Neural bases of social deficits in ADHD: A systematic review. Does the Theory of Mind matter?

5) was measured by a modified version of The Newcastle-Ottawa Scale and by measures specific for our study. This systematic review was registered on PROSPERO (CRD42020139847). Results: Results indicated that impairments in performing of the ToM tasks were negatively associated with the grey matter volume in the bilateral amygdala and hippocampus in both, ADHD and control group. In EEG studies, a significantly greater electrophysiological activity during ToM tasks was observed in the, frontal, temporal, parietal and occipital lobes in participants with ADHD as compared to healthy subjects. Conclusion: More research is needed to explore the ToM deficits in children with ADHD. Future research might focus on the neural circuits associated with attention and inhibition, which deficits seems to contribute to the ToM deficits in children and adolescents with ADHD.


Introduction
Theory of Mind (ToM -also known as cognitive empathy, mentalization, or social cognition) can be understood as the ability of an individual to infer the mental states of others (Tesfaye and Gruber, 2017).More specifically, ToM refers, besides others, to the following abilities: first, it enables the individual to "read" the emotional state in the face of another person, typically measured by Reading the Mind in the Eyes test (RMET).This test aims to measure the degree to which participants are able to infer the emotions of another person from facial expressionsfrom the orbital part of the face, respectively (Baron-Cohen et al., 2001).Second, it refers to false belief reasoning, which represents the ability to reason about the other person's beliefs, which differs from reality (Bernstein et al., 2017).
Several brain regions have been consistently implicated in ToM, including the temporoparietal junction (TPJ) and the superior temporal sulcus -STS - (Schurz et al., 2014;Van Overwalle, 2009).Furthermore, studies using Transcranial Magnetic Stimulation (TMS) have demonstrated that disrupting the activity of the TPJ, dorsolateral prefrontal cortex (dlPFC), and medial prefrontal cortex (mPFC) affects ToM ability (Schuwerk et al., 2014a).Moreover, Magnetic Resonance Imaging (MRI) studies have provided insights into how white matter connectivity relates to ToM.For example, research has shown that ToM is associated with white matter connectivity between the inferior frontal gyrus (IFG) and the STS (Wang et al., 2018).Lesion studies have also been used to investigate the neural bases of ToM.Damage to the ventromedial prefrontal cortex (vmPFC) has been associated with deficits in ToM, particularly in the emotional aspects of ToM (Shamay-Tsoory and Aharon-Peretz, 2007).Similarly, damage to the TPJ has been associated with impairments in the ability to understand others' beliefs (Samson et al., 2004).
There is also evidence suggesting that the deficits in ToM can be observed in people suffering from Attention Deficit Hyperactivity Disorder, ADHD (Caillies et al., 2014;Mary et al., 2016;Pineda-Alhucema et al., 2018;Saeedi et al., 2014).For instance, a study of Bozkurt et al. (2024) found that children with ADHD showed significant ToM impairments when compared to typically developing peers.Moreover, children with ADHD had significant difficulties with emotional facial recognition and attention to emotional cues, which are essential components of social cognition (Pineda-Alhucema et al., 2018;Saeedi et al., 2014).Some studies also emphasize that decreased ToM in individuals with ADHD can contribute to a lower degree of social functioning (Pineda-Alhucema et al., 2018;Sasayama et al., 2010).
Deficits in social functioning in individuals with ADHD often manifest as difficulties in forming and maintaining relationships, inappropriate social behaviours, frequent conflicts with peers (Caillies et al., 2014;Pineda-Alhucema et al., 2018), increased social impairment and higher degree of anger (Sacchetti and Kefler, 2017) or impaired in assertiveness (Solanto et al., 2009).One of the key factors contributing to impairments in social functioning might be a lower degree of executive functions (Rinsky and Hinshaw, 2011;Tseng and Gau, 2013).
Executive functions (EFs) are processes that control and regulate thoughts and actions and are used in situations where suppressing habitual responses is needed (Friedman et al., 2006).The EF can affect social functioning in two significant ways: first, by impairing the ability to plan, organize, and control social behaviours, leading to inappropriate interactions and misunderstandings (Nijmeijer et al., 2008); (Sacchetti and Kefler, 2017).Second, it impacts the ToM (Pineda-Alhucema et al., 2018).Although recent studies have examined ToM deficits in ADHD in relation to executive functions (Pineda-Alhucema et al., 2018), review evidence regarding neural substrates underlying ToM deficits in ADHD is totally lacking.
Thus, our research question is as follows: What are the neural bases underlying the Theory of Mind deficits in children and adolescents with ADHD?This research question defines the aim of this study: to systematically review articles that examined neural correlates of ToM deficits in children and adolescents with ADHD.The theoretical framework for the interpretation of the results will be primarily based on the theory linking ToM ability to executive functions (see (Brock et al., 2019;Mutter et al., 2006;van der Meer et al., 2011).The results from neuroimaging and electroencephalography (EEG) studies will be explored.

