Mentalizing about others when
thinking about yourself?
Hanna Rozema (11007516)
Supervisor: Mariët van Buuren
Research Master Brain and Cognitive Sciences – University of Amsterdam
Abstract
Self- and other referential processing are important aspects of the self- and social development that continues during adolescence. Brain areas related to these processes, include the medial prefrontal cortex (mPFC) and the temporo-parietal junction (TPJ). In adults a division has been observed in the mPFC related to the similarity of the other. The ventral mPFC (vmPFC) has been shown to be involved in the evaluation of the self or a similar other, whereas the dorsal mPFC (dmPFC) and the TPJ were more engaged when thinking about a more distant other. However, not much is known about this division in adolescents. Interestingly, the mPFC and the TPJ are also core regions of mentalizing. In adolescence, higher activation of the dmPFC and the TPJ has been observed when thinking about the self, suggesting that adolescents use mentalizing in self-referential processing. The aim of this study is to determine what the effect is of peer similarity on the brain activity underlying self- and other-referential processing in young adolescents, and to determine whether inter-individual differences in mentalizing are related to brain activity during self-referential processing in young adolescents. In order to do this, brain activity was measured with fMRI when adolescents made judgements about themselves, a similar classmate and a dissimilar classmate. Additionally, mentalizing performance was determined with the Reading the Mind in the Eyes (RME) task and an association between
mentalizing ability and brain activity in the dmPFC and the TPJ was investigated. Within the vmPFC and the right TPJ differences in activation were found between the self and the two other conditions. However, the similar and dissimilar other condition did not differ significantly. For the dmPFC and the left TPJ, no differences were observed between the conditions. Additionally, correlation analysis revealed no association between the RME scores and activity in the dmPFC or the TPJ. These results suggest that although young adolescents show larger involvement of the vmPFC and the right TPJ during self-referential processing, the similarity of peers does not influence the involvement of the vmPFC or right TPJ. Furthermore, the current results show no support for the theory that adolescents rely on metalizing when thinking about themselves.
Key words: adolescence, self- and other referential processing, mentalizing, vmPFC, dmPFC, TPJ
Introduction
Adolescence is an important period of life for both self- and social development. During this period, adolescents continue to develop their self-concept, identity, goals (Crone & Fuligni, 2020) and social cognitive skills (Brizio, Gabbatore, Tirassa & Bosco, 2015). As they are presented into the new social environment of secondary school, peers and their opinions become more relevant (Guyer, McClure-Tone, Shiffrin, Pine & Nelson, 2009). The social relationships that youths engage in are often with peers that share similar behavior, attitudes and interests (Brechwald & Prinstein, 2011). One important aspect of social development is mentalizing, referring to the ability to understand what other people are feeling and thinking. Mentalizing has mostly been researched in children and developmental disorders, however it has been found that adolescents still continue to develop this
skill (Bosacki, Moreira, Sitnik, Andrews & Talwar, 2020; Dumontheil, Apperly & Blakemore, 2010). Additionally, improving their mentalizing skills will also make adolescents aware of others’ opinions about them, and this in turn may influence their self-concept (Sebastian, Burnett & Blakemore, 2008; Somerville et al., 2013).
The brain areas that are involved in both self- and other referential processing are the medial prefrontal cortex (mPFC), the posterior parietal cortex (PPC) and the temporo-parietal junction (TPJ) (Davey, Pujol & Harrison, 2016; D’Argembeau, 2013). In adults, a division in mPFC activity has been observed related to similarity (Murray, Schaer & Debanné, 2011). That is, the ventral part of the mPFC (vmPFC) has been shown to be involved in the evaluation of the self or a similar other, whereas the dorsal side of the mPFC (dmPFC) was more engaged when thinking about a more distant other (Denny, Kober, Wager & Ochsner, 2012; Murray et al., 2011), together with the TPJ (Denny et al., 2012). These brain regions are also involved in self- and other processing in adolescents. However, as adolescents still develop their self-concept and representation of others, differences in the height of activity are expected in comparison to adults. Indeed, in adolescents, both the dmPFC (Davey et al., 2019; Pfeifer et al., 2013) and the TPJ (Pfeifer et al., 2009) show greater activation during self-referential processing compared to adults. As adolescents transition into adulthood, this activation decreases (Davey et al., 2019). Furthermore, although peers have a large influence on the behavior and opinions of teens, not much is known about the influence of peer similarity on adolescents’ self-concept and other representations. A study performed by Romund et al. (2017), did show that adolescents display greater overlap in neural responses to self and friend related judgements, compared to judgements about teachers or politicians. Additionally, this study also revealed higher activation in the vmPFC when making judgements about friends. It therefore seems likely that adolescents depend on similar underlying neural networks when judging a similar or distant other, compared to adults.
