A machine-predicted brain sex score explains individual differences in cognitive intelligence and genetic influence in young children; sex identified with a 93pct success

The sexual brain, genes, and cognition: A machine-predicted brain sex score explains individual differences in cognitive intelligence and genetic influence in young children. Kakyeong Kim, Yoonjung Yoonie Joo, Gun Ahn, Hee-Hwan Wang, Seo-Yoon Moon, Hyeonjin Kim, Woo-Young Ahn, Jiook Cha. Human Brain Mapping, April 26 2022. https://doi.org/10.1002/hbm.25888

Abstract: Sex impacts the development of the brain and cognition differently across individuals. However, the literature on brain sex dimorphism in humans is mixed. We aim to investigate the biological underpinnings of the individual variability of sexual dimorphism in the brain and its impact on cognitive performance. To this end, we tested whether the individual difference in brain sex would be linked to that in cognitive performance that is influenced by genetic factors in prepubertal children (N = 9,658, ages 9–10 years old; the Adolescent Brain Cognitive Development study). To capture the interindividual variability of the brain, we estimated the probability of being male or female based on the brain morphometry and connectivity features using machine learning (herein called a brain sex score). The models accurately classified the biological sex with a test ROC–AUC of 93.32%. As a result, a greater brain sex score correlated significantly with greater intelligence (rho-fdr < .001, (eta sub p)**2= .011–.034; adjusted for covariates) and higher cognitive genome-wide polygenic scores (GPSs) (rho-fdr < .001,  (eta sub p)**2 < .005). Structural equation models revealed that the GPS-intelligence association was significantly modulated by the brain sex score, such that a brain with a higher maleness score (or a lower femaleness score) mediated a positive GPS effect on intelligence (indirect effects = .006–.009; p = .002–.022; sex-stratified analysis). The finding of the sex modulatory effect on the gene–brain–cognition relationship presents a likely biological pathway to the individual and sex differences in the brain and cognitive performance in preadolescence.

4 DISCUSSION

We report the novel relationship between brain sex difference, cognitive performance, and shared genetic influence in an admixed American population of prepubertal children. As trained on the grey matter morphometric and white matter connectomes, our machine learning models showed the accurate classification of sex with over 93.32% ROC–AUC in a replication set. Furthermore, the individual variability of the sexual brain development, indexed by the brain-based sex score, showed significant correlations with general intelligence and the inherited genetic influence on general intelligence, the cognitive GPSs. Moreover, the SEM showed that the effect of the cognitive GPSs on cognitive outcomes was modulated by the brain sex score significantly in females and with a similar trend in males. Thus, this study indicates the critical role of brain sex in cognitive performance in prepubertal children, influenced by genetic factors, providing a biological account for the individual variability of neurocognition.

Our study departs from the prior literature on sex differences in intelligence in children by showing the role of the continuum of brain sex on cognitive performance. Literature shows that the group sex differences in mind and behaviors, such as hormonal influences (Vuoksimaa, Kaprio, Eriksson, & Rose, 2012), brain differences (Ostatníková et al., 2010), cultural influences (Penner & Paret, 2008), gender stereotypes (Stoet & Geary, 2012), and biopsychosocial interactions (Haier, Karama, Leyba, & Jung, 2009; Miller & Halpern, 2014). In intelligence, however, literature shows mixed findings of sex differences (Dykiert, Gale, & Deary, 2009). Some show that males have advantages (Irwing & Lynn, 2005; Jackson & Philippe Rushton, 2006; van der Linden, van der Linden, Dunkel, & Madison, 2017) in general intelligence over females, while others show females have advantages over males (Keith, Reynolds, Patel, & Ridley, 2008). These mixed findings may allude to large individual variability in intelligence within sex. Indeed, a recent seminal study shows the biological underpinnings of the individual variability in behavioral phenotypes in adolescents (Vosberg et al., 2020). This study presents an estimate of the continuum of sex based on the brain and body traits, which predicts within each sex the individual variability in sex hormones, personality traits, and internalizing–externalizing behaviors. In line with this, our study further demonstrates the utility of multimodal brain imaging combined with machine learning in estimating an individual status of brain sex. For example, our method permitted the accurate estimation of an individual's developmental status of the brain sex and revealed that the brain sex estimates varied across individuals even within the narrow age range. The discovery of the correlation of the brain sex variability with the genetic and cognitive variables further reflects that this novel estimate may represent a critical neurobiological process.