Search procedure
To assure the methodological quality of this review, we conducted it in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al., 2009).PRISMA represents methodological directions, which are considered as the gold standard for conducting systematic reviews (Oyinlola et al., 2016).Unlike methodological guidelines for conducting systematic reviews provided by Cochrane collaboration -which are in general more focused on the effectivity of interventions (Chandler et al., 2019), PRISMA guidelines are applicable to a broader type of studies (Pati and Lorusso, 2018) including cross-sectional studies relevant to our systematic review aims.Our search was done in four databases: PubMed, Web of Knowledge, Scopus, and PsycInfo.The search in every database was based on the four following keywords, their related terms, abbreviations, or extensions: "ADHD" AND "Theory of Mind" AND "Neuroimaging" AND "EEG".In order to target relevant studies, we collected derivations of keywords from systematic review studies examining ToM deficits in children (Benarous et al., 2015;Pineda-Alhucema et al., 2018) as well as from the studies systematically reviewing neural correlates of social cognition (Sugranyes et al., 2011), emotions (Bastin et al., 2016) face processing (Collin et al., 2013;García-Sancho et al., 2015) and emotion regulation (Cibralic et al., 2019;Weiss et al., 2014).From keywords obtained in previously mentioned studies, we created a search syntax (see Fig. 1), which we report in the search filters for each database, so that future studies can replicate results.In order to facilitate our search in the databases, we used the Mesh terms in PubMed.Neuroimaging studies utilizing magnetic resonance imaging (MRI) began being published to a larger extent in approximately the early 1980s (Edelman, 2014), and in the case of computed tomography (CT) the first scans were utilized in the 1970s (Ghonge, 2013); we thus decided to limit our search from January 1970 to July 2022.The database search was conducted in June 2023 by the first author (LN) and by the fifth author (TS).This systematic review was preregistered on PROSPERO (Novák et al., 2020) under the following ID: CRD42020139847.

Inclusion and exclusion criteria
In the first step of the screening procedure, two authors (LN) and (PM) independently reviewed 638 articles and removed duplicates (n = 224).In the following part the same researchers screened the studies based on study abstracts, which had to meet the following inclusion criteria: 1) ADHD diagnosed by either DSM-IV/DSM-V or ICD-10; 2) age from 3 years to 18 years; 3) a healthy (i.e., without a neurodevelopmental disorder) control group included; 4) a reported relationship between either brain structure or relevant brain activity (e.g.fMRI) and the ToM (i.e., inferring intentions, beliefs and emotions of another person including face processing i.e. inferring emotions from facial expressions); 5) studies examining the relationship between brain structure or function and emotional regulation if related to ToM.
Studies were excluded according to the following criteria: 1) irrelevant literature: book chapter, theoretical article, qualitative study, review article, case study; 2) not published in a peer-reviewed journal; 3) absence of an ADHD diagnostic system (e.g.DSM-IV) or comorbidity with Autism spectrum disorder; 4) not relevant age, i.e. over 18 years old or below 3 years old; 5) absence of a healthy control group; 6) not examining the ToM; 7) irrelevant measure of the ToM, i.e., questionnaire on social abilities (e.g.pragmatic language usage); 8) examining only emotional regulation without relationship to both ToM and brain structure or function; If there was a disagreement between the authors, a discussion between (LM) and (PM) took place until consensus was reached.If consensus between (LN) and (PM) could not be reached, then the other authors would be asked to decide.However, because 100 % consensus between (LN) and (PM) was reached, it was not necessary to ask the other authors.The degree of agreement between the two authors was assessed as follows: in line with the recommendations regarding the  assessment of inter-rater reliability (McHugh, 2012), the percentage of agreement as well as Gwet's AC1 statistic (Gwet, 2001) were both calculated.The AC1 values below 0 indicate a poor agreement, 0.00-0.20 slight agreement, 0.21-0.40fair agreement, 0.4-0.60 moderate agreement, 0.61-0.80substantial agreement and 0.81-1.00an almost perfect agreement (Landis and Koch, 1977).The formula for calculating Gwet's AC1 is as follows: In this formula, the p a represents agreement probability, while p eγ is the chance-agreement probability -for more details see: Viswanathan & Berkman (Viswanathan and Berkman, 2012).The main reasons for exclusion were: Missing or irrelevant ToM task (n= 229) and Missing examination of neural correlates in ADHD (n = 70).For other exclusion reasons, see Table 1.In the final step, one researcher (LN) assessed the full texts of the remaining articles (n = 11) for eligibility.During this procedure, reasons for the inclusion and exclusion of the remaining studies were discussed with the second author (PM) until 100 % agreement was reached.The LN and PM then performed data extraction from the remaining article (n=5).List of studies excluded during eligibility evaluation is presented in Supplementary Table 1.
Table 1 should be placed here.

