on functional affiliations thus may be a timely approach for the investigation of complex human cognitive and emotional behaviors.

One functional approach is that adopted by Mesulam (Mesulam, 1985, 1998). Based on studies of the phylogenetic development of the cortex (Sanides, 1970), on pathoarchitectonic ontogenetic studies (Yakovlev, 1959), and on anatomical, physiological and behavioral experiments in macaque monkeys, he divided the cerebral cortex into five major functional zones: primary sensory and motor, unimodal, heteromodal, paralimbic, and limbic cortices. Heteromodal cortices are characterized by three criteria: they receive convergent inputs from unimodal areas in more than one sensory modality, respond to stimulation in more than one modality, and lesions yield multimodal behavioral deficits.

Heteromodal and paralimbic and limbic, i.e. transmodal, cortices represent the highest synaptic levels of sensory-fugal processing, and also exert sensory-petal influences upon unimodal cortices. It is likely that human emotion processing substantially depends on transmodal cortices: they connect the primary and unimodal sensory cortices to the limbic system, and thus enable a link between the inside and the outside world so that needs can be discharged according to the opportunities and restrictions of the extrapersonal environment.

The aim of the present study was to clarify the role of functionally defined cortical areas for the decoding of emotional facial expressions and the modulation of mood state. Although it is known that heteromodal and paralimbic cortices are intensively interconnected, detailed knowledge about the role of these cortices in human emotion processing is missing. In the present study we investigated a consecutive sample of 68 subjects with focal cerebral cortical lesions using two different tasks to assess the decoding of emotional facial expressions and a questionnaire to assess the mood state. Lesions of each subject were thoroughly assessed on magnetic resonance imaging (MRI) and computerized tomography (CT) scans and classified according to the functional properties of the areas with lesions. The goals of the study were to analyze: (i) whether lesions of transmodal cortices impair emotional task performance stronger than lesions of other cortices, and (ii) whether deficits in emotional task performance are associated to specific transmodal lesion sites.

Materials and Methods

Subjects

A consecutive sample of subjects with focal cerebral cortical lesions caused by microsurgical tumor resection (TUM; n = 45) or intracerebral hemorrhage (ICH; n = 23) participated in the study. They were compared with 16 clinical control subjects (CCG) who had undergone surgery for slipped disks, as well as 15 healthy control subjects (HCG), who were recruited for the study via an advert in a local newspaper. Subjects of groups TUM, ICH and CCG were in or outpatients of the neurosurgical or neurological departments of the University of Göttingen. Exclusion criteria consisted of ages older than 75 years, or a history of psychiatric illness, or lesions covering the visual cortices. After complete description of the study to the subjects, informed consent was obtained. The demographic and clinical characteristics of subjects are summarized in Table 1.

Lesion Analysis and Lesion Groups

Before the analysis of the behavioral data, the lesions of the TUM and ICH subjects were transcribed from T1- and T2-weighted MR images to appropriate pages of the Talairach–Tournoux brain atlas (Talairach and Tournoux, 1988). For subjects who had undergone neurosurgery (all TUM subjects and 17 ICH subjects) the postsurgical lesions were assessed. The extent of the lesions was estimated (as a percentage) for cortical regions according to Brodmann areas (BA). A given Brodmann area was considered as affected when the lesion covered >5% of its extent. The total lesion volume was assessed using digitized CT scans according to a procedure described previously (Wowra et al., 1989).

Cortical lesion groups were formed according to the functional properties of the areas with lesions. According to Mesulam (Mesulam, 1998), the following five functional zones can be distinguished: (i) primary motor or sensory cortices include motor (BA 4), visual (BA 17), auditory (BA 41, 42), somatosensory (BA 3), and gustatory (BA 43) areas; (ii) unimodal cortices include visual (BA 18–21), auditory (BA 22) and somatosensory (BA 1, 2, 5) association areas, as well as the premotor cortex (BA 6); (iii) heteromodal cortices are located in the frontal (BA 8–10, 45, 46, anterior BA 11, 12) and parietal (BA 7, 39, 40) lobe, as well as in portions of the lateral temporal (BA 21, 37) and parahippocampal (BA 35, 36) cortex; (iv) there are five paralimbic regions, i.e. caudal orbitofrontal cortex (BA 11–13), insula (BA 14–16), temporal pole (BA 38), parahippocampal gyrus (BA 27, 28), and the retrosplenial-cingulate-parolfactory complex (BA 23–26, 29–33); (v) Limbic core formations include the amygdala, hippocampus, prepiriform olfactory cortex, septal area and substantia innominata.

The following lesion groups were formed: group H (subjects with lesions of heteromodal cortices; n = 11), group L (subjects with lesions of limbic/paralimbic cortices; n = 8), group H/L (subjects with lesions of heteromodal and limbic/paralimbic cortices; n = 14), group MS (subjects with lesions of unimodal or primary motor or sensory cortices; n = 3), group H/MS (subjects with lesions of heteromodal, and unimodal or primary motor or sensory cortices; n = 17), and group H/L/MS (subjects with lesions of heteromodal, limbic/paralimbic, and unimodal or primary motor or sensory cortices; n = 15). The lesion characteristics of subjects are described in Table 2. Representative lesions of the six lesion groups are illustrated in Figures 1 and 2.

Neuropsychological Testing

Neuropsychological examination included a short form of the German version of the Wechsler Adult Intelligence Scale — Revised (WAIS-R) which comprised the subtests Information, Similarities, Block Design and Picture Completion. Mnemonic functions were assessed using the Logical Memory I and II subtests and the Verbal and Visual Span subtests of the Wechsler Memory Scale — Revised (WMS-R). The Trail Making Test (TMT; parts A and B) was used as a measure of attentional performance and psychomotor speed. We also applied the Benton Facial Recognition Test (BFRT) to control for the ability to recognize unemotional faces.

Experimental Tasks

The stimulus set contained 16 black-and-white pictures of faces with prototypical emotional expressions. We included two neutral facial expressions, and each two pictures of Anger, Sadness, Disgust, Fear and Surprise. Furthermore, we included each two pictures showing happiness with high (Joy) and low intensity (Pleasure). Each pair of identical facial expressions contained one picture of a female face and one of a male face expressing the emotion in question. All expressions have been shown to be identified by normal subjects at >80% success rate (Ekman and Friesen, 1976).

Sorting Task

Subjects were explicitly instructed to sort the pictures with identical emotional expressions into groups, and not to sort according to gender or physiognomic aspects. There was no restriction as to the number of groups to be formed or the number of pictures to be sorted into a group. The correct solution of the task was to sort the eight pairs of identical emotional expressions into eight groups. Because Joy and Pleasure represent different intensities of the same emotion, sorting these two pairs in only one group was also regarded as a correct solution. For each subject, the Sum of Errors, representing the total number of errors within the individual sorting matrix, was calculated. Furthermore, we calculated for each group of subjects Proximities, expressing the percentage of subjects sorting identical emotional expressions into the same group.

Rating Task

After finishing the sorting task, subjects were asked to rate each facial expression on a seven-point Likert scale according to the dimensions of valence (1: pleasant–unpleasant: 7) and arousal (1: unaroused–aroused: 7), which have been suggested to be the most relevant attributes for the judgement of emotional facial expressions. The average rating of valence and arousal was compared for each pair of identical emotional expression, as well as for all expressions with positive valence (Joy, Pleasure), neutral or ambiguous valence (Neutral, Surprise), or negative valence (Anger, Sadness, Disgust, Fear).

Furthermore, Rating Errors on the valence and arousal dimension were calculated separately for each pair of identical emotional expressions: rating positive expressions not as pleasant (>3 on a seven-point Likert scale), or rating ambiguous expressions as pleasant (<3) or unpleasant (>5), or rating negative expressions not as unpleasant (<5) was considered as wrong. Rating highly aroused expressions (Joy, Surprise, Anger, Fear) not as aroused (<5), or rating mildly aroused expressions (Pleasure, Disgust) as highly aroused (>5) or unaroused (<3), or rating unaroused expressions (Neutral, Sadness) not as unaroused (>3) was also considered as wrong.