Another pattern to note is the greater association of crystallized intelligence (the ability that is acquired throughout life: i.e., knowledge, facts, and skills) with the brain-based sex, as well as GPSs for cognitive capacity, compared with fluid intelligence (the ability to reason and solve problems in novel situations; a trend towards significance). These findings are partially in line with prior genetic research showing that crystallized intelligence is greatly associated with genetic influence than fluid intelligence (Christoforou et al., 2014; Genç et al., 2021). Furthermore, since learning attitude (i.e., reading books) may be genetically inherited (Krapohl et al., 2014; Olson, Vernon, Harris, & Jang, 2001), it adds to the genetic propensity of crystallized intelligence. Taken together, these empirical findings including ours may challenge the historical conceptualization that fluid intelligence may be more driven by genes and crystallized intelligence by the environment (Cattell, 1971).

Our structural equation models show the potential relationships among the genes, brain sex, and cognition. The results indicate that a higher brain maleness score (a lower femaleness score) positively modulates the positive effect of the cognitive GPS on general intelligence significantly in both sexes. Considering that the modulatory effect remains significant after controlling for several potential confounding factors of the brain and cognitive performance, this GPS-brain sex-intelligence pathway has a significant statistical association. These results thus suggest the novel role of brain sex in children, linking the genetic influence to cognitive performance.

Then, what is the biological account of the modulatory effects of the brain sex on the genetic influence on cognitive performance: that is, positive toward maleness and negative toward femaleness? Literature shows that sex chromosomes play a crucial role in cognitive performance (Bender, Puck, Salbenblatt, & Robinson, 1990; Hong & Reiss, 2014; Warling et al., 2020). However, since we did not include the sex chromosomes when constructing the GPSs (following the common practice of the GWAS designs to boost statistical power), it might not fully explain the differences in the mediation effects across sex. Alternatively, we speculate that the different expression patterns of autosomal variants across sex (Boraska et al., 2012; Wijchers & Festenstein, 2011; Zuo et al., 2015) may account for the modulatory effects of sex. Indeed, in line with this speculation, recent literature highlights sex differences in brain transcriptomes related to schizophrenia and alcohol effects (Hitzemann et al., 2021; Hoffman et al., 2022). Future research may test the association between sex differences in genetic expression in the brain and neurocognitive development.

Note that only females showed a significant correlation between brain-based sex score and cognitive GPSs, whereas males showed a marginally significant correlation after correction for multiple comparisons. We think this should not be interpreted as the female-only effect of the cognitive GPSs in the brain sex development. Rather, it should be noted that their effect sizes were similar across sex (in educational attainment GPS) and the models combining males and females showed the significant correlations of the brain-based sex score and cognitive GPSs (in educational attainment and cognitive performance GPSs). Furthermore, regardless of the modulatory effects of sex, in both females and males, the influence of the cognitive GPSs on cognition was positive. This is in line with the literature in adults (Lee et al., 2018; Savage et al., 2018). Taken together, we think that the genetic underpinnings of cognitive development might be related to sex differentiation in the brain. Therefore, our integrative analysis reveals the subtle relationships among sex, genes, brains, and cognition, otherwise undetectable. We suggest this is a novel biological pathway to individual differences in brain sex. It may be interesting to test whether this pathway is related to epigenetic effects of environmental factors, such as early life stress.

This study confirms that biological sex can be classified accurately based on morphometric and white matter connectivity. A recent study with ABCD data show that the biological sex was classified with 89.6% accuracy in the replication set using a deep neural network trained on ABCD T1-weighted structural MRI (Adeli et al., 2020). Our study extends this prior work by showing the additive classification performance increase with the diffusion white matter connectomes. This performance increase perhaps presents that the multimodal MRI effectively accounts for the heterogeneous developmental trajectories of grey and white matter (Giedd et al., 1999). It further shows the importance of the multimodal MRI approach in accurate delineation of brain development status.

