Sex dimorphic cortical brain volumes associated with antisocial behavior in young adults

Abstract Background Although sex differences in antisocial behavior are well-documented, the extent to which neuroanatomical differences are related to sex differences in antisocial behavior is unclear. The inconsistent results from different clinical populations exhibiting antisocial behaviors are mainly due to the heterogeneity in etiologies, comorbidity inequality, and small sample size, especially in females. Objective The study aimed to find sexual dimorphic brain regions associated with individual differences in antisocial behavior while avoiding the issues of heterogeneity and sample size. Methods We collected structural neuroimaging data from 281 college students (131 males, 150 females) and analyzed the data using voxel-based morphometry. Results The gray matter volume in three brain regions correlates with self-reported antisocial behavior in males and females differently: the posterior superior temporal sulcus, middle temporal gyrus, and precuneus. The findings have controlled for the total cortical gray matter volume, age, IQ, and socioeconomic status. Additionally, we found a common neural substrate of antisocial behavior in both males and females, extending from the anterior temporal lobe to the insula. Conclusion This is the first neuroanatomical evidence from a large non-clinical sample of young adults. The study suggests that differences in males and females in reading social cues, understanding intentions and emotions, and responding to conflicts may contribute to the modulation of brain morphometry concerning antisocial behavior.


Introduction
Antisocial behavior is a spectrum of behaviors ranging from less se v er e, non-a ggr essiv e rule-br eaking beha viors , suc h as l ying, to the very critical, violent, and a ggr essiv e behaviors as reported in criminal records .T hese beha viors ha ve added burdens to the educational, medical, and judicial systems, as well as bringing afflictions to the victims , families , and society.T herefore , antisocial beha vior has always been under the spotlight of public concerns; and it has attracted research interests from man y div erse disciplines suc h as peda gogy, psyc hology, sociology, and psychiatry (Bennett et al., 2005 ;Blair, 2001 ;Dabbs & Morris, 1990 ;Moffitt et al., 2001 ;Patterson et al., 1989 ;Raine et al., 2011 ).
Of all the studies from those diverse disciplines, one consensus was r eac hed: antisocial behavior is mor e pr e v alent in males than in females throughout lifespan; and such male preponderance was found consistently in terms of behavioral frequency, quantity, and se v erity (Bennett et al., 2005 ;Blair, 2001 ;Dabbs & Morris, 1990 ;Moffitt et al., 2001 ;Patterson et al., 1989 ).In particular, males have an earlier onset of antisocial behavior and commit more antisocial behavior in adolescence; mor eov er, life-course-persistent antisocial behavior is 10 to 14 times more prevalent in males than in females (Moffitt et al., 2001 ;Yildirim & Derksen, 2012 ).
Although the observed behavioral differences are welldocumented, the etiology of the sex differences on antisocial behavior is still unclear.For example, most twin-family studies have reported no sex differences on the genetic and environmental contributions to antisocial behavior (Burt et al., 2019 ).Specifically, for the neur oanatomical substr ates, onl y fiv e studies so far hav e r eported the structur al differ ences between males and females r ele v ant to antisocial behavior (Fairchild et al., 2013 ;Ibrahim et al., 2021 ;Michalska et al., 2015 ;Raine et al., 2011 ;Smar a gdi et al., 2017 ).One study was on adults with antisocial personality disorder, and the rest were on children and adolescents with disruptive behavior disorders, which included conduct disorder and oppositional defiance disorder.Studies hav e r eported sex-by-dia gnosis interactions within various brain regions.Notably, in the frontal cortex, inter actions hav e been observ ed in the superior frontal gyrus (Smar a gdi et al., 2017 ), middle fr ontal gyrus, orbitofr ontal cortex (Raine et al., 2011 ), and v entr omedial pr efr ontal cortex (Ibr ahim et al., 2021 ).Similarly, in the temporal and parietal cortex, the superior temporal sulcus (Michalska et al., 2015 ) and supramarginal gyrus (Ibrahim et al., 2021 ) exhibited such interactions.Furthermor e, additional r egions suc h as the anterior insula (Fairchild et al., 2013 ), amygdala (Ibrahim et al., 2021 ), and parahippocampal cortex (Smar a gdi et al., 2017 ) have also been reported to display the interactions.Ho w ever, there is little consensus on the brain structures specific to the sex differences relevant to antisocial behavior, and the div er gence on a same finding is hard to explain.This could be due to the differences in disease-specific etiologies, comorbidities, medication, age, small sample size of the female patients, and other control factors such as IQ and socioeconomic status (SES).
Ther efor e, in the current study, we aim to examine the associations between sex differences in brain structure in males and females and antisocial behavior se v erity in a college sample.In particular, we ar e inter ested in finding br ain r egions in whic h gr ay matter volume covaries with the individual differences in selfreported antisocial behavior in males and females differ entl y.To further validate the findings, we tested whether the effects were still there after regressing out other variations in age, IQ, SES, and total cer ebr al gr a y matter volume .Besides , we also c hec ked the common neural correlates of antisocial behavior for males and females.