Not only are the mPFC and the TPJ involved in self- and other referential processing, they are also core regions of mentalizing. A meta-analysis, performed by Schurz, Radua, Aichhorn, Richlan and Perner (2014) investigated the neural network involved in mentalizing during different types of mentalizing tasks. They revealed that both the TPJ and the mPFC are activated in every task and therefore crucial brain areas for mentalizing. Especially the dorsal component of the mPFC is highly engaged during these tasks (Amodio & Frith, 2006; Blakemore, 2008). As mentioned before, mentalizing is also still developing in adolescents. Studies investigating this development, have compared the brain activation of adolescents with those of adults when performing a mentalizing task and have observed higher activation levels in the mPFC in adolescents, especially in the dorsal component (Blakemore, 2012; Blakemore, den Ouden, Choudhury & Frith, 2007).
Together, these studies show overlap in the brain areas involved in self- and other referential processing as well as mentalizing, suggesting that thinking about the self or others, may not be two separate developments, but rather an intertwined process (Crone & Fuligni, 2020). Moreover, the increased involvement of the dmPFC and the TPJ during self-referential processing in adolescence may suggest that adolescents rely on mentalizing and the opinions of others during self-evaluation (Romund et al., 2017). These ideas are in line with the heightened sensitivity to peers during this period of life (Guyer et al., 2009). However, it remains unclear if mentalizing indeed plays a role in self-referential processing and the underlying brain activity in adolescents.
Therefore, the current study aims to understand the effect of peer similarity on brain activity underlying self- and other-referential processing. Furthermore, this study investigates whether inter-individual differences in mentalizing are related to brain activity during self-referential processing in adolescents. To answer these questions, brain activity will be measured with fMRI in 70 young
adolescents during the trait judgement task. During this task trait adjectives will be presented and the participants need to indicate whether a trait applies to them (self-condition), a similar classmate (similar condition), a dissimilar classmate (dissimilar condition) or whether the trait contains the letter a (control condition). Additionally, mentalizing will be measured with the Reading the Mind in the Eyes (RME) task, in order to determine whether mentalizing ability is related to brain activation in the areas engaged in self-referential processing. Based on adult studies and the study performed by Romund et al., (2017), I hypothesize that the involvement of the vmPFC and the TPJ will be
dependent on the reference condition. More specifically, for the vmPFC I expect more involvement during the self-condition versus the two other condition, as well as more activation in the similar other condition compared to the dissimilar other condition. Within the TPJ I expect the opposite effect of reference condition, thus more involvement during other referential processing compared to self-referential processing and especially during the dissimilar other condition. Furthermore, I hypothesize that adolescents use mentalizing skills during self-evaluation tasks and therefore their mentalizing ability would be related to brain activity during self-referential processing. Specifically, as adolescents become better at mentalizing, the associated neural activation will decrease, because less cognitive effort will be needed to perform the same task. Therefore, I expect higher scores on the RME test to be associated with less activity within the dmPFC and the TPJ during the self-condition.