Mood State

An adjective checklist was used [Eigenschaftswörterliste (Janke and Debus, 1978)]. The questionnaire consists of 123 adjectives constituting five factors: Vigor, Fatigue, Extraversion, Irritability/Anger and Anxiety/Depression. These five factors can be summed up to a negative and positive factor, including all items containing negative or positive emotional valence respectively. Only tumor subjects completed the mood questionnaire.

Statistical Analysis

Statistical computations were performed using the Statistical Analysis System (SAS for Windows, Version 6.12) and the Statistical Package for the Social Sciences (SPSS for Windows, Version 6.0.1). For the computation of the statistical significance of Proximities we applied a software package developed by Eckes (Eckes, 1986). For the analysis of small samples, non-parametric statistical methods (Kruskal–Wallis one-way ANOVA, Mann–Whitney U-test, Fisher's exact test, Spearman rank correlation) were used. All analyses were two-tailed and the alpha was defined at 0.05.

Results

Analysis According to Functional Properties of the Areas with Lesions

Sorting Task

Errors.

The six lesion groups (groups H, L, MS, H/L, H/MS and H/L/MS) and the two control groups (HCG, CCG) differed significantly in the amount of sorting errors (Kruskal–Wallis; P < 0.05). Table 3 shows that subjects with lesions of limbic/ paralimbic (L) or unimodal or primary motor/sensory (MS) cortices alone showed only mild deficits if any. Subjects with heteromodal (H) lesions or combined lesions of heteromodal and unimodal or primary motor/sensory cortices (H/MS) showed moderate deficits. However, subjects with combined lesions of heteromodal and limbic/paralimbic cortices (H/L), and subjects with combined lesions of heteromodal, limbic/paralimbic, and unimodal or primary motor/sensory cortices (H/L/MS) showed strong deficits. Thus, combined lesions involving both heteromodal and limbic/paralimbic cortices seem to be most strongly related to deficient performance in the sorting of emotional facial expressions.

Considering only subjects with right sided lesions, an analysis of variance (Kruskal–Wallis) revealed significant differences between groups H/L + H/L/MS, H/MS and H + L + MS for the Sum or Errors (P < 0.05). In contrast, left side lesioned subjects of these groups did not differ significantly. U-tests showed that right side lesioned subjects of groups H/L + H/L/MS performed significantly more errors than right side lesioned subjects of groups H + L + MS (P < 0.01) and H/MS (P < 0.05), and than left side lesioned subjects of groups H + L + MS (P 0.05) (Table 3).

We calculated the distribution of the Sum of Errors for the merged control groups (HCG and CCG), resulting in a critical value of 24 errors at the first percentile. Classifying the lesioned subjects according to the first percentile, a significant effect emerged (Fisher's exact test; P = 0.001): 12 subjects of groups H/L + H/L/MS ( = 41.4%) scored at the first percentile (i.e. made 24 or more errors), but only one H/MS subject (= 5.9%) and one subject of groups H + L + MS ( = 4.5%) (Table 4). Considering only subjects with right-sided lesions, 50% of H/L + H/L/MS subjects scored at the first percentile, but only 11.1% of H + L + MS subjects and no H/MS subject (Fisher's exact test; P = 0.02). In contrast, left side lesioned subjects were less impaired and performed more equally, with 22.2% of H/L + H/L/MS subjects, 14.3% of H/MS subjects and no H + L + MS subject scoring at the first percentile (Fisher's exact test; P = 0.15).

Specific Emotions.

Group differences were also seen for the Proximities, which expresses the percentage of subjects sorting identical emotional expressions into the same group (Table 5). For groups HCG, CCG, L and H/MS the Proximities of all eight pairs of emotional expressions reached statistical significance, indicating that subjects of these groups were able to identify all facial emotions. All other lesion groups displayed moderate or marked impairments. H and H/L subjects were not able to identify Fear, Surprise, Sadness and neutral expressions. H/L/MS subjects showed the strongest impairments, being just able to identify the expressions Joy and Anger (Table 5).

Rating Task

The six lesion groups (groups H, L, MS, H/L, H/MS and H/L/MS) did not differ significantly from groups HCG and CCG in the valence and arousal rating of positive, neutral or negative emotional expressions (Kruskal–Wallis; U-tests; Ps > 0.05). The same is true when groups H/L + H/L/MS, H/MS, and H + L + MS were compared to groups HCG and CCG (Ps > 0.05).

Considering only left side lesioned subjects or only right side lesioned subjects, no significant differences were obtained between groups H/L + H/L/MS, H/MS and H + L + MS for the valence and arousal rating (Kruskal–Wallis; Ps > 0.05). However, comparisons of each two lesion groups (U-tests) indicated that right side lesioned subjects of groups H/L + H/L/MS rated negative expressions less aroused than left side lesioned subjects of groups H/L + H/L/MS (P < 0.01).

Specific Emotions.

The analysis of the valence and arousal rating of pairs of identical emotional expressions revealed no significant differences (Kruskal–Wallis, U-tests) between the various lesion groups and the control groups. The analysis of Rating Errors indicated that only the Rating Error of Fear on the arousal dimension differed significantly between groups H/L + H/L/MS, H/MS and H + L + MS (Kruskal–Wallis, P <0.05). Post hoc analyses (U-tests, Fisher's exact test) indicated that subjects of groups H/L + H/L/MS and H + L + MS committed significantly more Rating Errors (i.e. rated Fear as unaroused) than control subjects (Ps < 0.05).

Mood State

Only tumor subjects completed the mood questionnaire. Comparing the six lesion groups to groups HCG and CCG, analyses of variance (Kruskal–Wallis) revealed no significant effects for the factors Vigor, Fatigue, Extraversion, Irritability/ Anger, Anxiety/Depression, and for the factors containing items with positive or negative emotional valence. However, confirmed items of the subscale Anxiety differed significantly between groups (P < 0.01). U-tests revealed that all lesioned subjects except MS subjects confirmed significantly more items of this subscale than control subjects. H/L subjects confirmed significantly more items than MS, L or H/MS subjects, and H subjects confirmed more items than MS subjects (Table 6).

Considering only subjects with left-sided lesions, analyses of variance (Kruskal–Wallis) revealed significant differences between groups H/L + H/L/MS, H/MS and H + L + MS for the factors Fatigue, Irritability/Anger, Anxiety/Depression, for the subscale Anxiety, and for the factor containing all items with negative emotional valence (Ps < 0.05). In contrast, right side lesioned subjects of these groups did not differ significantly. U-tests showed that left side lesioned subjects of groups H/L + H/L/MS confirmed significantly more items than left side lesioned subjects of groups H + L + MS, and than right side lesioned subjects of groups H/L + H/L/MS on all these measures (Ps < 0.05) (Table 6).

Analysis According to Topographical Lesion Location

Sorting Task

A two-way analysis of variance for the Sum of Errors with the factors Group (H/L + H/L/MS; H/MS; H + L + MS) and Location (frontal vs, temporal/parietal lesions) revealed a significant effect for Group (F = 5.4; P < 0.01), but nor for Location (F = 0.1; P > 0.05) or the interaction of Group and Location (F = 0.2; P > 0.05). H/L + H/L/MS subjects with frontal lesions (n = 20; median: 22 errors) and temporal/parietal lesions (n = 9; median: 20 errors) performed very similarly (U-test; P > 0.50). The same is true when only H/L + H/L/MS subjects with right-sided lesions are considered (frontal lesions, n = 7, median: 22 errors; temporal/parietal lesions, n = 7, median: 29 errors; P > 0.50). Dividing all lesioned subjects into precisely localized subgroups (Table 7) clearly illustrates that topographical lesion location cannot explain the deficits of subjects with combined lesions of heteromodal and limbic/paralimbic cortices.

Rating Task

H/L + H/L/MS subjects with right-sided lesions were divided according to lesion location (frontal lesions: n = 20; temporal/ parietal lesions: n = 9). The two groups did not differ in the valence and arousal rating of any specific emotional expression, as well as in the rating of groups of expressions with positive, neutral or negative valence respectively (U-tests; Ps > 0.30). Especially negative expressions were rated as similarly aroused by both groups (frontal lesions, median: 4.3 points; temporal/ parietal lesions, median: 4.4 points). The Rating Error did not differ between groups as well.

Mood State

H/L + H/L/MS subjects were divided according to lesion location (frontal lesions: n = 8; temporal/parietal lesions: n = 3). The two groups did not differ significantly on the various factors of the mood questionnaire, nor on the subscale Anxiety (U-tests; Ps > 0.05).

