This month’s interview by Anaïs Thijssen and Jana Hirzinger with Dr. Dirk Smit, Assistant Professor Electrophysiology and Genetics at the department of Psychiatry at Amsterdam UMC.

What drew you to work at the intersection of psychiatry, genetics, and the neurosciences?

That’s a good question because this is not what I was trained for. I was trained as a cognitive psychologist doing vision research, and my first EEG experience came from a thesis project combining EEG with cognitive tests. After that, I moved into psychophysics, which was very experimental. Later, I was a computer programmer, but I decided to go back to academia and do a PhD. It was a bit of a risk trying to get a PhD position at 35. I ended up finding a PhD opening in genetics, an area I had no formal training in, but it sounded fascinating. Combining EEG with genetics felt like a unique approach. It’s a happy marriage, looking at things from a different perspective, combining cognitive neuroscience, EEG and genetics. Nowadays, with large fMRI databases, more people are starting to see the benefits of combining these kinds of databases with genetics.

You mentioned that you’re interested in the noisy nature of human behaviour, and you’re exploring it through equally noisy brain measures. At first glance, that seems counterintuitive – trying to explain one type of noisy data with another. How do you approach that challenge, and where do you see the benefits?

In a way, we try to simplify everything by having this clear separation between noise and signal. Noise is what we don’t want, and we try to get rid of it. But geneticists and developmental psychologists are more interested in noise than we think. In the simplest form, we are interested in variation: individual differences, but also changes within a person at the root level. So when I talk about this kind of noise, I’m thinking more about variation. The brain is different. The dynamics of the brain are actually very noisy. EEG researchers usually try to cancel out noise by averaging across trials, but we look at noise to see how stable or unstable the brain is as a marker of having stable or unstable behaviour. This approach fits within the study of individual differences while also accepting that the brain and behaviour are quite dynamic. 

Follow-up question: How do you distinguish between true variability and variability inherent to the measurement?

I think the short answer is genetics. Part of the dynamics in the brain can be explained by genetics, and the good thing about genetics is that if you find these associations, they are very stable. At first glance, combining genetics and EEG might seem like a strange idea – genetics is mostly stable over the lifetime, while EEG signal changes every millisecond. But it turns out that certain aspects of the EEG signal are very highly genetic. So the trick to linking EEG and genetics is to find EEG parameters that are genetic and stable, although they reflect the highly dynamic organization of the brain.

What type of phenotypes can you extract from EEG to use in genetics?

One of the most important EEG features is oscillations, or the rhythms of the brain. These oscillations reflect the excitability and sensitivity of groups of neurons and enable communication between different brain regions. For example, beta activity, a specific frequency at about 20 Hertz, plays a crucial role in communication between the cortex and subcortical brain areas. In Parkinson’s disease, beta activity is elevated, which is directly linked to the rigidity symptoms. In our GWAS of brain oscillations, we found that healthy people with a higher genetic risk for epilepsy had increased brain oscillations—essentially, they had a more excitable brain. This aligns with neural models of excitation and inhibition, where heightened excitability leads to more pronounced oscillatory activity. This imbalance in excitation and inhibition could represent a shared mechanism underlying various brain disorders. EEG phenotypes like these allow us to investigate such mechanisms through genetics, opening research avenues to better understand a range of behavioural and neurological phenotypes.

Is the genetic architecture of EEG phenotypes different from behavioural traits?

It’s not as oligogenic as you would think, but the effect sizes of EEG phenotypes are bigger than those we normally see in complex behavioural traits. We find significant hits with sample sizes of 15,000, so there is loads and loads of signal. There are also fewer genes that explain more SNP heritability, which is now up to 30%. Some of the effects are quite explainable. We, for instance, find genes that are related to postsynaptic potential maintenance, which is exactly the signal that EEG picks up. At the moment, we can relate this genetic signal to neuronal traits like epilepsy, but the potential link with behavioural traits like ADHD or other psychiatric disorders is still hard to establish.

What is the biggest strength of EEG in uncovering the genetic basis of behaviours?

Well, there seems to be some truth to the endophenotype concept. When GWAS became common practice, we went straight from genes to complex behaviour, skipping everything that happens in between, even though we knew that there had to be something happening in the brain. Right now, we can find the link from genes to EEG and the link from EEG to neurological traits. But for psychiatric phenotypes, finding this link is a lot harder, probably because of the complexity of the phenotypes. All kinds of different pathways could lead to a complex behaviour like ADHD, meaning that it is very hard to find the aetiology of such a trait. Genetics can be a promising step in uncovering the basis of such complex traits. With complex modelling like genomic SEM, we could find genetically homogenous phenotypes and then potentially find more systematic differences in the brain. I truly believe there is promise in this approach.

You are involved in many different research consortia. What do you find the most rewarding, and what are the biggest challenges of being part of a consortium?

With ENIGMA EEG, the main hardship was a lack of funding. Now, with funding in place, the focus is on studying genetics and EEG, which is exciting. Another challenge is, of course, to get all the cohorts to do their work. The other consortium I’m a part of is the PGC, where we work on OCD genetics. That was actually a lot of fun because the OCD PGC group is not that big, and the cooperative nature of consortium work is really rewarding. This level of cooperation is probably the biggest strength of genetics research overall, and the same is true for ENIGMA.