Saturday, February 18, 2023

Contemporary sex differences in partner preferences are maintained by several psycho-biological mechanisms that evolved in conjunction, including not only sexual but also romantic attraction

The Role of Sexual and Romantic Attraction in Human Mate Preferences. Meike Scheller et al. The Journal of Sex Research, Feb 16 2023.

Abstract: Sex differences in mate preferences are ubiquitous, having been evidenced across generations and cultures. Their prevalence and persistence have compellingly placed them in the evolutionarily adaptive context of sexual selection. However, the psycho-biological mechanisms contributing to their generation and maintenance remain poorly understood. As such a mechanism, sexual attraction is assumed to guide interest, desire, and the affinity toward specific partner features. However, whether sexual attraction can indeed explain sex differences in partner preferences has not been explicitly tested. To better understand how sex and sexual attraction shape mate preferences in humans we assessed how partner preferences differed across the spectrum of sexual attraction in a sample of 479 individuals that identified as asexual, gray-sexual, demisexual or allosexual. We further tested whether romantic attraction predicted preference profiles better than sexual attraction. Our results show that sexual attraction accounts for highly replicable sex differences in mate preferences for high social status and financial prospects, conscientiousness, and intelligence; however, it does not account for the enhanced preference for physical attractiveness expressed by men, which persists even in individuals with low sexual attraction. Instead, sex differences in physical attractiveness preference are better explained by the degree of romantic attraction. Furthermore, effects of sexual attraction on sex differences in partner preferences were grounded in current rather than previous experiences of sexual attraction. Taken together, the results support the idea that contemporary sex differences in partner preferences are maintained by several psycho-biological mechanisms that evolved in conjunction, including not only sexual but also romantic attraction.


While sex differences in human mate preferences have been observed across cultures and generations (Buss, Citation1989; Shackelford et al., Citation2005; Walter et al., Citation2020) the proximate factors contributing to its generation and maintenance remain less well understood (but see Alexander et al., Citation2011; Balthazart, Citation2011Citation2016; Little, Jones et al., Citation2011; Scheller et al., Citation2021). Attraction, especially sexual attraction, is a psycho-biological mechanism that is thought to guide interest, desire or the affinity toward specific features of others and establishes an individual’s preference profile. In order to better understand how sex and sexual attraction interact, and whether other forms of attraction (i.e., romantic) may explain persistent patterns in mate preferences, we assessed how partner preferences differ across the spectrum of sexual attraction intensity.

In individuals that experience high sexual attraction we replicated well-documented sex differences in partner preferences that are linked to sexual reproduction. Here, heterosexual men rated physical attractiveness higher than women, while the latter placed higher importance on social status and financial prospects than men. This is in line with a large body of research showing sex differences in preferences for these traits (Bech-Sørensen & Pollet, Citation2016; Buss, Citation1989; Walter et al., Citation2020). The data-driven nature of the factor composition indicated two further partner characteristic factors that were described as conscientiousness (ambition, emotional stability/maturity, reliability, diligence, humor, and sociability) and intelligence/education (be educated, similar education, intelligence). Here, individuals that experienced high sexual attraction showed significant sex differences in preferences, with women rating conscientiousness as more important, while men gave higher importance ratings for intelligence and education. Indeed, previous studies have found conscientiousness, ambition and emotional stability to be more highly valued by women (Botwin et al., Citation1997; Buss & Schmitt, Citation2019; Furnham, Citation2009); however, the enhanced importance ratings of intelligence and education in allosexual men seem to contradict previous findings in hetero-allosexual samples in which women valued education more than men (Buss & Schmitt, Citation2019; Shackelford et al., Citation2005). Lippa (Citation2007), however, reported that heterosexual women rated intelligence lower than lesbian women and gay men, which may suggest that sexual intensity and orientation may indeed mediate intelligence importance ratings. As women show lower average sexual attraction levels than men and often show higher levels of non-exclusivity in their sexual orientation (e.g., bisexual preferences; Diamond, Citation2016), the effect of diversity in sexual intensity and orientation in women may not have been captured in previous studies, but impacted the absolute importance ratings. However, this would likely have affected all partner characteristics, and future research into the effects of sexual orientation and intensity on the importance of specific partner traits is needed to elucidate whether these effects replicate in a different sample.

Effects of Sexual and Romantic Attraction on Sex Differences in Partner Preferences

Notably, our results show that sexual and romantic attraction both help explain the maintenance of sex-specific mate preference differentiation. Self-reported preferences for all four partner traits were significantly or marginally significantly modulated by sex as well as either sexual or romantic attraction. Furthermore, the results support the notion that both forms of attraction can function independently (Diamond, Citation2003Citation2004; Fisher et al., Citation2005Citation2006) and do not affect partner preferences in the same way. While sex-specific differences in preferences for a high social status and good financial prospects (women > men), as well as intelligence and education (men > women) were reduced in individuals with low or no sexual attraction, the increased preferences for a conscientious partner in women was more strongly modulated by romantic than sexual attraction. Surprisingly, the only sex-specific difference that sexual attraction could not account for was the heightened preference for a physically attractive partner in men. Even individuals that reported to have never experienced sexual attraction before showed a sex difference in the preference for this partner trait that was of similar magnitude (β = −0.41) as that present in allosexual individuals (β = −0.39). Instead, sex-specific physical attractiveness preferences were more dependent on the degree of romantic attraction, highlighting the importance of physical attractiveness in romantic contexts. Note that, while the interaction of romantic attraction and sex was only marginally significant (p = .06), effect sizes suggested that the average explanatory power of romantic attraction was 25 times higher than that of sexual attraction.

This suggests that physical attractiveness may serve more functions than providing good genes. In fact, the heightened preference that men express for physical attractiveness even extends outside of the mating context, affecting other social bonds such as friendships. Here, men place more importance than women on physical attractiveness in opposite-sex friends (Lewis et al., Citation2011). Perhaps this is due to the indirect benefit gained from being associated with physically attractive individuals. That is, partnering with a physically attractive woman may enhance the perception of someone’s own social status and mate value, thereby attaching higher chances of partnering with other attractive woman in the long term (C. Anderson et al., Citation2001; R. C. Anderson & Surbey, Citation2014; see supplement S3 for indirect appearance belief effects).

