Wednesday, 7 December 2016

Brain Activity: Is Less More?

Areas of the brain that show a decrease in activity upon learning a task (from Chein and Schneider, 2005)

Imagine an experiment where scientists are using fMRI to look at activity in a person’s brain before and after the person learns how to do a particular task. (The task itself doesn’t matter too much right now, but it could be something like tapping a particular sequence of fingers, or reading backwards, or picking out pictures of dogs hidden in a complicated landscape.) The scientists scan the person’s brain while she is doing the task for the very first time. Then they have the person practice doing the task until she's really good at it. Finally, the person gets scanned again, while performing this well-learned task. What do you think the difference will be in the person’s brain activity? Would you expect to see more activity or less activity?

I’ve been slogging through scientific papers that look at changes in brain activity (using either fMRI or PET) when we learn skills, and growing increasingly puzzled. In some cases, when people learn to do something, there is more activity in their brains, and the researchers say: “See, that’s because they’re using more of their brains for this!” But sometimes there is less activity, and the researchers conclude that when we are good at something, our brains are more efficient.

Don’t those two things sound contradictory to you? Which is it? Do we use less brain power or more brain power when we’re good at doing a task? I’ve spent some time looking into this question, and when you get into the details, the answer is… it depends.

But it does make sense, trust me.

The best explanation that I found was in The Cambridge Handbook of Expertise and Expert Performance. There’s a chapter by Nicole Hill and Walter Schneider entitled “Brain Changes in the Development of Expertise: Neuroanatomical and Neurophysiological Evidence about Skill-Based Adaptations”. They suggest that when we learn a skill, there are a number of different patterns of changes in brain activity that are seen.

One of the most common patterns is a decrease in activity in parts of the brain that make up the control network. These are the parts of the brain responsible for working memory, attention, decision-making, and sequencing steps in a task. They’re active when we perform any task that isn’t well-learned, whether it’s a motor task, a perceptual task, or a reasoning task. When we gain experience with a task, we don’t need to devote as much concentration to it. We learn what steps follow which, and what is required to efficiently get the job done. Once we’re experienced at a task, the control network is not required to do as much work, so activity decreases in these areas of the brain.

A second pattern that is commonly seen is an increase in activity in parts of the brain specifically related to performing the task. For a motor task such as a sequence of finger taps, there is an increase in activity in the primary motor cortex of the brain (as shown by Avi Karni and colleagues in 1995). This is believed to be due to the recruitment of more neurons into the representation of the movement and supports the idea that networks of neurons in the primary motor cortex can code for sequences of movements. So when musicians are playing, there is a larger part of the primary motor cortex that is active, causing their hands to move in well-learned sequences.

A third pattern is known as functional reorganization, in which different areas of the brain are seen to be active when comparing novices vs. experts. For instance, in motor learning tasks such as learning a sequence of key-presses, when we initially are trying to learn the sequence, there is a lot of activity in the cerebellum, but once the sequence is well-learned, the cerebellum is much less active. Instead, there is an increase in activity in the striatum, a part of the brain believed to be responsible for (among other things) sequences of movements. Julien Doyon and colleagues, who reported this in 2002, conclude that the cerebellum has an important role in learning a motor task, but much less of a role in performing the task once it is well-learned.

All three patterns can be seen when learning different aspects of music, depending on what the learning task is. And sometimes all three patterns occur at the same time, so that we see a decrease in activity in control regions of the brain, an increase in regions specifically related to a task, and also some transfer of activity to regions that are not initially active. This is part of why it is so difficult to interpret data about activity in the brain. Understanding what each region of the brain does in relation to the task at hand allows us to tease apart the differences we see. And conversely, seeing how the activity changes in a particular area of the brain helps us understand how it contributes to learning and to performance of a skilled task.


Hill, N.M., and Schneider, W. (2006). Brain Changes in the Development of Expertise: Neuroanatomical and Neurophysiological Evidence about Skill-Based Adaptations. In The Cambridge Handbook of Expertise and Expert Performance, ed. Ericsson, K.A., Charness N., Feltovich, P.J., and Hoffman, R. R. (Cambridge University Press).