Participants
A total of 281 under gr aduates fr om Beijing Normal Univ ersity participated in the study, including 131 males and 150 females.All participants had no diagnosed mental disorder history, such as conduct disorder and antisocial personality disorder, neurological diseases (e pile psy , brain injury , neur odegener ativ e diseases, and cer ebr ov ascular diseases), or any intellectual disability before taking the tests.All behavioral and magnetic resonance imaging (MRI) experiments were approved by the Ethics Committee of Beijing Normal University, and informed consents were obtained prior to the experiments.One participant was excluded from data analysis because of incomplete data on the antisocial questionnaire, and six other participants were removed because they did not participate in the following IQ test or the SES questionnaire.One participant was excluded because his score on antisocial behavior fell out by three standard deviations; similarly, another participant was excluded because the IQ score fell out by three standard deviations.In addition, six other participants had excessive head movement during the scan and therefor e wer e not included in the anal yses.Taken together, we r eported the results from 266 participants, including 120 males, with an av er a ge a ge of 22 (21.7 ± 1.01), and 146 females with an aver a ge a ge of 22 (21.6 ± 1.04).Ther e wer e no significant differences between males and females in age ( t = 0.77, P = 0.44), IQ ( t = −0.90,P = 0.37), and SES ( t = −1.44,P = 0.15), as seen in Table 1 .

Antisocial behavior
The self-reported antisocial behavior was measured using the antisocial behavior subscale of the Behavioral Indicators of Conscientiousness (BIC) by Jackson et al. ( 2010 ).The BIC is used to measure behaviors related to conscientiousness, including impulsivity, organization, cleanliness, antisocial, and other seven subscales .T he antisocial subscale contains 23 items in total, sampling antisocial behaviors in daily life, such as breaking rules in a game .T he antisocial subscale has a significant negativ e corr elation with industriousness, organization, impulsive control, and r esponsiv eness, as well as a significant negativ e corr elation with the a gr eeableness in the Big Fiv e personality ( r = −0.45)(Jackson et al., 2010 ).In this study, we used 15 items from it because the other eight items do not describe the the sample we tested and ther efor e could affect the test validity: it is r ar e to see Chinese students drinking during working hours and swearing when talking to the boss.As such, the internal consistency coefficient of the 15item questionnair e r emains high: Cr onbac h's α = 0.8.During the test, participants were asked to answer the frequency of their enga gement in man y differ ent antisocial behaviors using a six-point scale to score from 1 "never" to 6 "alwa ys ."A higher score means more antisocial behavior, and a lo w er score means less antisocial behavior.

IQ
The intelligence test was performed with the Raven's Advance Pr ogr ess Matrices (Carpenter et al., 1990 ).It is suited to college students as they have proven abo ve-a verage intelligence .T he test included a series of geometric gr a phs arr anged in a logical sequence and the participants had to choose the consequent gr a ph.Ther e were 48 items in the test: 12 practice items and 36 test items.We took the normalized scores in the subsequent behavior al-br ain corr elational anal ysis.

