Previous studies showing the association of exposure to peer (PeVA) and parental verbal abuse in childhood with structural alterations in the young adult brain suggest functional changes in adolescence. In this functional MRI study, we investigated the effects of exposure to PeVA, during elementary and middle school periods, on brain response to emotional words, in high school students.
An emotional Stroop task consisting of swear, negative, positive, and neutral words was performed during functional MRI scan for 23 subjects who were divided into low- and high exposure groups to PeVA.
High-PeVA group had a higher depression score, greater left ventrolateral prefrontal cortex (VLPFC) activity, and higher left VLPFC-left hippocampus connectivity in swear word conditions. The VLPFC activity and left VLPFC-left hippocampus connectivity was negatively related to the severity of anxiety and depressive symptoms, respectively.
These preliminary findings support the hypothesis that exposure to PeVA, during childhood, is an aversive stimulus associated with meaningful functional change in emotional regulation network, showing hypersensitivity to swear words, at middle adolescence.
Childhood maltreatment is closely linked to increased vulnerability to behavioral problems related to self-control, such as impulsivity, substance abuse, and depression.
Although verbal abuse creates detrimental effects, adolescents seems to use swear words without awareness of its seriousness. Approximately two out of three Korean middle and high school students use swear words more than once a day without recognizing the exact meaning.
In neuroimaging studies with victims of childhood abuse, the disruption of the frontolimbic network (FLN) has been a frequently reported problem,
To verify our hypotheses, an emotional Stroop paradigm was developed to induce an implicit emotional processing to swear words. During this task, functional magnetic resonance imaging (fMRI) was performed to measure the brain activities, especially focused on the VLPFC, amygdala, and hippocampus and the interaction among these regions.
Our study had three main purposes: 1) ascertain the effects of the implicit exposure to swear words, compared with other emotional effects, on brain function, 2) explain how previous PeVA modulates brain activities and the functional connectivity within the FLN, focused on the VLPFC, during the implicit processing of emotional words, especially swear words, 3) explore the relatedness between these changes and maladaptive behaviors. This study might provide the background needed to understand the effects of swearing on brain networks and modulatory impacts of previous PeVA experiences.
To recruit subjects that would represent the distribution of peer verbal abuse (PeVA) in the general student population, we first introduced an investigation of language use in Korean adolescents. In our study, only male students were recruited since the prevalence of verbal bullying is higher males than in females
Exposure to parent or peer verbal abuse during elementary and middle school periods was assessed using the Verbal Abuse Questionnaire (VAQ), which evaluates 15 kinds of verbal abuse. A VAQ score of 40 or higher represents substantial exposure to PeVA.
We recruited 32 first year healthy male students from several high schools. All subjects were asked to complete a survey that contained questions related to behavior measurements using web-based questionnaires. Thirty participants completed the survey and 25 of them finally agreed to have fMRI scanning.
All 25 subjects accomplish our experiments along identical process. The experimenters explain fMRI task paradigms, and subjects conducted fMRI scanning. Also, two psychologists interviewed with participants to get diagnostic information of psychiatric disorders using the Kiddie-Schedule for Affective Disorders and Schizophrenia-Present and Lifetime Version-Korean Version (K-SADS-PL-K). After psychologists interviewed the participants, three authors in our study (one child psychiatrist and two psychiatrists) shared the information of participants to exclude participants who have psychiatric disorder. The Wechsler Intelligence Scale for Children-Revised (WISC-R) was used on participants to evaluate the intelligence of the subjects in the study. Among the 25 subjects, one subject with teeth braces was excluded and the data of one subject who exhibited much movement during fMRI due to rhinitis was excluded in the analysis. Finally, we obtained fMRI results from 23 subjects. All subjects joined voluntarily, and we received written informed consent from both the subjects and their parents for each step of study (behavioral measurements and fMRI). The study was approved by the Ethics Committee of the Korea Advanced Institute of Science and Technology (KAIST).