Influence of Lesion Size

The six lesion groups did not differ with respect to lesion size (Kruskal–Wallis; P > 0.05). Lesion size was not significantly related to the amount of errors performed in the sorting task (rs = 0.16; P = 0.18), nor to the mood state of the subjects (rs between 0.00 and 0.11; Ps > 0.50). However, subjects with larger lesions rated negative expressions more negative (rs = 0.28; P < 0.03) and more aroused (rs = 0.30; P < 0.02) than subjects with smaller lesions. We calculated an analysis of covariance for the Sum of Errors with the between-subjects factor Group (H/L + H/L/MS; H/MS; H + L + MS) and lesion size as covariate. After partialing out the covariate, the factor Group remained significant (F = 4.49; P < 0.02). Bonferroni-adjusted comparisons of each two of the three groups indicated that after having partialed out lesion size, H/L + H/L/MS subjects were still significantly impaired when compared with H/MS or H + L + MS subjects (Ps < 0.05).

Influence of Neuropsychological Variables

Considering lesioned subjects, significant correlations were obtained between the amount of errors performed in the sorting task and all neuropsychological measures; this indicates more sorting errors in subjects with worse neuropsychological performance. The Benton Facial Recognition Test (BFRT) was significantly related to the sorting error as well (rs = –0.44; P < 0.01), indicating more sorting errors in subjects with worse BFRT performance.

Analyses of covariance were performed for the Sum of Errors with four groups (HCG + CCG; H/L + H/L/MS; H/MS; H + L + MS) and the various neuropsychological measures as covariates. After partialing out the various covariates, the factor Group stayed significant (Fs between 5.4 and 11.4; Ps < 0.01). The same is true when the BFRT or Block Design were partialed out (BFRT: F = 9.9, P < 0.001; Block Design: F = 9.6, P < 0.001). Bonferroni-adjusted comparisons of each two of the four groups indicated that after partialing out the BFRT or Block Design subjects of groups H/L + H/L/MS were still significantly impaired when compared with subjects of the other three groups (Ps < 0.05).

Influence of Etiological Variables

A two-way analysis of variance for the Sum of Errors with the factors Group (H/L + H/L/MS; H/MS; H + L + MS) and Etiology (TUM; ICH) revealed a significant effect for Group (F = 4.9; P < 0.01) but not for Etiology (F = 0.72; P > 0.05) or the interaction of Group and Etiology (F = 2.5; P > 0.05). The Sum of Errors did not differ significantly between TUM and ICH subjects of groups H/L + H/L/MS, H/MS or H + L + MS (U-tests; Ps > 0.10). The same is true when TUM and ICH subjects of single lesion groups (groups H, H/MS, H/L and H/L/MS) were compared (Ps > 0.20).

Discussion

Summary of Findings

The present study indicated that combined lesions of heteromodal and limbic/paralimbic cortices disturb the decoding of emotional facial expressions. Lesions restricted to limbic/ paralimbic cortices, or lesions restricted to unimodal or primary motor/sensory cortices were followed by only minor deficits if any. Moderate deficits were found after lesions restricted to heteromodal cortices and after combined lesions of heteromodal and unimodal or primary motor/sensory cortices.

Our data further demonstrate an interaction between lesion laterality and functional properties of the areas with lesions. Only combined lesions of heteromodal and limbic/paralimbic cortices of the right hemisphere were related to strong deficits in the decoding of emotional facial expressions (cf. Table 3). Extremely negative mood states were observed after left-sided, but not right-sided lesions of heteromodal and limbic/paralimbic cortices (Table 6).

We found differential performances in the decoding of specific emotions. Joy seemed to be most easiest to decode, followed by anger and disgust, and sadness. Surprise and fear were most difficult to decode (Table 5). Subjects with combined lesions of heteromodal and limbic/paralimbic cortices were the most impaired, being unable to decode fear, sadness and surprise.

The deficits of subjects with combined lesions of heteromodal and limbic/paralimbic cortices cannot be attributed to topographical lesion location (Table 7). Furthermore, lesion size, lesion etiology and neuropsychological performance, especially face discrimination and visuoconstructive abilities, could not explain the deficits of subjects with combined lesions of heteromodal and limbic/paralimbic cortices.

Importance of Transmodal Cortices for the Decoding of Emotional Facial Expressions

Severe deficits in the decoding of emotional facial expressions were observed only after multiple (i.e. combined heteromodal and paralimbic) but not single lesions of transmodal cortices. These results suggest that complex human emotional behaviors may only be fundamentally disturbed when substantial amounts of transmodal tissue are lost. Table 7 clearly illustrates that the more transmodal areas were affected, the stronger were the behavioral deficits. The results cannot be attributed to a mere effect of lesion size, as there was no significant relation between lesion size and the number or sorting errors (cf. Results: Influence of lesion size).

Our results further suggest that transmodal cortices may be considered to a certain degree as equipotential (Lashley, 1929). A meta-analysis of lesion studies in Old World monkeys (Irle, 1990) also pointed to highly specific as well as equipotential functioning of transmodal cortices. However, the detection of more specific lesion effects in our study may have been hindered by the complexity of the behaviors being measured. It is likely that many transmodal areas differentially contribute to a complex emotional-cognitive behavior, and that abolishing many transmodal areas will result in a strong and global impairment of this behavior.

Functional imaging studies also underline the idea that many transmodal cortices are important for the decoding of emotional facial expressions. Processing of emotional facial expressions produced regional activity changes in the ventral prefrontal, cingular, retrosplenial, temporopolar and fusiform cortices (George et al., 1993; Dolan et al., 1996; Blair et al., 1999; Critchley et al., 2000).

A recently published lesion study by Adolphs et al. (Adolphs et al., 2000) emphasized the importance of somatosensory-related cortices for the decoding of emotional facial expressions. The lesion analysis was based on statistical analyses of joint volumetric lesion densities resulting in a visualization of lesion overlaps. The advantage of this approach is that small lesion compartments being most often associated with a certain deficit can be identified. However, the effects of other aspects of a lesion, which are possibly also shared by many subjects, are not considered in this approach. A close inspection of the results of Adolphs et al. (Adolphs et al., 2000) suggests that many of their subjects with somatosensory cortex-related lesions may have also presented with lesions of heteromodal and limbic/ paralimbic (i.e. temporo-parietal) cortices.

Cerebral Areas and Specific Emotions

Case studies of subjects with lesions of the amygdala and surrounding temporal cortices have indicated that the decoding of fear may be most strongly impaired in these subjects (Adolphs et al., 1995; Calder et al., 1996; Weniger et al., 1997; Broks et al., 1998). Functional imaging studies to a large degree confirm these results (Breiter et al., 1996; Morris et al., 1998), but also stress the relevance of additional cortical (insula) and subcortical (substantia innominata, striatum) areas (Phillips et al., 1997; Whalen et al., 1998). We could not find strong deficits in the decoding of fear in subjects with limbic/paralimbic temporal lesions (see Table 5). However, previous studies reported deficits in the decoding of fear consistently only after bilateral lesions of the amygdala and temporal cortex (Adolphs et al., 1995; Young et al., 1995; Calder et al., 1996; Rapcsak et al., 2000).

The attempt to relate the processing of emotions other than fear to specific cortical areas has proven much more difficult. Functional imaging studies (George et al., 1995; Lane et al., 1997; Paradiso et al., 1997, 1999; Reiman et al., 1997; Teasdale et al., 1999) failed to establish consistent and specific patterns of regional activity changes during the generation of sadness, or disgust, or happiness. However, all studies found that the medial heteromodal frontal cortex (largely corresponding to medial parts of BA 9) enhances its activity while experiencing any kind of emotion. Our results confirm these findings while showing that subjects with transmodal lesions were impaired in the decoding of fear, surprise, sadness and disgust (Table 5). We could not find evidence that the decoding of a specific emotion was independently impaired in a given lesion group. Similar findings were obtained by Adolphs et al. (Adolphs et al., 2000).

