Search

Intelligence and mental health are the core pillars of individual adaptation, growth, and opportunity. Here, we charted across childhood and adolescence the developmental interplay between the p-factor of psychopathology, which captures the experience of symptoms across the spectrum of psychiatric disorders, and the g-factor of general intelligence that describes the ability to think, reason, and learn.Our preregistered analyses included 7,433 twin pairs from the Twins Early Development Study (TEDS), who were born 1994 to 1996 in England and Wales. At the ages 7, 9, 12, and 16 years, the twins completed two to four intelligence tests, and multi-informant measures (i.e., self-, parent- and teacher-rated) of psychopathology were collected.Independent of their cross-sectional correlations, p- and g-factors were linked by consistent, bidirectional, and negative cross-lagged paths across childhood and adolescence (from −.07 to −.13 with 95% CIs from −.03 to −.15). The cross-lagged paths from intelligence to psychopathology were largely due to genetic influences, but the paths from psychopathology to intelligence were driven by environmental factors, and increasingly so with age.Our findings suggest that intelligence and psychopathology are developmentally intertwined due to fluctuating etiological processes. Understanding the interplay of g- and p-factors is key for improving children’s developmental outcomes.

Mental ill health is the largest single cause of disability in the UK (1). Research suggests that lifetime trajectories of common mental disorders are established by mid-adolescence, with half of all mental health problems starting by the age of 14 and three-quarters by 25 (2). According to the Mental Health of Children and Young People Survey, between 2017 and 2020, rates of probable mental disorders increased from 11.6% to 17.4% in the age group of 6-16 year-olds and by over 7% among 17-19 year-olds (10.1% vs 17.4%) in the UK (3). Although WHO estimates that roughly 10% of children and adolescents between 5 and 16 years have a diagnosable mental disorder (4), a recent study among 28,000 adolescents reported that the prevalence of mental health problems in children across many schools in England is much higher than previous estimates, with roughly two in five young people scoring above clinical cut-offs for emotional problems, conduct problems, and hyperactivity (5). However, only a fraction of them receives appropriate care. The Office of the Children's Commissioner found that from nearly 400,000 children and young people who were referred to NHS mental health services in 2018-2019, only 19% of children entered treatment within six weeks after the referral, while 33% stayed on the waiting list for over a year and 34% were not accepted into treatment (6). Mental health services further report that even among children with an official mental disorder diagnosis, only one in four children could access the necessary support in 2020 (7), while young people with psychological and emotional difficulties who do not meet the clinical criteria for a formal diagnosis are not eligible even for that, despite presenting a number of symptoms of physical and behavioural problems (8). It is unclear whether this increase in rates of mental disorders in youth is due to a genuine decline of mental health in children and young people, or it is instead due to greater awareness and less stigmatisation of reporting mental health problems for both children and their parents. Yet, the public mental health crisis is evident and tackling the causes and the development of effective early intervention and prevention programmes is urgent. The Children's Society's Good Childhood Report, the longest-standing report on the wellbeing of children in the UK, indicates that in 2022 a larger proportion of children aged 10 to 17 were unhappy with school (14.2%) than with the other nine aspects of life (e.g., home, family, friends) (9). Among the areas of school life, young people were most happy with the feeling of safety and relationships with their peers, and they felt least pleased with the amount of schoolwork and how much they were listened to in school (9). Similarly, Childline, a free counselling service for children in the UK, reported that children's primary concern was stress and anxiety related to schoolwork and exam performance (10,11). However, there is a lack of research exploring the role of the school context in young people's mental health. A systematic review of 48 studies reports strong evidence of the importance of the school climate, comprised of social connectedness, school safety, school connectedness, and the academic environment, for adolescents' mental health (12). Correspondingly, a qualitative study among adolescents and school staff across three secondary schools in the UK identified four dimensions of school culture (structure and context, organisational and academic, community, and safety and support) as important contributors to student mental health (13). A large longitudinal study of young people across 20 schools in England found strong and consistent associations between student-level and school-level (average student responses in each school) reports of positive school climate at baseline and better mental health outcomes at 36 months follow-up (14). However, these associations were compromised to borderline significant for student-level measures and non-significant for school-level climate after adjustment for baseline measures of student mental health (14). Thus, while the existing studies show that students' subjective perceptions of the school environment are significantly associated with their mental health, these reports are prone to bidirectional bias, so mental health outcomes measured after children and adolescents had been at the school for some time, may already have been influenced by the school climate, while mental disorders may, in turn, corrupt the perception of school environment. Furthermore, according to the NHS reports, children with a higher risk of having a mental disorder were twice as likely to miss more than 15 days of school than those