Thursday, 26 April 2012

Linking movement and sound

Lately my 6-year-old has been difficult about his piano practice.  His new trick goes like this:  I say something along the lines of, “Okay, why don’t you start with The Wild Horseman today?”

He shoots me a rebellious look and replies. “Fine, I’ll play The Wild Horseman”.  He puts his hands in the correct starting position, and proceeds to sing the entire song, note perfect, while moving his fingers over the correct keys, but not pressing them down.  He then grins saucily at me.

I put on my “mom” voice.  “Very funny.  Now play it properly, with your fingers, not your voice.”  What I don’t tell him is that this is actually not a half-bad way of practicing. 

Practice forms mental representations of the music
When we practice music what we’re actually doing is forming a mental representation of the song.  When we play it for the first time, we read it note-by-note (or maybe chord-by-chord, if we’re more experienced), but as we practice, we stop having to focus on the individual notes, and instead they become encoded in our brain as a whole sequence of notes. Once we have practiced the song enough, we just have to start it, and the notes  follow one another, like beads on a string. This is true for both the movements we make while playing and the sounds that are produced, because we form both a motor representation of the song AND an auditory representation of the song.  That is, we learn the movements we need to make, and we learn what the song sounds like.  These two representations are closely tied together in our brains and they support each other.  Scientists have a special name (don’t they always?) for this connection between the movements we make and the sensations that are produced: “sensorimotor integration”.

Imagine playing your favourite piece of music on an electronic keyboard.  Now imagine playing it with the keyboard turned off, so there is no sound.  It would be much harder, wouldn’t it?  We need that auditory feedback to help keep our motor program running properly.  In fact, the best way to hit all the right notes on the soundless keyboard is to play a mental soundtrack of the song while performing the movements.  This works because the parts of our brain that store and produce the motor pattern are intricately linked to the parts of our brain that listen to the sounds we produce by playing.  And these auditory parts of the brain are activated during mental imagery of music.

The connection between movement and sound works the other way around too.  It’s been shown that if you are listening to a piece of music that you know how to play, motor parts of your brain are activated, as if you were playing along.

Sensorimotor integration aids musical memory
A recent paper from McGill University explores the role of sensorimotor integration in musical memory.  The researchers, Rachel Brown and Caroline Palmer, had pianists learn short melodies in one of four different ways:  1) by simply listening to them, 2) by practicing the songs on a soundless keyboard  3) by practicing them on a keyboard with sound or 4) by practicing them along with recorded version of the songs, but unable to hear their own playing.  The pianists were then tested to see whether they recognized the melodies from among a pool of other melodies they had to listen to.  Pianists were also tested to see how good their auditory and motor imagery was. 

The researchers found that practicing without any auditory feedback (i.e. on a soundless keyboard) made it quite hard to recognize the melodies after.  It was much worse than normal practicing (which was the best), practicing with a recording (2nd best), or just listening to the tunes (3rd best).  However, pianists with good auditory imagery were the most successful at recognizing melodies they had practiced without sound.  In other words, if the pianists were better at mentally “singing along” with their soundless practice, they were better able to recognize those tunes later.

Direct auditory feedback makes for the strongest sensorimotor associations
Another interesting result from this paper came from comparing practice where the movements and sounds were either “strongly coupled” or “weakly coupled”.  Strongly coupled meant that the pianists could hear their own playing, so there was a complete and direct connection between the movements the pianists made and the sounds they heard.  In weakly coupled practice, the pianists could not hear their own playing, but they heard a recorded version of the melody.  What this meant was that as long as they played exactly correctly (in terms of both pitch and rhythm), the sounds they heard were connected to the movements they made.  But if they hit the wrong key on the keyboard or were a little slow in their rhythms, this was not reflected in the sounds they heard.  What the researchers found was that strongly coupled practice made for stronger memories of the melodies than weakly coupled practice.  The conclusion was that direct feedback of the effects of the movements seemed to be required for the strongest auditory-motor associations. 

Practicing the mental representation 
My son, while intending to be silly, is practicing his mental representation of the melody by singing it.  And moving his fingers at the same time practices his representation of the motor task of playing the song.  What’s lacking is the direct feedback: if he makes a mistake with his fingers, it won’t result in a wrong note in his singing.  So if he makes a lot of mistakes, this isn’t going to help his motor representation.  But since in this case he’s playing a song that he actually knows quite well, it’s not a terrible way to practice (and certainly better than not practicing at all!)

Another way to think about this type of “practicing” is that it’s a good way to warm up the brain for the physical practicing of this song.  In fact, a really good warm-up might just be to sit and look over the music and imagine playing it, thinking about how the hands would move and what the song would sound like.  This is mental practicing… but I think that’s a topic for another post.

Brown RM, Palmer C. Auditory–motor learning influences auditory memory for music. Memory & Cognition. 2012. Available at: Accessed April 23, 2012.

Thursday, 12 April 2012

Cellular Mechanisms of Learning

In a comment on a recent post, BusyB asked if I had read the book The Talent Code by Daniel Coyle.  I had not at the time, but quickly requested it from the good ol’ Vancouver Public Library.  The author has a pretty straightforward premise:  to get good at something, you have to engage in what he calls “deep practice”, and this increases your talent by causing the growth of myelin in your brain.

