Role of Sex or gender in biomedical studies

Effect of Sex

Researchers from Wellcome Trust Sanger Institute have found that sex differences could affect the results of a number of mouse model studies and haved implications for human biomedical research based upon mouse model work, Reuters reports.

The researchers examined phenotypic data from 14,250 wildtype and 40,192 mutant mice — from 2,186 single-gene knockout lines — to find that a number of key traits are influenced by sex. As they report in Nature Communications, the Sanger-led researchers found that sex affected some 56.6 percent of quantitative traits such as bone mass and cholesterol levels and nearly 10 percent of qualitative traits like head shape in the wildtype mice. Additionally, in the mutant mice, sex modified the effect of the mutation in 13.3 percent of qualitative traits and 17.7 percent of quantitative traits.

First author Natasha Karp from the Sanger tells New Scientist that she and her colleagues findings suggest that therapeutics developed using only males might not work as well in females, and that drugs that would work better in females might not have made it far in the drug development process if only males were used.

“Unless there’s a really good reason not to, we should be using both sexes in biomedical research,” Karp adds.

The US National Institutes of Health has required clinical trials to include women since 1994, and in 2014, required preclinical studies to report whether they were using male or female cells or animals.

The role of sex in biomedical studies has often been overlooked, despite evidence of sexually dimorphic effects in some biological studies. Here, we used high-throughput phenotype data from 14,250 wildtype and 40,192 mutant mice (representing 2,186 knockout lines), analysed for up to 234 traits, and found a large proportion of mammalian traits both in wildtype and mutants are influenced by sex. This result has implications for interpreting disease phenotypes in animal models and humans.


A systematic review of animal research studies identified a vast over-representation of experiments that exclusively evaluated males. Where two sexes were included, two-thirds of the time the results were not analysed by sex1,2. Furthermore, sex is often not adequately reported, despite the majority of common human diseases exhibiting some sex differences in prevalence, course and severity3. Fundamental differences exist between males and females that may influence the interpretation of traits and disease phenotypes4,5 and their treatment6. Some, however, have argued that considering both sexes lead to a waste of resources and underpowered experiments7, while others have questioned the value of preclinical research into sex differences8.

Here we quantify how often sex influences phenotype within a data set by analysing data from 14,250 wildtype animals and 40,192 mutant mice, from 2,186 single gene knockout lines, produced by the International Mouse Phenotyping Consortium (IMPC)9. The phenotyping performed by the IMPC explores a range of vertebrate biology, and aims to collect data from seven males and seven females from each mutant line with data from strain-matched controls accumulated over time. Data are collected at 10 phenotyping centres, providing a unique opportunity to explore the role of sex on a phenotype within an experimental data set, and the role of sex on a treatment effect, where the treatment is a gene disruption event, analogous to a Mendelian genetic disease. Our findings show that regardless of research field or biological system, consideration of sex is important in the design and analysis of animal studies. All data are freely available at


Sex as a biological variable within an experiment

We first assessed the contribution of sex using linear modelling to determine how often sex contributed to the variation in the phenotype in wildtype mice (control data) for an individual data set (a phenotypic test/trait at an individual phenotyping centre) (Supplementary Fig. 1a,b). Phenotypes were classified as either continuous, such as creatine kinase levels, or categorical, such as vibrissae shape. Because body size is dimorphic between male and female mice, and many continuous traits correlate with body weight, we included weight as a covariate in our analysis for continuous traits. Using this approach, our analysis revealed that 9.9% of data sets from categorical traits (54/545 data sets) were significantly influenced by sex at a 5% false discovery rate (FDR) (Fig. 1a). Many of these cases included phenotypes that would not a priori be assumed to be sexually dimorphic (SD). For example, abnormal corneal opacity occurred at a higher rate in female wildtype mice at most phenotyping centres. Looking at the SD rate by institute, we find that within categorical data the rate was relatively consistent (average percentage of traits which were SD 8.9% (s.d.=5.9)) (Fig. 1c).

Figure 1: Sex as a biological variable in control data.
Figure 1

The role of sex in explaining variation in phenotypes of wildtype mice as assessed using data from the IMPC. (a,b) The proportion of experiments where sex had a significant role in wildtype phenotype. (a) Categorical data sets (n=545). (b) Continuous data sets (n=903). (c,d) The distribution of classifications when analysed by institute (c: categorical data sets, d: continuous data sets). BCM: Baylor College of Medicine, HMGU: Helmholtz Zentrum Munich, ICS: Institut Clinique de la Souris, JAX: The Jackson Laboratory, Harwell: Medical Research Council Harwell, NING: Nanjing University, RBRC: RIKEN BioResource Centre, TCP: The Centre for Phenogenomics, UC Davis: University of California, Davis, and WTSI: Wellcome Trust Sanger Institute.

For continuous traits, a far higher proportion of data sets (56.6%, 511/903) exhibited sexual dimorphism at a 5% FDR (Fig. 1b). As expected, this proportion was higher when the absolute phenotypic differences were considered without taking body weight into account (73.3%, 662/903 data sets, Fig. 2a). With the continuous data set, the inter-institute SD rate was more variable (average percentage of traits which were SD 44% (s.d.=14)) (Fig. 1d). Variation in sensitivity is to be expected, arising from the observation that variance for a trait depends on the institute10 and the size of the control data set (Fig. 2b,d). Regardless of biological area studied, sex was found to have a role (Figs 2c and 3a,b) and where calls could be compared across institutes the effect of sex was in general reproducible, with only 8.7% of variables having opposing effects across the phenotyping centres (Fig. 3c,d). Variation in husbandry, diet and other environmental factors will contribute to this variability11. Previous manuscripts have conducted extensive analysis assessing consistency across institutes and found good agreement in findings10,12. It could be argued that the consistency is surprising; for example, considering how critical the microbiome is to phenotypic outcome13,14. While these studies are contained within facilities with high biosecurity, the microbiomes will differ from institute to institute. In fact, microbiomes will differ between individual litters depending on the maternal microbiome. This study comparing control data across many litters in effect accounts for this variation, which might go some way to explaining the consistency of the findings across sites.

Figure 2: Sex as a biological variable in wildtype phenotypic continuous data when exploring absolute difference in phenotypes.
Figure 2

Exploration of how often sex was significant at explaining variation at a 5% FDR in an individual experiment using IMPC wildtype data for continuous traits. The analysis assessed the role of sex in the trait of interest, at a centre level, as an absolute phenotype since weight was not included as a covariate. For all sections, green indicates the phenotype was greater in the female, magenta indicates the trait was greater in the males, white indicates missing data, and grey indicates there was no significant sex effect. (a) Pie chart showing the proportion of data sets where sex was a significant source of variation (n=903). (b) Comparison of the reproducibility of the sex differences in the traits monitored within the intra-peritoneal glucose tolerance test across ten phenotyping centres. (c) Bar graph showing the proportion of data sets where sex was a significant source of variation by procedure. CSD indicates combined SHIRPA and dysmorphology screen, DEXA: dual-energy X-ray absorptiometry, and PPI: acoustic startle and pre-pulse inhibition. (d) Comparison of the consistency of the role of sex in the traits monitored within the DEXA procedure across ten phenotyping centres. BCM: Baylor College of Medicine, HMGU: Helmholtz Zentrum Munich, ICS: Institut Clinique de la Souris, JAX: The Jackson Laboratory, Harwell: Medical Research Council Harwell, NING: Nanjing University, RBRC: RIKEN BioResource Centre, TCP: The Centre for Phenogenomics, UC Davis: University of California, Davis and WTSI: Wellcome Trust Sanger Institute.

