Overall, there were significantly more aligned males than females in the sample (1,172 vs. 873), and unaligned females showed lower Youth Self-Report Gender Questionnaire scores than unaligned males — both patterns also observed in the full ABCD dataset (Potter et al., 2022, Developmental Cog Neuro)

Here is the frequency distribution of the (A) Youth Self-Report and (B) Parent-Report Gender Questionnaires from the Supplementary Materials.

This categorization was used for the SVM sex classification analyses. In contrast, the SVR analyses, which aimed to predict continuous sex/gender alignment scores, did not involve any categorization.

Just to clarify — I prefer “unaligned” since it simply describes a lack of alignment, without implying that something went wrong.

In my study, 46% of girls scored below the highest possible score on the Youth Self-Report Gender Questionnaire, placing them in the unaligned category.

I am grateful to everyone who contributed to this effort, for their hard work and the thoughtful discussions we shared throughout the process.
@roselynechauvin.bsky.social @gordonneuro.bsky.social @benjaminkay.bsky.social Babatunde Adeyemo Aristeidis Sotiras @ndosenbach.bsky.social

We show that rsFC captures the relationship between sex and gender more effectively that cortical thickness or cortical volume and challenge binary models of brain-sex differences, advocating for more inclusive and nuanced approaches to studying the relationship between gender and neurobiology.

Lastly, rsFC, cortical thickness, and cortical volume did not predict sex/gender alignment. While rsFC effectively captured the sex/gender relationship, the inability to predict sex/gender alignment suggests that gender is a complex construct harder to discern in brain function or structure.

One explanation is that males may face lower social tolerance for gender nonconformity, leading boys to feel pressured to answer gender-related questions less truthfully, resulting in high sex/gender alignment scores despite rsFC patterns less similar to the typical male profile.

In males, however, there was an unexpected negative correlation: Those with higher rsFC sex SVM scores showed lower sex/gender alignment.

In females, the extent to which their brain rsFC matched a female or male sex profile was positively associated with the degree of sex/gender alignment, with higher rsFC sex SVM scores corresponding to greater sex/gender alignment.

The rsFC sex classifier, trained solely on youth with sex/gender alignment, was significantly better at classifying individuals with sex/gender alignment than unalignment. This suggests that gender may have a greater influence on functional connectivity patterns than neuroanatomy.

Key brain networks involved in predicting sex include association networks (default mode, dorsal attention, and parietal memory) and visual networks (visual and medial visual).

We found that rsFC significantly outperforms cortical thickness and cortical volume in predicting sex. This suggests that even in a young cohort, resting neural activity patterns capture subtle, sex-specific functional dynamics that structural measures like cortical thickness/volume may not reveal.

We leveraged data from ~3,200 youth in the Adolescent Brain Cognitive Development Study (ABCD) in the USA and Support Vector Machine learning to predict both sex (assigned at birth) and sex/gender alignment (the congruence between sex and gender) from rsFC, cortical thickness, and cortical volume.

In this study we moved beyond previous work by evaluating whether rsFC, cortical thickness or cortical volume is more effective in predicting sex and gender, while also exploring the extent of their interrelationship within the brain of preadolescents.

Although many neurological and psychiatric conditions vary by sex in prevalence, onset, and symptomatology, possibly due to differences in brain structure or function, evidence on brain differences related to sex and gender remains inconclusive.