A modified Emotional Stroop task was designed to measure the implicit processing-related brain activity to emotional words. The task consisted of four different conditions of word stimuli: swearing (SWE), negative (NEG), positive (POS) and neutral (NEU). To choose appropriate word stimuli, sixty words per each condition were screened. Swear words were selected from a survey on the use of language in adolescence by the National Institute of the Korean Language in 2011.
During the fMRI session, the subjects performed the modified Emotional Stroop task. The four different conditioned words (SWE, NEG, POS and NEU) with three different colors (red, blue, and yellow) were represented for 0.75 s followed by a black screen for 3.25 s in a pseudorandomized order; subjects were asked to press buttons of an MRI-compatible response box with the right index, middle and ring finger according to color of the words. In addition, 45 color words with the same color (COLOR) were inserted to maintain attention during the intervals of the target stimuli.
Reaction time and error rate date were analyzed using 2 by 4 repeated measures one-way analysis of variance (RANOVA) to measure the effects of group, condition and group by trial. Incorrect responses and individual delayed response time (>2SD from a subject's mean) were excluded for reaction time analysis.
A total 200 blood oxygen level dependent based functional MRI images were acquired at 3.0 Tesla (Siemens) with 12 channel head coil using T2*-weighted gradient echo, echo-planar imaging pulse sequence (35 axial slices, TR=2200 ms, TE=30 ms, flip angle=90 degrees, FOV=192×192 mm, acquisition matrix=64×64, slice thickness=3 mm, voxel size=3×3×3.3 mm). A high resolution structural image was obtained with a T1 weighed sequence (TI=900 ms, TR=1800 ms, TE=2.52 ms, Flip angle=9 degrees, FOV=256×256 mm, slice thickness=1 mm, voxel size=1×1×1 mm) before fMRI scanning.
All images were preprocessed using FMRI Expert Analysis Tool (FEAT) in the FMRIB Software Library (FSL).
After preprocessing, single and group level fMRI analyses were performed with FEAT in the FMRIB Software Library (FSL).
Additionally, an unpaired t-test applying the mixed effects model was performed using the parameter estimates of three emotional contrasts (SWE>NEU, NEG>NEU, and POS>NEU) of each subject to assess differences between high-PeVA and low-PeVA group, and frequency of swear words exposure was included as covariate in this analysis.
In the study, we used a beta series correlation method
The results of the Mann-Whitney U test of the 23 subjects who participated in fMRI scanning are shown in
In addition, reaction time and error rate during emotional Stroop task were evaluated using RANOVA. The data of 22 subjects (10 high-PeVA and 12 low-PeVA) were included and one high-PeVA data was missed by the error of button box. The trial effect on reaction time was significant [F (3, 60)=5.518, p=0.002]. In post hoc paired t-test, all emotional stimuli showed a delayed response time compared to NEU (all ps<0.05) (
The activity of several brain regions that are related to emotion processing showed hyper-activation during SWE compared to NEU (Z>2.3, corrected p<0.05) (
In the high-PeVA group, SWE stimuli induced significant activation in several regions including the bilateral VLPFC, left insular, bilateral amygdala, bilateral dorsolateral prefrontal cortex (DLPFC) and intracalcarine cortex (SWE>NEU contrast, Z>2.3, corrected p<0.05) (
To find the significance of the left VLPFC hyper-activation during swear word processing, the correlations between various behavioral measurements and the left VLPFC activity were evaluated. In the high-PeVA group, the activity of the left VLPFC was negatively associated with an anxiety trait (STAIT) (spearman's rho=-0.705, p=0.015) (
We assessed group differences in the left VLPFC–left hippocampus using RANOVA. The group effect on the left VLPFC–left hippocampus was significant [F (1, 21)=7.777, p=0.011], and the effects of trials and group by trial interaction did not have a significant effect on left VLPFC–left hippocampus connectivity [F (3, 63)=1.565, p=0.21 and F (3, 63)=1.032, p=0.39, respectively]. In post-hoc analysis using a t-test with Bonferroni correction for multiple comparisons (corrected p=0.05/4 or 0.0125), the high-PeVA group had a significantly higher left VLPFC–left hippocampus connectivity than the low-PeVA group in SWE conditions [t (21)=2.759, p=0.012] and a trend level difference in connectivity in NEG compared to the low-PeVA group [t (21)=2.323, p=0.03] (
