Grand averages of waveforms generated by facial expressions at the reference electrodes, with an enlargement of the waveforms at the Pz electrode. EOG indicates electro-oculogram.
Amplitudes of N400 generated at the Pz electrode by serotonin transporter promoter polymorphism (5-HTTLPR) genotype and type of facial expression.
Scatterplot of subjects in the experiment ordered by serotonin transporter promoter polymorphism (5-HTTLPR) genotype. The graph shows the degree of shyness or behavioral inhibition (BI) and the N400 amplitude evoked by anger at the Pz electrode for each subject.
Battaglia M, Ogliari A, Zanoni A, Citterio A, Pozzoli U, Giorda R, Maffei C, Marino C. Influence of the Serotonin Transporter Promoter Gene and Shyness on Children’s Cerebral Responses to Facial Expressions. Arch Gen Psychiatry. 2005;62(1):85-94. doi:10.1001/archpsyc.62.1.85
Childhood shyness can predate social anxiety disorder and may be associated with biased discrimination of facial expressions of emotions.
To determine whether childhood shyness, or the serotonin transporter promoter polymorphism genotype, can predict participants’ visual event-related potentials in response to expressions of children of similar ages.
Study group drawn from an inception cohort of 149 subjects characterized 1 year before the present study by their degree of shyness.
Third- and fourth-grade schoolchildren.
Forty-nine of the inception cohort children, randomly selected.
Main Outcome Measures
Latencies and amplitudes of the N400 waveform in response to happy, neutral, and angry expressions.
Shyness predicted significantly smaller N400 amplitudes in response to anger (at Pz: P≤.04) and to a neutral expression (at Pz: P≤.047). Shyness was significantly different across the 3 genotypes, the SS genotype being associated with higher shyness levels (analysis of variance: F2,42 = 4.47, P≤.02; Tukey honestly significant difference, SS vs LL, P≤.01). An analysis of covariance showed that neither the type of expression nor the genotype per se influenced the N400 amplitudes, but a significant expression×genotype interaction was found (F4,72 = 3.57, P≤.01), sustained by the difference in amplitude of the SS and S carrier subjects compared with the LL subjects when exposed to the anger expression (Tukey honestly significant difference, P≤.02).
Children who manifest higher levels of shyness or have 1 or 2 copies of the short allele of the serotonin transporter promoter gene appear to have a different pattern of processing affective stimuli of interpersonal hostility.
Neuroimaging studies are beginning to clarify the relationship between the brain’s cortical and subcortical activity in regulating the emotional and cognitive functions of behavior. We know that the orbitofrontal, prefrontal, insular, temporal, cingulate, parietal, and occipital cortices are important neural network nodes in human anxiety,1- 3 and that the amygdala plays a pivotal role in the perception and evocation of emotions related to threat and danger, such as fear or anger.3,4 The proportion of cortical and subcortical activation differs depending on the type of emotional and cognitive tasks. For instance, when normal subjects process perceptually angry or fearful faces, a bilateral amygdala response is observed, whereas labeling of the same expressions is associated with an attenuated amygdala response and an increased right prefrontal cortex response.5
Moreover, the patterns of activation and deactivation of brain regions in response to affective stimuli or in the course of mildly anxiogenic tasks vary quantitatively across subjects and can be predicted in part by individual differences in proneness to experience negative emotionality and anxiety, and by some polymorphic genes that influence behavior.
Several studies (see Tillfors et al1 for review) of perceptually induced anxiety concord broadly by showing altered patterns of cortical activation in patients with diverse anxiety disorders compared with control subjects. Abnormal responses have been reported in cortical areas involved in the emotional evaluative (eg, visuospatial) processes, namely, the secondary visual,1,6 parietal,1 and temporal,7 in addition to the prefrontal and orbitofrontal,6 cortices.
In social anxiety disorder, the investigation of brain activity in response to affective stimuli of social relevance and in simulated anxiogenic social contexts appears of special interest.