It could be argued that the sorting task used in this study demands subjects to distinguish between specific emotions rather than to identify these emotions, and that transmodal lesions may impair the ability to sort while leaving the ability to identify intact. However, we think that our subjects sorted the expressions on the basis of having identified them. We could not find sorting patterns indicating that subjects used sorting categories (i.e. gender, physiognomic aspects) other than emotion identity. Subjects were also explicitly taught to sort according to emotion identity. Furthermore, we found differential performances in the sorting of specific emotions, with joy being most easiest to decode, followed by anger, disgust, sadness, surprise and fear (Table 5).

We hold the opinion that distinguishing and identifying emotions may not depend on separate neural systems. Rather, categorical or dimensional task demands may use different neural subsystems. Adolphs et al. (Adolphs et al., 2000) used a categorical approach for both sorting and identifying emotional expressions and found that subjects impaired in the sorting of emotional facial expressions were also impaired in the identification of emotional facial expressions. In our study, the categorical approach of the sorting task (which is very similar to the sorting task of Adolphs et al.) and the dimensional approach of the rating task yielded different results, with subjects impaired in the sorting task showing no impairment in the rating task (Rating Error; cf. Results: Rating Task).

Role of the Amygdala for the Decoding of Emotional Facial Expressions

It is difficult to evaluate the role of the human amygdala in the decoding of emotional facial expressions because exclusive lesions of the amygdala are extremely rare. In most of the published cases with bilateral amygdala lesions (Jacobson, 1986; Broks et al., 1998; Adolphs et al., 1999) substantial lesions of temporal or extratemporal cortical or subcortical (i.e. striatal and thalamic) tissue is evident. Even the few cases with surgical or congenital lesions show evidence of an affection of the entorhinal or temporopolar cortex. However, a recent study investigating 22 subjects after anterior temporal lobectomy or selective amygdalo-hippocampectomy (Boucsein et al., 2001) showed that deficits in the associative learning of emotional facial expressions depend on the extent of amygdala damage.

Nevertheless, it cannot be ruled out that severe disturbances in the decoding of emotional facial expressions depend on the presence of multiple lesions affecting the amygdala and transmodal cortices. Functional imaging studies showed that viewing emotional faces, as compared with neutral faces, activated not only the amygdala and basal ganglia regions, but also transmodal (i.e. ventral and dorsolateral frontal, cingular, temporopolar, insular and ventral and lateral temporal) cortical areas (Breiter et al., 1996; Phillips et al., 1997; Morris et al., 1998; Whalen et al., 1998; Blair et al., 1999; Critchley et al., 2000).

Transmodal Cortices and Mood State

Subjects with transmodal lesions described themselves more negatively, and in particular were more anxious than control subjects (Table 6). These results are supported by functional imaging studies. Using various methods to generate emotions, activity changes in both limbic/paralimbic and heteromodal cortices could be found (George et al., 1995; Lane et al., 1997; Paradiso et al., 1997, 1999; Reiman et al., 1997; Teasdale et al., 1999).

To a great extent, our data replicate the findings of Irle et al. (Irle et al., 1994), who used the same mood questionnaire as this study to investigate 141 subjects with cortical tumor lesions. They found that transmodal lesions were followed by the worst mood states, but could not find an effect of hemispheric side of lesion. It is possible that a sampling bias, i.e. the exclusion of aphasic left side lesioned subjects may explain this discrepancy (Gainotti et al., 1997). In the study of Irle et al. (Irle et al., 1994) only half as many subjects with left-sided as compared with right-sided lesions entered the neurosurgical department in a state allowing formal neuropsychological testing and were thus included in the study. In the present investigation 20 subjects with left-sided lesions were compared to 16 subjects with right-sided lesions.

Conclusions and Directions for Future Research

Our study emphasizes the importance of transmodal cortices for complex human cognitive-emotional behaviors. These behaviors may provide a further example of ‘top-down processing’ (McClelland and Plaut, 1993). Transmodal cortices are not only capable of establishing multimodal convergence and integration, but also can exert sensory-petal influences upon unimodal cortices, and may provide a directory pointing to the distributed sources of related relevant information (Mesulam, 1998). Our results suggest that transmodal cortices play a crucial role, and may be even considered as ‘bottleneck structures’ for many human cognitive and emotional behaviors.

The functional classification used in the present study may provide a useful tool for the investigation of various neuro-cognitive disorders, which to a large degree may be unique to humans. For example, evidence is increasing that schizophrenia represents a disorder in which disruption of transmodal cortical functioning is a central event [for a review see Pearlson et al. (Pearlson et al., 1996)]. The volumes of heteromodal, but not other cortices have been shown to be reduced in schizophrenia (Schlaepfer et al., 1994). Behavioral and anatomical evidence points to a prefrontal dysfunction in schizophrenia (Weinberger et al., 1986; Goldman-Rakic & Selemon, 1997; Stevens et al., 1998). Andreasen proposed a model defining schizophrenia as a misconnection syndrome involving transmodal cortices, thalamus and cerebellum, resulting in a fundamental and global cognitive deficit (Andreasen, 1999). The application of imaging tools and neuropathological surveys could help to define the relationship of transmodal cortical abnormalities with the clinical syndrome, neurochemistry and pathogenesis of schizophrenia.

Notes

The authors thank all participating subjects, and D. Schwarze and B. Schmidt for performance of neuropsychological testing. Research was supported by grant Ir 15/6-1 from the Deutsche Forschungsgemeinschaft.

Table 1

Demographic and clinical characteristics of all subjects

Group n Age1,a (years) Sexb (female/male) Education1,2,a Etiology of lesiond (TUM/ICH) Neurological Impairments1,3,c 
TUM, tumor; ICH, intracerebral hemorrhage; HCG, healthy control group; CCG, clinical control group; H, subjects with lesions of heteromodal cortices; L, subjects with lesions of limbic/paralimbic cortices; H/L, subjects with lesions of heteromodal and limbic/paralimbic cortices; MS, subjects with lesions of unimodal or primary motor/sensory cortices; H/MS, subjects with lesions of heteromodal, and unimodal or primary motor/sensory cortices; H/L/MS, subjects with lesions of heteromodal, limbic/paralimbic, and unimodal or primary motor/sensory cortices. 
1Medians (and ranges of the 75th–25th percentile) are given. 
2Educational scores vary between 1 (<8 years of schooling) and 6 (academic achievement). 
3Assessed at time of testing. Scores may vary between 0 (no), 1 (mild), 2 (moderate), 3 (substantial) and 4 (strong impairment). 
aThe groups did not differ significantly (Kruskal–Wallis one-way analysis of variance; P > 0.05). 
bThe groups did not differ significantly (Fisher's exact test; P > 0.05). 
cThe lesion groups did not differ significantly (Kruskal–Wallis one-way analysis of variance; P > 0.05). 
dThe groups differed significantly (Fisher's exact test; P < 0.01). 
HCG 15 54 (25)  5/10 3 (3) – – 
CCG 16 39 (10)  8/8 3 (2) – – 
11 47 (17)  5/6 2 (1)  5/6 0 (2) 
49 (22)  4/4 2 (2)  7/1 0 (1) 
H/L 14 45 (21)  4/10 2 (1)  4/10 0 (1) 
MS 39 (16)  1/2 6 (3)  3/0 1 (3) 
H/MS 17 44 (32) 11/6 2 (2) 14/3 1 (2) 
H/L/MS 15 49 (22) 10/5 2 (2) 12/3 1 (1) 
Group n Age1,a (years) Sexb (female/male) Education1,2,a Etiology of lesiond (TUM/ICH) Neurological Impairments1,3,c 
TUM, tumor; ICH, intracerebral hemorrhage; HCG, healthy control group; CCG, clinical control group; H, subjects with lesions of heteromodal cortices; L, subjects with lesions of limbic/paralimbic cortices; H/L, subjects with lesions of heteromodal and limbic/paralimbic cortices; MS, subjects with lesions of unimodal or primary motor/sensory cortices; H/MS, subjects with lesions of heteromodal, and unimodal or primary motor/sensory cortices; H/L/MS, subjects with lesions of heteromodal, limbic/paralimbic, and unimodal or primary motor/sensory cortices. 
1Medians (and ranges of the 75th–25th percentile) are given. 
2Educational scores vary between 1 (<8 years of schooling) and 6 (academic achievement). 
3Assessed at time of testing. Scores may vary between 0 (no), 1 (mild), 2 (moderate), 3 (substantial) and 4 (strong impairment). 
aThe groups did not differ significantly (Kruskal–Wallis one-way analysis of variance; P > 0.05). 
bThe groups did not differ significantly (Fisher's exact test; P > 0.05). 
cThe lesion groups did not differ significantly (Kruskal–Wallis one-way analysis of variance; P > 0.05). 
dThe groups differed significantly (Fisher's exact test; P < 0.01). 
HCG 15 54 (25)  5/10 3 (3) – – 
CCG 16 39 (10)  8/8 3 (2) – – 
11 47 (17)  5/6 2 (1)  5/6 0 (2) 
49 (22)  4/4 2 (2)  7/1 0 (1) 
H/L 14 45 (21)  4/10 2 (1)  4/10 0 (1) 
MS 39 (16)  1/2 6 (3)  3/0 1 (3) 
H/MS 17 44 (32) 11/6 2 (2) 14/3 1 (2) 
H/L/MS 15 49 (22) 10/5 2 (2) 12/3 1 (1) 
Table 2

