Brain structure links trait conscientiousness to academic performance

In the long history of identifying factors to predict academic performance, conscientiousness, a so-called ‘big five’ personality trait describing self-regulation and goal-directed behavior, has emerged as a stable predictor for this purpose. However, the neuroanatomical substrates of trait conscientiousness and the underlying brain mechanism linking trait conscientiousness and academic performance are still largely unknown. Here, we examined these issues in 148 high school students within the same grade by estimating cortical gray matter volume (GMV) utilizing a voxel-based morphometry method based on structural magnetic resonance imaging. A whole-brain regression analysis showed that trait conscientiousness was positively associated with the GMV in the bilateral superior parietal lobe (SPL) and was negatively associated with the GMV in the right middle frontal gyrus (MFG). Furthermore, mediation analysis revealed that trait conscientiousness mediated the influences of the SPL and MFG volume on academic performance. Importantly, our results persisted even when we adjusted for general intelligence, family socioeconomic status and ‘big five’ personality traits other than conscientiousness. Altogether, our study suggests that the GMV in the frontoparietal network is a neurostructural marker of adolescents’ conscientiousness and reveals a potential brain-personality-achievement pathway for predicting academic performance in which gray matter structures affect academic performance through trait conscientiousness.

In the long history of identifying factors to predict academic performance, conscientiousness, a socalled 'big five' personality trait describing self-regulation and goal-directed behavior, has emerged as a stable predictor for this purpose. However, the neuroanatomical substrates of trait conscientiousness and the underlying brain mechanism linking trait conscientiousness and academic performance are still largely unknown. Here, we examined these issues in 148 high school students within the same grade by estimating cortical gray matter volume (GMV) utilizing a voxel-based morphometry method based on structural magnetic resonance imaging. A whole-brain regression analysis showed that trait conscientiousness was positively associated with the GMV in the bilateral superior parietal lobe (SPL) and was negatively associated with the GMV in the right middle frontal gyrus (MFG). Furthermore, mediation analysis revealed that trait conscientiousness mediated the influences of the SPL and MFG volume on academic performance. Importantly, our results persisted even when we adjusted for general intelligence, family socioeconomic status and 'big five' personality traits other than conscientiousness. Altogether, our study suggests that the GMV in the frontoparietal network is a neurostructural marker of adolescents' conscientiousness and reveals a potential brain-personality-achievement pathway for predicting academic performance in which gray matter structures affect academic performance through trait conscientiousness.
In the fields of education and psychology, there is a long history of exploring the predictors of academic performance 1 , which is generally assessed using grade point average (GPA) or standardized examinations (e.g., the Stanford Achievement Test and the Achievement College Test) 2,3 . It is well established that a wealth of psychosocial factors, such as family background (e.g., socioeconomic status, SES), general cognitive ability (e.g., intelligence level), motivation and personality traits, are reliable predictors of academic performance [4][5][6][7][8][9][10] . Conscientiousness, which is one's tendency to be goal directed, disciplined, organized and achievement focused, is an aspect of the well-known 'big five' personality model 11 and has been repeatedly shown to be related to academic performance 5,6,8 . For example, a meta-analysis based on 138 samples from 70,926 participants has revealed that trait conscientiousness is moderately associated with students' academic performance and that this association is independent of general intelligence 5 . Moreover, a recent systematic review of 38 meta-analyses found that among the 16 personality variables related to academic performance, trait conscientiousness showed the largest absolute effect size in the prediction of academic performance 8 . Furthermore, evidence from many longitudinal studies has suggested that trait conscientiousness plays a predictive and causal role in academic performance [12][13][14] .