Materials and method
ParticipantsFor the current study, data was obtained from the longitudinal #SO CONNeCT project. This project investigates the development of social cognition, behavior and social networks during adolescence. Students partaking in the project are tested at their schools twice a year for three consecutive years, starting in their first year of secondary education. As a part of this project, 86 adolescents agreed to participate in the fMRI study, and did not have contraindications for MRI or a self-reported
neurological disorder. Additional written informed consent was given by both the participant and their parent before participation. MRI data was obtained of 84 adolescents, because two participants got anxious before or shortly after starting the scanning session. In total, 14 participants were
excluded from the current study; one participant was excluded due to incomplete data caused by a technical failure, nine participants were excluded due to excessive motion (more than 3 mm), three participants pressed the wrong buttons during the task, and one participant did not perform the RME test. Thus, in total, 70 adolescents, aged 11.61-14.22 years (mean 12.87 ± 0.38 years; 28 female and 42 male; 10 left-handed) were included in the current research analysis. After participating in the MRI part of the study, participants received €20 as monetary compensation and a picture of their brain. Ethical approval of the protocol was provided by the institutional review board (VCWE, Faculty of Behavioral and Movement Sciences, VU Amsterdam, The Netherlands).
General procedure
Upon arrival, both the participant and their parent were first asked to sign an informed consent form. After that, the participant had to fill in two questionnaires, one to check for contraindications for the MRI and one about two classmates. In this last questionnaire, participants were asked to pick two classmates, one that was seen to be similar (similar hobbies, interests and believes) and another that was perceived as dissimilar to themselves. The given names were used as the cues for the similar and dissimilar condition in the trait judgement task. Participants were asked to name classmates that they both liked and knew, in order to focus on similarity levels and limit the effects of likeability and familiarity on the results of other referential processing. In the questionnaire, participants were asked to rate the two classmates on their similarity, likeability and familiarity on a scale from one to ten.
The similarity score between the similar and dissimilar classmate had to differ at least four points. When this was not the case, subjects were asked to think of another classmate that was thought to be more similar or more dissimilar. After all questionnaires had been filled out, participants received instructions for the MRI scanner and for the trait judgement task. Participants were then placed in the mock MRI scanner, in which the MRI sounds were played and another task was practiced, this task is not part of the current study. Next, participants were placed in the actual MRI scanner. They had to wear headphones and earplugs and received two button boxes, one for the right and one for the left hand. These boxes were placed on their lower abdomen. With a mirror on the head coil, participants were able to see the screen on which the instructions were presented. During the scanning session, participants were first scanned during a resting-state period of seven minutes. After this the trait judgement task was presented.
Experimental tasks
Trait judgement task
In the scanner, participants performed the trait judgement task (Craik et al., 1999; Van Buuren, Gladwin, Zandbelt, Kahn, & Vink, 2010). They were shown a total of 160 trait adjectives. Depending on the condition they had to indicate whether the trait described themselves (self-condition), the classmate considered to be similar (similar other condition) or the classmate considered to be dissimilar (dissimilar other condition). Additionally, the task contained a control condition, in which the participant had to indicate whether the trait contained the letter “a”. The adjectives that were presented in this task consisted of 80 positive and 80 negative, which were equally and randomly distributed among the conditions. All traits were randomly distributed for each participant in blocks of five per condition, with no more than three adjectives of the same valence. In every condition, participants were first shown a cue of one second, indicating the condition for that block. For the similar and the dissimilar condition, this cue contained the name of the similar classmate or the dissimilar classmate respectively. After the cue, five traits were presented for three seconds or until the subject responded. The conditions were presented in a way that each condition was shown once per block, in a randomized order, followed by a 17 second resting period. Participants could indicate “yes” by pressing a button with their left index finger and “no” by pressing a button with their right index finger. After responding, a fixation cross was presented for the remaining time of that trial.
Figure 1: Trait judgement task. The task consisted of four conditions (self, similar, dissimilar and control). Each condition
consisted of a block of 5 trait adjectives. First a cue was presented for 1 second, followed by the first stimulus, consisting of a trait adjective with a reminder for the condition on the top and response options below. The stimulus was presented for 3 seconds or until the participant responded, in which case a fixation cross appeared for the remaining time of that trial.
Reading the Mind in the Eyes task
At the schools, all the participants were asked to fill in several questionnaires, including the Reading the Mind in the Eyes (RME) test (Baron-Cohen, Wheelwright, Spong, Scahill & Lawson, 2001). This is a test used to determine mentalizing ability. In the RME task, participants were shown 28 photos of the eye region of the face. The participants are asked to choose the best of four brief descriptions that represents the mental state of the person in the photo. Answers were scored as correct (one point) or incorrect (zero points). Performance was measured as total score of correct answers on the 28 items.