Characteristics of the lesion groups

Group (nLesion Extenta,b Lesion size (ml)a,c Cases with subcortical side damage Lesion lateralityd (left/right/bilateral) 
 Heteromodal cortex (BA 7–12,39, 40,45,46) Limbic/paralimbic cortex (BA 11–16, 23–36,38, Am,Hi,OC,SI) Unimodal/primary motor/sensory cortex (BA 1–6, 20–22)    
BA, Brodmann area; Am, amygdala; Hi, hippocampus; OC, olfactory cortex; SI, substantia innominata. 
For abbreviations of the lesion groups see Table 1
aMedians (and ranges of the 75th–25th percentile) are given. 
bPercentage of destroyed tissue. The functional properties of cortical areas were defined according to Mesulam (Mesulam, 1998). Anterior parts of areas 11 and 12 were defined as heteromodal, and posterior parts of these areas as paralimbic. 
cLesion size included in TUM subjects: residual tumor, area of hypodensity (edema, injured tissue, infiltration zone); in ICH subjects: hemorrhage, area of hypodensity (edema, injured or resected tissue). The groups did not differ significantly in lesion size (Kruskal–Wallis; P > 0.05). The strong variability of lesion size in groups H and H/L is due to some subjects with subtotal unilateral frontal lobectomies. 
dThe groups did not differ significantly (Fisher's exact test; P > 0.05). 
H (n = 11) 20 (30)  0 (0)  0 (0)  3.5 (12.4) 6/5/0 
L (n = 8)  0 (0) 20 (20)  0 (10)  5.2 (5.4) 5/3/0 
H/L (n = 14) 35 (70) 20 (20)  0 (0) 11.8 (15.8) 5/5/4 
MS (n = 3)  0 (0)  0 (0) 20 (10)  1.0 (1.7) 2/1/0 
H/MS (n = 17) 20 (20)  0 (0) 30 (20)  5.5 (6.4) 7/8/2 
H/L/MS (n = 15) 30 (20) 10 (10) 20 (20)  8.7 (6.4) 4/9/2 
Group (nLesion Extenta,b Lesion size (ml)a,c Cases with subcortical side damage Lesion lateralityd (left/right/bilateral) 
 Heteromodal cortex (BA 7–12,39, 40,45,46) Limbic/paralimbic cortex (BA 11–16, 23–36,38, Am,Hi,OC,SI) Unimodal/primary motor/sensory cortex (BA 1–6, 20–22)    
BA, Brodmann area; Am, amygdala; Hi, hippocampus; OC, olfactory cortex; SI, substantia innominata. 
For abbreviations of the lesion groups see Table 1
aMedians (and ranges of the 75th–25th percentile) are given. 
bPercentage of destroyed tissue. The functional properties of cortical areas were defined according to Mesulam (Mesulam, 1998). Anterior parts of areas 11 and 12 were defined as heteromodal, and posterior parts of these areas as paralimbic. 
cLesion size included in TUM subjects: residual tumor, area of hypodensity (edema, injured tissue, infiltration zone); in ICH subjects: hemorrhage, area of hypodensity (edema, injured or resected tissue). The groups did not differ significantly in lesion size (Kruskal–Wallis; P > 0.05). The strong variability of lesion size in groups H and H/L is due to some subjects with subtotal unilateral frontal lobectomies. 
dThe groups did not differ significantly (Fisher's exact test; P > 0.05). 
H (n = 11) 20 (30)  0 (0)  0 (0)  3.5 (12.4) 6/5/0 
L (n = 8)  0 (0) 20 (20)  0 (10)  5.2 (5.4) 5/3/0 
H/L (n = 14) 35 (70) 20 (20)  0 (0) 11.8 (15.8) 5/5/4 
MS (n = 3)  0 (0)  0 (0) 20 (10)  1.0 (1.7) 2/1/0 
H/MS (n = 17) 20 (20)  0 (0) 30 (20)  5.5 (6.4) 7/8/2 
H/L/MS (n = 15) 30 (20) 10 (10) 20 (20)  8.7 (6.4) 4/9/2 
Table 3

Performance in the sorting of emotional facial expressions

Group (nSum of errors 
 Median 75th–25th percentile 
HCG, healthy control group; CCG, clinical control group. For abbreviations of the lesion groups see Table 1 and Materials and Methods: Lesion analysis. 
aSignificantly different from group HCG (U-test; P < 0.05). 
bSignificantly different from group HCG (U-test; P < 0.001). 
cSignificantly different from group CCG (U-test; P < 0.05). 
dSignificantly different from group CCG (U-test; P < 0.0001). 
eSignificantly different from group L (U-test; P < 0.05). 
fSignificantly different from group MS (U-test; P < 0.05). 
gSignificantly different from group H/MS (U-test; P < 0.05). 
hSignificantly different from group H (U-test; P < 0.05). 
iSignificantly different from group H + L + MS (left-sided lesion) (U-test; P < 0.01). 
jSignificantly different from group H + L + MS (right-sided lesion) (U-test; P < 0.05). 
kSignificantly different from group H/MS (right-sided lesion) (U-test; P < 0.05). 
HCG (n = 15) 
CCG (n = 16) 
L (n = 8) 15 
MS (n = 3) 10a 
H/MS (n = 17) 15b,c,e,f 
H (n = 11) 16b,c 
H/L (n = 14) 20b,c,e 17 
H/L/MS (n = 15) 21b,d,e,f,g,h 22 
Left sided lesions   
    H + L + MS (n = 13) 13b 
    H/MS (n = 7) 16b,c 
    H/L + H/L/MS (n = 9) 16b,c 
Right sided lesions   
    H + L + MS (n = 9) 13a 10 
    H/MS (n = 8) 14b,c 
    H/L + H/L/MS (n = 14) 26b,d,i,j,k 17 
Group (nSum of errors 
 Median 75th–25th percentile 
HCG, healthy control group; CCG, clinical control group. For abbreviations of the lesion groups see Table 1 and Materials and Methods: Lesion analysis. 
aSignificantly different from group HCG (U-test; P < 0.05). 
bSignificantly different from group HCG (U-test; P < 0.001). 
cSignificantly different from group CCG (U-test; P < 0.05). 
dSignificantly different from group CCG (U-test; P < 0.0001). 
eSignificantly different from group L (U-test; P < 0.05). 
fSignificantly different from group MS (U-test; P < 0.05). 
gSignificantly different from group H/MS (U-test; P < 0.05). 
hSignificantly different from group H (U-test; P < 0.05). 
iSignificantly different from group H + L + MS (left-sided lesion) (U-test; P < 0.01). 
jSignificantly different from group H + L + MS (right-sided lesion) (U-test; P < 0.05). 
kSignificantly different from group H/MS (right-sided lesion) (U-test; P < 0.05). 
HCG (n = 15) 
CCG (n = 16) 
L (n = 8) 15 
MS (n = 3) 10a 
H/MS (n = 17) 15b,c,e,f 
H (n = 11) 16b,c 
H/L (n = 14) 20b,c,e 17 
H/L/MS (n = 15) 21b,d,e,f,g,h 22 
Left sided lesions   
    H + L + MS (n = 13) 13b 
    H/MS (n = 7) 16b,c 
    H/L + H/L/MS (n = 9) 16b,c 
Right sided lesions   
    H + L + MS (n = 9) 13a 10 
    H/MS (n = 8) 14b,c 
    H/L + H/L/MS (n = 14) 26b,d,i,j,k 17 
Table 4