In adult patients with social phobia, compared with control subjects during simulated public speaking, a decreased cortical (secondary visual, parietal, retrosplenial, temporal, and insular cortex) activity occurs typically in association with increased amygdala activation (regional cerebral blood flow).1,6 Likewise, exaggerated amygdala activation has been reported in controlled studies of adults with social phobia exposed to angry or contemptuous8 and neutral9 facial expressions, which are usually categorized as mildly hostile or ambiguous.10
There is also initial, but consistent, evidence of biased processing and discrimination of the facial expressions of emotions in adults and children at heightened risk for social phobia. A study of visual attention showed that adults with social phobia avoid salient facial features when they watch facial expressions,11 while children with high indexes of social anxiety or shyness have biased recognition of emotions, particularly the neutral and the angry expressions.12,13
The brain responses to basic elements of social communication in children at risk of developing social anxiety or avoidant personality disorder can thus help clarify the developmental pathways to social phobia.
A temperamental disposition toward the avoidance of novel and uncertain situations together with a set of behaviors that indicate shyness and discomfort in social interactions are comprehensively named childhood shyness, or behavioral inhibition (BI).14 Children with high indexes of shyness-BI are at a heightened risk of developing anxiety disorders, in particular social phobia,15 and subjects who fall within the BI–social phobia developmental continuum show specific patterns of neurophysiologic responses to pictures of facial expressions. Adults who had been categorized as behaviorally inhibited at the age of 2 years exhibited a higher amygdala activation in response to unknown vs familiar faces16 compared with adults who were uninhibited as children, consistent with the notion that novel or ambiguous environmental stimuli of potential biological relevance activate the amygdala.17 Turning to genes that can influence the neurobiological bases of the processing of emotions, 2 common alleles, the short (S) and the long (L), in a variable repeat sequence of the serotonin transporter (5-HTT) promoter polymorphism (5-HTTLPR) on human chromosome 17q11 have been differently associated with greater amygdala activity in response to angry or fearful faces18 in healthy adults. The L and S alleles influence 5-HTT transcription activity,19 with the S allele conveying reduced transcription, lower transporter levels, and diminished serotonin uptake. The presence of 1 or 2 copies of the S variant influences predisposition to anxiety, avoidant behaviors, and interpersonal negative emotionality according to several, but not all, genetic association studies of adults,20 while a recent study of childhood shyness-BI21 found an association in the opposite direction (ie, with the LL genotype).
The cerebral visual event-related potentials (ERPs) are scalp potentials that occur within a few hundred milliseconds after the presentation of a visual stimulus. The ERP waveforms contain components22- 24 that span a continuum between the earlier, “exogenous” potentials (reflex responses controlled by the physical properties of an external eliciting event) and the later, “endogenous” potentials (manifestations of information processing determined more by the nature of interaction between the subject and the event). Meta-analytic evaluations of ERP studies in twins show substantial heritability of some late (the P300 and N400 waveforms) ERP components.25
Children from the age of 10 years show the same ERP late components26- 28 found in adults when they look at pictures of facial affects.29 These include a characteristic negative waveform that occurs at about 400 milliseconds (N400) believed to reflect specific cognitive processing of human facial information30,31 in the absence of verbal-semantic information.32 More broadly, the N400 has been claimed to reflect a temporal correlate of a corticoamygdala pathway of emotional processing.33,34 In keeping with positron emission tomographic and functional magnetic resonance imaging studies that showed activation of primary and secondary visual areas and of the somatosensory cortex during the act of processing face expressions,3 ERP studies suggest that expression identification is associated with a centroparietal effect, indexing attention and motivation processes of adaptive significance.29- 31
The present study analyzes the N400 visual ERP responses of children to different emotional facial expressions. We hypothesized that individual differences in shyness-BI and genetic variability at the 5-HTTLPR could predict a pattern of diminished cortical activation in response to hostile compared with neutral or prosocial expressions, which could be put into a relationship with the finding of biased recognition of expressions of emotions in shy children,12,13 and with the patterns of central nervous system activation found by brain-imaging studies of social phobia in tasks of perceptually induced anxiety.1
One year before this study (time 0), an inception cohort of 149 schoolchildren were characterized by their degree of shyness and ability to discriminate facial expressions of emotions.12 Resources were available to study a fraction of the original sample through interviews and electrophysiologic evaluation; 70 of the 149 children were therefore randomly selected and invited to participate in the present study. After a complete description of the study, 55 of the 70 invited children and their parents agreed to participate, and parents signed a declaration of informed consent. The procedures of this study were accepted by the ethical committee of the participating institutions.