Sexual attraction is not a prerequisite for romantic attraction. In fact, many individuals with reduced sexual desire still experience romantic attraction and look to engage in romantic relationships with others (Antonsen et al., Citation2020). However, the specific preferences for such romantic, non-sexual relationships may differ. For instance, women with low sexual attraction placed more importance on intelligence and education, and less importance on physical attractiveness, status and financial prospects or conscientiousness than women with higher sexual attraction. At the same time, men with low sexual attraction placed less importance on all partner characteristics, except social status and financial prospects, which already received the lowest importance rankings amongst all character traits. Our results show that sex differences in preferences for conscientiousness and physical attractiveness are better explained by romantic attraction, while those in social status/resources and intelligence are better explained by sexual attraction. This pattern may be explained by the types of benefits accrued in different relationship contexts (Buss & Schmitt, Citation1993; Li, Citation2007; Regan et al., Citation2000). For example, although previous research has often noted potential good-gene benefits from short-term partnerships, a long-term romantic relationship with a conscientious and physically attractive partner would provide commitment, investment, and potentially good gene benefits to offspring. Indeed, as romantic attraction is a crucial factor facilitating pair-bonding and (parental) care in monogamous relationships we might also expect that physical attractiveness will be important if there are associated benefits to offspring. Our data may suggest that, if sexual attraction is associated with short term mating (Edlund et al., Citation2021; Li, Citation2007), perhaps the increased preferences for status/resources reflect more material direct benefits from such partnerships.

Furthermore, our data showed that, while sexual and romantic orientation (directionality) were mostly aligned in allosexual individuals, the correlation was markedly lower among those with lower sexual attraction ratings. This divergent sexual and romantic orientation again suggests that sexual attraction and romantic attraction operate independently not only via their intensity but also their orientation (Antonsen et al., Citation2020; Diamond, Citation2004; Tennov, Citation1998).


Firstly, the AIS score was used as an indicator of progressively decreasing sexual attraction in the present study. However, it was primarily developed to differentiate those identifying as asexual from those who do not (Yule et al., Citation2015). As the AIS incorporates further measures that go beyond mere sexual attraction (such as sex-related disgust, sexual behavior avoidance, or sexual identity), it captures a wider experience of reduced sexual attraction and interest. In order to ensure that the main feature of interest i.e., reduced sexual attraction, was the modulating factor, we included a second, isolated measure of sexual attraction intensity, along with a measure of romantic attraction intensity. Here, participants were asked to indicate how strongly they felt sexually or romantically attracted to men and to women on a 7-point Likert scale. Comparing our findings from the first analysis using AIS scores and the second analysis using raw sexual attraction intensity scores, we found that results were similar across both analyses: both AIS scores and sexual attraction intensity show no interaction with sex in explaining preferences for physical attractiveness, while they both explain sex differences in the preferences for social status, conscientiousness, and intelligence (note that the inverse coefficients result from low sexual attraction being indicated by high AIS scores). This suggests that, while the AIS score offers a more fine-grained and experience-dependent measure of sexual attraction, desire and interest, sexual attraction alone is a large contributing factor to the expression of specific partner preferences.

Secondly, while sexual attraction is considered here as a psycho-biological mechanism that evolved to maintain partner preferences that adapted to evolutionary reproductive pressures, reduced sexual attraction does not automatically equate to a lowered desire to procreate. That is, individuals with reduced sexual attraction may not seek sexual encounters for pleasure but may still express the desire to have and raise children. If the desire is not reduced, this can suggest that partner preferences in people with reduced sexual attraction may not be targeted at a partner with whom genetic material or resources are exchanged to provide for offspring. Instead, the alteration in partner preferences might be driven by other factors, depending on the function of the partnership. Indeed, previous research suggests that the function of a partnership, such as in short-term or long-term mating, alters mate preferences (Bode & Kushnick, Citation2021; Jonason et al., Citation2013; Li, Citation2007). However, we observed, on average, a reduced desire to have and raise children across sexual orientation groups with low sexual attraction (demi-, gray- and asexuals), compared to individuals with high sexual attraction (see supplement S6). This suggests that reduced sexual desire and reduced desire to have and raise children with a partner may shape the expression of specific partner preferences.

We appraise the empirical claims claiming that implicit bias deeply affects legal proceedings and practices, and that training can be used to reduce that bias; we find that these claims carry many indicia of unreliability

Jason Chin, Alexander Holcombe, Kathryn Zeiler, Patrick Forscher & Ann Guo, Metaresearch, Psychology, and Law: A Case Study on Implicit Bias (2023). SSRN Feb 15 2023.

Abstract: When can scientific findings from experimental psychology be confidently applied to legal issues? And when applications have clear limits, do legal commentators readily acknowledge them? To address these questions, we survey recent findings from an emerging field of research on research (i.e., metaresearch). We find that many aspects of experimental psychology’s research and reporting practices threaten the validity and generalizability of legally relevant research findings, including those relied on by courts and policy-setting bodies. As a case study, we appraise the empirical claims relied on by commentators claiming that implicit bias deeply affects legal proceedings and practices, and that training can be used to reduce that bias. We find that these claims carry many indicia of unreliability. Only limited evidence indicates that interventions designed to reduce prejudicial behavior through implicit bias training are effective, and the research area shows many signs of publication bias. To examine whether law journal articles are acknowledging these limits, we collected a sample of 100 law journal articles mentioning “implicit bias training” published from 2017-2021. Of those 100 articles, 58 recommend implicit bias training and only 8 of those 58 express any skepticism about its effectiveness. Overall, only 19 articles express skepticism about implicit bias training. We end with recommendations for law journal authors, researchers, and practitioners towards more credible application of psychology findings in law research and policy. Our focus is on how empirical research can be best used to solve our most important social issues including racism.

The fears about the menace of misinformation are primarily fed by the misconception that others are more gullible than oneself

People believe misinformation is a threat because they assume others are gullible. Sacha Altay, Alberto Acerbi. New Media & Society, February 17, 2023.

Abstract: Alarmist narratives about the flow of misinformation and its negative consequences have gained traction in recent years. If these fears are to some extent warranted, the scientific literature suggests that many of them are exaggerated. Why are people so worried about misinformation? In two pre-registered surveys conducted in the United Kingdom (Nstudy_1 = 300, Nstudy_2 = 300) and replicated in the United States (Nstudy_1 = 302, Nstudy_2 = 299), we investigated the psychological factors associated with perceived danger of misinformation and how it contributes to the popularity of alarmist narratives on misinformation. We find that the strongest, and most reliable, predictor of perceived danger of misinformation is the third-person effect (i.e. the perception that others are more vulnerable to misinformation than the self) and, in particular, the belief that “distant” others (as opposed to family and friends) are vulnerable to misinformation. The belief that societal problems have simple solutions and clear causes was consistently, but weakly, associated with perceived danger of online misinformation. Other factors, like negative attitudes toward new technologies and higher sensitivity to threats, were inconsistently, and weakly, associated with perceived danger of online misinformation. Finally, we found that participants who report being more worried about misinformation are more willing to like and share alarmist narratives on misinformation. Our findings suggest that fears about misinformation tap into our tendency to view other people as gullible.