The RANOVA analysis revealed a significant effect of group [F (1, 21)=4.801, p=0.04] on left VLPFC-left amygdala connectivity. There was no significant effect of trials [F (3, 63)=1.520, p=0.22] and group by trials interaction [F (3, 63)=0.875, p=0.46]. In post-hoc analysis, there was no significant group difference in functional connectivity between the left VLPFC and left amygdala among SWE, NEG, POS and NEU stimuli. There was only a trend of higher connectivity in the high-PeVA group than in the low-PeVA group in NEG and NEU contrast [t (21)=2.054, p=0.053 and t (21)=2.099, p=0.048, respectively] (
There was no significant effect of group [F (1, 21)=2.916, p=0.1], trials [F (3, 63)=0.741, p=0.53] and group by trials [F (3, 63)=0.539, p=0.66] on left amygdala–left hippocampus connectivity (
To find the possible relationship between the changes of functional connectivity and behavioral measurements, we evaluated the correlations between the left VLPFC-left hippocampus connectivity and behavioral measures. Within several behavioral measures, the ratings of BDI were highly correlated with the left VLPFC-left hippocampus connectivity in the high-PeVA group both at SWE and NEG conditions (Spearman's rho=-0.867, p=0.001 and Spearman's rho=-0.610, p=0.046, respectively). However, in the low-PeVA group, the left VLPFC-left hippocampus connectivity during negative emotional word processing did not show any significant correlation with BDI (
In our study, although subjects had no overt psychiatric symptoms, adolescents in the high-PeVA group presented more depressive symptom scores, and had a higher frequency of previous PaVA experiences compared to low-PeVA group. Moreover, the high-PeVA group had greater activity of the VLPFC, negatively related to the anxiety trait, and higher functional interaction between the left VLPFC and left hippocampus, negatively related to the severity of depressive symptoms, during implicit processing to swear word. Our preliminary results, showing the exposure to PeVA links to functional changes in the left frontolimbic network which are related to psychiatric symptoms, suggest that swear words from peers can harm children even if the hurt was unintended or not overtly used.
In behavioral results, although frequency of exposure to swear words showed no group difference, PeVA experiences were closely related with PaVA experiences. A previous study reported that parental maltreatment can be a risk factor for victimization by peers.
In the fMRI analysis, the brain regions engaged in emotion processing,
Within various brain regions that were activated in SWE>NEU, the VLPFC was significantly more activated in high-PeVA group than low-PeVA group. In recent meta-analysis, the left inferior frontal gyrus including the VLPFC was activated by basic negative emotional stimuli; disgust and anger stimuli induced a higher level of activation in this region compared to sadness.
In previous fMRI studies, increased amygdala-responsiveness to negative stimuli was commonly reported in the victims of childhood maltreatments.
During swear word processing, functional connection between the left hippocampus and the left VLPFC was more enhanced in the high-PeVA group than in the low-PeVA group. The hippocampal region and the lateral prefrontal cortex are anatomically connected;
There are some limitations that should be noted. First, our sample size was rather small for evaluating correlations between psychiatric symptoms and activity or connectivity of the brain regions. However, we recruited adolescents who had recently graduated middle school, a developmentally sensitive period,
In spite of these limitations, our preliminary results indicated that the VLPFC and left VLPFC–left hippocampus network were important in processing swear words in high-PeVA adolescents. The victims of PeVA seems to require more resources to control the stimuli that were related to previous negative experiences, even if it is implicitly presented, and the disruption of the regulatory processes may lead anxiety and depressive symptoms. We think these results may be able to provide the clues in understanding how verbal abuse experiences can induce depressive symptoms and be the possible answer to the question: “Why should adolescents not use swear words without any concern for their peers?”