from a lower risk group (3). Thus, having a mental disorder presents further risk of exacerbating worries about school and compromising the perception of school environments, making it more complicated to design effective intervention and prevention programmes. Additional objective measures of school-related risk factors, such as school-staff ratio, number of students eligible for free school meals (FSM), or Ofsted school rates, could provide valuable evidence for better understanding the associations between school environment and child and adolescent mental health. Indeed, a recent study across 97 schools in England which linked survey responses with the National Pupil Database (NPD) showed that having special educational needs (SEN) and being eligible for FSM significantly increased the odds of experiencing any of the four mental health problems, namely emotional symptoms, conduct problems, peer-relationship problems, and hyperactivity/inattention problems (5). Furthermore, students with a child in need status were also at significantly increased risk of experiencing all mental health problems except for emotional symptoms (5). However, the schools explained very little variance in mental health problems (intraclass correlations ≤0.02). Moreover, other studies have concluded that school composition and resources (e.g., staff-student ratio) were not associated with students' mental health (15). The mixed findings might indicate a complex relationship between the objective and subjective measures of school environment and their effects on young people's mental health; nevertheless, even fewer studies included both objective and subjective factors in one cohort. A large study that did use both the NPD data from 648 primary schools on school composition (number of children in the school, student gender and ethnicity ratio, and school deprivation indexed as the proportion of students eligible for FSM) and child-reported school climate explored their association revealed that among the school composition factors only school deprivation was associated with adolescents' mental health, such that young people in schools with higher proportions of students eligible for FSM had on average more behavioural symptoms and were more likely to have mental health difficulties above the clinical cut-off, whereas a subjectively-reported more positive school climate was associated with both lower levels of emotional and behavioural symptoms, and lower rates of clinically significant levels of mental health problems (16). The authors further explored the interaction between the school composition and school climate but found no significant effects on mental health outcomes (16). However, to date, there is no evidence whether the objective and subjective measures of school environment may complement each other and explain more variance of child and adolescent mental health problems. Even less is known about whether objective school-level characteristics provide different contexts and present different levels of risk to children with individual biological differences in mental health characteristics. Thus, studies exploring the association between school environment and child and adolescent mental health remain scarce and lack a comprehensive approach, considering both objective and subjective measures of school environment and the individual predisposition to mental disorders. Furthermore, they are often designed as cross-sectional studies, limiting the opportunities of exploring the cumulative risk of objective and subjective school environmental factors for young people's mental health across their development. Accordingly, the main goal of the present study is to clarify the longitudinal associations between school experiences and developmental psychopathology from childhood to emerging adulthood using both objective and subjective data. We have previously investigated mental health associations with subjective parent-, teacher-, and twin-reported measurements of school environment using the data from the Twin Early Development Study (TEDS), a cohort of twin pairs born in England and Wales between 1994 and 1996, utilising the data collected when twins were 7, 9, 12, 16, 18, and 21 years old (17). Our preliminary results show substantial associations between multiple school environment factors, such as classroom environment, homework and attitudes to school, and psychopathology (r~.25). They show additive effects over the years, possibly indicating that these environmental experiences might pose a cumulative risk for child mental health outcomes (Rimfeld et al., in preparation). Furthermore, prediction increased when previous school experience was added to the contemporary predictors in the model. We aim to extend these findings in the present study by focusing on the objective measurements of school environment, such as the number of absences, exam results, and school quality, by utilising data from the UK-wide registers, namely the NPD and the Ofsted ratings. Contrary to the subjective reports, these objective measures are not affected by the possible rater's bias due to existing mental health problems. Furthermore, decades of twin studies have shown that mental health problems are highly heritable (30-80%), thus, it is important to use genetically sensitive designs when exploring the links between school environment and mental health (18). Our preliminary results show that both mental health and educational experiences across the school years are highly heritable (~50%), and that genetic factors largely explain the correlation between subjectively reported school environment and psychopathology (~90%) (Rimfeld et al., in preparation). The present study will explore the links between objective school environment (National Pupil Database school records on school quality, Ofsted ratings) and mental health using a longitudinal genetically sensitive design (age 7-21) and will combine subjective (parent-, teacher- and self-reported school experiences) and objective school measures to explore the additive prediction from school experiences to mental health both contemporaneously and over time.