Men and women are less alike, especially in personality traits and basic human values, in countries that have invested the most in gender equality

Linking gender differences with gender equality: A systematic-narrative literature review of basic skills and personality. Marco Balducci. Front. Psychol., February 16 2023, Volume 14 - 2023.

Abstract: There is controversy regarding whether gender differences are smaller or larger in societies that promote gender equality highlighting the need for an integrated analysis. This review examines literature correlating, on a national level, gender differences in basic skills—mathematics, science (including attitudes and anxiety), and reading—as well as personality, to gender equality indicators. The aim is to assess the cross-national pattern of these differences when linked to measures of gender equality and explore new explanatory variables that can shed light on this linkage. The review was based on quantitative research relating country-level measures of gender differences to gender equality composite indices and specific indicators. The findings show that the mathematics gender gap from the PISA and TIMMS assessments, is not linked to composite indices and specific indicators, but gender differences are larger in gender-equal countries for reading, mathematics attitudes, and personality (Big Five, HEXACO, Basic Human Values, and Vocational Interests). Research on science and overall scores (mathematics, science, and reading considered together) is inconclusive. It is proposed that the paradox in reading results from the interrelation between basic skills and the attempt to increase girls’ mathematics abilities both acting simultaneously while the paradox in mathematics attitudes might be explained by girls being less exposed to mathematics than boys. On the other hand, a more nuanced understanding of the gender equality paradox in personality is advanced, in which a gene–environment-cultural interplay accounts for the phenomenon. Challenges for future cross-national research are discussed.

6. Discussion

The systematic narrative literature review investigated recent studies on gender differences in basic skills and personality to determine whether cross-national relationships can be found with gender equality. The goal was to assess whether theories predicting that gender equality is linked with smaller gender differences have empirical support or whether a gender equality paradox has emerged in recent years. The general trend considers gender equality as either being connected to an increase in gender variations or having no relation with them, with a gender equality paradox occurring for gender gaps in some cognitive domains (attitudes toward mathematics, mathematics self-efficacy, mathematics anxiety, and reading) and personality.

6.1. Summary of the review

Based on the foregoing literature review, it can be seen that research supporting reduced gender differences in more gender-equal countries is scarce and inconsistent. A negative correlation is generally detected when analyzing gender differences in mathematics skills utilizing PISA data, although the correlation is influenced by either the year considered in the study or the sample country (see below). Moreover, “women in research” is the only specific indicator consistently negatively linked to the mathematics gender gap, albeit with disagreement about the strength of the association. Lastly, no connection between gender differences in mathematics and gender equality indicators is found when analyzing the TIMMS assessment. However, many studies have focused solely on mean differences in mathematics abilities, which are small or non-existent. Only Bergold et al. (2017) and Hyde and Mertz (2009) assessed the right tail of the distribution, where gender differences are more pronounced. This lack of studies on top performers highlights a gap in the research that needs to be filled. Also important is analyzing intra-individual strengths when studying the mathematics gender gap, as Stoet and Geary (2018) have emphasized.

Research supporting a positive link between gender variances and gender equality measures appears to be more robust and consistent. The literature on mathematics attitudes and anxiety shows that composite indicators predict a widening gender gap as equality between men and women advances. In addition, scholars agree that gender equality is connected with a larger advantage for women in reading and evidence further shows that gender personality differences are larger in more gender-equal nations. Men and women are less alike, especially in personality traits and basic human values, in countries that have invested the most in gender equality. Further support for a gender equality paradox in personality also emerges when examining other personality domains not included in this review. For example, wider gender gaps in self-esteem and narcissism (higher in men) exist in more gender-equal nations where women have more reproductive control, more executive positions, and their education is either similar to or higher than that of men (Bleidorn et al., 2016Jonason et al., 2020).

Specific indicators are either directly or inversely related to the mathematics gender gap, raising doubt about them being related to a general advantage (Table 4). In addition, findings on science and overall scores are uncertain, even though both science anxiety and science intra-individual strengths follow a trend opposite to that anticipated by theories predicting a link between gender equality and smaller gender differences. Interestingly, other skills, such as episodic memory and visuospatial ability, show the same widening tendency, strengthening the case for a possible paradox in this area (Lippa et al., 2010Asperholm et al., 2019).

Table 4. Summary of the papers included in the review.

6.2. Implications of the gender equality paradox

Understanding the possible reasons for the increase in gender differences in countries that promote gender equality is important and relevant since these countries may be leading men and women toward gendered trajectories, a path that is already observable in higher education. Charles and Bradley (2009) noted that the most advanced societies demonstrate more pronounced gender segregation in education. Stoet and Geary (2018) also observed that more gender-equal nations (measured by the GGI) have the widest gender gap among STEM graduates. Supporting these results, research has shown that gender differences using “interest in math careers” as a predictor of future major subjects are greater in countries with higher gender equality, with both men and women being, on average, less interested in mathematics than those in other countries (Goldman and Penner, 2016Charles, 2017Breda et al., 2020). The same pattern is observed in the job market, where horizontal segregation is more pronounced in more gender-equal environments (Blackburn and Jarman, 2006Wong and Charles, 2020). Several investigations have documented this phenomenon and concluded that “Scandinavian countries are notable for their exceptionally high degrees of segregation” despite their advancement in gender equality (Jarman et al., 2012). However, more recent findings have also detected desegregation patterns in more gender-equal nations (Hustad et al., 2020).

6.3. The gender equality paradox: Possible explanations

The question of why gender differences are sometimes higher in more gender-equal countries remains. Some have proposed that the paradox in mathematics anxiety and attitudes might originate from the better economic conditions needed for these emotions to emerge. In countries where women are highly oppressed, these are more concerned about meeting more basic needs. Conversely, where economic, political, and educational circumstances are more favorable for women, anxiety toward mathematics activities is more likely to emerge (Else-Quest et al., 2010). However, at the national level, both men and women are less anxious about mathematics in developed, gender-equal countries, indicating that alternative explanations are needed (Stoet et al., 2016). In fact, others have suggested that, in gender-equal nations, men and women set aside financial drives and follow more intrinsic career interests because of easier access to economic resources. Hence, women are less exposed than men to STEM activities, “giving them less opportunity to reduce their negative feelings about mathematics” (Stoet et al., 2016).