Our brain features exclude the total volumes of the brain, grey and white matter, of which the sex differences have been reported (Ruigrok et al., 2014). Though the whole brain volume difference between sexes may be a biological aspect, we reasoned that the measures of gross anatomy would confound the brain–cognition relationship. Therefore, beyond the sex difference in the gross anatomy, this study shows that the patterns of the grey matter and white matter fibers are associated with the continuum of brain sex.

In testing the relationships among the brain sex, cognitive ability, and the genetic influence on cognitive ability, we focused on the cognitive GPSs. However, our discovery of the significant tripartite correlation among the brain-based sex score, total brain volume, and intelligence may lead to a question whether the genetic underpinning of cognitive ability is related to that of the total brain size. Indeed, a recent GWAS meta-analysis reveals an overlap of GWAS hits between cognitive intelligence and brain size in 5 genomic loci (Jansen et al., 2020). We hope that future research may test the moderation effect of sex on the genetic influence on brain size and its impact on cognitive intelligence.

In our study, we found no significant relationship among our key variables with salivary measures of sex hormones. Given the prepubertal stages of the participants, the negative statistical findings may reflect that the gene–brain sex–cognition relationship is not significantly related to the effects of sex hormones. Literature shows a complex relationship between the level of sex hormones and cognitive intelligence (Castanho et al., 2014; Gurvich, Hoy, Thomas, & Kulkarni, 2018). Though different sex hormonal levels across the sexes are observed from the prepubertal ages (Courant et al., 2010), the actual effect of the sex hormones on cognitive intelligence (or its modulation) may not appear until puberty (Shangguan & Shi, 2009).

This study shows a novel relationship among genetic factors, brain sex, and cognitive intelligence. The link between genome-wide factors and cognitive ability has been shown in previous studies. Cognitive GPSs account for general cognitive ability up to 3.5% in pre-adolescence children (Allegrini et al., 2019), 11% of the variance in general intelligence, and 16% of the variance in educational achievement in adolescents (Selzam et al., 2017). Extending this literature, our study shows that an individual's degree of brain sex may modulate the impact of the genetic factor on cognitive intelligence. Since this modulatory effect is positive toward brain maleness and negative toward brain femaleness, it adds another source of sex and individual variability in intelligence. This inference also presents the benefit of using the brain data as an endophenotype in assessing the genotype–phenotype association (Glahn, Thompson, & Blangero, 2007). Taken together, brain sex is linked to the inherited genetic influence of cognition, accounting for a novel pathway to the individual difference in cognitive intelligence in preadolescence.

In contrast to the multiethnic participants, the SEM in the European-ancestry participants only showed nonsignificant indirect effects of brain sex score. The discrepant results may not be easily reconciled. It should be noted that the cross-ethnic transferability of our cognitive GPS based on the European-ancestry GWAS remains to be validated. However, our cognitive GPS was rigorously adjusted for the potential ethnic confounding. Our result of the significant modulatory effects in the admixed American participants needs to be interpreted with caution.

This study shows the novel relationships among brain sex, cognition, and cognitive GPSs. The brain sex score based on grey matter morphometric and white matter connectivity may represent a neurodevelopmental process in preadolescence related to the inherited genetic influence on cognitive intelligence and unrelated to sex hormonal levels. This study thus provides a novel framework for future research in neurocognitive development and mental disorders.

Hire ambitious people: Bright- and dark-side personality and work engagement

Furnham, A., Robinson, C., & Haakonsen, J. M. F. (2022). Hire ambitious people: Bright- and dark-side personality and work engagement. Journal of Individual Differences, Sep 2022. https://doi.org/10.1027/1614-0001/a000380

Abstract: Is work engagement, like job satisfaction, primarily a function of personality? In total, 397 working adults completed a short, reliable, three-facet model of work engagement, a short IQ test, various self-ratings, a Big Five (bright-side) personality scale, and a measure of the personality disorders (dark-side). Work engagement was related to age, intelligence, positive self-ratings, and all the personality variables. A regression analysis revealed six variables significantly related to total work engagement: sex, age, IQ, ratings of personal ambitiousness, trait Neuroticism and Cluster A personality disorders. Regressions onto each of the three facets of work engagement showed slightly different findings, yet in each, older people with lower Cluster A scores and who rated themselves as ambitious scored higher on all facets. Over a third of the variance was explained in each regression. In every analysis, the rating of ambitiousness was most strongly related to work engagement. Implications and limitations are acknowledged.