Monday, 14 November 2016

Multiple Rhythm Skills

Figure from Tierney & Kraus, 2015.

It’s a rainy Monday night and I’m giving Nicola her weekly lesson on the grand piano in the sanctuary of the church.  She’s a reticent girl with an engaging, shy smile, who practices consistently and plays well.  In particular, she has an impeccable sense of rhythm, always performing her pieces with accurate and consistent timing. 

Nicola is working towards her RCM Grade 4 exam, so we take some time in the lesson to practice for the ear-training requirements she will face.  She correctly names intervals and chord qualities as I play them, and flawlessly copies back a short melody at the keyboard.  But when I turn to the rhythm clap-backs, her face falls and she gets a worried wrinkle in her forehead.  We both know that this is her weak spot.  I count in and then play a three-bar melody for her, emphasizing the strong beats to help her maintain a sense of meter.  I encourage her to tap along silently as I play the tune a second time.  Then it’s her turn to clap the rhythm back to me.  She claps the first measure correctly, but falters in the second measure and just stops.  She clearly has no idea what comes next.

I never could quite make sense of why this is so difficult for her.  Her sense of rhythm is fantastic, but clap-backs stump her every time.  She just can’t keep the rhythm in her head long enough to clap it back.  So when I saw a research paper entitled “Evidence for Multiple Rhythmic Skills”, I immediately thought of Nicola and was drawn in.

The paper, published in PLOS ONE in 2015, comes from the lab of Nina Kraus, who has done a lot of work looking at how musical training affects language development in children, and in particular how it affects reading skills.  Research has shown a link between rhythm skills and reading ability, and so it seems like it would be important to understand what we mean by “rhythm skills”.  Are all rhythm-related abilities grouped together in the brain so that if you’re good at one rhythm task you’ll be good at another?  Or can rhythm skills be divided into different groups that rely on different brain networks?

To answer these questions, Tierney and Kraus tested 67 volunteers on four different rhythm tasks.  In the first, the volunteers simply had to tap along (on a conga drum) to a metronome, and the researchers kept track of how closely the tapping matched the metronome.  In the second task, the volunteers again had to tap to a metronome, but this time the tempo was variable, so they had to adjust their tapping to the changing speeds.  In the third task, the volunteers were presented with a repeating rhythm, and had pick up the rhythm and tap along with it.  And the fourth task was similar to the rhythm clap-backs my students have to do:  the volunteers heard a rhythm and had to tap it back.

The researchers looked for correlations between the tasks.  If people’s performance on two tasks is correlated, it means that when people are good at one, they are usually good at the other. The researchers present this type of data as a scatterplot, with each dot representing the performance of an individual person.  If the points are fit by a diagonal line, this shows a positive correlation.

Figure from Tierney & Kraus, 2015

It turns out (as you can see in the figure above) that performance on the metronome tapping and tempo adaptation tasks was correlated.  The researchers conclude these are both based on the same skill of beat tapping.  The third and fourth task, tapping along to rhythms and remembering rhythms were also correlated, both based on rhythm memory skill.  But there wasn’t any real correlation between the two groups of tasks.

The researchers state their conclusions pretty clearly:  These results support the theory that there are multiple dissociable rhythm skills, but do not support the existence of a single overarching rhythmic competence or ‘rhythm IQ’.”

This means that it’s actually completely reasonable for Nicola to have a good sense of beat and to be able to play rhythms well in her pieces, but have problems with rhythmic memory.  The two are based on completely different rhythm skills.  And what that means for me as her teacher is that I need to focus on practicing musical memory with her, and not try to improve her clap-backs by working on other types of rhythm tasks.  It’s a mistake for me to think of rhythm as one thing, and then just have in mind that Nicola needs help with rhythm, when in fact most of her rhythm skills are excellent.  So I’ll be working specifically on rhythm memory with her – practicing formal clap-backs as well as simple “repeat after me” exercises, and looking for patterns in the rhythms that she can grab on to.