SES
The SES was measured with the MacArthur Scale of Subjective Social Status (Adler et al., 2000 ;Goodman et al., 2001 ).The scale is presented to participants with a 10-ste p lad der.The highest le v el of the ladder r epr esents the famil y with the highest standard of living in the area where the tested family is located-the highest income, the highest le v el of education, and the highest social status; the lo w est le v el of the ladder r epr esents the family with the lo w est standar d of living in the region.The higher the le v el of the family, the closer it is to the family with the highest standard of living; the lo w er the le v el of the famil y, the closer it is to the family with the lo w est standar d of living.Participants were asked to dr aw a cr oss on the corr esponding le v el based on their families' le v els in the local areas during high school, as previous literature (Moffitt et al., 2001 ) has shown that there is a high incidence of antisocial behavior during adolescence, when teenagers are vulnerable to environmental factors such as their SES.
All the statistical analyses of the behavioral data were performed with MATLAB v.R2013a (MathWorks, Natick, MA, USA).

MRI data collection
The MRI data were collected at the Brain Imaging Center of Beijing Normal University.The scanner was a 3 T Siemens magnetic resonance scanner (Siemens 3 T scanner, MA GENTOM Trio , a Tim system), with a 12-channel phased array head coils.The magnetic resonance structural images were acquired using a threedimensional ma gnetization pr epar ed r a pid gr adient ec ho (MP-RAGE) T1 weighted sequence (TR/TE/TI = 2530/3.39/1100ms, flip angle = 7 • , FOV = 256 × 256 mm).One scan covering the whole brain yielded 128 consecutive sagittal slices with a 1 × 1 mm planar resolution and 1.33 mm thickness.
First, the image quality was checked to avoid the artifacts caused by excessive head movements and the abnormal brain anatomy.Second, the origin was manually marked at the anterior commissur e for eac h participant for better quality control.Third, we use a unified segmentation a ppr oac h (Ashburner & Friston, 2005 ) to segment the images into gray matter, white matter, cerebrospinal fluid, and the other parts (e.g.skull and scalp).Fourth, the gray matter images were realigned, resampled, and spatially normalized in the MNI152 space using the study-specific template gener ated thr ough the "diffeomor phic anatomical r egistr ation through exponential lie algebra method" (Ashburner, 2007 ).Fifth, the gray matter volume value on each v oxel w as adjusted by m ultipl ying the Jacobian coefficient r esulting fr om the normalization process to retain the volume of the tissues at a r elativ el y constant le v el after war ping.Sixth, the gener ated v alue ma p was Gaussian smoothed with a full-width at half-maximum at 8 mm.Finally, as none of the reported brain regions from previous antisocial behavior studies were located in the cerebellum, we masked out the cerebellum using a gray matter probability threshold of 0.2.T hus , for each voxel in the cer ebrum, one v alue of the gray matter volume was pr epar ed for the subsequent r egr ession analyses.
First, to find out whether there were any neuroanatomical correlates of antisocial behavior, regardless of sex differences, we mixed male and female participants and constructed a general linear model (GLM).In the GLM model.We set the antisocial behavior as the only variable of interest; age, IQ, SES, and the aver a ged total cer ebr al gr ay matter volume wer e v ariables of noninter est.The r egional gr ay matter v olume w as taken as the dependent variable.
Mor e importantl y, as we hypothesized that males and females hav e differ ent antisocial behavior-r elated br ain mor phometry, two identical independent GLMs were made to fit the relationships between antisocial behavior and regional gray matter in males and females, r espectiv el y.The antisocial behavior was set as the v ariable of inter est, and a ge , IQ, SES, and the a v er a ged total cer ebr al gr ay matter v olume w er e set as v ariables of non-inter est as well.
T hus , the coefficients of the antisocial behavior in the GLMs stand for the degree to which the regional gray matter volume covaries with the antisocial behavior.As w e w er e mainl y interested in finding br ain r egions sensitiv e to the differ ences in males' and females' antisocial behavior, the antisocial coefficients for the males and females were compared using the F -test.For the multiple comparison correction, first, we applied a threshold of P < 0.01 voxel-wise for the whole brain.
We conducted an explor atory whole-br ain anal ysis r ather than functional region of interest analysis due to the inconsistent results in pr e vious studies (Fairc hild et al., 2013 ;Mic halska et al., 2015 ;Raine et al., 2011 ;Smar a gdi et al., 2017 ).For the cluster-le v el correction, first, we performed the standard Monte Carlo simulation (Ward, 2000 ) using the Analysis of Functional NeuroImages ( http:// afni.nimh.nih.gov/afni/ ) 3dClustsim to determine the cluster-le v el thr eshold.According to the corr ection, onl y when a cluster is bigger than the minimum cluster size determined by the Monte Carlo sim ulation, whic h was 322 voxels, could it be reported as significant, and the probability of making a type I error is < 0.01.All the reported clusters appeared after the multiple comparison correction in the whole cerebrum, and the Z scores for each cluster were reported as well as the cluster sizes.