Peer ratings and behavioral analyses
First, differences in similarity, likeability and familiarity ratings between similar and dissimilar classmates were compared with the Wilcoxon matched-pairs test.
Behavioral responses during the trait judgement task for the three reference conditions were analyzed. Analysis of endorsement of positive and negative responses (i.e. answering yes to positive or negative traits respectively) was performed, using a two way repeated-measures ANOVA, with condition (self, similar or dissimilar) and endorsement (positive or negative) as factors. When interaction effects between the factors were found this was followed by a one-way ANOVA and Bonferroni corrected post-hoc pairwise comparisons. Additionally, analysis of reaction time was performed applying a one-way repeated measures ANOVA, with condition as factor. This was followed with pairwise comparisons, with Bonferroni corrections, in case of significant results.
MRI data acquisition
Neuroimaging was performed with a 3.0 Tesla Philips Ingenia CX MRI scanner equipped with a 32-channel-phased array head coil (Spinoza Centre for Neuroimaging; Philips Medical Systems, Best, The Netherlands). During the trait judgement task, a two-dimensional echo planar imaging-sensitivity encoding (EPI-SENSE) sequence was used to obtain 326 functional images. For this, the following parameters were used: voxel size 3 mm isotropic; repetition time (TR): 2000 ms; echo time (TE): 27.63 ms; flip angle = 76.1°; matrix 80 x 80; field of view 240 x 240 x 121.8; 37-slice volume with a 0.3 mm gap. Furthermore, a T1-weigthed structural brain image was made using a three-dimensional fast
field echo sequence, with the following parameters: voxel size 1 mm isotropic, TR = 8.2 ms; TE = 3.7 ms; flip angle = 8°; matrix 240 x 188; field of view 240 x 188 x 220; 220 slices in total.
MRI data preprocessing
MRI data preprocessing of the functional and anatomical images was performed using SPM12 (http://www.fil.ion.ucl.ac.uk/spm). First, realignment of the functional scans to the reference image was performed. This was followed by co-registration of the structural image to the mean functional image. Next, unified segmentation was applied to the co-registered structural image, using matched probability maps to control for the age and gender of the participants. These maps were created with CerebroMatic toolbox (Wilke et al., 2017) and used in SPM12. Normalization parameters were estimated to change the functional and structural images to Montreal Neurological Institute (MNI) space. Finally, a Gaussian filter (6-mm full width at half maximum) was applied to the functional images for smoothing of the normalized data.
Definition of regions-of-interest
The regions of interest (ROI) used in the MRI data analysis are based on the meta-analysis from Denny et al. (2012) on studies investigating self- and other-referential processing. A 10-mm radius sphere was created around the peak coordinates of activation in the vmPFC, the dmPFC and the left and right TPJ during self- and other-referential processing. The following coordinates were used: vmPFC (x, y, z = -6, 56, 10), dmPFC (x, y, z = -6, 54, 32), left TPJ (x, y, z = -50, -62, 22), right TPJ (50, -62, 22).
MRI data analyses
A generalized linear model regression was applied to the functional scans to analyze effects of condition. For this analysis, four variables were used for the different conditions (self, similar, dissimilar and control). These variables were modelled with a box-car function, with a duration of 15 seconds and the onset of the first trial of each block as the start. Additionally, the cue periods were modelled as a variable of no-interest, with a duration of one second. All variables were convolved with a canonical hemodynamic response function (Friston et al., 1995). To correct for head motion, the six realignment parameters were included as variables of no-interest. In order to remove low-frequency fluctuations (cut-off at 128 s) a high pass filter was applied to the data. Furthermore, each reference condition was contrasted to the control condition and to each other, to construct the contrast images.
Next, ROI analyses were performed to investigate the differences in brain activity during self- and other-referential processing. Contrast estimates for every participant and every ROI were obtained using the MarsBar toolbox (version 0.44, http://marsbar.sourceforge.net/). The neural activation levels, relative to the control condition, were analyzed using a two-way repeated measures ANOVA , with reference condition (self, similar, dissimilar) and ROI (vmPFC, dmPFC, left and right TPJ) as factors. In case of significant effects, this was followed by one-way repeated measures ANOVA for each ROI and post-hoc pairwise comparisons, which were Bonferroni corrected.