Mathematics anxiety (MA) is a partially overlaps with general anxiety, but is an independent phenotype, characterised by negative emotional reactions experienced in anticipation of or during mathematics-related activities (Malanchini et al., 2017). Research has shown that, independent of the effects of general anxiety, students who report high levels of MA tend to exhibit lower levels of maths achievement (Zhang et al., 2019). While this association (~ r = -0.3) is now well-replicated, research suggests that the relationship between MA and maths achievement is more complex, with the effects of MA on performance also dependent on levels of mathematics motivation (MM). Individuals with high levels of MM value mathematic abilities highly, are willing to embrace maths challenges, and are keen to perform well in the subject (Gottfried et al., 2007). Some studies have demonstrated that, alongside high reported levels of MM, moderate levels of MA might be beneficial for students’ mathematics achievement (Wang et al., 2015, 2018). Research has also suggested that MA might also lead to lower participation in STEM (Science, Technology, Engineering and Mathematics), regardless of achievement levels (Ahmed, 2018; Choe et al., 2019). In both compulsory and post-compulsory education settings, MA and MM have been found to predict STEM avoidance in course enrolment (Daker et al., 2021; Meece et al., 1990), but fewer studies have considered the potential effects of MA and MM on STEM avoidance in wider career choices. In a UK sample of twin pairs aged 16 to 21 from the Twins Early Development Study (TEDS), we will investigate the relationships between MA, MM and STEM career choices, and whether these are affected by previous mathematics achievement. Evidence also suggests that both motivation and anxiety levels are associated with socioeconomic status (SES), with research indicating that a disproportionate number of students with low educational motivation may be from low SES backgrounds (De Clercq et al., 2017; Rogelberg et al., 2021), and that children from lower SES backgrounds exhibit higher levels of anxiety than those from higher SES backgrounds (Zhu et al., 2019). This study will, therefore, also account for the effects of SES on the relationships between the stated variables. There are approximately 10,000 twin pairs actively involved in the TEDS. We will use all data available at 16-21 years for our chosen variables, but the final sample will be lower than this. First, we will assess and report the relationships between all variables (self-reported MA at 18 (both corrected and uncorrected for general anxiety), self-reported MM at 16, GCSE mathematics achievement at 16, STEM career choice at 21 and SES at 16). Then, we will investigate the extent to which self-reported MA (both corrected and uncorrected for general anxiety) and self-reported MM predict STEM career choice and take a stepwise approach to adding covariates into this model (maths achievement at 16 and SES). As previous studies have suggested that the highest achieving students might exhibit both high levels of MM and moderate to low levels of MA, we hypothesise that both MA and MM will be significant independent predictors of STEM career choice, with the expectation that higher levels of MM will predict a greater likelihood of STEM career choice, and higher levels of MA will predict a lower likelihood of STEM career choice. We also hypothesise that both previous maths achievement and SES will attenuate the associations between MA, MM and STEM career choice. Several studies have reported that female students tend to report higher levels of MA (Malanchini et al., 2017, 2020) and lower levels of MM (Wang et al., 2018). One study investigating whether the outcomes of MA differed by sex found that MA and self-perceived mathematics ability predict each other only in male students, suggesting that MA and MM may have a negative reciprocal relationship in boys, but not girls (Wang et al., 2020). Boys have been shown to identify more with mathematics, and feel more obligated to perform well than girls (Cvencek et al., 2011). Experiencing MA may therefore be more threatening to male student identity than female student identity, and might result in lower MM. Conversely, it may be harder for boys to reconcile lower MM with the societal expectation that they should perform well in mathematics, and this conflict could result in increased MA compared to girls with low MM (Wang et al., 2020). As a result, differences in MA and MM might go on to have a greater impact on the STEM-related career choices of male participants compared to female participants, with lower MM and higher MA