With respect to reading abilities, the paradox might result from the interaction of two factors: the interrelation between basic skills and Western societies’ strong efforts to equalize boys’ and girls’ mathematics performance that has instead, paradoxically, increased reading skills in girls. Notably, where mathematics gender differences are reduced, the reduction is mainly due to an improvement in women’s reading (Guiso et al., 2008). It follows that countries with smaller mathematics gender differences have the largest reading gaps (Stoet and Geary, 2013). As mathematics is promoted in girls, their reading skills appear to benefit. However, because boys’ disadvantage in reading is, on average, less of a concern among policymakers, gender variations in this dimension have widened.

Some researchers have explained the gender equality paradox in personality by arguing that only differences in self-reported domains are increased (Eagly and Wood, 2012). Here, the reference-group effect (Heine et al., 2002) might conceal variances in less gender-equal countries, where men and women compare themselves with others of their own gender (Guimond et al., 2007). If this explanation holds true, the gap in gender-equal nations would be a better estimate of personality differences between the genders because in these nations both women and men have a more accurate comparative term that includes the whole population rather than just a subset (Schmitt et al., 2017).

Another explanation may be that personality is strongly culturally influenced. According to this view, individualism and self-expressive values act in tandem with gender stereotypes, promoting gender variance as individuals act out their “gendered self” (Charles and Bradley, 2009Breda et al., 2020). This explanation of the gender equality paradox corresponds to the findings in gender-equal nations that cultural mechanisms are at play accommodating women-typical roles, such as job flexibility and high parental care—roles that encourage women to embark on gendered paths and experience more communal traits (Levanon and Grusky, 2018). Thus, it should not be surprising that, in gender-equal countries, men and women appear to differ more than in non-gender-equal countries and that this difference is expanding as women-typical roles are becoming more prevalent. Rather than expressing intrinsic gender differences, in these nations, there is a reinforcement of gender essentialist beliefs, which constitute an artifact of social expectations about how men and women should comply with gender stereotypes (England, 2010).

While this argument is somewhat persuasive, research aiming at linking gender stereotypes with gender equality suffers from several theoretical and methodological limitations. Often scholars apply broad assumptions and rely on a limited, as well as unreliable, set of items to capture latent dimensions of implicit stereotypes hidden in survey data. For instance, in their recent article Napp and Breda (2022) used solely one item to grasp an alleged stereotype that girls lack talent by arguing that systematic gender difference in answering the question would highlight “the magnitude of the (internalized) stereotype associating talent with boys rather than girls.” In addition, several studies have argued that stereotypes about group features, when measured reliably, appear to be accurate (Jussim et al., 2015Moè et al., 2021). Löckenhoff et al. (2014) observed that perceived gender differences in personality substantially match those found in self- and observer-rated personality tests. The authors concluded that gender stereotypes constitute “valid social judgments about the size and direction of sex differences” that are more relevant than socialization processes and ascribed cultural gender roles (Löckenhoff et al., 2014). This is not to say that culture plays no role in the emergence of gender differences, but that the social mechanisms amplifying gender variances—mechanisms that social-role theorists have identified—also capture intrinsic gender differences.

Evolutionary theorists propose a different explanation for the gender equality paradox. As they argue, some gender variations are sensitive to context-related fluctuations, demonstrating a gene–environment interplay. In societies in which conditions are favorable, gender-specific genes flourish due to a lower prevalence of diseases, lower ecological stressors, and lower starvation rates. Per this view, wider gender gaps in gender-equal nations most likely “reflect a more general biological trend toward greater dimorphism in resource-rich environments” (Schmitt et al., 2008). If this explanation holds true, then heritability estimates will be higher in developed societies than in less-advanced cultures. Some evidence in this direction has recently emerged (Selita and Kovas, 2019); however, the “WEIRD” gene problem—that nearly all twin studies have been conducted among Western, educated, industrialized, rich, and democratic societies—represents an obstacle for generalizing results and making inferences about cross-cultural heritability differences (Henrich et al., 2010).

6.4. A novel socio-cultural evolutionary account of the gender equality paradox in personality

The present review proposes that the evolutionary explanation for the gender equality paradox might be more complex than it appears due to the presence of socio-cultural elements in the evolutionary process. As previously noted, genetic effects depend on the environmental conditions (diseases and ecological stress) under which they occur, yet the environment is embedded into society. Thus, the gene–environment interplay is enclosed within a cultural context with specific social norms and, by itself, cannot encompass all involved elements (Figure 2). Stated otherwise, the gene–environment interplay is a function of culture (Uchiyama et al., 2022). Therefore, gender-specific genes can be expected to be emphasized in societies embracing cultural values that would favor the expression of these genes. Consider, for example, individualism and self-expression. It is unsurprising that these values are related to the gender equality paradox, as Charles and Bradley (2009) have highlighted. In resource-rich environments that also value individualism and self-expression, intrinsic gender differences are more likely to emerge. This thesis points toward interpretation of Kaiser (2019), which states that both cultural individualism and pathogen levels confound the gender equality paradox in personality (see below). Also, Murphy et al. (2021) reached similar conclusions. A coherent, yet opposite, prediction might see gender differences as remaining stable or even decreasing in those resource-rich environments that culturally constrain self-expression. Accordingly, favorable cultural values would trump social mechanisms that amplify gender-based genes to emerge via a feedback-loop effect or “reciprocal causation” (Dickens and Flynn, 2001) according to which social structures adjust to distinct gender traits and vice versa, thus increasing gender differences.

Figure 2. Socio-cultural evolutionary explanation of the gender equality paradox. The gears show the interrelations between gender-specific genes, social structures, and environmental components mediated by cultural values.

6.5. Challenges for future cross-national research

While searching and analyzing the literature, this review also highlighted some challenges that researchers might face when conducting cross-national studies relating gender differences to gender equality measures. For mathematics ability, results could depend on outlier countries such as Scandinavian and gender-segregated, Muslim countries. In addition, the restricted country samples in international student assessments might be problematic. Despite the strong effort of PISA and TIMMS to be more inclusive, wealthy countries have traditionally been overrepresented, although the latest rounds have had very high coverage, including over 75 participating nations worldwide. Nevertheless, researchers, when assessing gender differences in mathematics abilities, should pay close attention to the countries included in their study because either the inclusion of outliers or a lack of heterogeneity might lead to biased estimations.

Another possible source of bias in research linking gender differences to gender equality on a cultural level is participant sample sizes, with some nations being overrepresented in comparison to others. How countries are clustered may also be problematic since countries are not independent data points and, “as such, they are like members of the same family or pupils of the same classroom” (Kuppens and Pollet, 2015). Therefore, appropriate statistical methods, multilevel modeling, for example, should be utilized to account for both unbalanced sample sizes and data structure.