Six of the 55 participating children were not included in the final sample for different reasons, such as minor physical illnesses on the day of the ERP recording, or unavailability of an adult person to accompany the child to the laboratory. This left 49 white children of Italian ancestry with normal or corrected-to-normal visual acuity to take part in the study. Post hoc comparisons showed no differences in demographic and psychometric characteristics between 49 participating vs 100 nonparticipating children (Table 1). Although there were 49 participants, results are shown for numbers of participants that vary between 34 and 49 because of removal of subjects who had artifacts like eyeblinks in some recordings, or who had an unsuccessful DNA collection (which occurred in 4 children).
At time 0, the children’s degree of shyness-BI was evaluated by a questionnaire that was filled in by appropriately trained teachers, and by direct observation of the number of spontaneous comments made in the presence of an unfamiliar adult, based on previous descriptions of children with BI.12,14 The questionnaire included a set of items seeking to identify temperamental disposition to BI and symptoms of possible social anxiety disorder proper, and included the Italian translations of the Stevenson-Hinde and Glover Shyness to the Unfamiliar,35 Cloninger and coworkers’ Harm Avoidance Scale,36 and the Liebowitz Social Anxiety Scale37 adapted for children. The teachers were asked to base their judgment on the instructions received from our group at seminars and on direct observation. The scales had satisfactory indexes of internal consistency (Cronbach α, 0.89, 0.85, and 0.88, respectively), moderate to high cross-correlations (Harm Avoidance Scale–Shyness to the Unfamiliar, 0.64; Harm Avoidance Scale–Liebowitz Social Anxiety Scale, 0.62; Shyness to the Unfamiliar–Liebowitz Social Anxiety Scale, 0.63; all P≤.01), and negative covariation with children’s number of spontaneous comments.12 Moreover, higher scores on the Liebowitz Social Anxiety Scale predicted a significantly higher number of misclassifications of expressions.12 Therefore, we used part of these scales’ items to build a more succinct index of BI-shyness.
A principal component analysis of the items collected by questionnaire at time 0 provided 3 factors with the following eigenvalues (and percentages of explained variance): 11.99 (59.9%), 2.1 (9.8%), and 1.2 (6.1%). Among the items belonging in the first factor, the following 7 had maximal factorial loading (range, 0.80-0.89), maximal reciprocal correlations (r = 0.58-0.86, all P≤.001), and maximal internal consistency (Cronbach α≥0.89): (1) the child fears or avoids being at the center of attention; (2) the child fears or avoids raising his or her hand and answering questions in front of the class; (3) the child fears or avoids unfamiliar people; (4) the child speaks easily with unfamiliar adults (negative loading); (5) the child tends to avoid new visitors and strangers; (6) the child plays readily with new children (negative loading); and (7) when the child meets new children, it takes him or her a long time to start talking. These items were selected to build a concise index of shyness-BI that could be associated with the variables used in the present study. The shyness-BI index ranged from 0 to 30, with a mean value of 10.86 ± 6.6 (minimum, 0; maximum, 26; skewness, 0.34 [SE, 0.33], and kurtosis, −0.76 [SE, 0.67]); it was normally distributed (Shapiro-Wilk W = 0.96, P = .10) without significant sex differences (mean for 26 boys, 9.5 + 6.0, vs mean for 23 girls, 12.8 + 5.9; t47 = 1.93, P = .06).
Children’s behavioral profiles were also assessed on the day of the ERP experiment, based on 3 different measures. First, while electrodes were placed for the ERP recording (average time to complete, 20 minutes; range, 17-22 minutes), the number of the spontaneous comments made by every participant was counted. Second, after the ERP recording, all mothers and children were interviewed individually by trained clinical psychologists with the Italian version of the Schedule for Affective Disorders and Schizophrenia for School-age Children (K-SADS)38 interview to collect the children’s lifetime DSM-IV symptoms of social phobia, simple phobia, depression, enuresis, generalized anxiety disorder, separation anxiety disorder, panic disorder, attention-deficit/hyperactivity disorder, obsessive-compulsive disorder, conduct disorder, oppositional disorder, and tic disorder. The presence of symptoms of DSM-IV childhood disorders was established on the basis of blind review sessions chaired by a senior child psychiatrist (Dr Marino) of all the information obtained by both the mother and the child’s K-SADS interview protocols. All children were included in the study, regardless of the presence or absence of a full psychiatric diagnosis. Third, mothers filled in the Child Behavior Checklist 4-1839 and answered questions on parental occupations to calculate the socioeconomic status on the basis of the Hollingshead scale.40
Genomic DNA was extracted from mouthwash samples collected in 4% sucrose by means of a reagent for isolation of genomic DNA (DNAzol Genomic DNA Isolation Reagent; Molecular Research Center Inc, Cincinnati, Ohio).