Classification of lesion groups according to their performance in the sorting task

Group (nSorting of emotional facial expressions (Sum of Errors) 
 ≤1st percentile >1st percentile 
The first percentile of the distribution of the Sum of Errors for groups HCG and CCG is 24 errors. Percentages of lesioned subjects scoring above and below the first percentile are given respectively. 
Fisher's exact test revealed for the comparison of the three lesion groups a significant effect (P = 0.001). 
For abbreviations of the lesion groups see Table 1 and Materials and Methods: Lesion analysis. 
H/L + H/L/MS (n = 29) 41.4 58.6 
H/MS (n = 17)  5.9 94.1 
H + L + MS (n = 22)  4.5 95.5 
Group (nSorting of emotional facial expressions (Sum of Errors) 
 ≤1st percentile >1st percentile 
The first percentile of the distribution of the Sum of Errors for groups HCG and CCG is 24 errors. Percentages of lesioned subjects scoring above and below the first percentile are given respectively. 
Fisher's exact test revealed for the comparison of the three lesion groups a significant effect (P = 0.001). 
For abbreviations of the lesion groups see Table 1 and Materials and Methods: Lesion analysis. 
H/L + H/L/MS (n = 29) 41.4 58.6 
H/MS (n = 17)  5.9 94.1 
H + L + MS (n = 22)  4.5 95.5 
Table 5

Proximities of identical emotional facial expressions

Group (nCritical valuea Joy Anger Disgust Pleasure Sadness Surprise Neutral Fear 
The values represent percentages of subjects sorting identical emotional facial expressions into the same group. Values surpassing the critical value are in italics. Subjects of group MS (n = 3) are excluded in this analysis because of the small sample size 
HCG, healthy control group; CCG, clinical control group. For abbreviations of the lesion groups see Table 1
aCritical percentage of proximities with an experimentwise error rate of α = 0.05 based on an adjusted alpha level of α* = 0.05/120. 
HCG (n = 15) 40 93 100 93 87 100 67 80 67 
CCG (n = 16) 44 100 83 94 78 73 72 89 72 
L (n = 8) 50 75 75 63 75 63 50 75 50 
H/MS (n = 17) 53 88 59 53 76 53 71 59 59 
H (n = 11) 64 91 73 82 64  27 55 46 46 
H/L (n = 14) 57 64 57 57 57  36 36 36 29 
H/L/MS (n = 15) 60 80 73 47 40  40 33 47 40 
Group (nCritical valuea Joy Anger Disgust Pleasure Sadness Surprise Neutral Fear 
The values represent percentages of subjects sorting identical emotional facial expressions into the same group. Values surpassing the critical value are in italics. Subjects of group MS (n = 3) are excluded in this analysis because of the small sample size 
HCG, healthy control group; CCG, clinical control group. For abbreviations of the lesion groups see Table 1
aCritical percentage of proximities with an experimentwise error rate of α = 0.05 based on an adjusted alpha level of α* = 0.05/120. 
HCG (n = 15) 40 93 100 93 87 100 67 80 67 
CCG (n = 16) 44 100 83 94 78 73 72 89 72 
L (n = 8) 50 75 75 63 75 63 50 75 50 
H/MS (n = 17) 53 88 59 53 76 53 71 59 59 
H (n = 11) 64 91 73 82 64  27 55 46 46 
H/L (n = 14) 57 64 57 57 57  36 36 36 29 
H/L/MS (n = 15) 60 80 73 47 40  40 33 47 40 
Table 6

Mood states

Group (nItems containing anxiety Items containing negative emotional valence 
 Median 75th–25th percentile Median 75th–25th percentile 
Only tumor subjects completed the mood questionnaire. Percentages of confirmed items are given. 
HCG, healthy control group; CCG, clinical control group. For abbreviations of the lesion groups see Table 1 and Materials and Methods: Lesion analysis. 
aSignificantly different from group HCG (U-test; P < 0.05). 
bSignificantly different from group HCG (U-test; P < 0.01). 
cSignificantly different from group CCG (U-test; P < 0.05). 
dSignificantly different from group CCG (U-test; P < 0.01). 
eSignificantly different from group MS (U-test; P < 0.05). 
fSignificantly different from group L (U-test; P < 0.05). 
gSignificantly different from group L (U-test; P < 0.01). 
hSignificantly different from group H/MS (U-test; P < 0.05). 
iSignificantly different from group H/L + H/L/MS (right-sided lesions) (U-test; P < 0.05). 
jSignificantly different from group H + L + MS (left-sided lesions) (U-test; P < 0.01). 
kSignificantly different from group H/MS (right-sided lesions) (U-test; P < 0.05). 
HCG (n = 15) 14 13 18 
CCG (n = 15) 14 13 32 
MS (n = 3) 19 14 
L (n = 5) 17a 17 14 
H/MS (n = 13) 17a,c 17 23 20 
H (n = 5) 25b,d,e 36 22f 35 
H/L/MS (n = 7) 25a,c 35 32 21 
H/L (n = 4) 33b,c,e,g,h 17 37a,f 22 
 
Left sided lesions     
    H + L + MS (n = 9) 14a,c 25 16 
    H/MS (n = 5) 25a 17 31 10 
    H/L + H/L/MS (n = 6) 38b,d,i,j,k 17 38b,c,i,j 15 
Right sided lesions     
    H + L + MS (n = 4) 15 36 17 27 
    H/MS (n = 6) 17 17 22 
    H/L + H/L/MS (n = 4) 17 26 15 18 
Group (nItems containing anxiety Items containing negative emotional valence 
 Median 75th–25th percentile Median 75th–25th percentile 
Only tumor subjects completed the mood questionnaire. Percentages of confirmed items are given. 
HCG, healthy control group; CCG, clinical control group. For abbreviations of the lesion groups see Table 1 and Materials and Methods: Lesion analysis. 
aSignificantly different from group HCG (U-test; P < 0.05). 
bSignificantly different from group HCG (U-test; P < 0.01). 
cSignificantly different from group CCG (U-test; P < 0.05). 
dSignificantly different from group CCG (U-test; P < 0.01). 
eSignificantly different from group MS (U-test; P < 0.05). 
fSignificantly different from group L (U-test; P < 0.05). 
gSignificantly different from group L (U-test; P < 0.01). 
hSignificantly different from group H/MS (U-test; P < 0.05). 
iSignificantly different from group H/L + H/L/MS (right-sided lesions) (U-test; P < 0.05). 
jSignificantly different from group H + L + MS (left-sided lesions) (U-test; P < 0.01). 
kSignificantly different from group H/MS (right-sided lesions) (U-test; P < 0.05). 
HCG (n = 15) 14 13 18 
CCG (n = 15) 14 13 32 
MS (n = 3) 19 14 
L (n = 5) 17a 17 14 
H/MS (n = 13) 17a,c 17 23 20 
H (n = 5) 25b,d,e 36 22f 35 
H/L/MS (n = 7) 25a,c 35 32 21 
H/L (n = 4) 33b,c,e,g,h 17 37a,f 22 
 
Left sided lesions     
    H + L + MS (n = 9) 14a,c 25 16 
    H/MS (n = 5) 25a 17 31 10 
    H/L + H/L/MS (n = 6) 38b,d,i,j,k 17 38b,c,i,j 15 
Right sided lesions     
    H + L + MS (n = 4) 15 36 17 27 
    H/MS (n = 6) 17 17 22 
    H/L + H/L/MS (n = 4) 17 26 15 18 
Table 7

Performance in the sorting of emotional facial expressions in localizational subgroups