Quality measures
We assessed the quality of the studies via the widely used Newcastle-Ottawa Scale (NOS) (Wells et al., 2000), respectively, its adapted form for observation and cross-sectional studies (Bawor et al., 2014).The NOS has been identified by the Cochrane organization as one of the most useful tools for measuring study quality (Higgins, 2011).The modified form of NOS aims to evaluate several sources of bias: 1) selection biasmethods used to select a study sample; 2) performance biasthe ways in which the study deals with confounders; 3) detection biasstatistical approaches used; 4) information biasthe ways in which the study measures dependent and independent variables.Bias risk is rated on a four-point scale ranging from 0 to 3, where 0 represents a high risk of bias and 3 represents a low risk (Bawor et al., 2014).
1) Besides the NOS, we included four additional specific quality measures.The rationale for these criteria was as follows: 1) performance in some ToM tasks (e.g.The Reading Mind in the Eyes Test -RMET) is influenced by the participant's IQ (see meta-analysis of Baker (Baker et al., 2014)).Therefore, as a quality measure, we assessed whether a study controlled for IQ, if the RMET was used.2) Many neuroimaging studies, especially those using fMRI, lack sufficient power, which indirectly contributes to findings that cannot be trusted (Poldrack et al., 2017).However, strategies have been developed that would prospectively prevent the underpowering of fMRI studies, e.g. by power analysis.We decided to assess whether a study used a power analysis prior to the experiment or not.3) Because of the anatomical properties of the frontal brain region, a loss of signal in the orbitofrontal cortex (OFC) is frequent in fMRI studies.However, some strategies exist that can be used to prevent such signal loss from the frontal brain region (see the study of Moccia (Moccia et al., 2017)).We also evaluated whether the fMRI studies used this orbitofrontal correction strategy or not.4) Fluctuation of attention is one of the core symptoms in ADHD.It is therefore possible that healthy participants outperform those with ADHD in the ToM task not because they have better ToM abilities, but because their attention does not fluctuate in the ToM tasks as much as that of the ADHD population.Therefore, we also evaluated whether the study reported procedures that controlled for the sustained attention in children and adolescents with ADHD during the ToM task (e.g. via eye tracking).All quality measures criteria were evaluated by the first author (LN).Due to the use of the two quality measures (i.e.NOS as well as the four specific quality measures) our study presents at least the methodological limitations of previous research.

Extraction procedure
After the final screening procedure, the data was extracted from the relevant studies (n=5) by the first author (LN).The following information was gathered: 1) publication information: year of publication, authors of the study, 2) study sample information: number of participants, age of participants, ADHD diagnostic system (e.g.DSM-IV); country of recruitment 3) Neurodiagnostic information: type of neuroimaging method used, power analysis method, orbitofrontal correction strategy, neuroimaging results, number of electrodes, the position of electrodes, type of EEG analysis 4) ToM task information: ToM confounders assessment (i.e., IQ, sustained attention).