Here, we employed structural magnetic resonance imaging (sMRI) to examine the neuroanatomical bases of trait conscientiousness and then explored the underlying brain mechanism linking trait conscientiousness and academic performance. We adopted a multidimensional approach (i.e., a brain-personality-achievement approach) to explore the relationships between brain structure, trait conscientiousness and academic performance in a group of high school students (N = 148).
www.nature.com/scientificreports www.nature.com/scientificreports/ Behavioral tests. NEO Five-Factor Inventory (NEO-FFI). The individual differences in trait conscientiousness were assessed using the conscientiousness subscale of the NEO-FFI, which is a popular measure for 'big five' personality traits 11 . As a test for the specificity of our results, the other NEO-FFI subscales (i.e., extraversion, neuroticism, openness and agreeableness) were also administered to the participants. The NEO-FFI is a 5-point Likert-type self-report questionnaire that includes 60 items, with 12 items for each subscale. This inventory has been repeatedly used in different Chinese populations and has been shown to have satisfactory validity and reliability [46][47][48] . In this study, the internal consistency (i.e., Cronbach's α) of each subscale of the NEO-FFI was adequate: conscientiousness (α = 0.79), extraversion (α = 0.80), neuroticism (α = 0.81), openness (α = 0.74) and agreeableness (α = 0.72).
Chinese National University Entrance Exam (CNUEE). The participants' academic performance was evaluated using their CNUEE scores 49,50 , which were collected from the databases of the high schools the students attended. The CNUEE is a standardized and scaled measure that consists of six curriculum subjects (i.e., Chinese, Mathematics, English, Physics, Chemistry and Biology). CNUEE scores range between 0 and 750, allowing comparison among students in the same grade and providing the sole criterion for admission to Chinese universities 50,51 . All participants took the CNUEE in June 2015.
Raven's Advanced Progressive Matrices (RAPM). The 36-item RAPM test 52 , a measure of abstract reasoning ability, was included in the current study to investigate whether general intelligence could affect the associations between trait conscientiousness, gray matter structure and academic performance. For a given item, participants were presented with a picture matrix that was missing one part and were required to select the missing part among 8 options. Participants were required to complete this pencil-and-paper test within 30 minutes 43,53 . RAPM scores were derived based on the number of correct answers, with higher scores suggesting higher levels of general intelligence. In this experiment, the internal consistency of the RAPM was acceptable (α = 0.80).

Socioeconomic Status Scale (SSS).
To exclude the potential effects of family SES on the relations among trait conscientiousness, gray matter structure and academic performance, we employed the SSS, which is a single-item scale presenting participants with a drawing of a 10-rung ladder 54 . This scale reflects participants' perceptions of their parents' socioeconomic status, including three aspects: education, occupational prestige and income 54 . During the testing, each participant was instructed to indicate the overall level of his/her parents' socioeconomic status, ranging from 1 (bottom rung) to 10 (top rung). Increasing evidence has revealed that compared to objective measures, SSS is a superior predictor of health-linked outcomes 55 . This scale has been widely used in Chinese populations 56,57 . image acquisition and preprocessing. Image acquisition. The sMRI scans were performed with a 3.0 T Siemens-Trio Erlangen scanner (Germany) with a 12-channel head coil. Using a whole-head magnetization-prepared rapid gradient-echo sequence, each participant underwent a T1-weighted structural scan with the following parameters: inversion time = 900 ms, repetition time = 1900 ms, echo time = 2.26 ms, slice thickness = 1 mm, flip angle = 9°, matrix size = 256 × 256, 1 mm isotropic resolution and 176 slices.
Image preprocessing. Preprocessing of images was conducted with SPM12 39 . All of the images were first displayed in SPM12 to check for gross anatomical abnormalities or artifacts, and two participants were excluded due to abnormal brain structure. For a more accurate registration, each image was manually reoriented, set to the anterior commissure, and then segmented into three tissue groups (i.e., gray matter, white matter and cerebrospinal fluid) by employing the new segmentation tool in SPM12. Afterwards, registration, normalization and modulation analyses were conducted with DARTEL (diffeomorphic anatomical registration through exponentiated lie) algebra 58 in SPM12. The gray matter data were aligned, resampled to 2 × 2 × 2 mm 3 , and then transformed to Montreal Neurological Institute (MNI) space. The inverse Jacobian of the local transformations was used to modulate the segmented gray matter data, which allows the volume measurements to be preserved. Subsequently, the normalized and modulated gray matter data were smoothed with an 8-mm full width at half maximum Gaussian kernel. Finally, the resulting data were masked with an absolute threshold masking of 0.2 to remove edge effects around the borders between gray matter and white matter 34,43,59 . In particular, to ensure that there were no crucial anomalies in GMV estimation, a medical radiologist who was blinded to the current study performed visual quality control of each participant's images in each step. No participants were excluded in this process.