Correlations between mathematics gender differences and gender equality might originate from a lack of country-level effects in the models. Anghel et al. (2019) argued that when time-invariant country unobserved heterogeneity is controlled for, no association between the two variables is found. Moreover, the link between gender equality and the gender gap in mathematics attitudes might be confounded by country-level academic achievements and socioeconomic status (Marsh et al., 2021).

Further, the gender equality paradox could be due to measurement error. Given that many international assessments and personality models have been developed in WEIRD countries, it is plausible that measurement error could be higher in non-WEIRD nations generating an illusory gender equality paradox. However, international assessments have been constructed to prevent such bias. For instance, PISA computes each student’s score based on a set of 5/10 plausible values designed to prevent measurement error and simplify secondary data analysis (Marsh et al., 2021). Also, the gender equality paradox in personality appears to hold even after correcting for measurement error (Kaiser, 2019Fors Connolly et al., 2020Tao et al., 2022). Nevertheless, when analyzing the link between gender differences in personality and gender equality, statistical procedures that control for measurement error should be applied (see for example Schmidt and Hunter, 2015).

Fors Connolly et al. (2020) highlighted the need for more temporal analyses of personality because an observed cross-national pattern may result from “a spurious relationship between gender equality and differences in personality” due to different country-level elements. Kaiser (2019) identified these elements as cultural individualism, food consumption, and historical pathogen prevalence levels. Other research has also agreed that cultural individualism could be a possible confounding factor as gender differences in personality are more pronounced in nations that highly regard individual self-expression (Costa et al., 2001Schmitt et al., 2008Tao et al., 2022).

Some scholars have called attention to the misuse of composite indicators of gender equality, raising several concerns thereof and arguing that they might not be suitable for empirical research (Else-Quest et al., 2010Hyde, 2012). One concern is that these indicators, which encompass various domains from politics to economics, do not measure opportunities (Richardson et al., 2020). Another concern is that they are not interchangeable since they are differentially constructed. Thus, comparisons between research relying on different measures of gender equality might not be suitable. Some of the disparate findings concerning math ability might be driven by computational differences in the indices included in the analysis. Nevertheless, the gender equality composite indicators most commonly utilized (GGI, GEI, and GEM) show very high correlation coefficients (r ≥ 0.84), while other indicators substantially relate to one another, suggesting that, although some differences occur, these indices are similar in their ability to capture the general dimension of gender equality (Else-Quest et al., 2010van Staveren, 2013Stoet and Geary, 2015). Lastly, composite indicators may present a biased view of society due to the way gender equality is understood in the models. Often, disadvantages pertaining mostly to men are not taken into account when computing the indicators (Benatar, 2012). As an example of this bias, the GGI from the World Economic Forum assumes perfect gender equality in areas where women have an advantage over men. Specifically, values higher than 1, which would assume a men’s disadvantage, in each sub-index are capped. Thus, a more simplified approach to measuring national gender inequality is preferred (Stoet and Geary, 2019).

In addition, methodological issues also arise when using these indices. Some scholars have pointed out that correlations between gender gaps and the indices of gender equality could be driven by the strong economic component in these indices (Fors Connolly et al., 2020). Therefore, it is important to control for appropriate economic indicators, such as GDP per capita and the Human Development Index, when linking gender differences with gender equality (Kuppens and Pollet, 2015). Another difficulty may arise when contrasting results between composite indices and specific indicators occur. For mathematics attitudes, for instance, although composite indices suggest a gender equality paradox, specific indicators are either positively or negatively related to the gender gap. This may suggest that composite indices either capture an overall influence of gender equality or are unsuitable for evaluating gender differences. However, evaluation may lie outside the scope of models using these indices. Research linking gender differences with gender equality indicators has not tried to explain the paradox emerging from the analysis on the basis of gender equality per se; instead, it has just highlighted a paradoxical pattern that would otherwise have remained concealed. Since no theory has been put forward that fully unravels the paradox, further studies are needed.

Theories considered in this review that predict that gender equality is linked with smaller gender differences do not offer a valid explanation of gender differences in basic skills and personality. In addition, for some dimensions, the gender equality paradox raises further questions about how gender variation emerges, which calls for a new approach. Based on these premises, this review explored both social-role and evolutionary hypotheses and suggested new insights that combine these views, while also highlighting explanatory variables that might cause bias in the results. Thus, specific research that more closely examines the explanations proposed is needed, especially studies with an interdisciplinary focus. Notably, Fors Connolly et al. (2020) highlighted the importance of cross-temporal analyses of the gender equality paradox because these may reveal a different path. Since country comparisons may be insufficient for fully grasping the evolution of the paradox, future research should include a thorough cross-temporal examination for a more comprehensive understanding.

Lastly, the gender equality paradox is an emerging phenomenon that has gained substantial scientific support across subjects (Falk and Hermle, 2018Campbell et al., 2021Block et al., 2022Vishkin, 2022). It requires attention from both the scientific community and the public because attempting to close gender gaps following traditional social-role theories and applying conventional methods, might end up exacerbating gender variations. In addition, the general pattern of increased gender differences in more gender-equal countries might inform that achieving equal opportunities does not go hand in hand with a reduction of gender gaps. Thus, policymakers should consider this trend when justifying interventions attempting to achieve equality of outcome between men and women.

Sex differences that grow larger in more gender egalitarian societies are probably sexually selected "luxuries," which are condition-dependent and blossom under favorable circumstances

Now you see them, and now you don’t: An evolutionarily informed model of environmental influences on human sex differences. David C. Geary. Neuroscience & Biobehavioral Reviews, Volume 125, June 2021, Pages 26-32.

Abstract: The contributions of evolutionary processes to human sex differences are vigorously debated. One counterargument is that the magnitude of many sex differences fluctuates from one context to the next, implying an environment origin. Sexual selection provides a framework for integrating evolutionary processes and environmental influences on the origin and magnitude of sex differences. The dynamics of sexual selection involve competition for mates and discriminative mate choices. The associated traits are typically exaggerated and condition-dependent, that is, their development and expression are very sensitive to social and ecological conditions. The magnitude of sex differences in sexually selected traits should then be largest under optimal social and ecological conditions and shrink as conditions deteriorate. The basics of this framework are described, and its utility is illustrated with discussion of fluctuations in the magnitude of human physical, behavioral, and cognitive sex differences.