The polymorphism in the transcriptional control region upstream from the 5-HTT coding sequence (5-HTTLPR) was analyzed by polymerase chain reaction according to the method reported by Lesch et al19 and Heils et al.41 Two fragments were generated: a short variant (S) of 484 base pairs and a long variant (L) of 528 base pairs.
All amplification reactions were performed on a thermocycler (Mastercycler; Eppendorf, Milan, Italy). The amplified products were analyzed on 2% agarose gels.
For 4 of the 49 children, the DNA extraction was unsuccessful, and therefore genetic data were available for only 45 subjects.
We chose to use standardized faces of other children of similar age (models aged 8-9 years), instead of adults, for 2 reasons: first, schoolchildren spend most of their time among other children, not adults; second, socially anxious children rate rejection and teasing from peers among the most feared situations.42 Stimuli consisted of 6 black-and-white pictures of a boy and a girl standardized for size, contrast, and luminosity, displaying 3 emotions: joy, anger, and a neutral expression. The boy pictures were drawn from a collection of Linda Camras, PhD.43 The girl pictures were selected under the direction of Camras from a pool of pictures that had been developed by showing the model Ekman and Friesen’s44 prototypical pictures as a reference. In accordance with Camras’ standard-validation procedure, the girl pictures were correctly classified in more than 80% of evaluations by 20 undergraduates of both sexes.12 Stimuli appeared on a 38-cm monitor placed at a distance of 41 cm and were presented in an oval aperture that occluded sex-specific features. Children were told that this was a video game in which they would get a gift if they carefully followed the instructions and performed well.
Joy was included as a prototype of a prosocialization expression that is almost never misidentified by children12 or neural networks that replicate human emotional-cognitive tasks.45 The joy expression is processed by brain networks at least partially independent from those involved in processing hostile expressions3 and evokes neurofunctional responses that can be in part predicted by the extroverted, but not by the introverted, temperamental dimensions.46
The neutral expression was included because it is usually categorized as mildly hostile or ambiguous10 and was significantly more often misclassified by children with higher indexes of social anxiety12 at time 0.
The angry expression was included because it is significantly more often misclassified by children with higher indexes of social anxiety,12,13 and it has been found to elicit heightened subcortical amygdala activity in adults with social phobia8 and in adults with 1 or 2 copies of the 5-HTT S allele.18
On each trial the children were first presented with a child’s face (total time on screen, 1300 milliseconds), which they were instructed to watch carefully until a blue circle appeared superimposed around the center of the picture. As soon as a blue circle appeared (700 milliseconds after the appearance of the stimulus), they had to click a mouse. Thus, the ERPs relevant to this study were all generated before the motor task, which was merely set up to stimulate children’s participation and attention. The monitor screen remained dark between trials for periods that varied randomly from 1200 to 1600 milliseconds. The stimuli were presented to all the children in a sequence that alternated male and female pictures and that avoided close repetition of the same expression. To simulate a real video game and reinforce participation, a screen picture with increasing scoring values appeared about every 6 pictures. Each stimulus was presented 20 times to ensure sufficient ERP acquisition (total, 120 presentations in a complete session).
Every child was exposed to a preexperiment trial of 6 pictures not belonging to the same set used for the experiment to make sure she or he understood the procedure well.
Each child received a gift of a value equivalent to €30.