Areas with lesionsa n Sum of Errorsb Type of cortex destroyedc Lesion group 
aAs defined by Brodmann. Am, amygdala; Hi, hippocampus; OC, olfactory cortex; SI, substantia innominata. 
bMedians (and ranges of the 75th–25th percentile) are given. 
cThe functional properties of cortical areas were defined according to Mesulam (Mesulam, 1998). Anterior parts of areas 11 and 12 were defined as heteromodal, and posterior parts of these areas as paralimbic. 
28, 34, Am, Hi  3 (16) limbic/paralimbic temporal 
14–16, 27, 28, 34–36, 38, Am, Hi  8 (9) limbic/paralimbic temporal 
3, 4, 6 10 (4) premotor, primary motor/somatosensory MS 
8–12, 45, 46 15 (7) heteromodal frontal 
4, 6, 8–10, 45, 46 15 (11) heteromodal frontal, premotor, primary motor H/MS 
7, 39, 40 16 (1) heteromodal parietal 
1–9, 39, 40, 45, 46 10 16 (9) heteromodal frontal and parietal, premotor, unimodal parietal, primary motor/somatosensory H/MS 
8–13, 45, 46, 24, 32, OC, SI 11 18 (33) heteromodal and limbic/paralimbic frontal, cingulate H/L 
1–3, 5, 7, 14–16, 20–24, 27–31, 35, 36, 39, 40 18 (13) heteromodal parietal, paralimbic temporal, cingulate, retrosplenial, unimodal parietal and temporal, primary motor/somatosensory/auditory H/L/MS 
4, 6, 8–13, 23, 24, 31, 32, SI 25 (22) heteromodal and limbic/paralimbic frontal, cingulate, retrosplenial, premotor, primary motor H/L/MS 
24, 27, 28, 38, 45, Am, Hi, OC, SI 31 (22) heteromodal and limbic/paralimbic frontal and temporal H/L 
Areas with lesionsa n Sum of Errorsb Type of cortex destroyedc Lesion group 
aAs defined by Brodmann. Am, amygdala; Hi, hippocampus; OC, olfactory cortex; SI, substantia innominata. 
bMedians (and ranges of the 75th–25th percentile) are given. 
cThe functional properties of cortical areas were defined according to Mesulam (Mesulam, 1998). Anterior parts of areas 11 and 12 were defined as heteromodal, and posterior parts of these areas as paralimbic. 
28, 34, Am, Hi  3 (16) limbic/paralimbic temporal 
14–16, 27, 28, 34–36, 38, Am, Hi  8 (9) limbic/paralimbic temporal 
3, 4, 6 10 (4) premotor, primary motor/somatosensory MS 
8–12, 45, 46 15 (7) heteromodal frontal 
4, 6, 8–10, 45, 46 15 (11) heteromodal frontal, premotor, primary motor H/MS 
7, 39, 40 16 (1) heteromodal parietal 
1–9, 39, 40, 45, 46 10 16 (9) heteromodal frontal and parietal, premotor, unimodal parietal, primary motor/somatosensory H/MS 
8–13, 45, 46, 24, 32, OC, SI 11 18 (33) heteromodal and limbic/paralimbic frontal, cingulate H/L 
1–3, 5, 7, 14–16, 20–24, 27–31, 35, 36, 39, 40 18 (13) heteromodal parietal, paralimbic temporal, cingulate, retrosplenial, unimodal parietal and temporal, primary motor/somatosensory/auditory H/L/MS 
4, 6, 8–13, 23, 24, 31, 32, SI 25 (22) heteromodal and limbic/paralimbic frontal, cingulate, retrosplenial, premotor, primary motor H/L/MS 
24, 27, 28, 38, 45, Am, Hi, OC, SI 31 (22) heteromodal and limbic/paralimbic frontal and temporal H/L 
Figure 1.

(a) Postoperative T1-weighted MR images (post GD) of a subject of group H with tumor lesions affecting the heteromodal parietal cortex of the left hemisphere. Scan thickness: 6 mm; distance between scans: 1.2 mm. (b) Postoperative T1-weighted MR images (post GD) of a subject of group MS with tumor lesions affecting the premotor and motor cortex of the right hemisphere. Scan thickness: 6 mm; distance between scans: 1.2 mm. (c) Postoperative T1-weighted MR images (post GD) of a subject of group L with tumor lesions affecting paralimbic and limbic temporal regions (including amygdala) of the right hemisphere. Scan thickness: 3 mm; distance between scans: 11.4 mm. The right hemisphere is represented on the viewer's left. Lesions are indicated by arrows.

Figure 1.

(a) Postoperative T1-weighted MR images (post GD) of a subject of group H with tumor lesions affecting the heteromodal parietal cortex of the left hemisphere. Scan thickness: 6 mm; distance between scans: 1.2 mm. (b) Postoperative T1-weighted MR images (post GD) of a subject of group MS with tumor lesions affecting the premotor and motor cortex of the right hemisphere. Scan thickness: 6 mm; distance between scans: 1.2 mm. (c) Postoperative T1-weighted MR images (post GD) of a subject of group L with tumor lesions affecting paralimbic and limbic temporal regions (including amygdala) of the right hemisphere. Scan thickness: 3 mm; distance between scans: 11.4 mm. The right hemisphere is represented on the viewer's left. Lesions are indicated by arrows.

Figure 2.

(a) Postoperative T1-weighted MR images (post GD) of a subject of group H/MS with tumor lesions affecting the heteromodal parietal and the primary motor/somatosensory cortex of the left hemisphere. Scan thickness: 6 mm; distance between scans: 8.4 mm. (b) Postoperative T1-weighted MR images (post GD) of a subject of group H/L with tumor lesions affecting the heteromodal and paralimbic frontal cortex of the right hemisphere. Scan thickness: 6 mm; distance between scans: 1.2 mm. (c) T1-weighted MR images (post GD) of a subject of group H/L/MS with an intracerebral hemorrhage affecting the heteromodal parietal, paralimbic temporal, and primary motor/somatosensory/auditory cortex of the right hemisphere. Scan thickness: 6 mm; distance between scans: 15.6 mm. The right hemisphere is represented on the viewer's left. Lesions are indicated by arrows.

Figure 2.

(a) Postoperative T1-weighted MR images (post GD) of a subject of group H/MS with tumor lesions affecting the heteromodal parietal and the primary motor/somatosensory cortex of the left hemisphere. Scan thickness: 6 mm; distance between scans: 8.4 mm. (b) Postoperative T1-weighted MR images (post GD) of a subject of group H/L with tumor lesions affecting the heteromodal and paralimbic frontal cortex of the right hemisphere. Scan thickness: 6 mm; distance between scans: 1.2 mm. (c) T1-weighted MR images (post GD) of a subject of group H/L/MS with an intracerebral hemorrhage affecting the heteromodal parietal, paralimbic temporal, and primary motor/somatosensory/auditory cortex of the right hemisphere. Scan thickness: 6 mm; distance between scans: 15.6 mm. The right hemisphere is represented on the viewer's left. Lesions are indicated by arrows.

References

Adolphs R, Tranel D, Damasio H, Damasio AR (
1995
) Fear and the human amygdala.
J Neurosci
 
15
:
5879
–5891.
Adolphs R, Tranel D, Hamann S, Young AW, Calder AJ, Phelps EA, Anderson A, Lee GP, Damasio AR (
1999
) Recognition of facial emotion in nine individuals with bilateral amygdala damage.
Neuropsychologia
 
37
:
1111
–1117.
Adolphs R, Damasio H, Tranel D, Cooper G, Damasio AR (
2000
) A role for somatosensory cortices in the visual recognition of emotion as revealed by three-dimensional lesion mapping.
J Neurosci
 
20
:
2683
–2690.
Aggleton JP (ed.) (2000) The amygdala. A functional analysis, 2nd edn. New York: Oxford University Press.
Andreasen NC (
1999
) A unitary model of schizophrenia. Bleuler's ‘fragmented phrene’ as schizencephaly.
Arch Gen Psychiat
 
56
:
781
–787.
Blair RJR, Morris JS, Frith CD, Perrett DI, Dolan RJ (
1999
) Dissociable neural responses to facial expressions of sadness and anger.
Brain
 
122
:
883
–893.
Blonder LX, Bowers D, Heilman KM (
1991
) The role of the right hemisphere in emotional communication.
Brain
 
114
:
1115
–1127.
Broks P, Young AW, Maratos EJ, Coffey PJ, Calder AJ, Isaac CL, Mayes AR, Hodges JR, Montaldi D, Cezayirli E, Roberts N, Hadley D (
1998
) Face processing impairments after encephalitis: amygdala damage and recognition of fear.
Neuropsychologia
 
36
:
59
–70.
Boucsein K, Weniger G, Mursch K, Steinhoff B, Irle E (
2001
) Amygdala lesion in temporal lobe epilepsy subjects impairs associative learning of emotional facial expressions.
Neuropsychologia
 