Statistical analyses. GMV-behavior correlation analysis.
To identify the brain regions for which GMV is linked with trait conscientiousness, we conducted a whole-brain regression analysis using NEO-FFI conscientiousness scores as the independent variable; voxelwise GMV as the dependent variable; and gender, age and total GMV as the control variables. Additionally, to investigate gender differences in the relations between trait conscientiousness and gray matter structure, we conducted a condition-by-covariate interaction analysis 43,59,60 using gender as a condition, NEO-FFI conscientiousness scores as the variable of interest and age and total GMV as the variables of no interest. Nonstationary cluster correction was employed to infer the regions of significance, as previous studies have demonstrated that the VBM data are nonstationary (e.g., not uniformly smooth) 37,61 . Specifically, the results were corrected with a threshold (cluster level: p < 0.05; voxel level: p < 0.0025; nonstationary cluster correction) that has been successfully applied to VBM data 34,42,62 . The above analyses were performed with SPM12.
Prediction analysis. To validate the predictability of trait conscientiousness from brain structures, we implemented a balanced fourfold cross-validation procedure utilizing a machine learning method 34,43,44,57,63,64 . For the www.nature.com/scientificreports www.nature.com/scientificreports/ analysis, a linear regression algorithm was performed using NEO-FFI conscientiousness scores as the dependent variable and the GMV of the brain region as the independent variable. To evaluate how well the dependent variable could be predicted by the independent variable, we calculated r (predicted, observed) using a balanced fourfold cross-validation method 34,43,44,57,63,64 . The data were first divided into four subsets to guarantee that there were no significant differences among the distributions of these variables across subsets. Then, the data from three subsets were used to build a linear regression model, with one subset left out. This model was further employed to predict the unused data subset. The value r (predicted, observed) , which represents the correlation of the actual observed data and the predicted data, was finally obtained after the data of all subsets had been predicted. The significance of r (predicted, observed) was determined using a nonparametric testing method 34,43,44,57,63,64 . By randomly shuffling the data of the independent variable, we used the original dependent variable and the shuffled independent variable to compute the r n(predicted, observed) . The null distribution of r (predicted, observed) was obtained by repeating this procedure 5000 times. By subtracting the percentile of the true r (predicted, observed) in the null distribution from one, we obtained the significance of r (predicted, observed) .

Mediation analysis.
To evaluate the indirect effect of brain structure on academic performance through trait conscientiousness, we conducted a mediation analysis with the SPSS macro PROCESS, including a bootstrapping approach 65 . For this analysis, the GMV of the brain region was considered the predictor variable (X), NEO-FFI conscientiousness scores were considered the mediator variable (M), and CNUEE scores were considered the outcome variable (Y). The indirect effect, referring to the product of path a (i.e., the relation between X and M) and path b (i.e., the relation between M and Y after adjusting for X), was estimated. The point estimates of the indirect effects were considered significant if the bootstrapped 95% confidence intervals (CIs) (5000 iterations) did not include zero.

Results
neurostructural correlates of trait conscientiousness. Table 1 displays the descriptive statistics (i.e., mean, standard deviation, range, skewness and kurtosis) for each behavioral variable included in the present study. The kurtosis and skewness values for all variables ranged between −2 and 2, indicating that the data fulfilled the normality assumption 66,67 . Importantly, trait conscientiousness had no significant relation with age (r = −0.02, p = 0.788), gender [t (146) = 1.92, p = 0.057] or total GMV (r = 0.12, p = 0.138). After controlling for gender and age, the association between trait conscientiousness and total GMV was still not significant (partial r = 0.02, p = 0.788). We then investigated the structural substrates of trait conscientiousness.