Figure 3: Sex as a biological variable after accounting for the potential confounding effect of body weight.
Figure 3

Assessment of the role of sex within an experiment using IMPC wildtype data at a 5% FDR. The analysis assessed the role of sex in the trait of interest, at a centre level. For continuous traits, this was computed as a relative phenotype since weight was included as a covariate. For all sections, green indicates the phenotype was greater in the females, magenta indicates the trait was greater in the males, white indicates missing data and grey indicates there was no significant sex effect. (a) Bar graph showing the proportion of data sets where sex was a significant source of variation by procedure for continuous traits. CSD indicates combined SHIRPA and dysmorphology screen, DEXA: dual-energy X-ray absorptiometry and PPI: acoustic startle and pre-pulse inhibition. (b) Bar graph showing the role of sex by procedure for categorical traits where CSD indicates combined SHIRPA and dysmorphology screen. (c) Comparison of the reproducibility of the sex differences in the traits monitored within the DEXA procedure across ten phenotyping centres. (d) Comparison of the consistency of the role of sex in the traits monitored within the intra-peritoneal glucose tolerance test across ten phenotyping centres. BCM: Baylor College of Medicine, HMGU: Helmholtz Zentrum Munich, ICS: Institut Clinique de la Souris, JAX: The Jackson Laboratory, Harwell: Medical Research Council Harwell, NING: Nanjing University, RBRC: RIKEN BioResource Centre, TCP: The Centre for Phenogenomics, UC Davis: University of California, Davis and WTSI: Wellcome Trust Sanger Institute.

Sex as a modifier of a treatment effect

We next looked at the role of sex in influencing phenotypes in the context of gene ablation (Supplementary Fig. 1c–e). Bespoke statistical analyses, distinct from those implemented on the IMPC portal, were used to assess sexual dimorphism and control the false positive rate. For this analysis, we used data collected from 2,186 mutant mouse lines, first assessing whether genotype significantly influenced phenotype, and if significant whether the effect was modified by sex. Of the categorical phenotypes that showed a significant genotype effect (0.46% 1,220/266,952 data sets at 5% FDR), 13.3% (162/1,220) were classed as SD at a 20% FDR (Fig. 4a). Our previous investigations15 found it necessary to use a higher FDR for categorical traits because of the conservative nature of this statistical pipeline and multiple testing burden. For continuous traits, 7.2% (7,929/110,586 at 5% FDR) had a significant genotype effect, of which 17.7% (1,407/7,929 at 5% FDR) were classed as SD (Fig. 4b). Increasing the stringency of the continuous data analysis by decreasing the FDR to 1%, reduced the number of phenotype calls (3.4%; 3,719/110,586 data sets) but we still observed a high proportion (12.0%; 446/3,719) of sexual dimorphism (Fig. 5a). For continuous traits, phenotypes ascertained using mice phenotyped in multiple batches are more robust as data is collected across multiple litters and modelling of environmental variation is more reliable, thereby giving better control of the false positive rate16. Focusing only on multi-batch data sets, 8.9% (4,177/46,925) had a significant genotype effect of which 13.8% were classed as SD (Fig. 5b).

Figure 4: Role of sex as a modifier of the genotype effect.
Figure 4

The role of sex in explaining variation in phenotypes of knockout mice as assessed using data from the IMPC. (a) Classification for categorical data sets (n=1,220) with a genotype effect at 20% FDR. (b) Classification for continuous data sets (n=7,929) with a genotype effect at 5% FDR. The ‘Cannot classify’ effect arises when statistically an interaction is detected between sex and genotype but the model output is insufficient to specify where the interaction arises. The ‘Genotype effect with no sex effect’ effect is the classification when the null hypothesis of no genotype*sex interaction was not rejected; this may be because the effect is the same across sexes, or due to a lack of power to detect the differential effect across sexes. (c,d) Comparison of SD hit rate for each screen with more than 35 genotype significant hits (c) Categorical traits. (d) Continuous traits. IPGTT: intra-peritoneal glucose tolerance test; ECG: electrocardiogram; DEXA: dual-energy X-ray absorptiometry; CSD: combined SHIRPA and dysmorphology; PPI: acoustic startle and pre-pulse inhibition.

Figure 5: Role of sex as a modifier of the genotype effect with more stringent criteria.
Figure 5

Exploration of the role of sex in modifying the genotype effect in studies of continuous traits of knockout mice data from the IMPC. (a) Distribution of sex effect in the genotype significant data sets when using a 1% FDR. Overall, 110,586 data sets were tested and 3.4% (3,719) were significant for the stage 1 genotype effect. Of these, 12% (n=466) were classed as SD. (b) Distribution of sex effect in the genotype significant data sets when processing only multi-batch data sets at a 5% FDR. A total of 46,925 data sets were tested and 8.9% (3,719) were significant for the stage 1 genotype effect. Of these, 13.8% (n=575) were classed as SD.

The experimental design and the statistical analysis used here were formulated to control the type-one error (false positive) rate, at the expense of sensitivity15 (Figs 6 and 7). The fact that we detected a significant number of SD genotype–phenotype relationships, despite this limitation and relatively small sample size, suggests that other traits may display more subtle sexual dimorphism. The primary impact of sex as a modifier of genotype effect for continuous traits was that of ‘one sex only’ (12.8% for all continuous traits using a 5% FDR) where only males or females showed a statistically significant phenotype (Fig. 4b), as demonstrated by the Usp47tm1b/tm1b mouse; the mutant that showed the largest proportion of SD calls (Fig. 8 and Supplementary Data 1). Of the SD calls in the IMPC data set, 3.5% demonstrated a phenotype that was significant in both sexes but with opposing phenotypic changes (Fig. 4b); for example a significant increase in the males and a significant decrease in the females (Fig. 8b, total protein and red blood cell). In 0.8% of cases, we observed phenotypes that were significant in both sexes, when compared to controls, but the phenotype was more pronounced in one sex when compared to the other (Fig. 4b). With the goal of assessing the prevalence, a simple summary has been used; however sensitivity will vary by trait. For categorical screens the hit rate by screen averaged 12.5% (s.d.=3.2%, Supplementary Table 1), while for continuous data the average SD hit rate by screen was 12.6% (s.d.=8.3% Supplementary Table 2). Co-correlation of phenotypes is expected and future research will need to focus on cross variable identification of phenotypic abnormalities, but at present is beyond the scope of this manuscript.

Figure 6: Control of type-one errors for continuous traits in the stage 1 assessment of genotype and stage 2 assessment of genotype*sex interaction.
Figure 6

Resampling studies of wildtype data, to build data sets with ‘mock’ knockout animals under various phenotyping workflows, were used to assess the control of type-one errors for the statistical pipeline. (a) The stage 1 type-one error rate at a trait level at the 0.05 significance threshold as a function of workflow. (b) The stage 1 average type-one error rate for various significance thresholds and workflows. (c) The stage 2 type-one error rate at a trait level at the 0.05 significance threshold as a function of workflow. (d) The stage 2 average type-one error rate at a trait level for various significance thresholds and workflows. For ad the one-batch workflow is represented by either B1 or a black line, a two batch workflow is represented by B2 or a red line, a three batch workflow is represented by B3 or a green line, a multiple batch workflow is represented by MB or an orange line, a dashed line represents the ideal rate, and finally a random workflow is represented by R or a blue line.

Figure 7: Control of type-one errors for assessing a genotype by sex interaction in the presence of a genotype effect for continuous traits.
Figure 7

A simulation study, to assess the stage 2 type-one error rate, where a genotype effect affecting both sexes has been added to the knockout mice as signal (as a function of standard deviation for four levels: 0.5, 1, 1.5, 2). (a) The average stage 2 type-one error rate as a function of the significance threshold. Only one level can be seen as the lines overlay each over. (b) The stage 2 type-one error rate at the trait level for the 0.05 significance threshold as a function of the signal added. Shown is a boxplot giving a five point summary (minimum, first quantile, mean, third quantile and maximum).

Figure 8: Role of sex as a modifier of a genotype–phenotype relationship.
Figure 8

(a) Looking across IMPC knockout lines identified as having 10% or more traits with a significant genotype effect, the proportion of traits classed as SD varied. Data sets are classified as SD (purple) or genotype effect with no sex effect (green) where a genotype effect in a mutant line is observed. (b) For all traits identified as having a significant genotype effect for the Usp47tm1b(EUCOMM)Wtsi line (MGI:5605792), a comparison is presented of the standardized genotype effect with 95% confidence interval for each sex with no multiple comparisons correction. Standardization, to allow comparison across variables, was achieved by dividing the genotype estimate by the signal seen in the wildtype population. Shown in red are statistically significant estimates. RBC: red blood cells; BMC: bone mineral content; BMD: bone mineral density; WBC: white blood cells. (c,d) Raw data visualization for two traits in Usp47tm1b(EUCOMM)Wtsi mice identified as having a significant SD genotype effect as the effect was specific to male mice. HDL: high-density lipoprotein–cholesterol; KO: knockout, WT: wildtype. While the Usp47tm1b(EUCOMM)Wtsi line had a high proportion of SD traits, the standardized effect size (ES) change leading to a SD call observed for each trait was typically ±1 s.d. unit from the average ES. Globally for all SD calls, the ES was 0.28 (s.d.=0.38) while for Usp47tm1b(EUCOMM)Wtsi the ES was 0.18 (s.d.=0.14).