Correlational analyses
To determine whether dmPFC or TPJ activity during self-referential processing and mentalizing ability are negatively related in adolescents, a correlation analysis was performed. The correlation
coefficient (Pearson’s r) was calculated between the RME scores and the activity in the dmPFC and the TPJ during the self-condition in contrast to the control condition. Finally, an exploratory whole brain analysis was done to determine whether RME scores were associated with brain activity outside
the ROIs during the self-condition. Correlational analysis was performed to determine an association between the brain activity during the self-condition contrasted to the control condition and the RME scores as covariate.
Results
Behavioral resultsTo determine whether the similar and dissimilar classmate were differently regarded by the
participants, ratings on similarity, likeability and familiarity were compared (table 1). The differences in ratings have been found to be significant for similarity (t = 28.7, p < 0.001), likeability (t = 8.5, p < 0.001) and familiarity (t = 6.3, p < 0.001) indicating that the similar and dissimilar classmate are perceived differently, based on all three classifications.
Similar classmate Dissimilar classmate
Similarity 8.1 ± 0.7 3.8 ± 1.3
Likeability 9.0 ± 0.9 7.4 ± 1.5
Familiarity 8.5 ± 1.3 6.7 ± 2.0
Table 1: Similarity, likability and familiarity ratings on a scale from 1 to 10 for the similar and the dissimilar classmate
presented as mean ± SD.
Furthermore, in order to check for behavioral differences between the reference conditions, response times and endorsement was analyzed (see table 2). A significant main effect of condition was found for the response times (F(2,138) = 5.73, p = 0.004). Bonferroni corrected pairwise comparisons revealed that the similar and the dissimilar condition (p = 0.009) were found to differ significantly, as participants responded quicker in the similar condition compared to the dissimilar condition.
Differences in reaction time were non-significant for the self- versus the similar condition (p = 0.118) or the self- versus the dissimilar condition (p = 0.401). Next, for endorsement, repeated measures ANOVA analysis showed a significant main effect for the conditions (F(2,138) = 6.72, p = 0.002), endorsement (F(1,69) = 409.9, p < 0.001) and an interaction effect (F(2,138) = 13.32, p < 0.001). Because Mauchly’s test of sphericity was violated, Huynh–Feldt correction was applied before determining significance. For endorsement, one way ANOVA revealed a main effect of condition for the positive (F(2,138) = 15.08, p < 0.001) and negative (F(2,138) = 5,81, p = 0.005) ratings. Post-hoc pairwise comparisons, with Bonferroni corrections, revealed significant effects of endorsement between the self and the dissimilar condition (p < 0.001) and between the similar and dissimilar condition (p = 0.001) for the positive traits. No significant differences in yes responses to positive traits were found between the self and the similar condition (p = 1). For the negative traits, significant differences were observed between the self and the similar condition (p = 0.024) and the similar and dissimilar condition (p = 0.009). Yes responses for negative traits did not differ significantly between the self and dissimilar condition (p = 0.887).
Self Similar Dissimilar
Response time (in ms) 1251 ± 179 1223 ± 189 1274 ± 191
Positive 14.3 ± 2.9 14.6 ± 3.2 11.9 ± 4.2
Negative 4.7 ± 2.8 3.7 ± 3.0 5.2 ± 3.5
Table 2: Response times and amount of yes responses to positive and negative traits for the self-, similar and dissimilar
Brain activity the ROIs during the trait judgment task
Brain activity was measured during the trait judgement task and was analyzed for each ROI and each condition (results are presented in figure 1). This revealed a significant main effect of ROI (F(3, 207) = 72.39, p < 0.001, Huynh–Feldt corrected), and an interaction effect between the ROI and condition (F(6, 414) = 14.21, p < 0.001, Huynh–Feldt corrected). No main effect of condition was observed (p = 0.308).