Post hoc visualization
To see how exactly males' and females' gray matter volumes covary with antisocial behavior, we scatter-plotted the correlation of antisocial behavior scores and the gray matter volumes for males and females separ atel y in the clusters acquired in the previous step.The correlational plots have also controlled for age, IQ, SES, and the av er a ged total cer ebr al gr ay matter volume.

Antisocial behavior
Ov er all, the individual differ ences among the total 266 participants w ere b y no means small: the minimum w as 15 and the maximum 49.The av er a ge scor e was 32.77, with a standard deviation of 7.06.The ov er all distribution follows a normal distribution, as the K olmogoro v-Smirno v test indicated [ D (266) = 0.07, P = 0.14].
For the 120 male participants, the minimum score was 19, the maximum was 49, and the average was 35.03, with a standard deviation of 6.61, following a normal distribution [( D (120) = 0.06, P = 0.81].
For the 146 female participants, the minimum was 15, the maximum was 47, the average was 30.91, with a standard deviation of 6.91.It follo w ed a normal distribution [ D (146) = 0.10, P = 0.13] as well.

Sex differences in antisocial behavior
Sex and antisocial beha vior ha ve shown significant interactions on the regional gray matter volume of the brain.From Table 2 , we can see that these brain regions were reported in important social cognitive and emotional functions, such as the posterior part of the temporal gyrus extending to angular gyrus, especially around the posterior superior temporal sulcus (pSTS) (peak coordinates: 58, −44,18; cluster size 4272 mm 3 ), the middle temporal gyrus (MTG), which extended to the inferior temporal part (peak coordinates: 66, −10, −26; cluster size 2648 mm 3 ), and the precuneus (PC) (peak coordinates: 6, −68, 36; cluster size 3296 mm 3 ).
First, the biggest interaction effect was around the pSTS.We can see a significant positive correlation ( r = 0.36, P = 6.50 × 10 −5 ) between males' antisocial behavior and the gray matter volume in this cluster (see Fig. 2 ).Ho w e v er, this corr elation in females was the opposite ( r = −0.14,P = 0.10).Ther efor e, incr eased gr ay matter volume in the pSTS was significantly related to increased antisocial behavior in males, yet a decreased trend for antisocial behavior in females.
Similarly, in the MTG, the correlation between males' antisocial behavior and the gray matter volume within the cluster was 0.27 ( P = 0.004), showing that a greater gray matter volume in the MTG cluster was associated with more antisocial behavior in males.On    the contrary, in females it sho w ed a negativ e tr end ( r = −0.13,P = 0.12).The plot is shown in Fig. 3 .
Ho w e v er, in the PC, the male antisocial behavior and regional gray matter volume showed a significant negative correlation ( r = −0.31;P = 7.89 × 10 −04 ); whereas in females, the correlation was positiv e ne v ertheless not significant ( r = 0.14, P = 0.11).As shown in Fig. 4 , such results sho w ed that in the pr ecuneus, mor e antisocial behavior was associated with reduced gray matter volume in males; yet for females more antisocial behavior was associated with increased gray matter volume as a trend.