Inter-rater reliability and sociodemographic results
Measuring the degree of agreement between the two raters revealed that the percentage of agreement was 96.45 %.Gwet's AC1 statistic also supported a high degree of agreement between the two raters: AC1=0.96,95 %CI [0.94-0.99],p <.001.Interestingly, our analysis revealed that across all three databases we found 5 studies that fit all of our inclusion criteria.Across these studies, data from 325 control and ADHD participants were analyzed.Study subjects diagnosed with ADHD (n= 178) were recruited in Canada, the USA, England, and Iran.Where exactly participants in control groups were recruited was not explicitly reported in most studies, however.The mean age was 11.47 (±2.62) years.

EEG markers of ADHD in ToM studies
We found four suitable studies using EEG methodology.One of them was focused on event-related potentials (Alperin et al., 2017), and three studies were focused on frequency bands (Dini et al., 2020;Gross et al., 2012;Sarraf Razavi et al., 2017).Alperin et al. (2017) observed that ADHD patients had a larger N170 wave to non-emotional than to emotional faces (neutral > fear = happy), while typically-developing adolescents showed the opposite pattern with a larger N170 to emotional than non-emotional faces (fear = happy > neutral).They also used P3b (~300-500 ms post-stimulus) and LPP (Late Positive Potential) as indicators of late or sustained emotional processing.The results

Table 1
Number of excluded studies and the reason for their exclusion.were that a larger P3b was in all participants to fearful versus other faces and a larger LPP to emotional vs. non-emotional faces.Dini et al. (2020) analyzed functional connectivity during the emotion recognition task in all frequency bands (from 1 to 80 Hz), and then they calculated the event-related phase coherence in each frequency band.They also used machine learning with logistic regression to discriminate between the ADHD and healthy control (HC) groups.According to the graph theory, two main features from connectivity matrices were extracted to evaluate brain networks.These features are the clustering coefficient (C) and the shortest path length (L) which indicate brain network integration and segregation.Their findings show that the beta band is the most crucial sub-band in the classification between ADHD and HC in all four emotions (Angry, Happy, Neutral, and Sad emotions).Significant differences in the intra-hemisphere in the right (FP2-AF8 (3-34)) and left (F7-T7 (4-13)) hemispheres, interhemispheric (P8-P5 (26-49)), and fronto-occipital (P8-AF4 (26-33)) hemispheres between the ADHD and HC groups in the main and interaction effects analysis were observed in all four emotions.
Other authors (Gross et al., 2012;Sarraf Razavi et al., 2017) observed only the gamma band (35-45 Hz).Razavi et al. (2017) showed in ADHD children a significant reduction in the gamma band activity, connected with early (0-250 ms) perceptual emotion discrimination for happy and angry emotions compared to HC.A significantly greater gamma band activity was observed, especially in the occipital lobe compared to frontal, central, and parietal areas in ADHD patients.Gross et al. observed the effect of induced gamma alignment in the emotion recognition task in ADHD, HC, and patients with autism spectrum disorder (ASD) but in the later time window (200-600 ms).Significant (Group x Condition x Alignment) three-way interactions were observed across the parietal and occipital channels using a model that compared the separate emotion recognition tasks and the gender recognition task.In models that compared the Anger/Disgust recognition task to the gender recognition task, significant interactions were seen in the P3-P4 electrodes and in the P7-P8 electrodes in different groups.The interaction effect in the aligned data set can be described as a higher power of induced gamma in the ADHD group as compared to the ASD group during the anger versus disgust differentiation task without any differences during the gender recognition task at the parietal sites (P3, P4, P7, P8).Taken together, higher electrophysiological activity in ADHD was observed in multiple brain regions, including the frontal, parietal, and temporal lobes.Interestingly, during the emotion recognition task, significantly greater electrophysiological activity was observed in the occipital lobe (Fig. 2) in ADHD as compared with healthy controls.

Neuroimaging methods and Theory of Mind task used
In the study of Baribeau et al., (2019), weighted T1 images were acquired in structural MRI in order to examine grey matter volume and its relationship to ToM deficits.In this study, both whole brain analysis, as well as Regions of Interest (ROI) analysis, were used.The complete list of ROI can be found in Supplementary Table 2.The study registered brain regions using the Automated Anatomical Labelling atlas (AAL).The ToM task consisted of the child version of RMET.Following behavioural and neuroimaging measures, statistical analysis was conducted.During the inferential statistics procedure, a logistic regression was used and the number of incorrect answers from RMET was regressed on grey matter volume and odds ratios calculated for control, ADHD and other groups.