Our study focused exclusively on mutants of autosomal loci finding a high proportion associated with one or more SD calls (33.2% of genes studied: 725/2,186). This result is in keeping with the view that once the sex determination cascade is initiated, genes exhibiting SD effects can be located anywhere in the genome17. Moreover, it illustrates the pervasive nature of sexual dimorphism that impacts a wide range of loci and genetic systems.

Sexual dimorphism and gene function

We considered whether our findings relate to similar examples of SD in humans. Evidence for SD in humans has typically come from complex disease and trait studies where the numbers of tested subjects are amenable to statistical analysis18. However, meta-analysis studies have typically failed to replicate findings with only the association of angiotensin-converting enzyme gene (ACE) and hypertension in men being consistently replicated18. Within the IMPC portal, we do relate the knockout phenotypes to Mendelian disease data19 using resources such as Online Mendelian Inheritance in Man (OMIM)20 and Orphanet21where we are most likely to reproduce the phenotypes observed in these single gene diseases. However, these human resources do not consistently document SD in the signs and symptoms and it is unlikely the numbers of patients recorded for these rare diseases would make detection of significant SD possible.

To determine whether prevalent sex differences are the result of a common biological process, we performed a functional analysis of a set of 29 genes for which sex differences were detected on more than 4% of all measures. The statistical analysis, and subsequent call of SD, is at the level of an individual trait for a genotype. Therefore, classifying a gene as SD is somewhat arbitrary as it involves accounting for the number of traits having a genotypic effect, and the prevalence of SD within these. Despite this limitation, an evaluation of this set of 29 genes in comparison to all curated and experimentally derived functional annotation sets in GeneWeaver22 revealed statistically significant overlap (J=0.0385; P<1.12 × 10−7) to 25 genes associated with ‘absence of the oestrous cycle’ (MP:0009009), based on representation of Kiss1r and Postn on both gene lists. A further review of genes for which any significant sex*genotype interaction was detected revealed additional genes associated with MP:0009009, absence of the oestrous cycle. This additional set includes Fshr, Lhcgr, Cyp27b1, Fancl and Foxo3. This result suggests that constitutive perturbations of oestrous cyclicity, including developmental absence or loss of cyclicity in adulthood, may broadly influence sex differences. The gene products encoded by Fshr (follicle stimulating releasing hormone receptor) and Lhcgr (Luteinizing hormone/choriogonadotropin receptor) have well known effects on reproductive cycles, and behaviour due to their role in maintaining hormonal cycles in females. Cyp27b1 is a steroid synthesizing enzyme, which is primarily involved in vitamin D metabolism, known to influence many sex-specific phenomena in autoimmune and other diseases (for a recent example23). Fancl (Fanconi anema complementation group L) causes male and female infertility and gonadal hormone abnormalities in Zebrafish24 through developmental signalling mechanisms via aromatase conversion of androgen. Foxo3 is associated with ovarian pathology in humans25 and premature ovarian failure in mice26. Therefore, each of these gene perturbations has the capability of influencing hormonal effects on behaviour and physiology, though it remains to be evaluated whether the sex differences herein are stable throughout the hormonal cycle or result from interference in a sex-specific gonadal steroid-regulated process. An evaluation of other genes with high sex difference hit rates may reveal additional pervasive effects on reproductive traits. Other sex differences identified in the IMPC analysis may be the result of more specific effects of gene perturbation on a sex-specific process in males or females.


Many authors have raised the need to address the role of sex in basic biological research and recommendations have been made, but with limited progress to date. Bespoke analysis of IMPC data, revealed that 9.9% of qualitative and 56.6% of quantitative data sets were SD in wildtype mice. Furthermore, as a mediator of a mutant phenotype, sex modifies the genotype effect in 13.3% of qualitative data sets and up to 17.7% of quantitative data sets. Our findings are consistent with a recently published study examining the role of sex in human genetic variation that found the effect was of modest magnitude but across a broad spectrum of traits27. Further studies to understand the biological mechanism of the interactions reported herein are challenging because of the difficulty of designing experiments with sufficient sensitivity to consistently detect those interactions. However, our findings also span a broad phenotypic spectrum and indicate that regardless of research field or biological system, consideration of sex is important in the design and analysis of animal studies for studies where sex differences could occur, thus supporting the recent National Institute of Health mandate to consider sex as a biological variable28.


Methodology consideration

Bespoke methods were developed to assess for prevalence of sexual dimorphism and are independent of the methodologies implemented on the IMPC portal.

Ethical approval

Institutes that breed the mice and collect phenotyping data are guided by their own ethical review panels and licensing and accrediting bodies, reflecting the national legislation under which they operate. Details of their ethical review bodies and licences are provided in Supplementary Table 3. All efforts were made to minimize suffering by considerate housing and husbandry. All phenotyping procedures were examined for potential refinements that were disseminated throughout the consortium. Animal welfare was assessed routinely for all mice involved.

Mouse generation

Targeted ES cell clones obtained from the European Conditional Mouse Mutagenesis Program (EUCOMM) and Knockout Mouse Project (KOMP) resource29,30 were injected into BALB/cAnN, C57BL/6J, CD1 or C57BL/6N blastocysts for chimera generation. The resulting chimeras were mated to C57BL/6N mice, and the progeny were screened to confirm germline transmission. Following the recovery of germline-transmitting progeny, for the majority of lines, heterozygotes were intercrossed to generate homozygous mutants10. A few knockout lines were generated on other genetic backgrounds, as detailed in the data output and presented on the IMPC portal. For these lines, control data from the equivalent genetic background was collected. All lines are available from

Genotyping and allele quality control

The targeted alleles were validated by a combination of short-range PCR, qPCR and non-radioactive Southern blot, as described previously31,32.

Housing and husbandry

Housing and husbandry data was captured for each institute as described in Karp et al.33 and is available on the IMPC portal (

Phenotyping data collection

We have used data collected from high-throughput phenotyping, which is based on a pipeline concept where a mouse is characterized by a series of standardized and validated tests underpinned by standard operating procedures (SOPs). The phenotyping tests chosen cover a variety of disease-related and biological systems, including the metabolic, cardiovascular, bone, neurological and behavioural, sensory and haematological systems and clinical chemistry. The IMPRESS database (, defines all screens, the purpose of the screen, the experimental design, detailed procedural information, the data that is to be collected, age of the mice, significant metadata parameters, and data quality control (QC).

Experimental design

At each institute, phenotyping data from both sexes is collected at regular intervals on age-matched wildtype mice of equivalent genetic backgrounds. Cohorts of at least seven homozygote mice of each sex per pipeline were generated. If no homozygotes were obtained from 28 or more offspring of heterozygote intercrosses, the line was classified as non-viable. Similarly, if <13% of the pups resulting from intercrossing were homozygous, the line was classified as being subviable. In such circumstances, heterozygote mice were analysed in the phenotyping pipelines. The random allocation of mice to experimental group (wildtype versus knockout) was driven by Mendelian inheritance. The individual mouse was considered the experimental unit within the studies. Further detailed experimental design information (for example, exact definition of a control animal) for each phenotyping institute, or the blinding strategy implemented is captured with a standardized ontology as detailed in Karp et al.33 and is available from the IMPC portal (

As a high-throughput project, the target sample size of 14 animals (seven per sex) per knockout strain is relatively low. This number was arrived at after a community-wide debate to find the lowest sample size that would consume the least amount of resources while achieving the goal of detecting phenotypic abnormalities10. At times, viability issues or the difficulty in administering a test might further limit the number of animals. As such, whenever data are visualized, the number of animals phenotyped is listed. In a high-throughput environment, replication of individual lines is not cost effective. Instead, we are generating and characterizing a common set of six ‘reference’ knockout lines that will present a wide range of phenotypes based on previously published research.