The current findings provide support for mild but robust cognitive dysfunction in first-degree relatives of late-onset Alzheimer's disease affected individuals

Cognitive Functioning of Unaffected First-degree Relatives of Individuals With Late-onset Alzheimer's Disease: A Systematic Literature Review and Meta-analysis. Ari Alex Ramos, Noelia Galiano-Castillo & Liana Machado. Neuropsychology Review, Sep 3 2022. https://rd.springer.com/article/10.1007/s11065-022-09555-2

Abstract: First-degree relatives of individuals with late-onset Alzheimer's disease (LOAD) are at increased risk for developing dementia, yet the associations between family history of LOAD and cognitive dysfunction remain unclear. In this quantitative review, we provide the first meta-analysis on the cognitive profile of unaffected first-degree blood relatives of LOAD-affected individuals compared to controls without a family history of LOAD. A systematic literature search was conducted in PsycINFO, PubMed /MEDLINE, and Scopus. We fitted a three-level structural equation modeling meta-analysis to control for non-independent effect sizes. Heterogeneity and risk of publication bias were also investigated. Thirty-four studies enabled us to estimate 218 effect sizes across several cognitive domains. Overall, first-degree relatives (n = 4,086, mean age = 57.40, SD = 4.71) showed significantly inferior cognitive performance (Hedges’ g = -0.16; 95% CI, -0.25 to -0.08; p < .001) compared to controls (n = 2,388, mean age = 58.43, SD = 5.69). Specifically, controls outperformed first-degree relatives in language, visuospatial and verbal long-term memory, executive functions, verbal short-term memory, and verbal IQ. Among the first-degree relatives, APOE ɛ4 carriership was associated with more significant dysfunction in cognition (g = -0.24; 95% CI, -0.38 to -0.11; p < .001) compared to non-carriers (g = -0.14; 95% CI, -0.28 to -0.01; p = .04). Cognitive test type was significantly associated with between-group differences, accounting for 65% (R23 = .6499) of the effect size heterogeneity in the fitted regression model. No evidence of publication bias was found. The current findings provide support for mild but robust cognitive dysfunction in first-degree relatives of LOAD-affected individuals that appears to be moderated by cognitive domain, cognitive test type, and APOE ɛ4.

Discussion

To our knowledge, this is the first meta-analysis to quantify the impact of family history of LOAD on cognition, summarizing 218 effect sizes from 34 empirical studies. The results provide compelling evidence that first-degree relatives show a mild but robust amount of overall cognitive dysfunction compared to controls without LOAD-affected relatives. Cognitive deficits in first-degree relatives were evident in executive functions, language, verbal IQ, verbal and visuospatial LTM, and verbal STM or IM. These outcomes indicate that, compared to controls without a family history of LOAD, first-degree relatives have higher chances of obtaining lower scores on neuropsychological measures across multiple cognitive domains. One plausible explanation for these findings relates to altered biomarkers in probands of LOAD-affected individuals. For instance, previous studies have indicated that unaffected offspring of individuals with LOAD show morphological and metabolic brain changes that resemble the preclinical manifestations of LOAD-related pathology (Dubois et al., 2016), including increased global brain atrophy rates (Debette et al., 2009), reduced medial temporal lobe activation (Donix et al., 2010; Johnson et al., 2006), higher levels of beta-amyloid deposition (Clark et al., 2016; Duarte-Abritta et al., 2018), and decreased gray matter volume (Berti et al., 2011; Honea et al., 2010). On the other hand, the lack of significant group differences in premorbid intelligence and visuospatial STM or IM, and especially the near null effects in performance IQ and visual perception, suggest that having a family history of LOAD does not seem to be associated with significant decline in these domains. Alternatively, first-degree relatives may exhibit distinct patterns of cognitive dysfunction related to phenotypic differences in LOAD (Carrasquillo et al., 2014; Ferreira et al., 2020; Snowden et al., 2007; Vogel et al., 2021). For example, recent research indicated that the limbic-predominant phenotype is strongly associated with the amnestic presentation of the disease (e.g., LTM dysfunction), whereas the posterior phenotype is characterized by visuospatial or perceptual abnormalities (Vogel et al., 2021).