Antisocial behavior
In line with pr e vious studies (Aoki et al., 2013 ;De Brito et al., 2009 ;Raine, 2019 ), we found that antisocial behavior was related to the gray matter volume in the anterior part of the temporal lobe and insula.Specifically, the cluster size is 2768 mm 3 (peak: −50 0 −8) as described in Table 3 .On the temporal lobe, the cluster extends from the planum polare, the anterior part of the superior temporal  gyrus, to the temporal pole; meanwhile, the cluster also extends to the insula from its posterior to the anterior part.The cluster centered at its peak location is illustrated in Fig. 5 .The results sho w ed that there is a common neuroanatomical basis for antisocial behavior in both males and females in a large non-clinical sample of young adults, and this basis is mainly located in the left anterior part of the superior temporal gyrus and the insula.
Post hoc analyses showed that, for males, after controlling for age, IQ, SES, and the total cerebral gray matter volume, there was a significant correlation ( r = 0.27, P = 0.0029) of antisocial behavior and the gray matter volume of the cluster.Similarly, for females such a correlation was also significant ( r = 0.18, P = 0.0337).This indicated that both male and female antisocial behavior could be modulated by this region.

Discussion
The current study provides the first evidence of sex differences in brain morphometry related to self-reported antisocial behavior in a large non-clinical sample of young adults.First, we hav e r eplicated the sex differ ences in antisocial behavior showing that males demonstrated more antisocial behavior than females .Moreo ver, further results revealed that the pSTS, MTG, and PC were the primary brain regions where males and females ex-hibited different patterns of correlations between antisocial behavior and regional gray matter v olume.Additionally, w e have also found the common neur al substr ates of antisocial behavior for both males and females around the insula.As these regions related to sex differences in antisocial behavior are generally involved in several crucial social functions, such as theory of mind, empathy, and moral judgment, it indicates that the sex differences in these functions may influence antisocial behavior by shaping the brain's gray matter volume during development.
The sex differences in antisocial beha vior ha ve been wellestablished and many studies have tried to explain the phenomenon by exploring the risk factors behind it.Basicall y, ther e ar e thr ee main explanations for why males ar e mor e antisocial than females (Moffitt et al., 2001 ).First, males ar e mor e susceptible to risk factors than females, as reflected in the higher correlation between antisocial behavior and risk factors in males than in females .T his is a similar to our findings the correlations between regional gray matter volume and antisocial behavior are al ways str onger in males than in females .T he second explanation is that males are exposed to higher levels of risk factors than females: for c hildhood pr oblematic beha viors , males ha ve higher pr e v alence of ADHD than females; in peer relationships, males hav e mor e misbehaving peers.Finall y, the differ ences in personalities and propensities between males and females may also be the cause.In personality traits closely related to antisocial behavior, ther e ar e significant sex differ ences: e .g., males ha v e mor e negative emotions and are worse at self-control.The differences between males and females in these two aspects can explain 96% of the differences in the antisocial behavior in adolescents, and 78% of the differences in the chance of de v eloping a conduct disorder.
For the common neural basis, we found that the brain region gener all y associated with antisocial behavior was around the insula and temporal pole .T he findings on the insula were consistent with a former VBM study in males but not females (Fairchild et al., 2013 ): males with conduct disorder had more gray matter volume in the left anterior insula than normal controls while females had less .T he differ ent r esult in females could be due to small sample size and high comorbidity of ADHD and major de pressi ve disorder (MDD) in adolescent girls with conduct disor der, accor ding to the authors .T her efor e, our r esults fr om a lar ge non-clinical sample might be mor e likel y to pr ovide a stable normative effect.The anterior part of the insula is a classic functional area in emotional processing.It is responsible for interoce pti ve feelings and experiences participating in various emotional pr ocessing suc h as anger and fear (Craig, 2009 ).It is also str ongl y connected with the am ygdala (Me yer-Lindenberg & Tost, 2012 ; Naqvi & Bec har a, 2009 ;Sescousse et al., 2013 ).Antisocial offenders have deficits in emotional processing: in fMRI studies, antisocial offenders have less activ e br ain activities in r egions involv ed in emotional processing than normal participants, and are less sensitive to the negative emotions of others (Herpertz et al., 2008 ;Sadeh et al., 2011 ).This neurobiological basis