What do I think about this?  I’ll talk about talent and practice (and the inevitable nature-vs-nurture question) at a later date, but today I want to talk about what happens in our brain when we learn something.  Let me say right off that I find Coyle’s claim that skill-building equals myelin growth such a gross oversimplification that I literally cringed every time myelin was mentioned.  The book was peppered with repeated sentences such as “Skill is myelin insulation that wraps neural circuits”.  And, really?  It’s just not that simple.  Sorry.

Let me back up and tell you a little bit about neurons and synapses and myelin and what we think happens to them during learning.

Neurons in a nutshell
You probably know that neurons are brain cells, and you have billions of them in your head and spinal cord.  Neurons “talk” to one another electrically through specialized connections called synapses.  Here’s a schematic view of two neurons (one blue, the other green) connected by a couple of synapses.

The axon is the part of the neuron that carries electrical impulses away from the cell body.  At the end of the axon, there are specialized endings where the electrical signal gets transferred onto the dendrites of another neuron.  These are the synapses.  In most neurons, the axon is wrapped up in an electrical insulator made of a substance called myelin.  Electrical impulses travel faster down axons that are insulated, and so the presence and amount of myelin on an axon alters the neuron's ability to transmit electrical signals.

Learning makes stronger connections between neurons
That’s, in a very small nutshell, how neurons work.  And although I’ve only shown two neurons in my diagram, every neuron is connected to many, many other neurons, forming a complex spider’s web of neuronal circuits.  Almost everything that happens in our brains comes down to circuits of neurons transmitting electrical signals.  So when we learn something new, or get better at doing something, what happens in our brains is that the neuronal circuits responsible for that fact or skill become stronger, better able to communicate with one another. 

There are several main ways in which this can happen:
1)  The synapses themselves become stronger and so transmit the signals more reliably
2)  New synapses form, so the neurons are more strongly connected
3)  Myelin growth leads to faster transmission of the electrical signals down the axon, and better timing of neuronal signals.

The synaptic mechanisms (numbers 1 and 2 above) have been studied in excruciating detail (or at least that's how it feels to people like me who have spent years in nitty-gritty synaptic research) and scientists as a group are slowly getting a handle on how changes in synapses happen and how this helps us learn things. 

Neurons that fire together wire together
Here’s how scientists think that learning works:  The basic idea is that every thought is encoded by the firing of a specific group of neurons, all connected in a circuit.  So a particular circuit fires when we think of the note middle C, for example.  And there’s another circuit that fires when we picture a note on the first ledger line below the treble clef staff.  When we learn that this position on the staff corresponds to middle C, both of these circuits fire at the same time.  And when neurons fire at the same time, the connections between the neurons get stronger.  The synapses get stronger, and/or new synapses form.  This means that the next time we fire the circuit that means “note on the first ledger line below the staff”, the circuit that corresponds to “middle C” is more likely to fire.  Neuroscientists have a saying for this:  "Neurons that fire together wire together”.  From a learning standpoint, it means that we have learned to connect those two ideas by physically altering the way the neurons in our brain are connected.

These changes in synaptic strength very clearly happen when we learn something, whether new facts or new skills.  Synaptic changes are an important part of learning during development, and relearning following brain injury.  There is a ton of research showing this.  The fact that Coyle doesn’t even mention these types of mechanisms as taking place during learning is kind of ridiculous.

Myelin and Learning
So what about myelin?  Does myelin growth aid in skill learning, as Coyle purports?  The answer, based on scientific research, is “probably”.  There are correlational studies showing that people who are more skilled at certain tasks (like reading, or playing music) have greater myelination in areas of the brain related to those tasks.  In particular, musicians have a larger and more myelinated corpus callosum, the axon bundle that connects the two halves of the brains.  This is especially true for musicians who began their musical training before the age of 7, which makes sense, because the myelination of the nervous system is something that occurs throughout childhood and is not complete until a person is in their mid-twenties.  Myelination of neurons during learning in adults is still a controversial idea, and research in this area is on-going.  I’m interested in this line of research, especially the thought that the amount of myelin helps to co-ordinate the arrival of signals from different neurons.  Perhaps myelination plays a greater role in learning of skills compared to learning of facts (implicit vs. explicit learning), but I was not able to find any evidence for this in the scientific literature.

What do I think about The Talent Code?  I agree with the (rather obvious) idea that hard works leads to the acquisition of skills, but I think there are better and more interesting books that address this topic (this one, for example).  However, I think the scientific side of his book is weak, oversimplified and kind of misleading.


Bengtsson SL, Nagy Z, Skare S, Forsman L, Forssberg H, Ullen F. (2005) Extensive piano practicing has regionally specific effects on white matter development. Nat. Neurosci. 8(9):1148-1150.

Fields RD. (2008) White matter in learning, cognition and psychiatric disorders. Trends Neurosci. 31(7):361-370.

Schlaug G, Jäncke L, Huang Y, Staiger JF, Steinmetz H. (1995) Increased corpus callosum size in musicians. Neuropsychologia. 33(8):1047-1055.

Ullén F. (2009) Is activity regulation of late myelination a plastic mechanism in the human nervous system? Neuron Glia Biol. 5(1-2):29-34.