Data QC

Pre-set reasons are established for QC failures (for example, insufficient sample) and detailed within IMPRESS to provided standardized options as agreed by area experts as to when data can be discarded. A second QC cycle occurs when data are uploaded from the institutes to the Data Coordination Centre using an internal QC web interface. Data can only be QC failed from the data set if clear technical reasons can be found for a measurement being an outlier. Reasons are provided and this is tracked within the database. QC is an ongoing process; therefore, changes in data composition can occur between different data set versions if an institute later identifies an issue with the data. Analysis within this manuscript used IMPC data set version 4.2, published 8th December 2015.

Wildtype data sets

Wildtype data sets were assembled for a trait by selecting wildtype mice that were collected at the same institute, on the same genetic background, the same pipeline and with the same metadata parameters (for example, instrument). The subsequent statistical pipeline required that data was available for both sexes and there were more than 100 data points per sex. The nearest body weight measure was associated with data provided it was within +/− 4 days of the collection of the trait of interest.

Wildtype-knockout data sets

Wildtype-knockout data sets were assembled by selecting data from wildtype mice to associate to the data from the knockout mice that were collected at the same institute, from the same genetic background, the same pipeline, and with the same metadata parameters (for example, instrument). The nearest body weight measure was associated with the data provided it was within +/− 4 days of the collection of the trait of interest. A data set was only assembled for a knockout line and trait if data was available on both sexes, there were greater than five readings for each sex for the knockout mice, and body weight data was available. The requirement of a minimum of five readings was to maintain sensitivity.

This process gave 110,586 wildtype-knockout data sets monitoring continuous traits from 10 phenotyping centres. For categorical traits, 266,952 wildtype-knockout data sets from 10 centres were returned. The raw data are available at the IMPC web portal and there is a page detailing the various methods by which data can be extracted from the portal ( For Fig. 8 of the manuscript, a gene set was determined by grouping data by phenotyping centre, pipeline, allele, background strain and zygosity. The number of mice that comprise the Usp47tm1b(EUCOMM)Wtsi(MGI:5605792) data set presented within Fig. 8 of the manuscript are shown in Supplementary Table 4.

Statistical analysis

The analysis methods used were developed specifically to answer the biological question of the prevalence of SD and therefore are distinct from the statistical output presented on the IMPC portal. For each statistical analysis a flow diagram summarizing the analysis pipelines is available in Supplementary Fig. 1.

For continuous variables, regression analysis is necessary to assess the effect after accounting for sources of variation such as batch. As such, the estimated effect observed in the regression model cannot always be seen when visualizing raw graphs (Supplementary Fig. 1f).

Sex as a biological variable for categorical wildtype data

For categorical traits, the data were recoded to 0 to represent ‘as expected’ phenotypes or 1 to represent ‘not as expected’ phenotypes. A Bias Reduction Logistic Regression34 was used to assess the impact of sex on the abnormality rate with a likelihood ratio test that compares a test model (Ysex) with a null model (Y1). The P values were adjusted for multiple testing using the Hochberg method to control the FDR to 5%. To assess the biological effect, the differences in two binomial proportions were calculated and the 95% confidence interval calculated utilizing the Newcombe’s method. The analyses assessing the role of sex on categorical data, assumes that batch to batch and litter variation is negligible as discussed in ref. 15.

Sex as a biological variable for continuous wildtype data

For continuous traits, to assess the role of sex after adjusting for potential body weight differences, a mixed model regression analysis was used with model optimization to select the covariance structure for the residual to the data with a likelihood ratio test that compares a test model (equation 1) with a null model (equation 2). The P values were adjusted for multiple testing using the Hochberg method to control the FDR to 5%.

The role of sex was also assessed as an absolute difference using a mixed model regression with a likelihood ratio test to compare a test model (YSex+(1|Batch)) with a null model (Y(1|Batch)). The analyses assume that batch is a source of variation that adds noise in an independent normally distributed fashion. When weight is included, the analysis assumes that there is a linear relationship between weight and the variation of interest. Analysis was restricted to data sets with more than 100 data points per sex, and thus would be a data set comprising multiple batches and therefore would be robust to the analysis15.

Using the output from the two pipelines assessing the role of sex as a source of variation, the reproducibility of the role of sex across institutes was assessed for a variable that had been measured at three or more institutes. As data sets have differing size, and sensitivity varied across institutes, discordant results were classed as those where the effect of sex was in opposing direction.

Sex as a modifier of genotype effect-in categorical data

For each trait of interest, the data were recoded to either 0 to represent ‘as expected’ phenotypes or 1 to represent ‘not as expected’ phenotypes. The statistical pipeline, comparing the abnormality rates in the knockout mice against the baseline population was optimized to maximize sensitivity whilst maintain control of the type-one errors. In summary, a two stage process was used where first the genotype role was assessed and, if statistically significant, then the genotype effect by sex was assessed. To reduce the multiple testing burden, potential filters were used to allow analysis of only data sets that have the potential for statistical significance to be queried. To assess potential for a genotype effect, the Mantel–Haenszel alpha star (the minimal attainable P value for a data set) was calculated. If a data set had potential (P<0.05), then the role of genotype was assessed using a one-sided Cochran–Mantel–Haenszel mid P value to compare the proportion of abnormalities events difference between the knockout and wildtype groups, stratified by sex. After multiple testing adjustments, using the Hochberg method to control the FDR to 5%, data sets were selected for stage 2 testing of an interaction. The interaction was assessed by comparing abnormality rates between the sexes of the knockout data only using a Bias Reduction Logistic Regression with a likelihood ratio test that compared a test model (Ysex) with a null model (Y1). Prior to the assessment, the potential was assessed using a LR_KO alpha star P value, defined as the most extreme P value possible arising as a function of the number of abnormal calls and number of readings within a data set was used as a filter to select data sets for statistical testing for stage 2 (P<0.05). The remaining P values were adjusted for multiple testing using the Hochberg method to control the FDR to 20%. Lines were selected on statistical significance. To assess the biological effect, the difference in two binomial proportions was calculated and the 95% confidence interval calculated utilizing the Newcombe’s method. For data sets with a significant effect at stage 1, but not stage 2, the change was classified as ‘genotype effect with no sex effect’ as there was evidence of a genotype effect but the genotype*sex interaction was not significant, whilst those which were significant at stage 2 the change was classified (‘Female greater’ or ‘Male greater’) by comparing the abnormality rates in the knockout mice by sex. The analyses assessing the impact of genotype ablation on categorical data, assumes that batch to batch and litter variation is negligible as discussed in ref. 15.

The C57BL/6NTac strain carries the Crb1Rd8 mutation35. This recessive single base pair mutation (Retinal degeneration 8) can lead to a mild form of retinal degeneration that affects vision. The onset of the phenotype appears to be between 2 and 6 weeks of age36. The IMPC consortium within the eye screen monitors abnormalities including various retina parameters. The statistical analysis compares the abnormality rate in the knockout to the wildtype within that institute to account for variation in the penetrance of the retinal degeneration in the baseline.

The SD hit rate comparison across screens excluded screens with <35 hits.

Sex as a modifier of the genotype effect in continuous data

A two stage pipeline was implemented; stage 1 assessed the role of genotype and stage 2 assessed whether sex interacted with genotype. The complexity of the model is limited by the low number of knockout mice used; as such key fixed effects have been selected and batch is treated as a random effect16,37. For stage 1, testing the role of genotype, a mixed model regression analysis was used with model optimization to selects the covariance structure for the residual. The genotype effect was assessed with a likelihood ratio test comparing a full model (equation 3) with a null model (equation 4). The resulting P values were adjusted for multiple testing using the Hochberg method to control the FDR to 5%. For stage 2, testing the role of sex, a mixed model regression analysis was used with model optimization to select a covariance structure for the residual. The interaction was assessed with a likelihood ratio test comparing a full model (equation 3) with a null model (equation 5). The resulting P values were adjusted for multiple testing using the Hochberg method to control the FDR to 5%.

