What extreme male brain theory claims about low empathising and high systemising
The extreme male brain (EMB) hypothesis, developed by Simon Baron-Cohen, posits that autism risk is mediated by high systemising combined with low empathising. Systemising represents interest in non-living systems like engineering, classification, pattern recognition, and rule-based processes. Empathising represents understanding and responding to others’ mental states and emotions.
Neurotypical females typically show higher empathising than neurotypical males, measured via the Empathy Quotient (EQ). Neurotypical males typically show higher systemising than neurotypical females, measured via the Systemising Quotient (SQ). The EMB hypothesis claims autism reflects extreme development of male-typical cognitive profiles — lower EQ scores combined with higher SQ scores compared to neurotypical populations.
The theory proposes higher fetal testosterone as primary biological mechanism. Males experience higher prenatal testosterone exposure, potentially explaining consistent male bias in autism diagnoses — approximately four males diagnosed for every one female overall.
A February 2026 meta-analysis from Simon Fraser University reviewed 34 studies comprising 1,234,560 participants (757,726 females, 476,834 males) to evaluate whether EMB theory applies equally to males and females. The analysis examined relationships between EQ, SQ, and Autism Quotient (AQ) scores among autistic and neurotypical individuals, comparing patterns across sexes.
Findings revealed the EMB hypothesis applies more strongly to females than to males. Autistic females show larger proportional differences in both empathising and systemising compared to neurotypical females than autistic males show compared to neurotypical males. This pattern supports hypotheses about female protective factors requiring greater phenotypic shifts to reach diagnostic thresholds.
Autistic females show larger deviations from neurotypical female patterns than males do
The differences in empathising and systemising between autistic and neurotypical individuals prove significantly larger in females than in males. For empathising, autistic females differed from neurotypical females by 29.8 percentage points on average, whilst autistic males differed from neurotypical males by 25.4 percentage points. For systemising, autistic females differed from neurotypical females by 9.8 percentage points, whilst autistic males differed from neurotypical males by 4.7 percentage points.
Statistical testing confirmed these sex differences in magnitude. Paired t-tests showed autistic-neurotypical group differences significantly larger in females than males for both empathising (p=0.018) and systemising (p=0.027). Effect sizes indicated moderate strength — Cohen’s d values of 0.28 for empathising and 0.29 for systemising.
This pattern means autistic females deviate further from typical female profiles than autistic males deviate from typical male profiles. To reach diagnostic thresholds, females require larger phenotypic shifts in both empathising and systemising traits.
The findings align with genetic evidence showing autistic females carry higher mutational burdens across autism-associated variant classes. Studies document greater genetic liability in autistic females, with family members displaying elevated genetic load and higher autism recurrence risk. If females require greater biological and genetic loading to manifest diagnosable autism, then relatives should carry more subclinical traits — exactly what research observes.
Sex differences in empathising and systemising also show pronounced attenuation among autistic individuals. Neurotypical males and females differ substantially in both traits — neurotypical females score higher on empathising, neurotypical males score higher on systemising. These typical sex differences largely disappear among autistic individuals.
Autistic males and females show highly similar empathising and systemising profiles. The difference in systemising between autistic males and autistic females reached only 0.14 Cohen’s d — very small effect size, statistically non-significant. Empathising showed larger sex difference among autistic individuals (0.30 Cohen’s d, p<0.001), but still substantially reduced compared to neurotypical populations.
This attenuation supports hypotheses that autism aetiology involves factors directly underlying sex differences, consistent with prenatal testosterone effects. Autistic females demonstrate more male-typical systemising values than neurotypical females, whilst autistic males show profiles similar to autistic females rather than to neurotypical males.
Empathising differences matter 3-5 times more than systemising for autism diagnosis
Proportional differences between autistic and neurotypical individuals proved substantially larger for empathising than for systemising. Among females, empathising showed 29.8 percentage point difference between autistic and neurotypical groups, whilst systemising showed only 9.8 percentage point difference. Among males, empathising showed 25.4 percentage point difference, systemising only 4.7 percentage points.
Statistical comparison confirmed these differences. Paired t-tests demonstrated empathising differences significantly larger than systemising differences for both females (t=5.49, p<0.001) and males (t=5.49, p<0.001). The magnitude difference ranged from threefold to fivefold — empathising contributing 3-5 times more to distinguishing autistic from neurotypical individuals than systemising.
These findings suggest empathising alterations represent more sensitive diagnostic marker than systemising changes, at least as measured by these particular questionnaires. To the extent unit changes in EQ and SQ can be considered equivalent in effects on autism risk, empathising may have larger overall effects than systemising in contributing to psychological differences characterising autism diagnoses.
Meta-regressions examining relationships between EQ, SQ, and AQ scores revealed significantly steeper slopes among autistic individuals compared to neurotypical individuals. For both autistic females and autistic males, autism quotient showed strong inverse relationship with empathising — lower empathising associated with higher autism quotient scores. The relationship proved significantly stronger in autistic groups than neurotypical groups (p<0.001 for interaction terms).
Systemising showed opposite pattern. Among autistic females and males, autism quotient increased significantly with higher systemising scores. But among neurotypical individuals, systemising showed no significant relationship with autism quotient. Again, slopes differed significantly between autistic and neurotypical groups (p<0.001).