As in other studies of ERP and facial expressions of children and adults,26,27,29,47 electroencephalographic activity was recorded at sites Fz, C3, Cz, C4, and Pz of the 10-20 system with the use of silver–silver chloride electrodes referred to linked mastoids with an amplifier (Neuroscan SynAmp; Neuroscan Labs, Sterling, Va), with head preamplified gain 150 and acquisition software (SCAN, version 4.2; Neuroscan Labs). The ground electrode was attached to the forehead. Electro-oculographic activity was recorded from electrodes placed above and below the right eye.
Electrode impedance was maintained below 5 kΩ. The electroencephalogram and electro-oculogram were amplified (gain 500), analogically bandpass filtered (1-30 Hz), digitized, and acquired at a 1000-Hz sampling rate. Electroencephalogram and electro-oculogram epochs between –50 and 1300 milliseconds from the stimulus onset were obtained by means of different trigger codes for each image, allowing for later off-line artifact rejection, sorting, and digital averaging with the Neuroscan EDIT software (Neuroscan Labs). All epochs from all electrodes were rejected if affected by artifacts (greater than +65 μV or less than –65 μV between –50 and 700 milliseconds). The ERP averages were constructed from artifact-free epochs for each trigger code and for each electrode. Amplitudes were measured according to the distance between peaks and troughs for each identified waveform.
Table 2 summarizes the cross-correlations of the behavioral indexes used in the study.
The shyness-BI index significantly predicted the number of spontaneous comments made by children (mean ± SD, 3.39 ± 4.87; range, 0-17) while the electrodes were being placed on their scalps, and the number of lifetime symptoms of social phobia (mean ± SD, 2.26 ± 2.72; range, 0-8) collected by the K-SADS interview, but no other symptoms of mental disorders assessed with the K-SADS (the prediction closest to significance pertained to separation anxiety, with P = .18). Moreover, among the 9 narrow-band dimensions of problem behaviors measured by the Child Behavior Checklist, the Withdrawn scale (possible scores, 0-16; mean ± SD, 3.02 ± 2.81; range, 0-11) was the only one to correlate significantly with the shyness-BI index.
Analyses of the waveforms generated by the facial expressions (Figure 1 and Table 3) showed an enhanced late negativity occurring after stimulus presentation at a mean of 256 milliseconds for the Fz, 368 milliseconds for the Cz, and 394 milliseconds for the Pz electrode: these 2 latter waveforms were identified as N400 with an amplitude that followed an anteroposterior gradient midline. According to analysis of variance (ANOVA) on 5 electrodes (Table 3), the mean N400 amplitudes at Pz and Cz were not significantly different from each other, and the mean N400 amplitudes at C3 and C4 were not significantly different from those obtained from the Cz and Pz electrodes. Such centroparietal maximum amplitudes for a face-evoked ERP N400 wave can be found in other studies following similar methods.29,48 Although the emotional expressions of anger and joy tended to elicit slightly larger N400 amplitudes than the neutral expression across the different electrodes, no significant difference for N400 amplitude or latency for type of expression was disclosed by repeated-measures ANOVA with the type of expression as factor, carried out on the 49 children participating in the study at the reference electrodes (Table 3). However, waveforms at Pz were characterized by an enhanced late negativity occurring exactly at the 400-millisecond expected time, with separation of 3 waveforms (corresponding to the 3 expressions) that was visually apparent from the very beginning of the ERP waves (Figure 1; see section enlargement of the Pz electrode).
To ascertain the degree of variability of measures across trials, we performed a repeated-measures ANOVA with 2 factors: (1) type of expression (on 3 levels) and (2) time (on 2 levels: first block, encompassing the first 50% of repetitions of each expression; and second block, with the remaining repetitions) for both amplitudes and latencies at all electrodes. The results showed the absence of significant effects of time or time×expression factors for the N400 waveforms (Table 3: values and statistics are given for Pz; for other electrodes, data are available from authors on request).
We also found no effect of sex or sex×expression on the N400 characteristics (Table 3), and therefore subjects of both sexes were pooled in all subsequent analyses.
To test the hypothesis that shyness-BI predicts N400 characteristics when children are presented with other children’s expressions of emotions, 2 multiple linear regression procedures were performed separately for N400 amplitudes and latencies, with the N400 amplitudes (or latencies) for each expression (joy, neutral, anger) at the different electrodes (Fz, Cz, Pz, C3, C4) as dependent variables, and the degree of shyness-BI as the independent variable.