A final model was fitted (equation 6) to estimate the genotype effect by sex which was used to classify the genotype effect. The estimated genotype effect for each sex, and associated standard error, was standardized by dividing the values by the average of the average wildtype male and female mice to allow comparison across traits. Data sets were also given a workflow classification depending on how the knockout data were collected. Multi-batch data sets were defined as those with four or more distinct batches consisting of three or more batches within one sex and two or more for the other sex. One-batch data sets were defined as those with knockout mice collected in one batch. All other workflows were classed as low-batch. When a genotype effect was detected for a data set, the effect was classified as described in Supplementary Fig. 1d. For example, if a data set was significant at stage 1 but not for stage 2, as there was evidence of a genotype effect but the genotype*sex interaction was not significant, the effect would be classified as ‘genotype effect with no sex effect’.

The calls were reviewed individually by biologists to validate the calls made by the computational pipeline. Where questions were raised on a computational call, if a statistical issue could be identified (for example, a continuous variable was bound and thus was not appropriate for a mixed model methodology) then all data sets for that variable were removed. See the list detailed in the available code38.

The analyses assume that batch is a source of variation that adds noise in an independent normally distributed fashion. When weight is included, the analysis assumes that there is a linear relationship between weight and the variation of interest and that the slope doesn’t not dependent on sex. To address the concern that an interaction between weight and sex could act as a confounder, the data was processed with a model with an additional weight*sex term. The results were equivalent to that seen without the inclusion of the term (data not shown, but available at ref. 38). To validate the analysis pipeline, the control of type-one errors was investigated by a series of resampling studies of wildtype data from Wellcome Trust Sanger Institute MouseGP pipeline under the null at both stage 1 and 2 as described in ref. 16. Wildtype data was taken from five procedures (clinical chemistry, dual-energy X-ray absorptiometry (DEXA), immunophenotyping, haematology and open field) giving 60 traits. The simulated wildtype-knockout data sets were then examined statistically to assess the type-one error rate control at stage 1 and stage 2.

The control of type-one errors for stage 2 was also assessed under the null for stage 2 in the presence of a genotype effect that affected both sexes equally. Simulated data was constructed based on the signal characteristics (mean, variance and sex effect) of five clinical chemistry traits to give 14 male and 14 female data points in 300 batches. Batch variation was simulated under the assumption it was normally distributed with mean zero and defined variance that was 25% of the estimated s.d. Body weight data was generated by random sampling from the average signal for a wildtype female mouse. Resampling studies mimicking a random workflow were run as described in ref. 16 to build wildtype-knockout data sets (iterations 2,000). Signal was added to the knockout mice as a proportion of standard deviation (0, 0.5, 1, 1.5, 2) to represent a main effect genotype effect. The resulting data set were then examined statistically to assess the type-one error rate at stage 2.

The SD hit rate comparison across screens excluded screens with <35 hits.

Enrichment analysis

A list of genes (GSID=GS248996) and their SD hit rate (per cent of measures showing a statistically significant sex*genotype interaction) were entered into the GeneWeaver database. Genes with a 4% or greater hit rate (GSIDS=GS248973) were stored in a gene set. A ‘search for similar gene sets’ was performed using Jaccard similarity of GS248973f against GeneWeaver’s database of >100,000 gene sets from multiple sources including gene expression studies, curated annotations and other genomic data resources22. A statistically similar gene set was compared to the larger set of all sex*genotype interactions (GS248996) using the Jaccard similarity analysis tool to find additional relevant genes.

Data availability

All data sets, scripts and output have been made available at and as a Zenodo repository at (ref. 38).

What are the biggest component of your metabolism?

Your metabolism refers to the millions of chemical processes that keep your body alive and functioning.

It is related to weight because it influences the amount of energy your body needs at any given point. Take in more energy than you need, and the excess will be stored as fat.

Nonetheless many people are quick to blame a “slow metabolism” for their weight gain, when in fact they need to make better food choices and exercise choices.

The biggest component of your metabolism – accounting for 50 to 80 per cent of the energy used each day – is your basal metabolic rate (BMR), which is the energy your body burns just to maintain functioning at rest.

(Other influences include how much physical activity you do, and the ‘thermic effect’ of the food you eat – that is energy you use to digest and absorb your food.)

While there are many pills, supplements and foods that claim to boost metabolism and burn fat, most of these claims are unproven, says Tim Crowe, associate professor in nutrition at Deakin University.

Even if they did work, they might come with unintended side effects, such as increasing your heart rate, he says.

Nonetheless, it can be helpful to know what factors do affect your metabolism, as some of them are within your control. And even knowing you have factors you cannot control may nonetheless be useful as it can motivate you to take extra care to compensate for the issue, perhaps by being more vigilant about your diet and exercise.

Here are 10 factors that affect BMR and metabolism:

1. Muscle mass – that is, the amount of muscle tissue on your body. Muscle requires more energy to function than fat. So the more muscle tissue you carry, the more energy your body needs just to exist. (While most forms of exercise will help boost muscle, resistance or strength training is most effective: for example lifting weights and exercises that work against the resistance of your body weight such as push-ups, squats and lunges.)

2. Age – As you get older, your metabolic rate generally slows. This is partly because of a loss of muscle tissue, and also because of hormonal and neurological changes. When babies and children go through periods of growth, their metabolism speeds up.

3. Body size – People with bigger bodies tend to have a larger BMR because they usually have larger internal organs and fluid volume to maintain. Taller people have a larger skin surface, which means their bodies may have to work harder to maintain a constant temperature.

4. Gender – As men are usually larger than women, they generally have faster metabolisms.

5. Genetics – This can also play a role in whether you have a slower or faster metabolism, and some genetic disorders can also affect your metabolism.

6. Physical activity – Regular exercise increases muscle mass and encourages your body to burn kilojoules at a faster rate, even when at rest.

7. Hormonal factors – Hormonal imbalances caused by certain conditions, including hypo- and hyperthyroidism, can affect your metabolism.

8. Environmental factors – The weather can also have an effect on your metabolism; if it is very cold or very hot, your body has to work harder to maintain its normal temperature and that increases the metabolic rate.

9. Drugs – Caffeine and nicotine can increase your metabolic rate, while medications including some antidepressants and anabolic steroids can contribute to weight gain regardless of what you eat.

10. Diet – Certain aspects of your diet can also affect metabolism. For instance if you don’t have enough iodine for optimal thyroid function, it can slow down your metabolism.

Sex differences in human lifespan and healthspan

Aging is characterized by decreasing physiological integration, reduced function, loss of resilience and increased risk of death. Paradoxically, although women live longer, they suffer greater morbidity particularly late in life.

These sex differences in human lifespan and healthspan are consistently observed in all countries and during every era for which reliable data exist.

While these differences are ubiquitous in humans, evidence of sex differences in longevity and health for other species is more equivocal. Among fruit flies, nematodes and mice, sex differences in lifespan vary depending on strain and treatment.

In this review, we focus on sex differences in age-related alterations in DNA damage and mutation rates, telomere attrition, epigenetics, and nuclear architecture.

We find that robust sex differences exist, for example the higher incidence of DNA damage in men compared to women, but sex differences are not often conserved between species.

For most mechanisms reviewed here, there are insufficient data to make a clear determination regarding the impact of sex, largely because sex differences have not been analyzed. Overall, our findings reveal an urgent need for well-designed studies that explicitly examine sex differences in molecular drivers of aging.

Mounting Challenges to Brain Sex Differences

Summary: A new meta analysis reveals no significant differences in the size of the amygdala between males and females.

Source: Rosalind Franklin University of Medicine and Science.

How different are men and women’s brains?

The latest evidence to address this controversy comes from a study at Rosalind Franklin University of Medicine and Science, where a meta-analysis of human amygdala volumes found no significant difference between the sexes. Meta-analysis is a statistical approach for combining the results of multiple studies, in this case dozens of brain MRI studies.

“Despite the common impression that men and women are profoundly different, large analyses of brain measures are finding far more similarity than difference,” said Lise Eliot, PhD, principal investigator and associate professor of neuroscience at RFU’s Chicago Medical School. “There is no categorically ‘male brain’ or ‘female brain,’ and much more overlap than difference between genders for nearly all brain measures.”