Notably, subgroup analyses revealed that the APOE ɛ4 genotype moderates performance differences between first-degree relatives and controls without a family history of LOAD, which makes sense given that the APOE ɛ4 genotype is the most replicated risk factor for LOAD in genetics studies (Cacabelos, 2003; Yang et al., 2021). Specifically, relative groups documented as ɛ4 carriers exhibited more significant dysfunction in cognition (g = -0.24) compared to relative groups documented as non-ɛ4 carriers (g = -0.14). This finding is consistent with preliminary research (Debette et al., 2009; Tsai et al., 2021) demonstrating that first-degree relatives with both risk factors (APOE ɛ4 genotype and a family history of LOAD) are more likely to present with deficits in cognition (e.g., executive dysfunction and verbal and visuospatial LTM difficulties). Evidence also suggests that first-degree relatives with both risk factors exhibit greater beta-amyloid deposition (Yi et al., 2018), higher brain atrophy rates (Debette et al., 2009), and reduced gray matter volume (Ten Kate et al., 2016) compared to those with only one risk factor. Nevertheless, the current systematic synthesis revealed that few studies on the topic document separate scores for ɛ4 carriers verses non-carriers. Hence, the lack of control for APOE ɛ4 status might help account for the contradictory findings from empirical studies on cognition of first-degree relatives of LOAD-affected individuals previously noted in the introduction, and if factored in to analyses of cognitive domains, could potentially paint a different picture with regard to the domains that did not reach statistical significance. Moving forward from the current outcomes, a major challenge for future research on the topic is to determine the combined effects and parse out the unique contributions of APOE ɛ4 carriership and a family history of LOAD in profiling cognitive dysfunction in first-degree relatives. Importantly, the APOE ε4 effect on cognition reported here is based on a specific sample (first-degree relatives of LOAD-affected individuals) and hence our results do not apply to the general population of APOE ε4 carriers.

Although relative group mean age was not a significant moderator and the null hypothesis on the equality of effect sizes in the subgroup analysis on age category was not rejected, the dysfunction effect size for samples intermixing middle-aged (40–65 years) and older (> 65 years) first-degree relatives (g = -0.23, 95% CI [-0.37, -0.09], p = 0.002) was statistically significant and nearly twice the size of the dysfunction effect for samples including only middle-aged individuals (g = -0.12, 95% CI [-0.26, 0.02], p = 0.081). This suggests that the inclusion of a large percentage of middle-aged individuals in the studies analyzed here may have led to an overall smaller dysfunction effect size (g = -0.16, 95% CI [-0.25, -0.08], p < 0.001) than might be expected in older cohorts, thus calling into question the generalizability of the current findings. This conjecture seems in line with findings from a previous study noted in the introduction (Zeng et al., 2013), in which, compared to controls, family members of LOAD-affected individuals showed substantial differences on neuropsychological measures only quite late in life (70 or more years).

The effects of a family history of LOAD on cognition remain poorly understood. Cognitive dysfunction in first-degree relatives of AD-affected individuals has gained attention only in the last two decades. Figure 2 shows that out of 34 empirical works, only three studies (Green & Levey, 1999; La Rue et al., 1995, 1996) were published before the current century, and all of the studies were published within the past 30 years. As previously noted, LOAD-related neuropathological changes precede the clinical diagnosis of LOAD by many years, hence, an increasing number of studies has attempted to longitudinally follow cognitive changes and brain abnormalities in earlier first-degree relatives. In this meta-analytic review, some included studies were drawn from ongoing prospective studies, thus, follow-up research on these cohorts as they grow older is expected. This will allow investigation of cognitive dysfunction in older cohorts of first-degree relatives with a family history of LOAD.