For each ROI, differences in activity between the three references conditions were analyzed using one-way ANOVA’s. In the vmPFC, activity levels differed significantly between the conditions (F(2,138) = 6.68, p = 0.004). Post-hoc pairwise comparisons, with Bonferroni corrections, showed that the vmPFC was more activated during the self-condition, compared to the similar other condition (p = 0.012) and dissimilar other condition (p = 0.031). No significant differences were found between the similar and the dissimilar conditions (p = 1). Statistical analysis of the right TPJ revealed similar results, in which the reference conditions differed significantly in activation levels (F(2,138) = 11.27, p < 0.001, Huynh–Feldt correction). Pairwise comparison showed more activation in the right TPJ during the similar other condition (p = 0.014) and the dissimilar other condition (p < 0.001), compared to the self-condition. However, no distinction in the level of neural activation was found between the similar and the dissimilar other condition (p = 0.536). Furthermore, for both the dmPFC and the left TPJ, no significant main effect of condition was found (p = 0.053 and p = 0.315, respectively).
Figure 2: ROIs. Sagittal (A) and transverse plane (B) of the dmPFC (Light blue, x, y, z: -6, 54, 32), the vmPFC (Yellow, x, y, z: -6,
56, 10), the right TPJ (Red, x, y, z: 50, -62, 22) and the left TPJ (Dark blue, x, y, z: 50, -62, 22). (C) Signal changes in the four ROIs during the self, similar and dissimilar condition contrasted to the control condition, in a.u. (arbitrary units). Significant differences are found between the self and the similar condition (p = 0.012) and between the self and the dissimilar condition (p = 0.031) in the vmPFC. Activity in the right TPJ differed significantly between the self-condition and the similar condition (p = 0.014) as well as between the self-condition and the dissimilar condition (p < 0.001). No significant results were observed between the similar and the dissimilar condition in the vmPFC or the right TPJ. Additionally, no significant differences between the reference conditions were found in the dmPFC or the left TPJ. Results are presented as mean ± SE. * = p < 0.05, *** = p < 0.001.
Mentalizing during self-referential processing
To investigate whether mentalizing is involved in self-referential processing, correlation analyses were performed between the dmPFC and the TPJ activation levels during the self-condition, and the RME scores. No significant association was found between RME scores and activity within the dmPFC (p = 0.284), the right TPJ (p = 0.646) or the left TPJ (p = 0.446).
In addition, whole brain analysis was performed to determine if activity during self-referential processing outside the ROIs was associated with mentalizing ability. However, no significant association was found between activity during the self-condition and RME scores.
Figure 3: Association between the activity in the dmPFC p = 0.284 (A), the right TPJ p = 0.646 (B) and the left TPJ p =0.446 (C)
during the self-condition in a.u. (arbitrary units) and the RME scores.
Discussion
The aim of this study was to investigate self- and other referential processing during adolescence. Specifically, this study investigated whether involvement of the mPFC and TPJ depends on the similarity of the other and whether mentalizing is involved in self-referential processing in adolescents. For the first research question, the results show distinct activation levels within the vmPFC and the right TPJ depending on the condition. However, no differences were found between the similar and the dissimilar other condition, even though peer ratings did show that the two classmates differed significantly based on similarity, likeability and familiarity. Moreover, the
behavioral results showed differences between the similar and the dissimilar condition. In fact, faster responses were observed in the self and similar condition, compared to the dissimilar condition. In addition, endorsement of positive or negative traits differed significantly between the reference conditions. More yes responses were given for positive traits in the self and similar condition,
compared to the dissimilar condition. For the negative traits, participants gave more yes responses to themselves and their dissimilar classmate, compared to the similar classmate. This would support the evidence that the two classmates are regarded differently by the participant. Furthermore, for the dmPFC and the left TPJ no significant differences in activity were found between the self and other-referential conditions. Finally, to test whether mentalizing was involved in self-other-referential processing, brain activity during self-referential processing was associated with the mentalizing scores. However, in contrast to my hypothesis, the results showed no significant association between activity within the dmPFC or TPJ and mentalizing ability.
