Neuroanatomical research has long established that there are statistical differences at the macroscopic level in male and female brains. There are also functional differences in terms of memory and cognition. However, these differences are largely statistical, and exist in a mosaic of different traits. This means that, if we look at specific features (whether anatomical or functional) there are male-female differences, but individuals each have their own mosaic pattern of more male or more female characteristics. There is no one typical female or typical male brain.
One additional important caveat to this research is that it generally focuses on unambiguously cis male or cis female subjects. This is necessary to address the questions at hand, but is says nothing about people who do not fit into this paradigm. Studies are beginning to apply what we have learned about male and female brains to nonbinary individuals, but that is a separate line of research.
There is also a meta-question we can ask about this research – as our technology to image and analyze the brain improves, are male-female differences coming into sharper focus, or getting more blurry? If there are truly neurological differences based on biological sex, then as our technology improves these differences should not only hold up scientifically, we should get a more clear and accurate picture of them. That seems to be what is happening.
A recent study is a great example of where this research is heading. There are three features of interest to this study. The first is that it uses advanced MRI scanning technology to look at the microstructure of living brains. It relies on diffusion MRI scanning, which measures how quickly water molecules move through different tissues in various directions to differentiate different tissues and create detailed images. This technique reflects differences in the microstructure of tissues and is especially useful in imaging brain tissue, where the connections between brain cells create a very specific tissue microstructure. All those axons running in the same direction allow for directional water diffusion.
In fact this study mainly looks at the white matter part of the brain – the part where all the axons are running, making connections. Further, they placed the diffusion maps onto a standard brain template, so that the AI did not know how big the brain was or other gross details like cortical thickness. This way the AI algorithms could not use those details to determine biological sex, and had to rely only on the pattern of white matter connections.
The second advanced technology leveraged by this study is artificial intelligence (AI). I know there has been a lot of hype surrounding recent advances in AI, but don’t let the hype distract you. AI algorithms are powerful, and are increasingly a boon to many different types of research. There is a genuine AI revolution in research taking place, and this study is just one example.
The researchers used three main AI models, 2D convolutional neural networks, 3D convolutional neural networks and Vision Transformer, in order to analyze thousands of MRI scans from 471 male and 560 female healthy subjects (age range, 22–37 years). The AI models were first told which biological sex the scans were from and learned the diffusion patterns that were more likely to be associated with each. Then they were given unlabeled scans and asked to predict if they were from a male or female subject.
Across the different AI models, they were all able to accurately tell the difference between male and female brains between 92% and 98% of the time. That is remarkable accuracy. This is not the first study to use AI to examine biological sex differences in brain function, but so far it is the most accurate.
This latest crop of studies also differ from older studies (not using AI) in that they can actually tell individual male and female brains apart. Older studies mostly were able to determine statistical differences only. This is likely saying that men are taller than women on average, but knowing a person’s height does not tell you if they are biologically male or female. But AI can look at a suite of these statistical differences and learn to make accurate predictions about the sex of an individual.
This research, however, does not tell us what the ultimate causes of these differences are. Because brain development is strongly influenced by the environment and the experience and learning of an individual, even clear anatomical and functional differences can be due in part to socialization, for example. It’s also highly likely that hormonal differences strongly influence brain development. It’s difficult to determine the relative balance of biological vs environmental influences on brain structure, and this balance likely varies considerably among individuals.
Developing functional maps of male and female brains is incredibly useful for further research. It has long been established, for example, that many neurological and psychiatric conditions differ between males and females. Men are more likely to have autism while women are more likely to have depression, for example. If we have a detailed map of male-female brain differences, this can shed light on not only why these condition risks differ, but on the nature of these conditions themselves.
These same tools can also be used to tell the difference between a schizophrenic brain and a neurotypical brain. But of course, we first have to determine what a neurotypical brain looks like, and that means knowing what a typical male and typical female brain looks like. It would also be fascinating to use this technique to look at trans or non-binary individuals to help us understand this phenomenon better as well.
In other words – this research goes far beyond the basic question of male-female brain differences. It’s critical to neuroscience research in general, and is likely to inform our understanding of many neurological conditions.