Implications

Findings from the current quantitative review may have important clinical and theoretical implications. LOAD is an age-dependent dementing disease with cognitive symptoms that appear after a lengthy period of evolving neuropathophysiological abnormalities, and thus the effect sizes for between-group differences in several cognitive domains reported here may assist in establishing sensitive cognitive markers for first-degree relatives. This assertion builds on previous empirical research indicating that impairments in cognitive abilities such as premorbid intelligence, memory, and language are deemed potential markers for future development of LOAD (Blacker et al., 2007; Chen et al., 2000; Rapp & Reischies, 2005; Yeo et al., 2011). Equally important, executive dysfunction can be detected in middle-aged offspring many years before the affected parent develops dementia (Debette et al., 2009; Eyigoz et al., 2020). Hence, developing cognitive-based interventions for first-degree relatives, especially APOE ɛ4 carriers, is a pressing need. In relation to this, recent randomized controlled trials have shown that cognitive training benefits individuals at the early stages of LOAD (Cavallo et al., 2016; Kang et al., 2019; Lee et al., 2013). To our knowledge, however, no study has addressed the potential benefit of such a therapeutic strategy in buffering against cognitive decline in unaffected first-degree relatives of LOAD-affected individuals.

Emerging work has found that imagining mildly harming an individual (stealing, pushing) increased the participants' perceived likelihood of harming

How Imagination and Memory Shape the Moral Mind. Brendan Bo O’Connor, Zoë Fowler. Personality and Social Psychology Review, September 3, 2022. https://doi.org/10.1177/10888683221114215

Abstract: Interdisciplinary research has proposed a multifaceted view of human cognition and morality, establishing that inputs from multiple cognitive and affective processes guide moral decisions. However, extant work on moral cognition has largely overlooked the contributions of episodic representation. The ability to remember or imagine a specific moment in time plays a broadly influential role in cognition and behavior. Yet, existing research has only begun exploring the influence of episodic representation on moral cognition. Here, we evaluate the theoretical connections between episodic representation and moral cognition, review emerging empirical work revealing how episodic representation affects moral decision-making, and conclude by highlighting gaps in the literature and open questions. We argue that a comprehensive model of moral cognition will require including the episodic memory system, further delineating its direct influence on moral thought, and better understanding its interactions with other mental processes to fundamentally shape our sense of right and wrong.

Keywords: episodic simulation, imagination, memory, moral cognition

U-shape around middle age: Happiness initially increases after the age of 50, but commonly stagnates afterwards and eventually reverts at high age; this pattern does not emerge for all countries, and is not always observed for women

Does Happiness Increase in Old Age? Longitudinal Evidence from 20 European Countries. Christoph K. Becker & Stefan T. Trautmann. Journal of Happiness Studies, Sep 2 2022. https://rd.springer.com/article/10.1007/s10902-022-00569-4

Abstract: Several studies indicate that happiness follows a U-shape over the life cycle: Happiness decreases after the teenage years until reaching its nadir in middle age. A similar number of studies views the U-shape critically, stating that it is the result of the wrong controls or the wrong model. In this paper, we study the upward-pointing branch of the U-shape, tracing the happiness of European citizens 50 and older over multiple waves. Consistent with a U-shape around middle age, we find that happiness initially increases after the age of 50, but commonly stagnates afterwards and eventually reverts at high age. This pattern is generally observed irrespective of the utilized happiness measure, control variables, estimation methods, and the consideration of selection effects due to mortality. However, the strength of this pattern depends on the utilized happiness measure, control variables, and on mortality effects. The general pattern does not emerge for all countries, and is not always observed for women.