The whole-brain regression analysis revealed that after nonstationary cluster correction for multiple testing, trait conscientiousness showed a positive association with the GMV in the bilateral superior parietal lobe (SPL; see Table 2 and Fig. 1). Additionally, trait conscientiousness had a negative association with the GMV in the right MFG (see Table 2 and Fig. 2). No other significant clusters were obtained in this analysis. Given that a previous study has shown gender differences in the association between trait conscientiousness and regional GMV 25 , a condition-by-covariate interaction analysis was performed. The results indicated that there were no gender www.nature.com/scientificreports www.nature.com/scientificreports/ differences in the association between trait conscientiousness and regional GMV after correcting for multiple comparisons.
We next implemented prediction analyses to check the robustness of the relation between trait conscientiousness and the GMV in the brain regions identified from the whole-brain regression analysis. The results showed that trait conscientiousness could be stably predicted by the GMV in the bilateral SPL [for the left SPL: r (predicted, observed) = 0.26, p < 0.001; for the right SPL: r (predicted, observed) = 0.27, p < 0.001] and right MFG [r (predicted, observed) = −0.27, p < 0.001] after controlling for gender, age and total GMV. Given the high correlation between the left SPL and right SPL (r = 0.72, p < 0.001) and the similar results found in these two regions, we subsequently condensed our analyses by using only the bilateral average SPL volume.  www.nature.com/scientificreports www.nature.com/scientificreports/ Brain regions linking trait conscientiousness and academic performance. After assessing the structural substrates of trait conscientiousness, we further investigated the underlying brain-trait conscientiousness mechanism in predicting academic performance by collecting CNUEE scores. First, we verified the positive correlation of trait conscientiousness with academic performance (r = 0.30, p < 0.001). Further regression analysis revealed that trait conscientiousness explained additional variance in academic performance after adjusting for gender, age and total GMV ( R 2 = 7.4%, β = 0.28, p < 0.001). Second, we tested whether academic performance could be predicted by the GMV in the identified brain regions. We found a significant correlation between bilateral SPL volume and academic performance (r = 0.24, p = 0.003). However, we did not find a significant correlation between right MFG volume and academic performance (r = −0.08, p = 0.346). Further regression analysis revealed that the bilateral SPL volume ( R 2 = 3.9%, β = 0.20, p = 0.012) but not the right MFG volume ( R 2 = 0.8%, β = −0.09, p = 0.278) explained additional variance in academic performance after adjusting for gender, age and total GMV.
We then carried out mediation analyses to test whether trait conscientiousness could mediate the link between regional GMV and academic performance. Interestingly, trait conscientiousness played a mediating role in the relationship between bilateral SPL volume and academic performance (indirect effect = 0.088, 95% CI = [0.030, 0.166], p < 0.05). Given that many studies have suggested that direct effects are not an appropriate premise for testing indirect effects 68 , we also investigated the mediating role of trait conscientiousness in the association between right MFG volume and academic performance. The results showed that trait conscientiousness can also mediate the relationship between right MFG volume and academic performance Furthermore, we performed several other mediation analyses to explore the directionality of the relations between regional GMV, trait conscientiousness and academic performance. In particular, given that several previous longitudinal studies have suggested that trait conscientiousness is a stable antecedent for academic performance [12][13][14] , there are three possible mediation models for the association between brain structure, trait .044], p > 0.05). Gender, age and total GMV were controlled for in these analyses. Therefore, these findings suggest that the GMV in the bilateral SPL and right MFG might affect academic performance through trait conscientiousness.

Supplemental analyses.