resulting in a lower probability of STEM career choice for males than females. We hypothesise that, when doing sex-stratified analyses of the associations above, MA and MM are better predictors of STEM career choice in males than females. We will also use genetically sensitive designs (twin analyses) to understand the aetiology of the relationships between each of the stated traits. We will assess the heritability of all traits, and use multivariate genetic analyses (Cholesky decomposition) to explore the aetiology of covariance between the traits. Using a genetically sensitive framework will allow us to investigate the origins of individual differences in MA, MM and STEM degree choice, as well as the origins of the co-variation between these traits. We will assess the extent to which genes, shared environments, and individual-specific environments contribute to the aetiology of these traits, and determine whether the phenotypic correlations found between traits are reflected in their aetiology. Malanchini et al. (2020) found moderate estimates of heritability across of mathematics-related traits including mathematics anxiety, attitudes, ability and achievement, as well as considerable genetic overlap between these traits. We therefore hypothesise that all variables in the current study will be moderately heritable, and that genetic factors will largely explain the covariance between traits. The study results would advance our understanding of the complex emotion-motivation processes underlying STEM avoidance in young adults and might provide important insights for interventions targeted at increasing participation in STEM fields.

Educational and occupational success, jointly referred to as socioeconomic status, are associated with many important life outcomes, including physical and mental health, life satisfaction, happiness, and longevity (1–4). Individual differences in socioeconomic outcomes (combined socioeconomic status, or employment status, educational attainment and income separately) are often assumed to be environmental in origin, however, twin and family-based research has shown that both genetic and environmental factors play an important role in explaining individual variation (5–8). The heritability (the proportion of individual differences explained by inherited DNA differences) of educational attainment as well as occupational status is around 50%, with the rest of the variation explained by environmental factors (5,8,9). Heritability has traditionally been calculated by comparing the similarity of identical and non-identical twin pairs. It is also possible to estimate heritability directly from DNA, by comparing the phenotypic, or trait, similarity between people with their genetic similarity, calculated using hundreds of thousands of DNA variants, or SNPs (single nucleotide polymorphisms). These SNP-based heritabilities have also been shown to be substantial for socioeconomic outcomes, around 20% on average (7,10–12). SNP heritability for complex traits, like educational and occupational outcomes, is roughly half of that of family-based heritability (13). The reasons for this are that SNP heritability is calculated using genotyped genetic markers, so can only index the variation captured by genetic markers genotyped on a SNP array (SNP chip). As these arrays only uses common SNPs, this excludes rare and structural variants, and it only estimates additive genetic effects (14). The missing heritability, the gap between the family-based heritability and SNP heritability, has been shown to be almost recovered for height and body-mass-index (BMI) by including rare variants in a model using whole-genome sequence data (15). It remains to be seen if the missing heritability in other complex traits, for example for educational status and occupational status, is also explained by the rare variants, which would indicate that the heritability of complex traits is explained mainly by additive genetic effects. Alternatively, it has been argued that twin heritability overestimates the true heritability (16), and this explains (or, rather, explains away) the missing heritability. Another way to estimate genetic influence on complex traits is to use genome-wide polygenic scores (GPS). GPS use summary statistics of GWA (Genome-wide Association) studies that aim to identify specific genetic markers associated with a phenotype of interest, by aggregating thousands of SNP associations across the genome to predict the trait of interest (17). This prediction using GPS serves as an index that summarizes the current prediction of additive genetic influences of the trait of interest (18). These GPS heritabilities for socioeconomic