Discussion

Studies measuring happiness and well-being over the life cycle have found mixed results, and in particular the U-shape of happiness is a controversial finding. Consistent with a U-shape around middle age, we find that happiness increases after the age of 50, irrespective of the specification used. Furthermore, our results indicate that happiness tends to stagnate or even decrease at very high age. When conducting our analysis on country- or gender-specific subsamples, a more varied picture emerges. Where we find significant results in these subsamples, however, it is always consistent with a U-shape. These findings are also robust when accounting for differences due to mortality selection effects. While selection effects are indeed at work, with happier respondents being more likely to be alive at the time the next wave is elicited, CASP-12 is the only measure where the pattern is affected: selection makes the observed pattern more pronounced in this case. The result could potentially stem from the CASP-12 measuring control and agency, which decrease towards the end of one’s life (Oliver et al., 2021; Ribeiro et al., 2020; Rodríguez-Blázquez et al., 2020). This might also help to explain why we find lower turning points for CASP-12 and EURO-D in Table 4 in contrast to life satisfaction, when including additional controls. One reason why life satisfaction might continue to increase in high age is that older people might give up on aspiration and enjoy life more (Blanchflower & Oswald, 2004; Frey & Stutzer, 2010). CASP-12 and EURO-D, on the other hand, measure elements related to control and mental health, which might be more negatively affected by age. Different happiness measures might capture different aspects of life, highlighting the importance of looking at multiple measures at the same time.

Importantly, the observed age-happiness relation is consistently obtained using different approaches that have been used in both research that found and did not find the happiness dip in middle age. Additionally, the happiness-age relationship does not only hold for measures of subjective well-being (life satisfaction), but also for affective/eudemonic (CASP-12) and mental health measures (EURO-D). We are thus confident that our findings are meaningful for a substantial number of European countries.

Naturally, we can make no predictions about the trajectory of the happiness-age relation under the age of 50, as the SHARE data set only provides data for older Europeans. However, as other studies have indicated, there is support for the overall U-shape in various European countries (Blanchflower, 2021). We find that happiness indeed increases after middle age, compared to other studies finding a decrease after middle age (Easterlin, 2006; Mroczek & Spiro, 2005) or an overall decrease (Frijters & Beatton, 2012; Kassenboehmer & Haisken-DeNew, 2012). These differences could reflect regional differences, as Easterlin (2006) and Mroczek and Spiro (2005) use US data. Alternatively, methodological differences might drive these divergences. Kassenboehmer and Haisken-DeNew (2012) utilize respondents leaving the survey panel temporarily, to differentiate between age and years in the survey. Both should still be correlated, however. Frijters and Beatton’s (2012) main result is based on fixed effects regressions, which might ultimately not be reliable enough to deal with the age-period-cohort problem (Heckman & Robb Jr, 1985; Yang & Land, 2008). Mrozcek and Spiro’s (2005) use of a demeaned variable in their specification might similarly be problematic (McIntosh & Schlenker, 2006).

Our results are in line with previous studies indicating an increase of happiness after 50 (Morgan & O’Connor, 2017) or an upward profile for affective measures (Mroczek & Kolarz, 1998). However, similar to other studies, our results also provide evidence that happiness, depending on the measure used, stagnates or even decreases later in life (Blanchflower, 2021; Blanchflower & Graham, 2020; Gwozdz & Sousa-Poza, 2010). Our results support the view that people go through a period of relatively low happiness (relative to happiness at older age) around the midpoint of their life. For policy makers, it is important to further explore why this dip occurs and how it can be alleviated.

Going forward, it is important to highlight that proving or disproving the U-shape of happiness, or as in our case components of it, should not be a goal in itself. While knowing the average path happiness takes over the course of a human life is important, even more so is understanding which life events affect the emerging trajectory (Bjørnskov et al., 2008; Galambos et al., 2020, 2021; Lachman, 2015; Morgan & O’Connor, 2020). Past research has shown the happiness effects of marriage (Grover & Helliwell, 2019), parenthood (Nelson et al., 2013), social networks in general (Becker et al., 2019), income (Easterlin, 1974), social support (Siedlecki et al., 2014), permanent employment (Piper, 2021), the quality of formal institutions (Bjørnskov et al., 2010), giving up on aspirations (Schwandt, 2016), and health (Bussière et al., 2021; Gwozdz & Sousa-Poza, 2010; Oliver et al., 2021). Mapping the evolution of these events over the life course may help to better understand the emergence of the U-shape of happiness.