Image shows an MRI brain scan with the amygdala highlighted.

The olive-sized amygdala is a key brain structure involved in all types of emotion and in social behaviors such as aggression and sexual arousal. Animal studies and early MRI reports indicated that the amygdala is disproportionately larger in males’ brains. Such a size difference has been suggested to contribute to sex differences in emotionality and in the prevalence of disorders such as anxiety and depression.

Biologists use the term “sexually dimorphic” (literally, “two different forms”) to describe male-female differences. This new study shows that the term does not apply to human amygdala volume. It joins other recent research that challenges the concept of binary “male” and “female” human brains, and may have relevance to understanding disorders including depression, substance abuse, and gender dysphoria.

The report is co-authored by RFU medical students Dhruv Marwha and Meha Halari, who worked with Eliot to systematically identify all MRI studies of the human amygdala over the past 30 years. They found 58 published comparisons of amygdala volume in matched groups of healthy men and women (or boys and girls). Studies reporting raw amygdala volume show that the structure is indeed about 10 percent larger in male brains. However, this difference is comparable to males’ larger body size, including the 11-12 percent larger volume of males’ brains overall. Among studies that reported amygdala volumes corrected for overall brain size, the volume difference was negligible (

Image shows the location of the amygdala.

The paper appears in the journal NeuroImage.

In a similar 2015 meta-analysis, Dr. Eliot headed a Chicago Medical School team that debunked the widely-held belief that the brain’s hippocampus, which consolidates new memories, is larger in women than men.

“There are behavioral reasons to suspect a sex difference in the amygdala,” Dr. Eliot said. “Emotion, empathy, aggression, and sexual arousal all depend on it. Also, the evidence from animal studies suggesting a sex difference in amygdala volume is stronger than it is for the hippocampus. So this finding is more surprising than our hippocampal result and suggests that human brains are not as sexually dimorphic as rats.”

This study strengthens the case for gender similarity in the human brain and psychological abilities and has implications for efforts to understand the transgender brain.


Funding: The study was funded by the Fred B. Snite Foundation.

Source: Judith Masterson – Rosalind Franklin University of Medicine and Science
Image Source: images are credited to Rosalind Franklin University.
Original Research: Abstract for “Meta-analysis reveals a lack of sexual dimorphism in human amygdala volume” by Dhruv Marwha, Meha Halari, and Lise Eliot in NeuroImage. Published online December 9 2016 doi:10.1016/j.neuroimage.2016.12.021

Rosalind Franklin University of Medicine and Science “Mounting Challenges to Brain Sex Differences.” NeuroscienceNews. NeuroscienceNews, 17 January 2017.


Meta-analysis reveals a lack of sexual dimorphism in human amygdala volume

The amygdala plays a key role in many affective behaviors and psychiatric disorders that differ between men and women. To test whether human amygdala volume (AV) differs reliably between the sexes, we performed a systematic review and meta-analysis of AVs reported in MRI studies of age-matched healthy male and female groups. Using four search strategies, we identified 46 total studies (58 matched samples) from which we extracted effect sizes for the sex difference in AV. All data were converted to Hedges g values and pooled effect sizes were calculated using a random-effects model. Each dataset was further meta-regressed against study year and average participant age. We found that uncorrected amygdala volume is about 10% larger in males, with pooled sex difference effect sizes of g=0.581 for right amygdala (κ=28, n=2022), 0.666 for left amygdala (κ=28, n=2006), and 0.876 for bilateral amygdala (κ=16, n=1585) volumes (all p values < 0.001). However, this difference is comparable to the sex differences in intracranial volume (ICV; g=1.186, p<.001, 11.9% larger in males, κ=11) and total brain volume (TBV; g=1.278, p<0.001, 11.5% larger in males, κ=15) reported in subsets of the same studies, suggesting the sex difference in AV is a product of larger brain size in males. Among studies reporting AVs normalized for ICV or TBV, sex difference effect sizes were small and not statistically significant: g=0.171 for the right amygdala (p=0.206, κ=13, n=1560); 0.233 for the left amygdala (p=0.092, κ=12, n=1512); and 0.257 for bilateral volume (p=0.131, κ=5, n=1629). These values correspond to less than 0.1% larger corrected right AV and 2.5% larger corrected left AV in males compared to females. In summary, AV is not selectively enhanced in human males, as often claimed. Although we cannot rule out subtle male-female group differences, it is not accurate to refer to the human amygdala as “sexually dimorphic.”

“Meta-analysis reveals a lack of sexual dimorphism in human amygdala volume” by Dhruv Marwha, Meha Halari, and Lise Eliot in NeuroImage. Published online December 9 2016 doi:10.1016/j.neuroimage.2016.12.021

Neuroscience of gender differences by Wiki

Neuroscience of sex differences is the study of the characteristics of the brain that separate the male brain and the female brain. Psychological sex differences are thought by some to reflect the interaction of genes, hormones and social learning on brain development throughout the lifespan.

Some evidence from brain morphology and function studies indicates that male and female brains cannot always be assumed to be identical from either a structural or functional perspective, and some brain structures are sexually dimorphic.[1][2]

Experts note that neural sexual dimorphisms in humans exist only as averages, with overlapping variabilities;[3][4] that it is unknown to what extent each is influenced by genetics or environment, even in adulthood;[4][5][6] and that it is impossible to identify whether a given human brain is from an XX or an XY solely by examination of its anatomy.[4]


Ideas of differences in the male and female brain circulated during the time of ancient Greek philosophers around 850 B.C. Aristotle claimed that males did not “receive their soul” until 40 days post-gestation and females did not until 80 days. In 1854, Emil Huschke discovered that “the frontal lobe in the male is all of 1% larger than that of the female.”[7] As the 19th century progressed, scientists began researching sexual dimorphisms in the brain significantly more.[8] Until around 21 years ago, scientists knew of several structural sexual dimorphisms of the brain, but they did not think that gender had any impact on how the human brain performs daily tasks. Through fMRI and PET scan studies, a great deal of information regarding the differences between male and female brains and how much they differ in regards to both structure and function has been uncovered.[citation needed]

Evolutionary explanations

Sexual selection

It is thought that male and female differences in learning ability have contributed to sexual selection and mate preference throughout evolution. The hippocampus has even been found to exhibit seasonal activity in some mammals where it is active during breeding periods but inactive during hibernation; this is because spatial learning is more present during the breeding season.[9]

Females show enhanced information recall compared to males. This may be due to the fact that females have a more intricate evaluation of risk-scenario contemplation, based on an prefrontal cortical control of the amygdala. For example, the ability to recall information better than males most likely originated from sexual selective pressures on females during competition with other females in mate selection. Recognition of social cues was an advantageous characteristic because it ultimately maximized offspring and was therefore selected for during evolution.[1]

Oxytocin is a hormone that induces contraction of the uterus and lactation in mammals. It is also a characteristic hormone of nursing mothers. Studies have found that oxytocin improves spatial memory. Through activation of the MAP kinase pathway, oxytocin plays a role in the enhancement of long-term synaptic plasticity, which is a change in strength between two neurons over a synapse that lasts for minutes or longer, and long-term memory. This hormone may have helped mothers remember the location of distant food sources so they could better nurture their offspring.[1]

Male vs. female brain anatomy

Hemisphere differences

Inter- and intrahemispheric connectivities are different between male and female.