The existence of human sex differences is no longer debated, at least for some traits, but their origin, magnitude, and practical significance remain areas of contention (Archer, 2019; Hyde, 2005). The variable expression of sex differences across contexts adds to the contention and is often interpreted as evidence for the cultural origin of these differences (Wood and Eagly, 2002). I outline here how an evolutionary perspective helps us to understand these contextual effects and at the same time places the study of human sex differences within the same unifying framework used to study them in nonhuman species, that is, Darwin’s (1871) sexual selection (for review see Andersson, 1994). In the following, I provide a brief introduction to sexual selection and condition-dependent traits in nonhuman species and then illustrate how these principles can be used to understand fluctuations in the magnitude of sex differences in human traits.

Darwin’s (1871) sexual selection provides the evolutionary framework for the study of sex differences and includes competition with members of the same sex over mates (intrasexual competition) and discriminative choice of mating partners (intersexual choice). These dynamics have traditionally focused on male-male competition and female choice but in recent decades it has become clear that female-female competition over access to mates or access to other resources (e.g., high-quality food) is common (Stockley and Bro-Jørgensen, 2011; West-Eberhard, 1983), albeit typically not as intense (e.g., resulting in serious injury) as that found among males of the same species. As with females, males often show discriminative mate choices, especially when they provide some level of investment in offspring (Berglund and Rosenqvist, 2001; Reynolds and Székely, 1997).

The result of competition and mate choices is the elaboration of the traits that contribute to them. One result, as shown in the top set of distributions in Fig. 1, is that sexually selected traits are typically larger or more elaborated (e.g., plumage color) in the sex that is subjected to more intense competition or more exacting mate choices (Janicke et al., 2016). These traits are often physical, such as the mandibles of the male and female beetles (Chalcosoma atlas) shown in Fig. 2 but can also include behavioral or cognitive traits. Behavioral traits range from courtship displays to the building of the elaborate wooden structures that male bowerbirds use to attract potential mates (e.g., Borgia, 1985). Birdsong is among the better studied brain and cognitive traits associated with female choice (Ball and Hulse, 1998), as is spatial ability associated with male-male scramble competition (Gaulin, 1992; Jašarević et al., 2012). For the latter, males expand their home range during the breeding season and search for prospective mates that are dispersed throughout the ecology. Males with better navigational abilities find more mates and generally have higher reproductive success than their less adventurous peers (Spritzer et al., 2005).

The dynamics of competition and choice not only lead to the evolution of sex differences in trait size or degree of elaboration (e.g., plumage color), they also result in a heightened sensitivity of these traits to social and ecological stressors. The basic idea is shown in the bottom set of distributions in Fig. 1, whereby current or developmental exposure to stressors will compromise the development and expression of sexually selected traits more severely than other traits. The result is that sex differences become smaller and more variable than they would be under better conditions. The heightened sensitivity of these traits makes their expression dependent on the condition (e.g., physical health) of the individual and reduces the ability of unfit individuals to bluff in the context of intrasexual competition or cheat in the context of mate choices (Zahavi, 1975). Across species, the most common stressors that compromise these traits are nutritional deficits, parasitic diseases, and chronic social stress (Geary, 2015). Many man-made toxins have similar effects, that is, they compromise sexually selected traits more severely than other traits (e.g., Bortolotti et al., 2003; Jašarević et al., 2011).

The mechanisms underlying the heightened vulnerability of these traits are not yet fully understood but might include the efficiency of mitochondrial functioning (Hill, 2014). Mitochondria are the primary source of cellular energy and thus the common currency for the development and functioning of all biological traits. They are also the source of cell-damaging oxidative stress, and contribute to hormone synthesis, immune functioning and other basic cellular processes (von Schantz et al., 1999; Weinberg et al., 2015). The stressors that typically compromise condition-dependent traits can directly or indirectly compromise one or several aspects of mitochondrial functioning and result in a reduction in cellular energy production (Koch et al., 2017; Picard et al., 2016). On this view, the vulnerability of condition-dependent traits follows directly from their exaggerated size or other elaborations because these require more cellular energy to build, develop, and maintain than do other traits. By analogy, consider that it takes more energy to heat a 300-square-meter house than a 100-square-meter house. A drop in available energy will be noticed first in the larger house and result in a more rapid drop in ambient temperature relative to the smaller one. In fact, if available energy is sufficient for the latter, then a change in energy availability will go unnoticed.

The main point is that any advantages in trait size or elaboration enjoyed by one sex necessarily come with the attendant costs of building, maintaining, and expressing these traits. On the basis of this proposal, conditions with abundant nutritional resources, low disease risk, and muted social competition will result in near maximal trait expression, within any genetic constraints, and large sex differences for sexually selected traits. As these conditions deteriorate, many members of the advantaged sex can no longer build and maintain exaggerated traits and thus the magnitude of any sex difference for these traits will become smaller and the variation among members of the advantaged sex will become larger.

Sex differences in physical size and strength are consistent indicators of male-male competition and a polygynous mating system in mammals (Andersson, 1994). Males of these species also grow more slowly than females, mature at a later age, and have a shorter lifespan (Clutton-Brock and Isvaran, 2007; Leigh, 1995). Human physical and developmental sex differences fit this general pattern (Tanner, 1990), in keeping with an evolutionary history of physical male-male competition. Examination of the likely size differences between our male and female ancestors suggests that intense reproductive competition among males stretches back at least four million years (Leakey et al., 1998). In traditional contexts, male-male competition includes fights for dominance within the ingroup and smaller- to larger-scale raids of competing groups, with 20 % or more of men dying as a result of such conflicts (e.g., Walker and Bailey, 2013). The historical record and population genetic studies indicate the intensity of conflict and the attendant variation in men’s reproductive success intensified with the emergence of agriculture and early empires (Betzig, 2012; Zeng et al., 2018). The exercise of dominance-related physical male-male competition was slowly suppressed in modern nations over the past few centuries (Pinker, 2011), and partially replaced by knowledge- and skills-based competition (Geary, 2021); also known as prestige-based competition (Henrich and Gil-White, 2001).

Female-female competition is common in mammals and especially among primates but does not reach the same intensity as that found in same-species males (Smuts, 1987). Competition is typically over access to high-quality foods rather than mates, and females or female kin-groups that achieve access to these resources have a higher reproductive success than their less competitive peers (Silk, 1993). Competition among women for social influence and access to resources–often competition among cowives in polygynous marriages–is well documented in the cross-cultural record (Stockley and Campbell, 2013). Rather than being physical, the competition typically involves relational aggression that includes attempts to sully the reputation and undermine the social-support network of competitors. Although it is not as well documented as for men, in many contexts socially dominant women often have healthier and more surviving children than do submissive ones (Jankowiak et al., 2005; Ji et al., 2013). Female and male choice are also evident in humans but beyond the scope of this article (see Geary, 2021).

