Looking at the dlPFC and Music Improvisation with Explosions and Swearing!

Looking at the dlPFC and Music Improvisation with Explosions and Swearing!



In case you enjoy reading about feuds with brain areas, here's a fun* story about a recent spat I had with the dorsolateral prefrontal cortex (dlPFC) . We’re writing a review paper on music improvisation (I study the neural correlates (brain areas involved in) music perception and creation) and this means it’s time for everyone’s favourite activity: an extensive review of the literature.

There are a couple of very, very cool studies on improvisation with classical and jazz pianists and beat boxers, and here’s where the dlPFC started to crop up. The dlPFC is here-ish:

[pictured: my “art”]

[pictured: my “art”]


Now, the dlPFC is an interesting brain area. It's relatively recently evolved (it's in humans and non-human primates) and only fully matures when we reach adulthood. The dlPFC is, essentially, the aged cheddar of the primate brain; slowly ripening away while the rest of your brain gets on with the business of thinking and doing. It's involved with cognitive tasks, which include such enjoyable activities as decision-making, planning, and executive functioning (making a plan, then waiting to implement it; imagining likely outcomes from your decisions, formulating contingencies, etc).

You know what else involves all of those things?

 <dramatic pause>


One way to think about music improvisation is as a loop where you come up with a plan, implement that plan, and rinse/repeat until the end of the piece. Each new iteration of the plan requires you to remember the old plan (with some exceptions), and must be flexible enough to survive errors, changes in direction, etc. Think about driving from A to B, but dealing with construction/traffic/Godzilla. You will get there, but your route might not look like the one you first came up with.

Back to my beef with the dlPFC. In a study by Bengtsson and colleagues [1], classically-trained pianists were given a simple but novel (aka unfamiliar) melody and told to improvise based on it while their brain activity was recorded using fMRI. They found activations in the dlPFC, which makes a certain amount of sense. You have an unfamiliar piece of music, and need to improvise without reproducing it, which means planning and making decisions. A later study by Limb and Braun [2] looked at the brain activity of jazz pianists improvising, but this time, the pianists had been given the music ahead of time and had memorized it before doing the improvisation. In science, this is called over-learning. In music, it's called tedium.

Before the beat drops, I need to talk a bit about brain activations and deactivations. Your brain has a baseline amount of activity when you're not doing anything (though daydreaming is brain activity, buy I'm oversimplifying because I eventually need to get back to work). When you do a task requiring your brain (which is to say, most things), areas involved in that task will activate, pulling in more oxygen-rich blood to power the bits of brain that make up those areas. Different parts of this process can be seen with different scanners, and most of the papers we’re looking at here use fMRI, which picks up the movement of oxygen-rich blood through the brain. It's also possible for brain areas to deactivate, or take up less energy than at rest. This happens when a brain area would get in the way and slow down a task. If you think about it like the climactic final dance scene in any teen movie ever made, the brain is the anti-Johnny Castle, overly fastidious and putting Baby(s) in the corner frequently. This makes things more efficient, but cuts down on large-scale dance numbers. (OR DOES IT?).

BEAT DROP TIME. In the Limb and Braun [2] study and in a more recent study by Liu and colleagues [3] on freestyle rappers, the dlPFC shows a pattern of deactivation during improvisation (if you find any of this interesting, check out these papers. They’re so good). So this, as you may realize, is incongruous. I like to respond at these times with subtlety and decorum:


Limb and Braun viewed this deactivation as evidence of a "flow state": a state at which you're coming up with plans and decisions, but need to act and react more quickly than the dlPFC might allow. Here, the dlPFC is the old ball and chain standing in the way of ART and ART does not care for such trivialities as additional cognitive monitoring!

Before we start the wailing/gnashing of teeth/rending of garments, there are two points that require consideration. The first is the nature of the task, and the second is the nature of the participants.

1)    The task. In the first study, participants were seeing the music they would need to improvise around for the first time. It might stand to reason that additional cognitive control would be required.

2)    The participants. Jazz musicians improvise much more than classical musicians. They are trained to improvise, and it's a common feature in jazz. Same for freestyle rappers (it's called FREEstyle for a very good reason). 