To test the specific nature of the above findings, we carried out supplemental analyses to exclude the possible effects of general intelligence, family SES and 'big five' personality traits other than conscientiousness (i.e., extraversion, neuroticism, openness, agreeableness). Table 3 lists the correlations of all behavioral measures included in this study.
First, we checked whether general intelligence could affect our results because converging evidence has shown that general intelligence is stably associated with academic performance 4,10 and brain structure 69 . Behaviorally, when general intelligence was added as a covariate, trait conscientiousness had incremental predictive ability for academic performance ( R 2 = 7.7%, β = 0.28, p < 0.001). Neuroanatomically, trait conscientiousness was still correlated with the GMV in the bilateral SPL (partial r = 0.33, p < 0.001) and right MFG (partial r = −0.31, p < 0.001), even after controlling for general intelligence, gender, age and total GMV. Importantly, after adjusting for general intelligence, gender, age and total GMV, the mediating effects of trait conscientiousness were still significant: for the relation between bilateral SPL GMV and academic performance, indirect effect = 0.084, 95% CI = [0.030, 0.158], p < 0.05; for the relation between right MFG GMV and academic performance, indirect effect = −0.094, 95% CI = [−0.179, −0.038], p < 0.05. Thus, our findings are independent of general intelligence.
Second, because SES has been found to be a stable predictor of academic performance and brain structure 9,32,70 , we examined whether our results were influenced by the participants' family SES. Behaviorally, when family SES was added as a covariate, trait conscientiousness had incremental predictive ability for academic performance ( R 2 = 7.6%, β = 0.28, p < 0.001). Neuroanatomically, trait conscientiousness was still correlated with the GMV in the bilateral SPL (partial r = 0.32, p < 0.001) and right MFG (partial r = −0.28, p < 0.001), even after controlling for family SES, gender, age and total GMV. Importantly, after adjusting for family SES, gender, age and total GMV, the mediating effects of trait conscientiousness were still significant: for the relation between bilateral SPL GMV and academic performance, indirect effect = 0.078, 95% CI = [0.026, 0.153], p < 0.05; for the relation between right MFG GMV and academic performance, indirect effect = −0.080, 95% CI = [−0.155, −0.028], p < 0.05. In summary, our findings are not affected by family SES. Third, we evaluated the possible influence of other 'big five' personality traits (i.e., extraversion, neuroticism, openness, agreeableness) on the associations between trait conscientiousness, brain structure and academic performance. Behaviorally, when the rest of the 'big five' personality traits were included as covariates, trait conscientiousness had incremental predictive ability for academic performance ( R 2 = 4.7%, β = 0.25, p = 0.006). Neuroanatomically, trait conscientiousness was still correlated with the GMV in the bilateral SPL (partial r = 0.31, p < 0.001) and right MFG (partial r = −0.27, p < 0.001), even after controlling for other 'big five' personality traits, gender, age and total GMV. Importantly, after adjusting for other 'big five' personality traits, gender, age and total GMV, the mediating effects of trait conscientiousness were still significant: for the relation between bilateral SPL GMV and academic performance, indirect effect = 0.060, 95% CI = [0.011, 0.132], p < 0.05; for the relation between right MFG GMV and academic performance, indirect effect = −0.062, 95% CI = [−0.134, −0.018], p < 0.05. Therefore, our results show some degree of specificity, although other 'big five' personality traits reduced the effect sizes of the results.