indicates that they are more likely to ignore others' emotions, hurt others, or violate socially accepted mor al standards.Furthermor e, the anterior insula and temporal pole are also the core brain regions of empathy (de Vignemont & Singer, 2006 ;F an et al. , 2011 ;Morita et al. , 2013 ;Ponz et al. , 2013 ;Vogt, 2005 ).Empathy is an important prosocial quality.It includes the cognitive ability to see the other's perspective, and also the emotional ability to share another's emotional experience .T he deficit in empathy is a common cause for many mental illnesses and social maladaptation.As antisocial behavior is characterized as callous and unemotional, showing a lack of concern for others' suffering, which is closely associated with the functions of the anterior insula and temporal pole, it could be that the antisocial behavior is the manifestation of the dysfunction in these cor e br ain r egions.Taken together, the functional deficits in emotional processing and empathy may affect antisocial behavior in general.
Mor e importantl y, we found sex dimor phic cortical br ain volumes associated with antisocial behavior around the pSTS, MTG, and precuneus.
First, the right pSTS is located at the posterior part of the superior temporal sulcus and is connected to the temporoparietal junction (TPJ) and the inferior parietal lobe.It is fr equentl y r eported in various social cognitive brain imaging studies.Fundamentally, the pSTS is part of the human mirror neuron system and responsible for understanding facial expressions , actions , and the recognition of action intentions (Rizzolatti & Craighero, 2004 ;Van Overwalle & Baetens, 2009 ).Based on this, it is also involved in higher-le v el social cognitiv e pr ocessing suc h as social perception, theory of mind, empathy, moral judgment, and decision-making.Our results sho w ed that there was a significant positive correlation between male antisocial behavior and the gray matter volume around the pSTS.This is consistent with pr e vious studies reporting increased gray matter volume in boys with conduct disorder and callous-unemotional traits (De Brito et al., 2009 ).Additionall y, Ibr ahim et al. ( 2021 ) and Smar a gdi et al. ( 2017 ) also reported a close-by brain region, the supramarginal gyrus, which exhibited a sex-by-diagnosis interaction on cortical thickness in children and adolescents diagnosed with disruptive or conduct disorders.Ne v ertheless, this inter action manifested as a r eduction in cortical thickness in males, which was contrary to our findings.The inconsistencies could be due to the comorbidity mentioned in the articles, such as ADHD, depression, and also the differences in substance use, medication, and age of onset of the diseases.Despite the differences, our study and these previous studies in patients commonly found the crucial location, which is around the pSTS.Mor eov er, this r egion was r eported in the r e vie ws of the neural basis of antisocial behavior in 2008 and 2013 (Aoki et al., 2013 ;Raine, 2008 ).According to the r e vie ws, male antisocial beha vior ma y be regulated by the pSTS thr ough se v er al social cognitive abilities such as social perception, theory of mind, empathy, and moral judgment.Deficits in these functions could lead to antisocial behavior.For example, poor social perception and theory of mind abilities may cause wrong attribution of others' intentions, resulting in aggressive behavior.Lack of empathy may lead to ignorance of others' sufferings and thus unable to stop hurting behavior.It also happened that males were less empathetic than females (Adenzato et al., 2017 ), which could be an explanation of the sex differences in brain structures related to empathy.The finding supported the neur omor al theory of antisocial behavior (Raine , 2019 ).Moreo ver, our findings refined previous reviewsby showing that this correlation was more specific to males, while females sho w ed the opposite trend and w ere less affected.One possible argument could be that, since females sho w ed less antisocial behavior than males, the non-significant correlation could be due to the lack of individual differences in the distribution of antisocial behavior in females.Ho w e v er, fr om the scatter plot in Fig. 2 , we can see that the individual differences in female's antisocial behavior were no less than in males: all the female participants shown in red dots are not densely concentrated in one ar ea.Besides, the standard de viations in males and females ar e also comparable .T herefore , the sex differences in pSTS gray matter volume are likely to reflect sex-selective social deficits such as empathy in males.