variables have also been shown to be substantial, around 11% on average (19,20). However, all these heritability estimates may be different in varied environments and may change with major changes in environment, for example with societal changes such as changes in policy (5,21). Paradoxically, heritability for educational outcomes tends to be higher when environmental variation is attenuated. That is, with less variation in the environment, a larger proportion of individual differences is explained by genetic factors (22). Thus heritability could be used as an index of equal opportunities (18,23). In addition, research has shown that individual differences in socioeconomic outcomes are explained both by inherited DNA differences but also by non-transmitted alleles operating via parental genotypes. That is, part of the variance in educational attainment is explained by so-called genetic nurture; that is, parents’ genetic variants not transmitted to their children still predict educational attainment in offspring (24,25). If parents’ non-transmitted genetics predict their offspring’s achievement, this link must happen through parent-driven environmental processes, independent of child’s own biology. The environmental experiences that parents create for their children are associated with their offspring educational outcomes, are a clear example of passive gene-environment correlation. The gene-environment correlation is a well-studied area in quantitative genetic research (26,27), but often limited to between family variance. However, DNA data is allowing us to get closer to this phenomenon and quantify the genetic nurture component in several social outcomes. Genetic nurture has been shown to be substantial for educational outcomes (25) and to a smaller degree for depressive symptoms (28). It is a plausible hypothesis meriting major research interest to study how genetic nurture is influenced by changes in the larger social environment. A related issue here concerns the locus of the heritability of socioeconomic outcomes. It is well-established, for example, that educational achievement is explained by a package of heritable cognitive and non-cognitive traits (29). However, it is not known what proportion of the putative change in heritability of socioeconomic outcomes is explained by cognitive and non-cognitive factors following a major societal change in the environment. Furthermore, it is not known how the change in social environment reflects genetic nurture, that is the proportion of variance in offspring social outcomes that is explained by parental genotypes, independent of child’s biology, and the proportion of this genetic nurture that is explained by cognitive or non-cognitive genetic variances in parental genotypes. A newly developed genomic method, GWAS-by-subtraction, allows researchers to investigate these outstanding research questions by partitioning the SNP associations of educational attainment into cognitive and non-cognitive parts of variance (30). Educational attainment has also been shown to be associated with fecundity (the ability to produce many offspring): a more educated people tend to have children later in life and as a result, or in addition to that, tends to have fewer offspring. The relationship between educational attainment and fecundity could result in a decline of educationally relevant genetic variants in the population. Although this effect in one generation is small, it could lead to a significant decline over many generations (31). We have previously shown a massive shift in the aetiology of social outcomes in Estonia following an abrupt social change - the shift from communism to capitalism following the collapse of the Soviet Union. DNA differences explained twice as much variance in social outcomes (occupational status, educational attainment) independent Estonia compared to when Estonia was occupied by the Soviet Union (32). We also reported interesting sex differences, where the shift was much larger for females compared to males, but we were cautious with our conclusions due to limited power in the sex-stratified analyses. We also tested the decline of educationally relevant genetic variants in the population, but did not find evidence for the hypothesis. In addition, we did not see a decrease in average EduYears GPS across birth cohorts, but we did not account for potential differences in fecundity (32). The Estonian Biobank sample, used in our previous study, has grown from 50,000 DNA donors, making up 5% of the adult population, to 205,000 DNA donors, making up 20% of the adult population in Estonia, thus offering an opportunity to replicate and extend our previous research.