39
:
231
–236.
Breiter HC, Etcoff N, Whalen PJ, Kennedy WA, Rauch SL, Buckner RL, Strauss MM, Hyman SE, Rosen BR (
1996
) Response and habituation of the human amygdala during visual processing of facial expression.
Neuron
 
17
:
875
–887.
Calder AJ, Young AW, Rowland D, Perrett DI, Hodges JR, Etcoff N (
1996
) Facial emotion recognition after bilateral amygdala damage: differentially severe impairment of fear.
Cogn Neuropsychol
 
13
:
699
–745.
Critchley H, Daly E, Phillips M, Brammer M, Bullmore E, Williams S, Amelsvoort TV, Robertson D, David A, Murphy D (
2000
) Explicit and implicit neural mechanisms for processing of social information from facial expressions: a functional magnetic resonance imaging study.
Hum Brain Mapp
 
9
:
93
–105.
Dolan RJ, Fletcher P, Morris J, Kapur N, Deakin JFW, Frith CD (
1996
) Neural activation during covert processing of positive emotional facial expressions.
Neuroimage
 
4
:
194
–200.
Eckes T (
1986
) A BASIC program package for the analysis of sorting data.
Behav Res Methods Instrum Comput
 
18
:
472
–474.
Ekman P, Friesen WV (1976) Pictures of facial affect. Palo Alto, CA: Consulting Psychologist Press.
Gainotti G, Azzoni A, Gasparini F, Marra C, Razzano C (
1997
) Relation of lesion location to verbal and nonverbal mood measures in stroke patients.
Stroke
 
28
:
2145
–2149.
George MS, Ketter TA, Gill DS, Haxby JV, Ungerleider LG, Herscovitch P, Post RM (
1993
) Brain regions involved in recognizing facial emotion or identity: an oxygen-15 PET study.
J Neuropsychiat Clin Neurosci
 
5
:
384
–394.
George MS, Ketter TA, Parekh PI, Horwitz B, Herscovitch P, Post RM (
1995
) Brain activity during transient sadness and happiness in healthy women.
Am J Psychiat
 
152
:
341
–351.
Geschwind N (
1965
) Disconnexion syndromes in animals in man.
Brain
 
88
:
237
–294.
Goldman-Rakic PS, Selemon LD (
1997
) Functional and anatomical aspects of prefrontal pathology in schizophrenia.
Schizophr Bull
 
23
:
437
–458.
Irle E (
1990
) An analysis of the correlation of lesion size, localization and behavioral effects in 283 published studies of cortical and subcortical lesions in old-world monkeys.
Brain Res Brain Res Rev
 
15
:
181
–213.
Irle E, Peper M, Wowra B, Kunze S (
1994
) Mood changes after surgery for tumors of the cerebral cortex.
Arch Neurol
 
51
:
164
–174.
Jacobson R (
1986
) Disorders of facial recognition, social behaviour and affect after combined bilateral amygdalotomy and subcaudate tractotomy — a clinical and experimental study.
Psychol Med
 
16
:
439
–450.
Janke W, Debus G. (1978) Die Eigenschaftswörterliste EWL. Göttingen, Germany: Hogrefe.
Lane RD, Reiman EM, Ahern GL, Schwartz GE, Davidson RF (
1997
) Neuroanatomical correlates of happiness, sadness, and disgust.
Am J Psychiat
 
154
:
926
–933.
Lashley KS (1929) Brain mechanisms and intelligence. Chicago, IL: University of Chicago Press.
McClelland JL, Plaut DC (
1993
) Computational approaches to cognition: top-down approaches.
Curr Opin Neurobiol
 
3
:
209
–216.
Mesulam M-M (1985) Patterns in behavioral neuroanatomy. In: Principles of behavioral neurology (Mesulam M-M, ed.), pp. 1–70. Philadelphia, PA: Davis.
Mesulam M-M (
1998
) From sensation to cognition.
Brain
 
121
:
1013
–1052.
Morris JS, Friston KJ, Büchel C, Frith CD, Young AW, Calder AJ, Dolan RJ (
1998
) A neuromodulatory role for the human amygdala in processing emotional facial expressions.
Brain
 
121
:
47
–57.
Paradiso S, Robinson RG, Andreasen NC, Downhill JE, Davidson RJ, Kirchner PT, Watkins GL, Boles Ponto LL, Hichwa RD (
1997
) Emotional activation of limbic circuitry in elderly normal subjects in a PET study.
Am J Psychiat
 
154
:
384
–389.
Paradiso S, Johnson DL, Andreasen NC, O'Leary DS, Watkins GL, Boles Ponto LL, Hichwa RD (
1999
) Cerebral blood flow changes associated with attribution of emotional valence to pleasant, unpleasant, and neutral visual stimuli in a PET study of normal subjects.
Am J Psychiat
 
156
:
1618
–1629.
Pearlson GD, Petty RG, Ross CA, Tien AY (
1996
) Schizophrenia: a disease of heteromodal association cortex?
Neuropsychopharmacology
 
14
:
1
–17.
Phillips ML, Young AW, Senior C, Brammer M, Andrew C, Calder AJ, Bullmore ET, Perretti DI, Rowland D, Williams SCR, Gray JA, David AS (
1997
) A specific neural substrate for perceiving facial expressions of disgust.
Nature
 
389
:
495
–498.
Rapcsak SZ, Galper SR, Comer JF, Reminger SL, Nielsen L, Kaszniak AW, Verfaellie M, Laguna JF, Labiner DM, Cohen RA (
2000
) Fear recognition deficits after focal brain damage.
Neurology
 
54
:
575
–581.
Reiman EM, Lane RD, Ahern GL, Schwartz GE, Davidson RJ, Friston KJ, Yun L-S, Chen K (
1997
) Neuroanatomical correlates of externally and internally generated human emotions.
Am J Psychiat
 
154
:
918
–925.
Sanides F (1970) Functional architecture of motor and sensory cortices in primates in the light of a new concept of neocortex evolution. In: The primate brain (Noback CR, Montagna W, eds), pp. 137–208. New York: Appleton-Century-Crofts.
Schlaepfer TE, Harris GJ, Tien AY, Peng LW, Lee S, Federman EB, Chase GA, Barta PE, Pearlson GD (
1994
) Decreased regional cortical gray matter volume in schizophrenia.
Am J Psychiat
 
151
:
842
–848.
Stevens AA, Goldman-Rakic PS, Gore JC, Fulbright RK, Wexler BE (
1998
) Cortical dysfunction in schizophrenia during auditory word and tone working memory demonstrated by functional magnetic resonance imaging.
Arch Gen Psychiat
 
55
:
1097
–1103.
Talairach J, Tournoux P (1988) Co-planar stereotaxic atlas of the human brain. New York: Thieme.
Teasdale JD, Howard RJ, Cox SG, Ha Y; Brammer MJ, Williams SCR, Checkley SA (
1999
) Functional MRI study of the cognitive generation of affect.
Am J Psychiat
 
156
:
209
–215.
Weinberger DR, Berman KF, Zec RF (
1986
) Physiologic dysfunction of dorsolateral prefrontal cortex in schizophrenia.
Arch Gen Psychiat
 
43
:
114
–124.
Weniger G, Irle E, Exner C, Rüther E (
1997
) Defective conceptualization of emotional facial expressions during T2 signal enhancement of the right amygdala.
Neurocase
 
3
:
259
–266.
Whalen PJ, Rauch SL, Etcoff NL, McInerney SC, Lee MB, Jenike MA (
1998
) Masked presentations of emotional facial expressions modulate amygdala activity without explicit knowledge.
J Neurosci
 
18
:
411
–418.
Wowra B, Irle E, Peper M, Zeller WJ, Kunze S (
1989
) Evaluation of lesion volumes in neuro-oncology: basic data for local carrier-mediated chemotherapy.
Reg Cancer Treat
 
2
:
178
–183.
Yakovlev PI (
1959
) Pathoarchitectonic studies of cerebral malformations.
J Neuropathol Exp Neurol
 
18
:
22
–55.
Young AW, Aggleton JP, Hellawell DJ, Johnson M, Broks P, Hanley JR (
1995
) Face processing impairments after amygdalotomy.
Brain
 
118
:
15
–24.