Composing Music is Like Learning to Ride a Bicycle

26 May, 2011
An explanation of why you can't just learn to "compose music". The reason being that the composition of each new musical item must be a separate learning experience in itself.

An Explanation of the Title ...

Notice that the title of this article is not:

Composing Music is Like Riding a Bicycle
nor is it:
Learning to Compose Music is Like Learning to Ride a Bicycle

The point being that composing music is a learning experience, for each and every item of music that you compose.

For every item of music that you compose, you have to learn how to compose that particular item of music.

Which means that, strictly speaking, composing your first item of music is like learning to ride a bicycle, and composing your second item of music is like learning to ride some other vehicle, and a full career in musical composition is like learning to ride an endless stream of different kinds of vehicle.

If we really wanted to say something about "Learning to Compose Music", we could say:

Learning to Compose Music is Like Learning to Learn to Ride a New Type of Vehicle

Composition is to Music Theory as Bicycle Riding is to Physics

Music theory "explains" how to compose music in the same way that physics "explains" how to ride a bicycle.

But you can't learn to ride a bicycle by reading books on physics. If you want to learn to ride a bicycle you have to actually get on a bicycle and try to ride it.

In the process of learning to ride, your brain and your body will develop intuitions about how bicycles work, and how to make the bicycle go forwards and not fall over. We can say that, in some sense, the parts of your brain and nervous system containing these new intuitions actually "know" some physics. They don't know all physics, but they know a subset of physics relevant to riding a bicycle.

The same applies to composing music. In principle "music theory" could tell you how to compose music, but in practice, if you want to compose music, you have to sit down and play notes on an instrument.

And just like the learning cyclist who falls over many times before learning to ride, your attempts at composition will consist of playing notes that don't sound so much like music. But if you keep on trying, and varying your existing attempts, and try even more, eventually you will get there.

And when you have composed some music, your brain will have acquired new intuitions about music, which will be a subset of a full theory of music – a subset that relates to the musical item that you composed.

One Difference Between Learning to Ride a Bicycle and Composing Music

One thing that makes musical composition much harder than learning to ride a bicycle is that you have to learn the tune you are composing, even though that same tune doesn't exist yet.

Whereas in the case of riding a bicycle, you know that other people can ride bicycles, and you can even watch them do it.

But you can't easily watch someone else composing music, and you can't compose music by copying the music that someone else plays, because that's not composing, that's just playing someone else's composition.

It's like you have to invent bicycle riding at the same time as you are learning to do it.

Theories Trapped Inside the Brain

In the bicycle riding case, even when your brain "knows" how to ride a bicycle, and "knows" the relevant physics, there isn't any easy way to write that knowledge down in the form of mathematical equations. The physics knowledge exists in the cyclist's brain, but it's trapped inside, and it can only be accessed when the cyclist actually rides his or her bicycle.

Of course if we wanted to teach a robot how to ride a bicycle, then we would probably start with the relevant physical equations, and use them to devise a bicycle-riding strategy which we could program into the robot. The same should apply to programming a "musical robot" so that it composes music. But ...

Another Difference Between Learning to Ride a Bicycle and Composing Music

Although the physical knowledge contained within a cyclist's brain is equivalent to some subset of Newtonian mechanics, there is no need to know Newtonian mechanics in order to learn to ride a bicycle. Indeed it would be quite possible to learn to ride a bicycle if Newtonian mechanics hadn't even been invented yet.

I'm ignoring the issue here that to manufacture a bicycle probably requires technology which wouldn't exist if we didn't yet know Newtonian mechanics. But, for the sake of argument, let's just pretend that we have a time machine which we can use to take a bicycle back in time, before Isaac Newton was born, and that we want to teach the people living in that time how to ride the bicycle. It seems likely that those people could learn to ride our bicycle, even though they don't know Newton's laws of physics.

In the case of music, we don't need to go back in time, because right now there is no "music theory" equivalent to the Newton's laws of physics. We don't know the "laws of music" that determine whether or not a possible item of music is musical. The person who is going to discover the equations of music hasn't discovered them yet. (A corollary of these observations is that, despite various claims to the contrary, all current "algorithms" for musical composition produce rather crappy results.)

Of course we do know some things about music, like music is formed from pitch values belonging to scales, and these pitch values are related to each other by consonant intervals, and the notes occur at times determined by regular beat patterns. But these things are not a complete theory of music, and they are more like a set of observations which are relevant, but not sufficient for the purpose of composing music which most people would judge to be "musical". (A bit like knowing that "things fall down", which is probably useful to know if you're learning to ride a bicycle, but that's not the same as knowing a proper theory of gravity.)

In other words, we have to develop intuitions in order to compose music, not because it's too hard to use the formal theory, but because we don't have a formal theory. Scientifically, we have no idea what music is, why it exists, or what type of information processing occurs when music is "appreciated" by it's listeners.

Is it Any Easier to Learn To Ride a Second Vehicle/Compose a Second Item of Music?

If you've learned to ride a bicycle, probably some of the information stored in your brain can be re-used to help you learn to ride a new type of vehicle. Like a unicycle, or a motorbike. Or even a car, given that bicycles and cars both have brakes, gears, and they're both required to follow traffic regulations on the road.

Similarly with music composition, if you've composed one item of music, you'll probably be able to re-use some of the intuition acquired composing that first item when you compose a second item.

But for both vehicle-learning and musical composition, any re-use of intuitions will only be partial, and there will be a significant amount of new learning required which will be unique to each vehicle type learned or musical item composed.

So yes, it will be easier, but not a whole lot easier.