Mor eov er, our r esults sho w ed that males exhibiting more antisocial beha vior ha v e mor e gr ay matter volume in the MTG.This is consistent with the pr e vious study (Yang et al., 2015 ) that adolescents with higher psychopathic traits have higher cortical thickness in MTG, and this is specifically found in males .Besides , functional neur oima ging studies also r eported abnormal neur al activities in the MTG in male violent offenders (Gregory et al., 2015 ;K umari et al. , 2009 ).Similar to the pSTS, the MTG has also been fr equentl y r eported in man y studies r ele v ant to social cognitiv e functions.In particular, a pr e vious meta-anal ysis (Bzdok et al., 2012 ) has r e v ealed that the right MTG is one of a few robust regions commonl y involv ed in mor al cognition, theory of mind, and empathy-the other regions are the dorsomedial prefrontal cortex and bilateral TPJ.As has been discussed before, moral reasoning, theory of mind, and empathy are three crucial abilities closely related to antisocial beha vior.T he sex differences in these functions may result in the divergence in the development of the MTG morphometry.
Notably, as the pSTS and MTG are both within the temporal lobe , we ha v e to ac knowledge that sex differ ences in the de v elopment of the structures of the temporal lobe could influence antisocial behavior (Michalska et al., 2015 ).In line with the positive correlation of antisocial behavior and gray matter volume in the temporal lobe (pSTS and MTG) in males, we also found similar results in other populations: children with callous-unemotional traits and conduct disorder have more gray matter in the temporal lobe than normal c hildr en (De Brito et al., 2009 ).Children and adults may have completely different patterns of correlation due to de v elopmental delay or functional deficits, showing increased or r educed gr ay matter volume in various locations in the temporal lobe .T his could explain why our results in young adults differ from some of those in children (Michalska et al., 2015 ) and adolescents (Fairchild et al., 2013 ;Smar a gdi et al., 2017 ).
The precuneus has been less discussed in previous neuroimaging studies of antisocial behavior, although it is an important part of the default mode network (DMN).Our study found that males sho w ed significant gray matter volume reduction in the precuneus with the increase of antisocial behavior, which is consistent with former studies on antisocial clinical samples reporting significant loss in the gray matter volume in the precuneus (Raine et al., 2000 ).The precuneus and the posterior cingulate gyrus are adjacent to each other and form one of the core nodes of the DMN.They also have strong functional connectivity with the medial pr efr ontal cortex and the later al tempor al cortex (Buc kner et al., 2008 ).The DMN is a set of brain regions that show spontaneous neural activities without performing any external tasks.It includes the medial pr efr ontal cortex, pr ecuneus and posterior cingulate gyrus, IPL, later al tempor al cortex, and par ahippocampal gyrus .T hese br ain r egions ar e well connected and ar e mainl y involved in functions involving self-r efer ential pr ocessing, suc h as autobiogr a phical memory and moral decision-making (Buckner et al., 2008 ).A recent study of sex differences in adult human brains (based on > 5000 individuals) sho w ed that males have weaker connectivity within the DMN nodes than females (Ritchie et al., 2018 ), whic h indir ectl y supports our finding in a male deficit perspective.
While the study may shed light on sex differences in brain mor phometry r elated to self-r eported antisocial behavior, se v er al limitations warr ant consider ation.First, the cr oss-sectional design restricts the ability to establish causal relationships between br ain mor phometry and antisocial beha vior.T he lack of longitudinal data impedes the identification of de v elopmental tr ajectories and precludes definitive conclusions about whether changes in br ain structur e pr ecede or follow the onset of antisocial behavior.Additionall y, the r eliance on self-r eported measur es for antisocial behavior may introduce biases or underreporting, as individuals might be hesitant to disclose sensitive or socially undesirable behaviors accurately.Furthermore, the generalizability of the findings might be limited since the study pr edominantl y focuses on young adults from a non-clinical population, which does not full y r epr esent individuals with clinicall y significant antisocial beha viors .Futur e r esearc h employing longitudinal designs, objectiv e measur es of antisocial beha viors , and a mor e div erse sample could provide a more comprehensive understanding of the nuanced relationship between brain morphometry and antisocial behavior across different populations and developmental stages.
Taken together, our study identified the neuroanatomical correlates of the sex differences in antisocial behavior in a large non-clinical sample, and the effect is independent of age, IQ, SES, and total cer ebr al gr a y matter volume .As these brain regions are mainl y involv ed in emotional pr ocessing, theory of mind, empathy, moral judgment, and self-referential processing, the sex differences in these functions might modulate antisocial behavior thr ough the de v elopment of the br ain mor phometry.This study pr ovides a neur oanatomical perspectiv e on the male preponderance of antisocial behavior and contributes evidence to the bio-logical basis of sex differences in antisocial behavior from a large non-clinical young adult sample.