A neurobiological underpinning of the Theory of Mind in ADHD
The study of Baribeau et al. (2019) found that a larger cortical grey matter volume in the bilateral amygdala and hippocampus was associated with lower performance deficits in RMET in children with ADHD.This association remained significant even after adjusting for IQ.They found no association between the RMET score and any other region in children with ADHD.The association of larger gray matter volume in the bilateral amygdala and hippocampus and higher RMET performance was observed not only in children with ADHD but also in healthy subjects and in subjects with an autism spectrum disorder.

Quality rating results
In this systematic review we assessed the quality of studies which passed all of our inclusion criteria with a modified version of NOS and with our own quality measures, as described in the method section.In the case of NOS, the quality rating score of the study of Baribeau et al. (2019) reached in most cases a score of 3 (Table 2), which indicates the absence of a high risk of bias.Even better results were obtained in a study of Alperin et al. (2017).For all other studies examined, at least one parameter was found to be at high risk of bias.Concerning our quality measures criteria, a study of Baribeau et al. (2019) was the only one that met the IQ controlling criterion.However, none of the examined studies met the criteria of Power analysis reported or Sustained attention controlling.Criterion of OFC correction strategy was assessed irrelevant in the case of the examined studies.(Table 2).
Table 2 should be placed here.
Fig. 2. image depicting localization of the occipital lobe that was frequently associated with ToM deficits in children and adolescents with ADHD across EEG studies.
L. Novak et al.

Discussion
In this study, we reviewed relevant literature on the neural bases of ToM deficits in children and adolescents with ADHD from 1970 to 2023, using neuroimaging and electrophysiological studies.We found 4 electrophysiological and 1 neuroimaging study that met inclusion criteria.In electrophysiological studies, a significantly greater electrophysiological activity in the frontal, parietal, temporal and occipital lobe was observed in participants with ADHD during emotion recognition task compared to healthy controls.The only one neuroimaging study that we found reported that the deficits in the ToM task, i.e., in the Reading the Mind in the Eye Test, were negatively associated with grey matter volume in the bilateral amygdala and hippocampus across both control, ADHD and other groups.Additionally, this study did not find any significant difference between children with ADHD and healthy controls with respect to RMET scores.However, as we found certain methodological issues in this study, it is difficult to put its findings into a meaningful context within the results of EEG studies.
Findings from most EEG studies indicate significantly higher electrophysiological activity in the frontal lobe during emotion recognition tasks in ADHD participants.This aligns with the results of other studies: parts of the frontal lobe, such as the medial prefrontal cortex, were found to be associated with ToM across different age groups (Döhnel et al., 2012;Schuwerk et al., 2014b;Sommer et al., 2007), including children and adolescents (Bowman et al., 2019).Although neural activity in this neural group is consistent with previous works, it remains to be explained why neural activity in the frontal lobe was higher in the ADHD group than in the control group.One possible explanation might rest on the role of the medial prefrontal cortex in attention regulation: several studies revealed that some parts of the frontal lobe, such as the middle frontal gyrus, have a pivotal role in attention regulation (Japee et al., 2015;Song et al., 2019).It is, therefore, possible that to keep a necessary degree of sustained attention to the task at hand, people with ADHD must make an additional mental effort.This additional effort is consequently reflected in increased activity of brain regions related to attention regulation.
Across most of the reviewed studies, pronounced electrophysiological activity in the temporal lobe was observed during emotion recognition tasks in ADHD participants.This finding is in line with the results of previous studies that also showed abnormal neural activity in the temporal lobe among ADHD participants (Rubia et al., 2007;Shafritz et al., 2004).In general, neural activity in the temporal lobe may reflect the processing of emotional expressions and non-verbal cues, crucial components of social interaction (Goghari et al., 2011).It is possible that in individuals with ADHD, the increased neural activity in the temporal lobe could signify an increased cognitive load needed to interpret these emotional cues accurately.Thus, this atypical activity might represent a compensatory response: people with ADHD need to exert more cognitive effort to interpret emotional cues accurately and achieve the same level of social understanding as their non-ADHD peers.
The results of our qualitative synthesis revealed abnormal electrophysiological activity in the occipital lobe during a ToM task in participants with ADHD.The function of this neural area during the ToM task might be the result of brain adaptation compensating the dysfunction of attention networks (Sörös et al., 2017).Taken together, the abnormal electrophysiological activity of the occipital cortex during the ToM task might be the result of this compensatory mechanism.
Our systematic review found that participants with ADHD show higher electrical activity in the parietal lobe when performing emotion recognition tasks.One possible reason for this increased activity could be related to how electrical signals spread throughout the brain.These signals can diffuse through brain tissues and the scalp, making it difficult to pinpoint the exact source of neural activity (Burle et al., 2015).Therefore, the higher activity observed in the parietal lobe in individuals with ADHD might actually be due to signals originating from nearby areas, such as the occipital cortex.
No abnormal activity was found in the anterior cingulate cortex across reviewed studies.This neural area is typically related to ToM across different tasks (Apps et al., 2016) and age groups, including children (Kim et al., 2016;Saxe et al., 2009).Therefore, it is surprising that none of the EEG studies included in this systematic review reported alterations in this neural area.One of the possible explanations for this finding might rest on the spatial resolution of EEG.More specifically, although electrophysiological measures can provide rough information about the source of neural activity in cortical areas, their ability to detect the source of neural activity is decreased in neural structures located deeper in the brain (Balconi and Vanutelli, 2017).Therefore, as the anterior cingulate cortex is a region located deeper under the brain surface, it is possible that EEG studies were not able to detect abnormal neural activity in this neural area Neurobiological results of the analyzed fMRI study of Baribeau et al., (2019) are not congruent with findings from similar study of Sato et al., (2016).Across all study groups, Baribeau (2019) found a negative association between RMET score and the cortical thickness in bilateral amygdala and hippocampus.However, Sato et al., (2016) found a positive association between the RMET score and grey matter volume in the left temporoparietal junction and the left praecuneus but not in the amygdala and hippocampus.One explanation for such a difference may lie in the fact that neural networks of ToM are developing during lifespan (Moriguchi et al., 2007).While the study of Sato et al., (2016) focused on adults, Baribeau et al., (2019) focused on ten year old children.Thus, the results of the latter study may partially reflect a specific developmental stage of ToM.
Although the study of Baribeau et al., (2019) found the association between cortical thickness and the RMET score across all study groups, it failed to find a difference in cortical thickness between children with ADHD and healthy children.Moreover, it also did not find a difference between healthy children and children with ADHD in the ToM task.Collectively, these findings suggest that the recognition of an emotional

