According to a study released in March by researchers at Frankfurt’s Max Planck Institute for Empirical Aesthetics, people engaged in making music are at a higher risk for mental illnesses such as depression and anxiety. Results of the study suggest there is an overlap between inherited genetic variants associated with a tendency to make music, and those that increase the risk for mental illness.

Laura Wesseldijk is an author of the study alongside Yi Lu, Robert Karlsson, Fredrik Ullén, and Miriam Mosing. She describes herself as a “former musician,” and took trombone lessons for ten years. “That was really uncool at 15, so I started to play electric guitar in a rock band too,” she told me when I reached her in London, where she is currently working remotely. Wesseldijk decided to pursue science instead of the trombone, and now conducts research for the Max Planck Institute in collaboration with the Psychiatric Institute of Amsterdam University. 

Laura Wesseldijk

VAN: We tend to assume that music is good for mental health, a fact also shown by a lot of studies that you cite in your papers. What kind of things do these studies say?

Laura Wesseldijk: I’m a musician as well, or I used to be. Nowadays I just play the drums, as an amateur. So I also expected that making music is good for your mental health, because this is what you often experience as a musician. It’s also what these studies report: There are a lot of studies that suggest that making music is good for well-being or social engagement, which [in turn] is good for mental health. And there is music therapy of course… So the baseline thought for me was: Music is good. At the same time, for ages, there have been a lot of suicides among musicians: Chris Cornell, who used to be my favorite rock artist, Keith Flint from The Prodigy, Kurt Cobain… but in other scenes (outside of the rock scene) there are also quite high rates of suicide, depression, anxiety, and substance abuse.

I was interested in what we’d see when we looked at the general population. There is quite a big difference between looking at a group of 30 people, half who play music and half who don’t, and then comparing it—like an interventional study looking at 10,000 individuals—to see what we see in the general population who have not been pushed into something.

How did you analyze the music engagement and the mental health of the general population in your studies?

In our first study from 2019, we used a big data set on Swedish twins [with information on around 10,500 people—Ed.]. In Sweden we can link the data of the twins—we had all this information on how much they play music, what age they started, how often they practice per week, what level they are at—to the national patient registry, so I knew if they ever got diagnosed with depression, anxiety, bipolar disorder, or schizophrenia. And I had self-reported data about depression, burnout, and schizotypal personality disorder. I didn’t see a [statistically] significant [correlation between music engagement and mental health] in the registry based outcomes. But there was a trend: There was more anxiety and more depression [among those who were engaged with music]. So I was surprised by that. And the self-reported symptoms were significant. The people who played music reported a higher rate of depressive symptoms. You can see associations in the general population.

Were you surprised by that?

It was unexpected. But I’m sure you are aware that correlation is not causation. So there can be a lot of reasons why we saw more depressive symptoms among the musicians. One of them we call familial factors, which includes genetics and the family environment. Genes are of influence for psychiatric disorders; making music is also partly heritable. But I have to emphasize that genes interact with the environment. Just because something is heritable doesn’t mean it is deterministic. Your environment can counteract it: It can break it or make it.

But I thought: It’s possible that genes partly overlap, meaning the ones that influence playing music also influence mental health problems. The same could be said for family environmental influences. For example: growing up in a family with a lot of instruments available might go together with talking a lot about mental health problems, being more open about it, more familiar with what it looks like… Things like this can influence the fact that there is more depression among musicians.

How did you find out that genetic and environmental factors didn’t explain the higher rates of depression among musicians?

We compensated for that by using twins. When we controlled for overlapping genes or overlapping family environmental influences, there was no more association. The trend of increased mental health problems disappeared.

Can you explain this controlling process?

Monozygotic twins are genetically identical and grow up in an identical family environment: same parents, same house, same neighborhood. So if we assume that making music causes depression, then we look at monozygotic twins that are discordant: One plays music, the other does not. 

As monozygotic twins are identical in DNA and share their family environment, studying identical twins excludes confounding, in case a genetic predisposition or shared environmental influence affects both outcome (depression) and exposure (music engagement). Therefore, if music engagement truly causes a higher risk for receiving a depression diagnosis, we would expect the twin that plays music to have a higher risk of depression than his or her twin that does not play music.