As hypothesized, activation of the vmPFC was greater in the self-condition, compared to the similar and the dissimilar other condition. This is in line with previous studies (Denny et al., 2012; Murray et
al., 2011), which suggest that the ventral part of the mPFC is involved in self-referential processing. Although a difference in vmPFC activation was found between the self and the two other conditions, no differences were observed between the similar and the dissimilar condition. These results are inconsistent with previous studies in adults, showing more involvement of the vmPFC when making judgements about a similar person compared to a dissimilar person (Benoit et al., 2010; Mitchell, Macrae & Mahzarin, 2006). The reason for these unexpected results can probably not be attributed to the likeability or the familiarity to the named classmates, since for both scores, the dissimilar classmate scored lower compared to the similar classmate. It is important to note however, that the current study was performed in adolescents, whereas most research investigating the underlying neural mechanisms of self- and other related processing is done in adults. There is one study, performed by Romund et al. (2017), that did investigate the effect of familiarity on the neural correlates related to other referential processing in adolescents. However in this study the other conditions were likely far more dissimilar from each other, compared to the classmates in the current study. In their study, Romund et al. asked participants to make trait judgments about themselves, their friends, their teachers and politicians. Great overlap was observed between the neural response to judging themselves or judging friends. Moreover, when comparing the neural response in the vmPFC, they found higher activation when judging friends compared to judging teachers or
politicians. These results could indicate that also in adolescence, the degree of familiarity of the other affects the underlying function brain processes. However, during adolescence, peers become of greater importance to a person, and adolescents will value the opinions of their peers more (Brown & Larson, 2009). Especially classmates form an important part of the social environment for adolescents and will have a large impact on an individual’s self and social development. Research has shown that peer interaction influences the behavior of an individual, as their behavior and attitudes becomes more similar over time (Brechwald & Prinstein, 2011). Therefore, choosing classmates for the similar and dissimilar condition might not allow for a large enough difference between the conditions, even though they vary on perceived similarity and participants show differences in behavioral responses to both conditions. This can also be supported by the peer ratings. Although the similar and dissimilar classmate differ significantly on likeability and familiarity, both classmates still score high on these dimensions, indicating that the classmates are likely close to the participant. Krienen et al. (2010) actually showed that closeness, rather than similarity is an important factor in the response of the vmPFC. They observed more activation in the vmPFC when individuals are judging a dissimilar friend compared to a similar stranger. Thus the reason for the lack of differences in vmPFC involvement between the similar and the dissimilar condition, could be that the classmates were equally close to the participants.
Furthermore, both the dmPFC and the TPJ have been suggested to be involved in mentalizing about others (Schurz et al., 2014). Therefore, I hypothesized that the TPJ would be more involved during other referential processing, compared to self-referential processing, especially in the dissimilar other condition. However, only a significant difference in activation was found in the right hemisphere. More specifically, as expected, participants showed greater activation in the right TPJ during the similar and dissimilar other condition, compared to the self-condition. No differences in involvement were observed between the similar and the dissimilar condition. Additionally, these differences between the three reference conditions were absent in the left TPJ. Previous studies investigating the neural network underlying mentalizing ability, commonly find bilateral TPJ activity, unlike the current results (Schurz et al., 2014). However, Denny et al. (2012) revealed that even though social processing involves bilateral activation of the TPJ, an overlap in brain activity exists in the left TPJ during self and other referential processing. This would suggest that the left TPJ is related to making judgements about people, irrespective of the person, whereas the right TPJ is more involved in taking another
persons’ perspective. In fact, previous research has shown that the left TPJ is mostly activated during self-referential processing (Pfeifer et al., 2009; Romund et al., 2017) and therefore not necessarily related to judging another. This theory is supported by Boccadoro et al., (2019) as well, who found that the right TPJ was more strongly activated during mentalizing tasks than the left TPJ.