SO. Is it the task or the participants?

If it's the task, we would expect classical musicians' dlPFCs to deactivate if they were able to memorize the music ahead of time. If it's the participants, we wouldn't. Enter in a glorious paper from Donnay and colleagues [4]. Here, they studied jazz musicians trading fours - a pattern of back-and-forth playing where each musician improvises four bars at a time before handing it over to the other musician. In this study, the musicians were playing memorized melodies, BUT THE DLPFC WAS ACTIVE.


Again, the tasks were different. We have jazz musicians improvising based on memorized pieces, but with the addition of another person. So let's talk about predictability. The brain, for all its charm and goo, is a lazy bastard who wants to conserve energy. One way of doing this is phoning it in when we're involved in highly repetitive, predictable tasks. Over time, especially if these tasks are physical or sensory in nature (think about walking or tuning out the annoying sound of your aging office furnace), they become automatic in the sense that your higher order brain areas don’t need to pay attention to them and can wait for more interesting things to happen.

For professional musicians, a lot of making their instrument work becomes automatic after years of practice. This extends to extremely sophisticated behaviours like 'make something up based on a melody you memorized' and, maybe, the dlPFC can relax...but then you add another person, and suddenly, things get complicated again. Now, in addition to paying attention to your music, you also need to pay attention to the other person and THEIR music. What started off as a decision-making problem is now also a social problem ("many layers of problems and also pictures of brains" could be neuroscience's motto).

So what does the dlPFC do? It's tricky to say for sure. Based on decision-making literature, it seems like it's there to yea or nay behaviour that is complex and highly changeable...but there's one more interesting thing. In addition to participating in decision-making, the dlPFC is also connected to areas for movement and sensory input in the brain.

Tangent time**! Making music is a physical activity. You need to physically vocalize or play an instrument, and how this embodiment fits into the plan-execute-repeat loop is not amazingly understood***. Some interesting findings from Sammler and colleagues [5] show that pianists can detect errors in videos of other pianists' hands, indicating that the physical positioning of the hand contains information that trained pianists can perceive. It’s NOW time for a DUBIOUS THEORY. Given that music is a physical process, could the suspension of the dlPFC in some cases point to the brain giving the body more control? The short answer is maybe, but we need more studies.

So where does that leave us with the dlPFC? In the great tradition of science, I wrote the contents of this blog post into a paragraph for our review paper calling for more research with 100% less shouting and MARKEDLY fewer curse words. And that's how I got into a disagreement with the dlPFC that ultimately led to a deeper understanding and appreciation of its functionality, bringing us closer together than ever before (figuratively. It’s still at the front of my brain).

*exceptionally nerdy

**this whole thing is a tangent. I'm sorry.

***nothing is. I'm sorry.

[1] Bengtsson, S.L., Csíkszentmihályi, M., & Ullén, F. ( 2007). Cortical regions involved in the generation of musical structures during improvisation in pianists. Journal of Cognitive Neuroscience, 19(5), 830–842.

[2] Limb, C. J., & Braun, A. R. (2008). Neural substrates of spontaneous musical performance: an FMRI study of jazz improvisation. PLoS one, 3(2), e1679.

[3] Liu, S., Chow, H. M., Xu, Y., Erkkinen, M. G., Swett, K. E., Eagle, M. W., ... & Braun, A. R. (2012). Neural correlates of lyrical improvisation: an fMRI study of freestyle rap. Scientific reports, 2, 834.

[4] Donnay, G. F., Rankin, S. K., Lopez-Gonzalez, M., Jiradejvong, P., & Limb, C. J. (2014). Neural substrates of interactive musical improvisation: an FMRI study of ‘trading fours’ in jazz. PLoS one, 9(2), e88665.

[5] Sammler, D., Novembre, G., Koelsch, S., & Keller, P.E. (2013). Syntax in a pianist’s hand: ERP signatures of “embodied” syntax processing in music. Cortex, 49, 1325-1339.

McLab Icon_C8A698 mini.png