If the association is still there in identical twins, and we continue to see more depression among the music-playing twin of the pair, then this is in line with causality. But if the association disappears, if we see no differences between the ones that play music and the ones that don’t, we conclude the reason for the association must be genetics or family environmental influences that influence both depression and music engagement. We saw the latter: there were no differences between the twins playing or not playing, and therefore the association was likely due to familial factors. 

This study design has also been used to establish that smoking causes lung cancer. The tobacco industry claimed that people with a biological basis for smoking also had a biological basis for lung cancer and that it wasn’t due to the smoking. Then they saw that in discordant monozygotic twins of which one smoked, the one who smoked really had a higher risk for lung cancer, and they could not attribute it to “familial factors” anymore. 


And that’s what you tested in your second paper, right?

Yes. We couldn’t ignore that there were more mental health problems among people who play musical instruments and among people who reach higher levels of music engagement. I wanted to see if it was because they were engaged in music, or because they were born with a package: of the tendency to engage in music, and a higher risk for mental health problems.

How can you find factors in the DNA data that indicate that it’s more likely for a person to become engaged in music and/or to have mental health problems?

DNA contains four to five million single nucleotide polymorphisms (SNPs). I used existing genome-wide association studies (GWAS). A GWAS takes, say, one group of depressed people, one group of not depressed people and compares all those single nucleotide polymorphisms. Then it checks: are there genetic variants found more often in the group with depression than in the other group? In the results, they can show all the chromosomes and say, for example: “On chromosome 18, this number, this genetic variant occurs more often in the group with depression, and on chromosome 21 this one …” I used results from a major depression GWAS.

And because the Swedish twins were genotyped, I could calculate for an individual where they are on the distribution of genetic risk of mental health problems, because I measured all of their 4 to 5 million single nucleotide polymorphisms and compared it to the GWAS. So I use a person’s DNA to actually measure how many of these genetic variants they have that are a risk for developing depression. And then a score comes out for each person. For example, I could have a score of .80, you might have a score of .40 and my neighbor might have a score of .20, and that would mean that, among the three of us, I am at the highest risk of developing depression. So everybody has a personalized score of their genetic predisposition to depression.

The latest from VAN, delivered straight to your inbox

Success! You're on the list.

Has there ever been a GWAS for musicality or music engagement?

No. For a GWAS you need extremely large samples. Hundreds of thousands of people need to be genotyped, and then we would need information on [their engagement with] music. GWAS studies have been done for schizophrenia, depression and bipolar disorder, a lot of diseases. Music is not on the priority list, because it is not life threatening. 

But people are working on it. There is one GWAS on the ability to clap to a beat. So we used that one in the study: rhythmic ability and beat synchronization. It’s not musical engagement, but it shows the genetic variants that are associated with whether you can clap in time to a beat. As you might realize, this is not a perfect measure, but it’s the only thing we have, so we did actually use it in our study to see if the genetic variants for this music DNA predicted diagnosis of depression or anxiety—or if people who have a genetic predisposition for music also have a higher risk for receiving a mental health diagnosis. We found that they did receive the diagnosis of depression more often. 

Five years ago, a study showed that people with a genetic risk for bipolar and schizophrenia are more often engaged in creative professions and creativity, so I thought: It could be genetic pleiotropy or shared genetics for creativity, schizophrenia and bipolar disorder, and maybe also music and depression.

Did you get different results for professional musicians compared with amateurs? 

I compared professionals, amateurs and non-players. Professionals are not worse off than amateurs. It’s just the people who do play music vs. people who do not. First I thought: Being a musician is a tough job, and it’s the professionals who drive this association. But it was also there among amateurs.

Isn’t it possible that people who have depression have a tendency to play music?  

We cannot definitively exclude that this is a possibility. It’s still possible that there is a causal relationship due to environmental factors. But importantly, apart from maybe depression causing people to play music—I wouldn’t speculate about that—there are genetic variants that influence both depression and music engagement.

You found out that it doesn’t make any difference if people really develop depression or bipolar disorders, or whether they just have the predisposition: both correlate with music engagement. 