This work was supported by grants (NRF-2006-2005372 to B. Jeong) from National Research Foundation, BK21 plus, the KAIST Future Systems Healthcare Project, and KI for Health Science and Technology Project Grant [N11160068, N11160074], and by in part grant from Brain research institute. The authors thank Y.-B. Lee and S.Cho for their help to our fMRI experiment.
The online-only Data Supplement is available with this article
at
Cluster analysis
The results of Mann-Whitney U test to compare degrees of mean head movement between high-PeVA and low-PeVA group
Brain regions that showed significant activation during SWE compared to NEU in one-way repeated measures ANOVA
The results of Mann-Whitney U test of behavioral measurements between high-PeVA and low-PeVA group in total subjects
The cluster plot using ‘fpc’ package of R. Our data was divided into two groups. All peer verbal abuse questionnaire (VAQ) scores ≥40 was assigned to one group, except one subject (peer VAQ=40, parental VAQ=31). The two groups explain 100% of the point variability.
The results of reaction time in emotional Stroop task. The trial effect on reaction time was significant [F (3, 60)=5.518, p=0.002]. The differences between NEG, POS and NEU were significant after Bonferroni correction [t (21)=3.356, *p=0.003 and t (21)=4.030, **p=0.001, respectively]. Also, SWE trial showed a trend level difference compared to the NEU [t (21)=2.323, p=0.046>0.05/6]. There was no significant difference between emotional stimuli (SWE, NEG, and POS). SWE: swearing, NEG: negative, POS: positive, NEU: neutral word.
Brain regions showed more activity during SWE than NEU in repeated ANOVA and linear patterns of activation between SWE, NEG and NEU. Several brain regions, the left VLPFC, left insular, bilateral DMPFC and left amygdala, were more activated during SWE compared to NEU (Z>2.3, corrected p<0.05). In these regions, there were linear patterns of activations between SWE, NEG and NEU. L: left, VLPFC: ventrolateral prefrontal cortex, DLPFC: dorsolateral prefrontal cortex, DMPFC: dorsomedial prefrontal cortex, VMPFC: ventromedial prefrontal cortex, LOC: lateral occipital cortex, PCC: posterior cingulate cortex, SWE: swearing, NEG: negative, POS: positive, NEU: neutral word, ANOVA: analysis of variance.
The group differences between high-PeVA and low-PeVA group in left amygdala functional connectivity with the left VLPFC and left hippocampus. A: Overall functional connectivity of the left VLPFC–left amygdala during the modified emotional Stroop task was greater in high-PeVA than in low-PeVA [F (1, 21)=4.801, p=0.04]. Although there was no significant difference between the two groups after Bonferoni correction, the high-PeVA group had marginally greater connectivity between the left VLPFC and the left amygdala than the low-PeVA group in the NEG and NEU contrast (p=0.053, 0.046, respectively). B: There were also not any significant effects of group (p=0.1), trials (p=0.53) and group*trials (p=0.66) on left amygdala–left hippocampus connectivity. VLPFC: ventrolateral prefrontal cortex, PeVA: peer verbal abuse, SWE: swearing, NEG: negative, POS: positive, NEU: neutral word.
The regression graphs between the left VLPFC–left hippocampus connectivity and depressive symptoms in the high-PeVA and the low-PeVA group. A: In the high-PeVA group, the ratings of BDI were negatively associated with the connectivity of the left VLPFC–left hippocampus in SWE (B) and NEG (Spearman's rho=-0.867, p=0.001 and Spearman's rho=-0.610, p=0.046, respectively). 95% of confidence interval is depicted using a dotted line. VLPFC: ventrolateral prefrontal cortex, PeVA: peer verbal abuse, BDI: Beck depression inventory, SWE: swearing, NEG: negative.
*frequency of swearing=sum of the frequency of hearing or speaking swear words, †the number of high-PeVA=10. VAQ: Verbal Abuse Questionnaire, STAI: Stait-Trait Anxiety Inventory, BDI: Beck Depression Inventory, RT: reaction time, ER: error rate, PeVA: peer verbal abuse group, MRI: magnetic resonance imaging, IQ: intelligence quotient, SWE: swearing, NEG: negative, POS: positive, NEU: neutral words