Table 2
Results from the quality assessment using the Newcastle-Ottawa Scale and quality assessment items defined in the present study.state does not seem to be a part of ToM which is impaired in children with ADHD.This suggestion is consistent with the results of the study of Mary (2016), which revealed that the performance of children with ADHD and a control group on the RMET was similar when inhibition and attention were statistically controlled.Results of studies (Baribeau et al., 2019;Mary et al., 2016) indicating no difference between ADHD and a control group in RMET score as well as in the underlying neural structure (Baribeau et al., 2019) are of great importance, because for a long time it was unclear which neural mechanisms may cause impairments in ToM in children with ADHD.However, we could expect that it could be caused by abnormalities in the brain structure and/or function underlying attention and subdomains of executive functions such as inhibition.Results support the above mentioned hypothesis.This is in line with findings of some behavioural studies (Mary et al., 2016), which suggest that impaired attention and/or inhibition processes may contribute to ToM deficits in ADHD.
The hypothesis that impairments in attention and/or inhibition processes contribute to deficits in ToM can also be traced on the neural level.Although not included in our review, other studies have linked the Inferior Frontal Gyrus (IFG) to attention and inhibition impairments in ADHD.This neural area processes both attentionn (Cools et al., 2002;Dove et al., 2000;Hampshire and Owen, 2006;Oh et al., 2014;van der Meer et al., 2011) and inhibition (Aron et al., 2014;Buchsbaum et al., 2005;Cai et al., 2014;Drummond et al., 2017;Huster et al., 2013;Jacobson et al., 2011;Swann et al., 2012;Swick et al., 2008) was found to be anatomically (Batty et al., 2010;Carmona et al., 2005;Nickel et al., 2018) and functionally (Dickstein et al., 2006;Janssen et al., 2015;Kumar et al., 2017) impaired in children with ADHD.Interestingly, the IFG also processes ToMfalse belief reasoning (Hartwright et al., 2016;van der Meer et al., 2011), which is also impaired in children with ADHD (Caillies et al., 2014;Saeedi et al., 2014).Taken together, there seems to be overlap in the IFG during ToM processes and attention and/or inhibition neural processes.Because of this neural overlap, it is possible that impairment of attention and/or inhibition of neural processes may negatively influence ToM processes as well which may in turn contribute to impairments in social functioning in children with ADHD.In other words, we may speculate that impairments in IFG contribute to both attention/inhibition and ToM problems in children with ADHD.The aim of the future research should be to test this hypothesis.
One of the strengths of this study is that we checked the risk of bias assessment using two quality measures, a widely acknowledged quality measure as well as a quality measure specific to the aims of this systematic review.Another strength of this study is that unlike many other studies, we calculated the degree of agreement between two raters.The high percentage of inter-rater agreement in our study suggested that inclusion and exclusion criteria were precisely formulated, which consequently decreased the risk of selection bias.The third strength is the fact that we looked for relevant studies in four big databases.Such a number of searched databases is in general twice as much as recommended for systematic reviews (Holly et al., 2016) and decreases the probability that we omitted some studies relevant to our research aims.However, this systematic review has some limitations.First, we included only those studies that examined neural correlates of ToM deficits in the child and adolescent population with ADHD.Therefore, on the systematic review evidence level, it remains unknown which neural mechanisms might stand behind ToM deficits in the adult population with ADHD.Second, the included studies were based on samples from various cultural/linguistic environments.As there are differences in neural activity during the ToM task based on the cultural/linguistic environment (Kobayashi et al., 2007), we cannot exclude the possibility that some of results might have been influenced by this effect.
Regarding research implications, future studies exploring neural bases of ToM impairments in ADHD should focus on neural abnormalities in the IFG.As we suggested, dysfunctions in the IFG can be causally related to ToM impairments in children with ADHD.Therefore, in order to examine this relationship, further studies might use methods such as Continuous Theta Burst Stimulation (cTBS), which allows researchers to infer a causal connection between cerebral functioning and behaviour.
Exploring the neural bases of ToM impairments in children with ADHD may have important therapeutic implications.For example, if ToM deficits in children with ADHD are associated with decreased neural activity in superficial neural areas, the application of repetitive transcranial magnetic stimulation (rTMS) could be used for treatment purposes.This neuromodulatory procedure, already used in the treatment of mood disorders, e.g., depression (Carpenter et al., 2012;Senova et al., 2019), may positively alter neural activity in regions associated with ToM, which could in turn increase their ToM ability and thus their overall social functioning.Also, if there is an association between IFG and ToM, therefore, treatment with attention/disinhibition may improve ToM.