In adolescents the dmPFC and the TPJ have also been observed to be more active during self-referential processing compared to adults (Davey et al., 2019; Pfeifer et al., 2009). As these are core regions of mentalizing I hypothesized that adolescents rely on their mentalizing ability when thinking about themselves. In line with this hypothesis, no activity differences were observed within the dmPFC or the left TPJ when comparing the self- with the other reference conditions. Nevertheless, these results do not imply that mentalizing skills are used when thinking about the self. Therefore, in addition, I expected that individual who are better at mentalizing, would rely less on this skill during self-referential processing and therefore show less activity in the dmPFC and the TPJ. This was examined by analyzing the association between RME scores and the activation of the dmPFC and both the TPJs during the self-condition. In contrast to my hypothesis, no correlation was found between mentalizing scores on the RME test and activity in the dmPFC or the TPJ. These results would suggest that mentalizing is not involved in self-referential processing in adolescents. However, mentalizing is complex ability with multiple dimensions (Turner & Felisberti, 2017). Therefore, mentalizing tasks can involve many aspects of this skill. For the RME task participants are asked to determine a mental state from only observing the eyes. Although this involves deducing a mental state from another individual, it mostly involves emotion recognition (Turner & Felisberti, 2017). It therefore might be the case that the RME task does not cover the aspect of mentalizing ability that is used in self-referential processing. Additionally, Schurz et al. (2014) show that compared to other mentalizing tasks, the RME shows less activation in the mPFC and other regions involved in self-referential processing. This could indicate that brain activity during self-self-referential processing and the RME task do not overlap due to differences in cognitive processes. This is also supported by the lack of activity outside the ROIs during the self-condition associated with the RME scores.
The results from the current study provide insight in the neural network underlying both self- and other referential processing. Young adolescents showed more activation in the vmPFC and less activation in the right TPJ when making judgements about themselves compared to others. Although the current results revealed no differences brain strategies for similar or dissimilar peers, additional research can be done to understand the involvement of the vmPFC in other-referential processing in adolescents. This was already done by Romund et al. (2017), however the other conditions in their study consisted of individuals with very distinct relationships to the adolescents. It would be useful for future studies to include an unknown peer condition or family members to determine the factor that influences the involvement of the underlying neural circuit. Furthermore, to my knowledge, this is the first study analyzing the association between mentalizing performance and involvement of the dmPFC or the TPJ during self-referential processing. Even though the current results show no
association, this does not necessarily suggest that adolescents do not utilize this skill when thinking about themselves. As previously mentioned, the RME task may not measure the same aspect of mentalizing that adolescents use during self-referential processing. Therefore, future studies should implement other mentalizing task that are based on different aspects of mentalizing in order to ensure that a more complete measurement of mentalizing performance is made. For example, the false belief vs. photo task is a different mentalizing task and has been shown to strongly engage the mPFC and other areas of the mentalizing network (Schurz et al., 2014) and would thus be interesting to include in a future study design. Additionally, it would be beneficial to include the mentalizing task in the MRI procedure, to demonstrate the explicit overlap with thinking about the self. Finally, as the involvement of the dmPFC during self-referential processing decreases when adolescents transition
into adulthood (Davey et al., 2019), it would be interesting to look at this association with mentalizing ability throughout adolescence. This to determine if indeed, activity in the dmPFC decreases with mentalizing ability or age.
Conclusion
In conclusion, the current study reveals that adolescents display greater involvement of the vmPFC during self-referential processing compared to other referential processing, while no differences were observed in the dmPFC. For the TPJ, only differences in the right hemisphere were observed, with lower activity in the self-condition. The similarity of peers did not seem to influence to what extent the brain areas were recruited when making judgements about others. This could suggest that because classmates are close to an individual during adolescence, not similarity but rather closeness influences the involvement of the underlying brain regions. Furthermore, the lack of differences in the left TPJ and dmPFC may suggest involvement of mentalizing in thinking about the self. However, no correlation was found between activation in these areas and RME scores. These results could indicate adolescents do not depend on mentalizing when thinking about themselves or that self-referential processing and the RME task rely on different aspects of mentalizing. More research is needed to uncover the neurological mechanism behind self- and social development during adolescence. Nevertheless, this study demonstrates that young adolescents engage the vmPFC mostly during self-referential processing and the right TPJ when judging others, however, the similarity of peers does not influence the involvement of the vmPFC or the TPJ in adolescents. Furthermore, the current study provides no support for the theory suggesting that adolescents rely on mentalizing during self-referential processing.