Children’s shyness-BI predicted the N400 amplitude for anger and the neutral expression (Table 3) at Pz, Cz, and C4, always in the direction of a reduction of amplitude, but not the amplitude for the joy expression on any of the electrodes. Similarly, no significant prediction for any of the expressions was provided by regression applied to the N400 latencies for any of the electrodes.
The allelic frequencies were 49 (54%) for the L allele and 41 (46%) for the S allele (Table 4); both were in Hardy-Weinberg equilibrium, without sex-related differences in distribution (respectively, χ22 = 1.09, P = .57; and χ21 = 0.43, P = .51). All ANOVAs dealing with the 5-HTTLPR genotype were conducted with the 3 genotypes separately and with the classification19 that combines the LS and SS genotypes into “S carriers.”
While socioeconomic status, grade, and sex were equally distributed across genotypes and in LL vs S carrier subjects, children with the SS genotype and S carriers were slightly, but significantly, younger (Table 4).
A polynomial univariate ANOVA, where the factor was the genotype at the 5-HTTLPR gene and the dependent variable was the shyness-BI index, showed that shyness was significantly different across the genotypes, with the SS genotype being associated with higher shyness-BI index (Table 4); similarly, S carriers vs LL subjects showed a trend toward significantly higher shyness-BI index (Table 4).
In light of the foregoing relationships, and after checking for the homogeneity of covariance matrices (Bartlett χ2 not significant on all dependent variables and the covariate; box M = 19.98, χ220 = 16.15, P = .70), we performed analysis of covariance where the independent variables were the genotypes at the 5-HTTLPR, and the dependent variables were the N400 amplitudes evoked by facial expressions at the Pz electrode, the repeated-measures factor was the expression (joy, neutral, anger), and the covariates were the shyness-BI index and age. Neither the type of expression nor genotype per se influenced the N400 amplitudes, but a significant expression×genotype interaction was found (Table 4). The post hoc comparison showed that the significance was sustained by smaller amplitude elicited by anger in the SS subjects compared with the LL subjects (Figure 2, Figure 3, and Table 4). The analysis of covariance, contrasting LL subjects with S carriers, provided very similar results, ie, the absence of genotype or expression effects, but a significant expression×genotype interaction effect, sustained by smaller N400 elicited by anger in S carriers (Table 4).
These results for the anger expression could not be explained by habituation effects on the different expressions that could have occurred between the first and the second block of repetitions, as shown by repeated-measures analysis of covariance on amplitudes with 3 factors: (1) time (on 2 levels: first and second blocks of repetitions), (2) expression, and (3) genotype, with shyness-BI index and age as covariates (Table 4).
Because the results of the ERP analyses suggested a role for both shyness-BI index and the 5-HTTLPR genotype in determining different patterns of information processing of hostile (angry) and/or neutral facial expressions, we tested post hoc whether the 5-HTTLPR genotype could predict a biased discrimination of expressions in the trial completed 1 year before the ERP recordings.12 A linear regression was then performed on the total number of misclassifications made at time 0 by the 45 children whose genotyping was available, with the 5-HTTLPR genotype (0, 1, or 2 S alleles) as the independent variable. The number of misclassifications was predicted significantly, in keeping with the previous findings. Similarly, an independent samples, 2-tailed t test of the number of misclassifications between S carriers and LL subjects showed a trend toward significantly worse performance for the former (Table 4).
To evaluate whether the 5-HTTLPR genotype had greater power to predict the phenotype of shyness-BI index, or the endophenotype of N400 amplitude at Pz in response to anger, we specified a linear trend, with weights of –1, 0, and 1 for the 0, 1, and 2 S alleles, and a quadratic trend, with weights of –1, 2, and −1 for the 0, 1, and 2 S alleles, respectively, and then entered these coded vectors simultaneously in the regression equations49,50 to predict shyness-BI index and the N400 amplitude. Results showed that the linear solution was the only one retained by the analyses, with a 25% greater R2 for the N400 amplitude (Table 5).