Figure 1 :
Figure 1: Sex differences in antisocial behavior.The y -axis shows the participants' total antisocial scores from the BIC, measuring the frequency of a variety of antisocial behaviors; the two bars on the x -axis denote male (in blue) and female (in red) groups.* * * P < 0.001.

Figure 2 :
Figure 2: Sex and antisocial behavior interaction in pSTS.The left panel of the figure is the right pSTS cluster (peak point: 58 −44 18, cluster size 4272 mm 3).On the right panel are the scatter plots depicting the correlations of individual differences in antisocial behavior and the av er a ge gr ay matter volume within the cluster in males and females.Each blue dot in the scatter plot represents each male participant given his scores on the x -and y -axes.Similarly, each red dot represents each female participant.The blue straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in males, controlling for age, IQ, SES, and the total cerebral gray matter volume ( r = 0.36, P = 6.50 × 10 −5 ).The red straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in females, controlling for the same factors as stated above ( r = −0.14,P = 0.10).

Figure 3 :
Figure 3: Sex and antisocial behavior interaction in MTG.The left panel of the figure is the right MTG cluster (peak point: 66 −10 −26, cluster size 2648 mm 3).On the right panel are the scatter plots depicting the correlations of individual differences in antisocial behavior and the av er a ge gr ay matter volume within the cluster in males and females.Each blue dot in the scatter plot represents each male participant given his scores on the x -and y -axes.Similarly, each red dot represents each female participant.The blue straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in males, controlling for age, IQ, SES, and the total cerebral gray matter volume ( r = 0.27, P = 0.004).The red straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in females, controlling for the same factors as stated above ( r = −0.13,P = 0.12).

Figure 4 :
Figure4: Sex and antisocial behavior interaction in precuneus .T he left panel of the figure is the precuneus cluster (peak point: 6 −68 36, cluster size 3296 mm 3 ).On the right panel are the scatter plots depicting the correlations of individual differences in antisocial behavior and the av er a ge gr ay matter volume within the cluster in males and females.Each blue dot in the scatter plot represents each male participant given his scores on the xand y -axes.Similarly, each red dot represents each female participant.The blue straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in males, controlling for age, IQ, SES, and the total cerebral gray matter volume ( r = −0.31;P = 7.89 × 10 −04 ).The red straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in females, controlling for the same factors as stated above ( r = 0.14, P = 0.11).

Figur e 5 :
Figur e 5: T he main effect of antisocial beha vior.T he left panel of the figur e shows the cluster in the planum polar e, extending to the insula and tempor al pole, whic h is depicted in gr een (cluster size = 2768 mm 3 ).The center of the cluster is at the peak point (MNI coordinates −50 0 −8).The scatter plots on the right panel depict the correlations of individual differences in antisocial behavior and the av er a ge gr ay matter volume within the cluster in males and females.Each blue dot in the scatter plot represents each male participant given his scores on the x -and y -axes.Similarly, each red dot represents each female participant.The blue straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in males, controlling for age, IQ, SES, and the total cerebral gray matter volume ( r = 0.27, P = 0.0029).The red straight line stands for the correlation of antisocial behavior and the cluster gray matter volume in females, controlling for the same factors as stated above ( r = 0.18, P = 0.03).

Table 1 :
Demogr a phic c har acteristics of participants.

Table 2 :
Brain regions showing sex differences relating to antisocial behavior.