A popular theory regarding language functions suggests that women use both hemispheres more equally, whereas men are more strongly lateralized to the left hemisphere.[10] This theory found initial support in a high-profile study of 19 men and 19 women, which found stronger lateralization in men during one of the three language tasks assessed.[11] In 2008, some researchers concluded that further studies have failed to replicate this finding, and a meta-analysis of 29 studies comparing language lateralization in males and females found no overall difference.[12]However, in 2013, researchers at the Perelman School of Medicine at the University of Pennsylvania mapped notable differences in male and female neural wiring. The study used diffusion tensor imaging of 949 individuals aged 8–22 years, and concluded that in all supratentorial regions of the brain inter-hemispheric connectivity was greater in women’s and girls’ brains, whereas intra-hemispheric connectivity was greater in the brains of men and boys. The effect was reversed in cerebellar connections.[13] The detected differences in neural connectivity were negligible up to the age of 13, but became much more prominent in the 14 to 17-year-olds.[13] In terms of the potential effect on behaviour, the authors concluded, “Overall, the results suggest that male brains are structured to facilitate connectivity between perception and coordinated action, whereas female brains are designed to facilitate communication between analytical and intuitive processing modes”.[13]


image of Amygdala

The amygdala (red) in a human brain

According to some researchers,[14] “… the research on sex differences in the amygdala has produced conflicting results. Multiple studies report increased amygdala activity during the processing of affective scenes in men relative to women (Schienle et al., 2005; Goldstein et al., 2010), and meta-analysis supports this view, showing larger effect sizes in studies of affective processing including only men compared with those including only women (Sergerie et al., 2008). However several studies using similar stimuli have reported a larger amygdala response in women (Klein et al., 2003; McClure et al., 2004; Hofer et al., 2006; Domes et al., 2010), and others have reported no sex difference at all (Wrase et al., 2003; Caseras et al., 2007; Aleman and Swart, 2008). A possible explanation for these inconsistent results is that sex differences in amygdala response are valence-dependent. Furthermore, according to other researchers,[15] “Correlation analyses revealed that gray matter thickness in left ventromedial PFC was inversely correlated with task-related activation in the amygdala. These data add support to a general role of the ventromedial PFC in regulating activity of the amygdala.”

Research has been done on post-traumatic stress disorder (PTSD), an anxiety disorder found in both sexes, which is particularly common in war veterans, assault victims and women who have experienced abuse. Emotional memory encoding varies in the amygdala on the right and left and occurs equally for both genders: the right triggers unpleasant and fear-related memories, both declarative (conscious) and episodic (nonconcious).[16]

Amygdala volume correlates positively with fearfulness in girls but not in boys.[17]


Several studies have shown the hippocampi of men and women to differ anatomically, neurochemically, and also in degree of long-term potentiation. Such evidence indicates that sex should influence the role of the hippocampus in learning. One experiment examined the effects of stress on Pavlovian conditioning performance in both sexes and found that males’ performance under stress was enhanced while female performance was impaired. Activation of the hippocampus is more dominant on the left side of hippocampus in females, while it is more dominant on the right side in males. This in turn influences cognitive reasoning; women use more verbal strategies than men when performing a task that requires cognitive thinking.[18] The hippocampus’s relationship with other structures in the brain influences learning and has been found to be sexually dimorphic as well.[1]

Oestradiol has been found to influence hippocampal development. Studies have shown neurogenesis, or the formation of new neurons, to be higher in the male hippocampus than in that of the female. This may be due to the lower levels of estradiol in the male brain compared to the female brain. providing a more optimal environment for neurogenesis.[19]

Frontal lobe

The ventromedial prefrontal cortex (VMPC), plays a key role in social emotional processing. In accordance with the sexual dimorphism of the amygdala, the right VPMC is more dominant in an active limbic system for males while the left is more dominant in females. These differences carry out to a behavioral level. For example, Koscik et al. wrote:

“A man with a unilateral right VMPC lesion, who was well educated and had worked successfully as a minister, was entirely unable to return to any form of gainful employment after his brain damage. He requires supervision for daily tasks and demonstrates severe disturbances in behavior and emotional regulation, including impulsivity and poor judgment. By contrast, a man with a unilateral left VMPC lesion was able to return to his job at a grain elevator and remains successfully employed there. He is remarkably free of disturbances to his social life and emotional functioning”

Orbital prefrontal corte

Positron emission tomography studies have shown that men and women ranging from the ages of 19 to 32 years old metabolize glucose at significantly different rates in the orbital prefrontal cortex. Infant males who exhibited lesions on their orbital prefrontal cortex struggled with object reversal experiments, but females exhibiting such lesions did not have impaired performance in object reversal.[20]

Other regions and not region-specific

There are sex differences in locus coeruleus dendritic structure that allow for an increased reception and processing of limbic information in females compared to males.[17]

Aggressive and defiant behavior is also associated with decreased right anterior cingulate cortex (ACC) volume in boys.[17]

According to the neuroscience journal review series Progress in Brain Research, it has been found that males have larger and longer planum temporale and Sylvian fissure while females have significantly larger proportionate volumes to total brain volume in the superior temporal cortex, Broca’s area, the hippocampus and the caudate.[21] The midsagittal and fiber numbers in the anterior commissure that connect the temporal polesand mass intermedia that connects the thalami is also larger in females.[21]

The journal review also found that male also have larger brain volume which can partly be accounted big bigger male body size. Researchers also found greater cortical thickness, cortical complexity and cortical surface area after adjusting for brain volume.[21] Given that cortical complexity and cortical features are positively correlated with intelligence, researchers postulated that these differences might have evolved for females to compensate for smaller brain size and equalize overall cognitive abilities with males.[21]

White/grey matter

Global and regional grey matter (GM) differs in men and women. Women have larger left orbitofrontal GM volumes and overall cortical thickness than men.[22] Behavioral implications of the greater volume have not yet been discovered. Women have a higher percentage of GM, whereas men have a higher percentage of white matter (WM) and of CSF (cerebrospinal fluid). In men the percentage of GM was higher in the left hemisphere, the percentage of WM was symmetric, and the percentage of CSF was higher in the right. Women showed no asymmetries. Both GM and WM volumes correlated moderately with global, verbal, and spatial performance across groups. However, the regression of cognitive performance and WM volume was significantly steeper in women.[23]

In a 2013 meta-analysis, researchers found on average males had larger grey matter volume in bilateral amygdalae, hippocampi, anterior parahippocampal gyri, posterior cingulate gyri, precuneus, putamen and temporal poles, areas in the left posterior and anterior cingulate gyri, and areas in the cerebellum bilateral VIIb, VIIIa and Crus I lobes, left VI and right Crus II lobes.[24] On the other hand, females on average had larger grey matter volume at the right frontal pole, inferior and middle frontal gyri, pars triangularis, planum temporale/parietal operculum, anterior cingulate gyrus, insular cortex, and Heschl’s gyrus; bilateral thalami and precuneus; the left parahippocampal gyrus and lateral occipital cortex(superior division).[24] The meta-analysis found larger volumes in females were most pronounced in areas in the right hemisphere related to language in addition to several limbic structures such as the right insular cortex and anterior cingulate gyrus.[24]

Amber Ruigrok’s 2013 meta-analysis also found greater grey matter density in the average male left amygdala, hippocampus, insula, pallidum, putamen, claustrum and right cerebellum.[24] The meta-analysis also found greater grey matter density in the average female left frontal pole[24]

Brain networks

A 2014 meta-analysis by researcher Ashley C.Hill found that although men and women commonly used the same brain networks for working memory, specific regions were gender specific.[25] For example, both men and women’s active working memory networks composed of bilateral middle frontal gyri, left cingulate gyrus, right precuneus, left inferior and superior parietal lobes, right claustrum, and left middle temporal gyrus but women also tended have consistent activity in the limbic regions such as the anterior cingulate, bilateral amygdala and right hippocampus while men tended to have a distributed networks spread out among the cerebellum, portions of the superior parietal lobe, the left insula and bilateral thalamus.[25]

Brain differences between homo- and heterosexuals

Brain wiring comparisons of homosexuals and persons of the opposite sex show that homosexuals may be born with a predisposition to be homosexual. Research at the Stockholm Brain Institute in Sweden found that homosexual men and heterosexual women have similar brain characteristics. Specifically, these similarities are in the overall size of the brain and the activity of the amygdala. The same is for heterosexual men and homosexual women. Molecular biologist at the National Institutes of Health, Dean Hamer, says, “this is from a series of observations showing there’s a biological reason for sexual orientation”.[26]

Ivanka Savic – Berglund conducted a study in which MRIs were used to measure the volume and shapes of the brain. She also used PET scans to view blood flow to the amygdala. Savic – Berglund found that in homosexual men and heterosexual women, the blood flowed to areas involved in fear and anxiety, whereas in heterosexual men and homosexual women, it tended to flow to pockets linked to aggression. When looking at hemisphere differences, the right hemisphere was found to be slightly larger than the left in heterosexual men and homosexual women, whereas those of homosexual men and heterosexual women were more symmetrical.[27]