Shyness-BI index and the presence of 1 or 2 copies of the 5-HTTLPR S allele predicted smaller ERP N400 amplitudes in response to overtly hostile and neutral facial expressions at centroparietal regions, which are interconnected with emotion processing networks encompassing the amygdala and the prefrontal cortex.3,51- 53 Inasmuch as the N400 reflects the neural processes involved in recognition of facial expressions,29- 31 a smaller amplitude in carriers of 1 or 2 copies of the 5-HTTLPR S allele suggests diminished cortical involvement and predicts partially impaired reading29,54 in response to some facial affects, consistent with the findings of biased discrimination of the hostile and ambiguous expressions at time 0 by the same subjects.12 Smaller ERP waveforms in response to angry faces have been reported in a controlled study of posttraumatic stress disorder and have been interpreted as a relative reduction of cortical activity, co-occurring with heightened subcortical activity,54 in response to hostile social stimuli. Indeed, a relatively decreased activity in the cortices that are involved in emotional evaluative processes (including the secondary visual and parieto-occipital areas) has been demonstrated to co-occur with exaggerated activation of the amygdala in adults with social phobia during simulated social interaction investigated by functional brain imaging techniques.1,6,7 A decreased cortical—as opposed to subcortical—perfusion is contrary to the pattern of activation in normal controls1 in conditions of simulated social interaction. Likewise, adults with social phobia show heightened subcortical, and specifically amygdala, activation in response to neutral9 and angry8 facial expressions, and carriers of 1 or 2 copies of the S allele have an increased amygdala response18 to anger. The increased subcortical, and decreased cortical, activation in response to emotional stimuli of social relevance has been interpreted as the excessive involvement of a phylogenetically “older” system1 of recognition of dangerous, novel, or ambiguous stimuli.17 Heightened amygdala activation in response to novel faces in adults who had been categorized as shy in childhood16 may be seen as broadly consistent with these data.
Consistent with the finding that the neurofunctional responses to happiness expressions can be partly predicted by the extraverted, but not by the introverted, temperamental dimensions,46 childhood shyness-BI and/or the 5-HTTLPR genotype did not predict the N400 potentials evoked in response to a prosocial (joy) expression.
The greater R2 found by regressing the 5-HTTLPR genotype on the endophenotype of N400 than on the shyness-BI phenotype is consistent with the suggestions55 that the effect of functional genetic polymorphisms is more robustly assessed by assays of brain physiology than with behavioral phenotypes. However, while the association between the SS 5-HTTLPR genotype or S carrier status and shyness-BI is consistent with most studies of adult human emotionality,20 a study found the LL genotype associated with childhood shyness-BI.21
Although these results need replication, they show that genetic variation at 5-HTTLPR contributes to shape the N400 waveform, a possible index of complex neuronal activity that occurs when expressions of emotions are being processed.29- 32,56 A pattern of decreased cortical activation in response to some specific social cues (including peers’ nonverbal signs of refusal or neutrality) may then constitute a heritable basis for biased discrimination of some forms of socially relevant information, which in turn may hamper social interactions and ultimately reinforce a child’s disposition to shyness-BI.
Shy children have been shown to provide relatively distinct physiologic responses in a variety of contexts.57,58 These data suggest that a biased pattern of processing emotional information of social relevance can be recognized and characterized psychophysiologically early in life in children with heightened indexes of shyness-BI or 1 or 2 copies of the 5-HTTLPR S allele.
Correspondence: Marco Battaglia, MD, Department of Neuropsychiatric Sciences, Istituto Scientifico San Raffaele, 20 via Stamira d’Ancona, 20127 Milan, Italy (firstname.lastname@example.org).
Submitted for Publication: February 24, 2003; final revision received May 27, 2004; accepted June 9, 2004.
Funding/Support: This study was supported by the Italian Ministry of University and Research, Rome (Co-Fin grant 11/2001-113555_004 awarded to Dr Battaglia), and the National Alliance for Research in Schizophrenia and Depression Independent Investigator Award, Great Neck, NY (Dr Battaglia).
Acknowledgment: We thank Massimo Molteni, MD, Flora Binaghi, MSc, Luca Bigotta, PhD, Linda Camras, PhD, Andrea Fossati, MD, Frederica Villa, MSc, Livia Brenna, Nino Stillitano, Silvana Villa, MSc, Raffaella Braga, MSc, Kim Sommerschield, PhD, and all of the children, parents, and teachers who took part in this study.