Discussion
The present research was conducted to investigate the structural brain substrates underlying adolescent conscientiousness and to explore the nature of the relationships among gray matter structure, trait conscientiousness and academic performance. We observed that the GMV in the bilateral SPL was positively linked to trait conscientiousness, which is in line with two previous VBM studies reporting gray matter variations in the SPL and individual differences in trait conscientiousness 18,20 . Furthermore, our result is consistent with the finding of a recent longitudinal experiment showing an association between annual percentage change in CT in the SPL and trait conscientiousness in adolescents 29 . The functioning of the SPL has also been found to be associated with trait conscientiousness [71][72][73] . For example, evidence from an electroencephalography study showed a positive relation between trait conscientiousness and photic driving in the beta frequency band (i.e., the mean of 13, 16 and 18 Hz stimulation frequency) in the parietal lobe 73 . In addition, through an independent component analysis based on resting-state functional MRI (fMRI), Sampaio et al. (2014) reported a positive association between trait conscientiousness and spontaneous activity in the SPL 72 . Furthermore, a task-based fMRI investigation found that during a 3-back working memory task, trait conscientiousness was related to the functional connectivity from the SPL to the MFG 71 . The SPL is generally considered a core brain region in the multiple-demand system 74 , which is hypothesized to be crucial for conscientious tendencies 27 . In summary, our finding regarding the relationship between trait conscientiousness and SPL volume may substantiate the role of SPL structure and function in trait conscientiousness.
Moreover, trait conscientiousness showed a negative association with the GMV in the right MFG. This finding fits well with several previous sMRI investigations that have demonstrated associations between trait conscientiousness and measures of brain structure (e.g., GMV and CSA) in MFG 18,19,27 . Evidence from lesion research has suggested that focal damage to the MFG leads to reduced levels of executive function and conscientiousness, which presents direct evidence for the causal role of MFG in trait conscientiousness 75 . Moreover, it has been reported that MFG activity during resting-state fMRI scans can predict individual differences in trait conscientiousness 76 . Broadly, the MFG is identified as crucial for behaviors related to self-regulation behaviors 77 , which have self-evident relevance to trait conscientiousness 78 . Evidence from several task-based fMRI studies has suggested that individual differences in trait conscientiousness are linked to the activity and connectivity in the MFG during self-regulation-related tasks such as delay discounting 79 , inhibitory control 80 and working memory 71 . Additionally, the negative relationship between trait conscientiousness and MFG volume may reflect the processing of synaptic pruning and myelination during development in some brain regions, which may lead to improved efficiency in certain cognitive and noncognitive capacities 31,81,82 . For example, decreased MFG volume has been linked to increases in high-order variables such as quality of life 41 , social well-being 83 and creative cognitive ability 84 . Therefore, synaptic pruning and myelination in the MFG during development may increase efficiency in self-regulation-related capacities (e.g., executive function, inhibitory control and delay discounting), which may increase conscientious tendencies.
Interestingly, we found that trait conscientiousness served as a mediator in the link between SPL volume and academic performance. Behaviorally, the association of trait conscientiousness with academic performance has been well established in previous investigations 5,6,8,[12][13][14] . This association was replicated in the current sample (r = 0.30, p < 0.001). Hierarchical regression analysis further revealed that even after we excluded the influences of general intelligence, family SES, and 'big five' personality traits other than conscientiousness, as well as gender, age and total GMV, trait conscientiousness still accounted for additional variance in academic performance (R 2 = 5.5%, β = 0.27, p < 0.001). Therefore, our study may present further evidence for the predictive role of trait conscientiousness in academic performance. Neuroanatomically, we observed that the variance in academic performance could be explained by the GMV in the SPL. Although there are currently no reports of an association between SPL structure and academic performance, there is some evidence suggesting a role of SPL function in measures of academic performance 85,86 . For example, SPL activity during a memory task has been correlated with educational attainment in a group of young adults 86 . One recent study further revealed that the SPL activity of adolescent students during working memory tasks can predict variance in academic performance on the mathematics section of the Massachusetts Comprehensive Assessment System 85 . In general, the SPL is associated with multiple high-level psychological functions, including top-down and goal-directed attention 87 , working memory 88 , executive function 89 , spatial ability 90 and problem solving related to planning and visuospatial reasoning 91 . Thus, the volume of the SPL may affect academic performance through these psychological attributes, which are regarded as important predictors of academic performance 6,92-95 . Altogether, our study indicates that trait conscientiousness may be a potential mechanism linking the GMV in the SPL to academic performance.