Yes, exactly. We looked at whether the genetic predisposition for depression predicted music engagement in the whole group. And then we thought, “What if we exclude the people that actually have depression? Does it still predict music engagement?” And it did. That indicates that it’s not driven by actually having to experience depressive symptoms, therefore the person starts playing an instrument. The genetic relationship is actually also present in people that do not experience depression.

What are possible criticisms of your method? 

It’s very important to keep in mind that the polygenic scores we’ve been using are very low in predictive value. These days, we know that depression or music engagement is not influenced by one gene—it is influenced by thousands of genes, all with very small effects.

To find all these genetic variants, we would have to do GWAS studies with millions of people. Obviously we don’t have them yet, because not everybody gives their genotype and does a music test or tells us about depression. Every year new GWAS studies come out with larger samples. But we are not there yet. If I calculate your polygenic score, it will have a very low predictive value, between five and ten percent.

What does that mean in practice? 

DNA representation • Photo: Andy Eick (CC BY 2.0)

It means it can only explain five to ten percent of actual depression outcomes. We’re only looking at genes here; there is still the whole environment. I might have a super high risk for depression, but if I grew up in a super comforting home maybe I’ll never develop depression. Or I might not have a risk for depression but I run into a traumatic event. 

So you have to keep in mind that this technique is limited. I can’t say who will eventually get depression. All I can show is that the genes that we know predispose to depression are more common among musicians. But we are talking about averages here. There are so many more factors that we need to be careful about drawing any conclusions about the effects.

Also, this is not applicable on an individual level. These things are interesting on a population level, when I’m looking at 10,000 people. But you can’t apply it to an individual. Everybody is like: “Oh, can I be tested?” Well, you can, but it won’t tell you a lot. If we test 2,000 people, we can give you probabilities.  

When I was a conservatory student, mental health was not much of a topic, but there is a lot more awareness of it now. Do your studies suggest therapies or treatments that might work well for musicians or music students?

Personally, I’m convinced that talking about it is helpful. But I can’t say anything about what therapy would be helpful, because I’m not a psychiatrist. So I wouldn’t presume to give any advice on therapeutic methods. But I’m sure awareness and openness is more helpful than ignoring [the issues].

Right now you are doing a study on the concept of flow while making music. What is that about?

This is actually mostly the work of Miriam Mosing. Flow is a state of ultimate concentration, and it has been associated with good mental health. We look at flow in general: both in work and in any leisure activity, not only in music making. We found that flow has a protective effect on mental health. We saw this in the sample with the monozygotic twins: the one who experiences more flow actually experiences less depression. That indicates that the reason is not overlapping genes, but that flow can be protective against mental health problems. So right now we are looking at whether therapy could tap into that: If you increase the prevalence of flow somebody experiences, might that help [fight] against depressive symptoms?

Socially, the image persists that great art requires crisis; that all real artists are a bit crazy or need to suffer for their art. Is it possible to reverse the conclusions of your study and say that a higher genetic risk for mental health problems makes it more likely that you become a great artist?

We do see that the predisposition more often leads to being engaged in music and also other art forms that I also looked at in my study: theater, dance, visual art, and writing. As was already shown, they have a higher genetic risk for schizophrenia, bipolar, and also for major depressive disorder. We didn’t see that for scientific achievements. In sports it was the opposite: If you’re at risk for mental health problems, you do sports less often. I didn’t want to interpret these results too much because it wasn’t what I was interested in. I just wanted to show with the same data that if we look at something different it is not like we see these effects everywhere. We see this for music engagement but not for sports.

For sure, you can also be a great artist without [being at risk for mental health problems]. It’s not a necessary evil. It’s also not like everybody with a major depressive disorder will become a great artist.

Are you planning any further research on this topic?

I would like to look more into listening to music. This study was about actively making music. But what about passive music engagement? Because there are so many people who do not necessarily play but are very much into music, and others who are just not into music. So I would like to do a follow-up study on passive music engagement. ¶

Subscribers keep VAN running!

VAN is proud to be an independent classical music magazine thanks to our subscribers. For just over 10 cents a day, you can enjoy unlimited access to over 800 articles in our archives—and get new ones delivered straight to your inbox each week.

Not ready to commit to a full year?
You can test-drive VAN for one month for the price of a coffee.