Research has indicated that the corpus callosum is larger in homosexual men than in heterosexual men. This is significant because the corpus callosum is a structure that is developed early. In the Journal Science Simon LeVay showed that the third interstitial nucleus of the hypothalamus has neurons that are packed more together in homosexual men than in heterosexual men.[28] Connections from the amygdala to other parts of the brain are similar between homosexuals and persons of the opposite gender as shown through PET and MRI scans. For example, in homosexual men and heterosexual women, there were more connections from the left amygdala. In homosexual women and heterosexual men, there were more connections from the right amygdala. LeVay’s results were not replicated in other studies. A 2001 study that attempted to replicate the findings concluded that “Although there was a trend for INAH3 to occupy a smaller volume in homosexual men than in heterosexual men, there was no difference in the number of neurons within the nucleus based on sexual orientation.”[29]

Neurochemical differences


Steroid hormones have several effects on brain development as well as maintenance of homeostasis throughout adulthood. One effect they exhibit is on the hypothalamus, where they increase synapse formation.[30] Estrogen receptors have been found in the hypothalamus, pituitary gland, hippocampus, and frontal cortex, indicating the estrogen plays a role in brain development. Gonadal hormone receptors have also been found in the basal forebrain nuclei.[31]

Estrogen and the female brain

Estradiol influences cognitive function, specifically by enhancing learning and memory in a dose-sensitive manner. Too much estrogen can have negative effects by weakening performance of learned tasks as well as hindering performance of memory tasks; this can result in females exhibiting poorer performance of such tasks when compared to males.[32]

It has been suggested that during development, estrogen can exhibit both feminizing and defeminizing effects on the human brain; high levels of estrogen induce male neural traits to develop while moderate levels induce female traits. In females, defeminizing effects are resisted because of the presence of α-fetoprotein (AFP), a carrier protein proposed to transport estrogen into brain cells, allowing the female brain to properly develop. The role of AFP is significant at crucial stages of development, however. Prenatally, AFP blocks estrogen. Postnatally, AFP decreases to ineffective levels; therefore, it is probable that estrogen exhibits its effects on female brain development postnatally.[33]

Ovariectomies, surgeries inducing menopause, or natural menopause cause fluctuating and decreased estrogen levels in women. This in turn can “attenuate the effects” of endogenous opioid peptides. Opioid peptides are known to play a role in emotion and motivation. β-endorphin (β-EP), an endogenous opioid peptide, content has been found to decrease (in varying amounts/brain region), post ovariectomy, in female rats within the hypothalamus, hippocampus, and pituitary gland. Such a change in β-EP levels could be the cause of mood swings, behavioral disturbances, and hot flashes in post menopausal women.[31]

Testosterone and the male brain

Testosterone has been found to play a big role during development but may have independent effects on sexually dimorphic brain regions in adulthood. Studies have shown that the medial amygdala of male hamsters exhibits lateralization and sexual dimorphism prior to puberty. Furthermore, organization of this structure during development is influenced by the presence of androgens and testosterone. This is evident when comparing medial amygdala volume of male and female rats, adult male brains have a medial amygdala of greater volume than do adult female brains which is partially due to androgen circulation.[34]

It also heavily influences male development; a study found that perinatal females introduced to elevated testosterone levels exhibited male behavior patterns. In the absence of testosterone, female behavior is retained.[30] Testosterone’s influence on the brain is caused by organizational developmental effects. It has been shown to influence proaptotic proteins so that they increase neuronal cell death in certain brain regions. Another way testosterone affects brain development is by aiding in the construction of the “limbic hypothalamic neural networks”.[30]

Similar to how estrogen enhances memory and learning in women, testosterone has been found to enhance memory recall in men. In a study testing a correlation between memory a recall and testosterone levels in men, “fMRI analysis revealed that higher testosterone levels were related to increased brain activation in the amygdala during encoding of neutral pictures”.[35]

Oxytocin and Vasopressin

Oxytocin is positively correlated with maternal behaviours, social recognition, social contact, sexual behaviour and pair bonding. Oxytocin appears at higher levels in women than in men.[36] Vasopressin on the other hand is more present in men and mediates sexual behavior, aggression and other social functions.[36][37]


Whole level 5-HT serotonin levels are higher in women versus men while men synthesize serotonin significantly faster than women. Healthy women also have higher 5-HT transport availability in the diencephalon and brainstem areas of the brain.[38] Dopamine function is also increased in women especially dopamine transporter which regulates the availability of receptors. Women before the onset of menopause synthesize higher levels of striatal presynaptic dopamine than age-matched men.[38] Other neurotransmitters like μ-opioids show significantly higher binding potential in the cerebellum, amygdala and the thalamus for women than it does so for men.[39] Women are also more dependent on norepinephrine in the formation of long term emotional memories than men are.[39]

Male vs. female brain functionality

Neural masculinization is a developmental process where different sex hormones assist in the expression of male behavior.[40]


image of stress regions in brain

Regions of the brain associated with stress and fear

Stress has been found to induce an increase in serotonin, norepinephrine, and dopamine levels within the basolateral amygdala of male rats, but not within that of female rats. Furthermore, object recognition is impaired in males as a result of short term stress exposure. Neurochemical levels in the brain can change under the influence of stress exposure, particularly in regions associated with spatial and non-spatial memory, such as the prefrontal cortex and the hippocampus. Dopamine metabolite levels decrease post stress in male rats’ brains, specifically within the CA1 region of the hippocampus.[41]

In female rats, both short term (1 hour) and long term (21 days) stress has been found to actually enhance spatial memory. Under stress, male rats exhibit deleterious effects on spatial memory, however female rats show a degree of resistance to this phenomenon. Stressed female rats’ norepinephrine (NE) levels go up by about 50% in their prefrontal cortex while that of male rats goes down 50%.[41]

Cognitive tasks

It was once thought that sex differences in cognitive task and problem solving did not occur until puberty. However, new evidence now suggests that cognitive and skill differences are present earlier in development. For example, researchers have found that three- and four-year-old boys were better at targeting and at mentally rotating figures within a clock face than girls of the same age were. Prepubescent girls, however, excelled at recalling lists of words. These sex differences in cognition correspond to patterns of ability rather than overall intelligence (although some researchers, such as Richard Lynn of the University of Ulster in Northern Ireland, have argued that there exists a small IQ difference favoring human males). Laboratory settings are used to systematically study the sexual dimorphism in problem solving task performed by adults.[42]

On average, males excel relative to females at certain spatial tasks. Specifically, males have an advantage in tests that require the mental rotationor manipulation of an object.[43] They tend to outperform females in mathematical reasoning and navigation. In a computer simulation of a maze task, males completed the task faster and with fewer errors than their female counterparts. Additionally, males have displayed higher accuracy in tests of targeted motor skills, such as guiding projectiles.[42] Males are also faster on reaction time and finger tapping tests.[44]

On average, females excel relative to males on tests that measure recollection. They have an advantage on processing speed involving letters,digits and rapid naming tasks.[44] Females tend to have better object location memory and verbal memory.[45] They also perform better at verbal learning.[46] Females have better performance at matching items and precision tasks, such as placing pegs into designated holes. In maze and path completion tasks, males learn the goal route in fewer trials than females, but females remember more of the landmarks presented. This shows that females use landmarks in everyday situations to orient themselves more than males. Females are better at remembering whether objects had switched places or not.[42]

Studies using the Iowa gambling task, or Iowa Card Task, have examined cognitive reasoning and decision-making in males and females. A study in which participants of various age groups who were asked to perform the Iowa Card Task produced data showing that males and females differ in their decision making processes on the neurological level. The study suggests that decision-making in females may be guided by avoidance of negativity while decision making in males is mainly guided by assessing the long term outcome of a situation. They also found that males outperformed females in the Iowa Card Task, but there was a negative correlation between elevated testosterone levels and performance in the card task which indicates gonadal hormones influence decision-making.[20]