The point is there is ample evidence that at least some currently observed sex differences are the result of sexual selection during human evolution. The goal here is to illustrate that an understanding of sexually selected traits as condition-dependent has the potential to expand our understanding of human sex differences, especially in terms of social and ecological factors that can influence the magnitude of these differences. An important prediction is that the magnitude of human sex differences, as in other species, will be largest for populations living in favorable conditions and smaller for populations living in more stressful conditions. This is not, however, a blanket statement about all sex differences, but rather those that have a clear evolutionary history related to competition for mates and other resources and related to mate choices. I provided a priori predictions regarding which traits will be most vulnerable to stressors in an earlier book (Geary, 2015, see Table 5.1, pp. 156–159) and more precisely define stressors elsewhere (Geary, 2017, 2019). The goal here is to illustrate the potential utility of this approach for understanding human sex differences and fluctuations in the magnitude of these differences, but before providing these illustrations I discussion some limitations to the approach.

As noted, the use of condition-dependent trait expression as a principled means to understand fluctuations in the magnitude of human sex differences does not directly apply to all of these differences. One important class of exceptions are evolutionarily novel academic abilities, such as reading, writing, and mathematics, that only emerge with the massive cultural intervention of formal schooling (Geary, 1995, 2007). Although there may be indirect relations between sex differences in sexually selected traits and academic abilities (e.g., language as related to reading acquisition), sex differences for the latter are not expected to be as strongly influenced by stressor exposure as sexually selected traits. This is because the development of academic abilities is heavily dependent on exposure in school, instructional approaches, and other experiences that will more strongly influence the development of these abilities, and any sex differences in them, than will any indirect influence of evolved abilities. Thus, the pattern of larger sex differences under favorable conditions might not manifest in these domains (e.g., Guiso et al., 2008).

There are in addition social dynamics and constraints that can influence the magnitude of some behavioral and psychological sex differences that are independent of or interact with condition-dependent trait expression (Kaiser, 2019; Salk et al., 2017; Schmitt, 2005, 2015). As an example, religious prohibitions and proscriptions are associated with increases in social cooperation and decreases in self-serving behaviors that in turn decrease the magnitude of the sex differences in the personality trait of agreeableness (favoring women) and in use of Machiavellian social strategies (favoring men). This is because these prohibitions and proscriptions influence men’s behaviors more strongly than those of women (Schmitt, 2015). These types of changes are sometimes correlated with changes in ecological conditions that could influence the expression of condition-dependent traits in ways similar to that found in other species (Kaiser, 2019), but this need not be the case.

The next section describes how improvements in living conditions resulted in an increase in many physical sex differences that can be linked to an evolutionary history of male-male competition and thus in line with condition-dependent trait expression in other species. These same changes are also associated with some evidence for a widening of the sex differences (favoring women) in anxiety and perhaps depression (Högberg et al., 2020; Salk et al., 2017; Schmitt, 2015; Thorisdottir et al., 2017). Women’s anxiety and depression are not condition-dependent traits in the same way as men’s physical competencies, although they may be condition-dependent social signals associated with men’s resilience in the face of physical male-male competition (see Geary, 2015, pp. 224–226). In other words, men have a bias to suppress behavioral indicators of anxiety and depression, because these will undermine their status in male dominance hierarchies; there are also hormonal mechanisms that reduce men’s reactivity to threat in competitive contexts (Stanton et al., 2009).

This perspective, however, is not a satisfactory explanation of why women’s rates of anxiety and depression remain elevated in the modern, low-risk world or, more critically, why they may have increased over the past several generations (e.g., Högberg et al., 2020; Thorisdottir et al., 2017), despite reductions in risk (e.g., crime) and improvements in overall living conditions (Pinker, 2011). Del Giudice (2018) proposed that anxiety and depression are evolved psychological defense mechanisms that reduce engagement in risky behaviors and withdrawal from risky social dynamics and are more easily triggered in women than in men. The benefits include reduced injury and death due to accidents and homicides at the extreme end, and lower social risks overall (Wilson and Daly, 1985). Sex differences would be expected to remain in low-risk contexts, but this leaves unanswered the recent secular increase in the magnitude of these differences. One possibility is concept creep, whereby reductions in serious threats (e.g., physical assault) are associated with increased sensitivity to what were once considered minor threats (e.g., unpleasant verbal statements; Levari et al., 2018). The latter are more common in the context of normal social dynamics and thus increased sensitivity to them could result in an increased triggering of the psychological defense mechanisms of anxiety and depression. The triggering would occur more often in women than in men and result in a corresponding increase in the sex differences in these areas.

Whatever is contributing to the sex differences in anxiety and depression, it is clear that the magnitude of many such differences can vary across contexts and time (for addition examples, Geary, 2021). It is also clear that such fluctuations are not necessarily evidence against sex differences in evolved biases, as is often assumed (e.g., Costa et al., 2001; Wood and Eagly, 2002), but rather reflect context-dependent plasticity in their expression (Geary, 2015; Kaiser, 2019; Schmitt, 2015). Much remains to be learned about contextual influences on the expression of evolved biases and any associated sex differences. The study of these interactions in the context of what is known about condition-dependent traits in nonhuman species provides a principled means to study some of these interactions, as illustrated in the following sections, but will not be sufficient to explain all of them.

The evolutionary echo of male-male competition includes sex differences in height, skeletal structure of the upper body, lean muscle mass, and cardiovascular fitness, among other traits (Tanner, 1990). Most of these differences are small to moderate during childhood and become quite large during pubertal development. If these traits follow the pattern illustrated in Fig. 1, then exposure to stressors and especially during puberty should more severely compromise them in boys than in girls. In other words, the magnitude of the male advantage in these areas should be largest in well-nourished populations with access to modern health care and sheltered from intense (i.e., life threatening) social competition and smallest in stressed populations, and this in fact the case.

As one example, men’s relative advantage in height is universal but the magnitude of this sex difference varies across populations and generations within populations, with the largest differences in the healthiest nations (Perkins et al., 2016). For the latter, adult stature is highly heritable but chronic poor nutrition and disease are important environmental contributors to adult height in suboptimal conditions and more so for men than women (Perkins et al., 2016; Zemel et al., 2007). Secular increases in height during the 20th century, a period of marked by substantive gains in overall health, illustrate the point (e.g., Kuh et al., 1991; Papadimitriou et al., 2002). From 1900–1958 in Great Britain, Kuh et al. found a 1.09 cm/decade increase in men’s height as compared to a 0.36 cm/decade increase for women. In 1900, the average British man was 11 cm taller than the average woman (d = 1.4), but this increased to 15 cm (d = 2.2) by 1958, a 36 % increase in less than three generations. The same pattern of a fluctuating sex difference is found in developing nations today. For young adults in nutritionally stressed regions of Nigeria, for instance, men are 7.5 cm shorter than their better-nourished peers, whereas women are 3.2 cm shorter (Omigbodun et al., 2010). The result is a sex difference in height that is 38 % smaller than it would be if these adults had received better nutritional and medical care during childhood and adolescence.