Additionally, we found that trait conscientiousness mediated the effects of the GMV in the right MFG on academic performance, although there was no significant association between right MFG volume and academic performance. This finding suggests that experimentally modulating right MFG structure and function, such as with transcranial direct current stimulation 96 and neurofeedback training 97 , may be a promising approach to promote conscientiousness in adolescent students, which might, in turn, enhance academic performance. Notably, further longitudinal or experimental evidence is needed to verify the directionality of the association in this process. Moreover, although no significant association between MFG volume and academic performance was observed in the present study, there is room for future studies exploring this association because some empirical evidence has shown a crucial role of MFG structure and function in academic performance 33,34,36,86,98 . In summary, our finding may present an underlying pathway in which the GMV in the right MFG indirectly affects academic performance by modulating trait conscientiousness.
Our research has several limitations that should be acknowledged. First, we used only a self-reported NEO-FFI questionnaire to assess trait conscientiousness, although the validity and reliability of this measure have been well established 11,46-48 . It is necessary for future studies to use multiple techniques (e.g., peer rating or experience sampling) to decrease the response bias and enhance measurement accuracy. Additionally, the www.nature.com/scientificreports www.nature.com/scientificreports/ measurement of academic performance relied on a single university entrance exam (i.e., CNUEE), although it is a well-standardized and authoritative test for assessing academic performance in Chinese high school students [49][50][51] . There is some evidence to suggest that in other countries, the university entrance exam may be heavily loaded to general intelligence and behave differently from measures such as GPA 2,6 . Thus, future studies will need to use other measures of academic performance (e.g., GPA) to validate our findings. Second, in our mediation analyses, we tested three possible mediation models, given that several previous longitudinal studies have suggested that trait conscientiousness is a stable predictor of academic performance [12][13][14] , and we found only one significant model. However, due to the correlational nature of the present study, we could not exclude other possible mediation associations in which academic performance is an antecedent for trait conscientiousness, which reflects hypotheses linking to cognitive development or self-perception. For example, in our dataset, academic performance could mediate the impact of bilateral SPL volume on trait conscientiousness (indirect effect = 0.046, 95% CI = [0.013, 0.102], p < 0.05). Thus, it is difficult to draw a causal conclusion regarding the relationships between brain structure, trait conscientiousness and academic performance because we based our research on cross-sectional data. Future studies using more sophisticated methods (e.g., experimental or longitudinal designs) are needed to establish the directions of the causal relationships between brain structure, trait conscientiousness and academic performance. Third, we revealed only that the GMV in the SPL and MFG is linked with trait conscientiousness; we failed to observe the relations found in prior investigations between trait conscientiousness and the GMV in other cortical and subcortical regions 17,19,20,22,25,27,29 . Given that we employed only one measure of brain structure (i.e., GMV) and no functional measures, other measures of brain structure (e.g., CT and CSA) and function (e.g., task-based and resting-state brain activity) may be used to further explore the neurobiological bases of trait conscientiousness and their relations to academic performance.
In conclusion, this research provides initial evidence for neurostructural markers underlying adolescent conscientiousness by revealing that the GMV in the bilateral SPL and right MFG is linked to trait conscientiousness in a sample of high school students. Additionally, our study presents pioneering evidence suggesting that the association between SPL/MFG volume and academic performance is mediated by trait conscientiousness. These findings suggest the important role of trait conscientiousness and the GMV in the frontoparietal network in the prediction of academic performance, and they also introduce new research directions for examining how brain measures affect academic performance through individual personality traits. Finally, our findings may add to the development of psychoradiology, a new field of radiology with the purpose of not only improving our understanding of the mechanisms underlying psychiatric disorders, but also have great potential to play the clinical role in guiding diagnostic and treatment planning decisions in psychiatric patients [99][100][101][102] .

Data and Code Availability
The data and code that support the findings of this study are available from the corresponding author upon reasonable request. And the data and code sharing adopted by the authors comply with the requirements of the funding institute, and comply with institutional ethics approval.