These steeper slopes indicate associations between empathising-systemising and autistic traits strengthen considerably among individuals diagnosed with autism compared to general population. Changes in empathising and systemising produce larger effects on autism quotient scores in autistic individuals than in neurotypical individuals.
Inverse empathising-systemising relationship unique to autism suggests cognitive trade-off
Autistic and neurotypical groups differed fundamentally in how empathising and systemising relate to each other. Among neurotypical individuals, empathising and systemising showed significant positive correlation — people scoring higher on one trait tended to score higher on the other. Among autistic individuals, these traits showed inverse relationship — higher systemising associated with lower empathising, and vice versa.
Previous research reported comparable directional difference, finding negative EQ-SQ correlation among autistic individuals (r=−0.29) but weaker negative correlation among neurotypical individuals (r=−0.09). The current meta-analysis confirmed and strengthened this pattern across multiple studies.
Combining autistic and neurotypical groups under dimensional model produces U-shaped relationship for both males and females. The statistical significance of quadratic terms (p<0.001) confirms non-linear pattern, with overall R² values of 0.25 in males and 0.36 in females.
This inverse correlation among autistic individuals suggests cognitive trade-off between empathising and systemising. Trade-offs arise from forms of incompatibility between expression of two beneficial traits. Physical or physiological incompatibilities typically express or intensify when relevant resources are limiting or constrained.
Cognitive trade-offs may involve specialisations driven by partial incompatibilities between different mental abilities. In autism context, increased systemising might interfere with empathising by generating cognitive style centring on non-social, rule-based stimuli. Alternatively, systemising cognitive style may associate with hyper-focused non-social attention patterns.
Neuroimaging evidence suggests autistic individuals predominantly rely on analytical neural pathways during social tasks, whilst neurotypical individuals preferentially recruit social-affective circuits. Sensory hypersensitivities may also contribute — studies show Sensory Perception Quotient correlates strongly and positively with systemising (r=0.47), whilst correlating negatively with empathising (r=−0.15).
The inverse relationship raises questions about why such trade-off manifests only in some individuals. If empathising and systemising operate through incompatible cognitive mechanisms, what factors determine when this incompatibility expresses? Why do neurotypical individuals show positive correlation whilst autistic individuals show negative correlation?
The researchers note inverse causal relationships between empathising and systemising would directly promote the low-EQ, high-SQ combination typifying autism under the EMB model. Understanding where, when and how cognitive trade-offs involving empathising, systemising and autism develop may prove key to understanding psychological aspects of autism spectrum conditions.
The opposite EQ-SQ associations between autistic and neurotypical groups indicate systemising still provides distinct clinical information beyond empathising measures. Elevated systemising may reflect alternative cognitive routes or compensatory strategies not captured by empathising assessments alone. Different cognitive architecture rather than simple deficit.
Female protective effect requires larger phenotypic shifts to reach diagnostic thresholds
The finding that autistic females show larger deviations from neurotypical norms than autistic males directly supports female protective effect hypotheses. Several nonexclusive mechanisms may explain sex differences in autism diagnosis rates.
First, autistic females may present with same core traits but different manifestations, leading to underdiagnosis and misdiagnosis. Autistic males tend to exhibit more externalising behaviour problems — aggression, hyperactivity, overt repetitive behaviours and restricted interests. Autistic females tend to display more internalising behaviours — anxiety, depression, demand avoidance — alongside subtler repetitive behaviours, heightened attention to socially salient stimuli, and social-communication styles appearing more reciprocal or socially attuned.
Male-typical autism presentations more likely prompt diagnostic evaluations, whilst female-typical manifestations may be noticed only upon assessment of intellectual disability. Historically developed diagnostic criteria based on male presentations contribute to sex-associated diagnostic discrepancies. Females may have same likelihood of developing autism but aren’t identified to same degree.
Second, camouflaging represents concealing autistic traits through masking, compensating, or attempting to hide differences. Autistic females camouflage more frequently and intensely than autistic males, thought to arise partly from societal pressures and gendered norms emphasising empathy, social reciprocity, politeness, and emotional expressiveness in girls. These expectations elevate motivation to fit in and hide autistic traits.
Camouflaging reduces visibility of autistic traits whilst carrying significant psychological costs — burnout, exhaustion, depression, anxiety — potentially contributing to misdiagnosis. Females may have similar trait severities as males but ascertainment probability reduces because they better conceal autistic phenotypes.
Third, biological factors may make females more resistant to autism, requiring higher genetic, phenotypic and environmental loading for diagnostic-threshold trait levels to manifest. The meta-analysis findings support this hypothesis directly — autistic females show larger phenotypic shifts in empathising and systemising than autistic males, consistent with greater perturbation from typical phenotypes needed to reach diagnostic thresholds.
The apparently required upward shifts to male-typical systemising values for females to develop autism may promote male bias, given such shifts involve increases and enhancements of psychological traits. Increases occur less readily through genetic and environmental effects than decreases and dysregulation leading to deficits.
Effects from increases in prenatal and postnatal testosterone would be more deleterious to female than male fitness, given relatively high testosterone has negative effects on female development and reproduction but generally positive effects in males. Female protection may relate to this fundamental biological asymmetry in testosterone’s effects across sexes.
Citations
Szakal, C., & Crespi, B. (2026) — Does the Extreme Male Brain Hypothesis of Autism Apply More to Females Than Males? A Systematic and Meta-Analytic Approach