In a review of the social play of mammals, Power (2000) found that young males of polygynous species with intense physical male-male competition nearly always engaged in more play fighting than females. This form of play likely results in improved social competencies and later social-competitive advantage (Graham and Burghardt, 2010), as well as establishing dominance relationships before play merges into potentially harmful fighting (Pellis and Pellis, 2007). The same is true with children. Boys engage in various forms of rough-and-tumble and competitive group play at least three times more frequently than girls (DiPietro, 1981; Lever, 1978), and by adolescence this form of play merges into physical intimidation and aggression and influences status among their peers (Pellegrini and Bartini, 2001). Boys who do not engage in these forms of play are often bullied and at risk for anxiety and depression (Fagot, 1977).

As with height, stressor exposure more strongly disrupts boys’ sex-typical play and these aspects of social behavior than that of girls. These stressors include prenatal exposure to man-made toxins (Swan et al., 2010), as well as chronic malnutrition during childhood (Barrett et al., 1982). Barrett and colleagues provided a unique and semi-natural assessment of the social play of 6- to 8-year-old Guatemalan children. These were children from a larger study of the benefits of prenatal and early postnatal (up to four years of age) nutritional supplements on physical growth and cognitive development. Girls and boys with higher levels of nutritional supplements were more active and socially engaged than their poorly-nourished peers. As shown in Fig. 3, five hours of observation of natural behavior revealed that the better-nourished boys’ social potency and thus dominance was consistently higher than that of poorly nourished boys, but there were few differences in the social potency of better- and poorly-nourished girls. During participation in a competitive game, poorly-nourished boys were the least engaged and well-nourished boys were the most engaged and competitive (Barrett and Radke-Yarrow, 1985). The engagement and competitiveness of the poorly- and better-nourished girls was in-between that of the two boys’ groups. In other words, the most active and socially potent children were well-nourished boys and the least potent were malnourished boys, with girls somewhere in between the boys’ groups independent of their nutritional status.

Male-male competition in traditional contexts often involves long-distance travel to raid competing groups or to hunt, as well as the use of projectile weapons (MacDonald and Hewlett, 1999). These activities are associated with male advantages in various areas of visuospatial cognition, including more accurate navigation, an enhanced ability to identify targets in large-scale 3-dimentional space, and more accurate tracking of the movement of objects as they travel through space. The sex differences in these areas are well documented and range from small (d = 0.2) to large (ds > 0.7; Peters, 1997; Peters et al., 1995; Voyer et al., 1995).

An example of a male-specific vulnerability in this area is provided by the accidental exposure of thousands of people in Taiwan to PCB-contaminated cooking oil, including 74 women who were pregnant at the time or became pregnant soon thereafter. A longitudinal assessment of these children from 6- to 9-years of age, inclusive, and relative to a group of demographically matched peers revealed that exposed boys’ but not girls’ spatial reasoning abilities were compromised (Guo et al., 1995), as shown in Fig. 4. One result was that healthy boys’ among the youngest group had a small spatial reasoning advantage over healthy girls (d = .09), but girls had an advantage among the exposed children (d = −.19). Among the oldest group, healthy boys had a moderate advantage over healthy girls (d = .57), but boys’ advantage was 70 % smaller among the exposed children (d = .17).

There are also indications that exposure to one or more toxins can compromise men’s visuospatial memory and performance on more complex spatial cognition tests (Farahat et al., 2003; Schwartz et al., 2000). Akila et al. (1999) assessed Finnish factory workers’ level of aluminum exposure (through welding) and performance on a variety of cognitive measures. With control of demographic and other factors, the primary deficits associated with men’s exposure were “in tasks requiring working memory, particularly that relating to processing of visuospatial information” (Akila et al., 1999, p. 632). The magnitude of the sex differences in spatial abilities also varies across nations (Lippa et al., 2010). As overall health improves, men’s advantages in visuospatial abilities increase modestly (r = .33) to substantially (r = .68), depending on the type of spatial competence assessed.

Female-female relational aggression may have contributed to girls’ and women’s advantages in interpreting nonverbal communication cues and facial expressions, and for theory of mind, that is the ability to infer the thoughts and feelings of others. The sex differences in these areas range from modest (ds = 0.2) to substantial (ds > 1.0; e.g., Hall, 1984; Thompson and Voyer, 2014). Girls and women also have small to moderate (d = 0.1 to 0.4) advantages in many basic aspects of language (Leaper and Smith, 2004; Majeres, 2007). The subtlety of women’s relational aggression may be one reason they rehash social episodes with a best friend (Rose et al., 2014); to evaluate and decipher ambiguous messages. Rehashing in turn is dependent on a strong episodic memory (i.e., memory for personal experiences) and a strong verbatim recall of what was said (memory for language) and how it was said (e.g., memory for faces). Girls and women do indeed have advantages (d = 0.2 to 0.3) over boys and men in the later recall of social information (Herlitz et al., 1997; Pauls et al., 2013).

Stressor exposure can reduce or eliminate girls’ and women’s advantages for these cognitive traits in the same way that stressors reduce boys’ and men’s advantages for other traits. An example is provided by the social-cognitive deficits that are associated with the acute of phase of anorexia nervosa (AN), which involves severe calorie and nutrient restriction. Women who eventually develop AN tend to have social-cognitive deficits independent of weight loss and thus contrasts of women in acute and recovered stages of AN are important (Zucker et al., 2007). Women with bulimia nervosa (BN) have similar psychological issues but do not have the severe weight loss that is associated with AN and thus provide a useful contrast group. As shown in Fig. 5, acute AN is associated with substantial deficits in women’s ability to make inferences about the thoughts of other people (theory of mind) and their ability to infer others’ emotions using facial cues (Bora and Köse, 2016). Deficits in making inferences about the emotions of others conveyed through voice and body posture are also common in AN (Oldershaw et al., 2010).

As with men’s spatial ability, the magnitude of women’s advantage in verbal episodic memory increases with improvements in the social and economic conditions of the population, as shown in Fig. 6. Asperholm et al.’s (2019) meta-analysis indicted that women have the largest advantage in verbal memory in countries with higher levels of gender equality, better educational and employment opportunities and higher income, with this advantage disappearing or reversing in more stressful contexts.