Conclusion
In our systematic review we revealed that despite its important implications, there are only few studies examining the neural bases of Theory of Mind deficits in ADHD.These studies found that children with ADHD exhibit electrophysiological abnormalities in the occipital cortex during a Theory of Mind task.There is only one fMRI study that investigated changes related to ToM in children with ADHD.Therefore, further research is needed to explore the link between Theory of Mind deficits and neural abnormalities in the inferior frontal gyrus in children with ADHD.

Study funding
This study was supported by the Palacký University Olomouc Young Researcher Grant, project The efficacy, effectiveness, and use of Emotion-Focused Therapy in counselling for university students: an experimental study (Contract No. JG 2020 006) and ERDF/ESF project DigiWELL (No. CZ.02.01.01/00/22_008/0004583).The founders did not influence the study design, collection of the data, their analysis or interpretation.

Declaration of Generative AI and AI-assisted technologies in the writing process
During the preparation of this work, the authors used a Generative pre-trained transformer (GPT) deployed on the OpenAI (ChatGPT) to increase the text's readability and for its text suggestions.After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Declaration of Competing Interest
The author(s) have declared that they have no competing or potential conflicts of interest.
L.Novak et al.
(Bawor et al#, 2014)escribed by Bawor et al#(Bawor et al#, 2014): Selection bias -selection of probands, Performance bias 1 -sample size used, Performance bias 2 -controlling for possible confounders, Detection bias 1proper statistical methods used, Detection bias 2 -dealing with missing data, Information bias 1 -methodology of outcome measurement, Information bias 2 -assessment of outcome of interest.Studies were scored on a 4-point scale (0=high risk of bias, 3=low risk of bias), IQintelligence quotient.