(Note: see also A Manifesto For Intuitive Music Composition, which summarises the most important ideas in this article, possibly in an easier to read form.)
An Example of Some Music I Have Composed
"Music theory" tells us something about music, but it only takes us so far: it does not tell us how to compose music good enough to be worth listening to. No one has yet succeeded in writing down a set of formal rules for writing original good music.
So how does anyone ever compose music?
One of the implications of the super-stimulus theory is that we have evolved the ability to appreciate music, but that any ability to create music is entirely a result of learning and practice.
If we reject the possibility that music appears fully formed in the composer's brain by some magical process, and if we can find no recipe for making music which can be written down and passed from one composer to another, then we are left with one plausible explanation for how music gets composed: composers compose music by developing their own internal intuitions about music.
The human brain is a powerful information processing system, and we might think that simple exposure to music would be enough to "teach" it how to compose music. After all, this works for language.
However language, unlike musical composition, is an essential human adaptation. Almost every person who is exposed to a reasonable sample of natural language at a young enough age will develop the ability to understand and generate natural language.
The human brain is not specially constructed to compose music. Most people will only learn to compose if they make a conscious effort to develop the required intuitions.
Not only is special effort required to develop intuition about how to compose, but very likely no one ever develops intuitions about music in general. Rather, successful composers are people who succeed in developing intuitions about how to compose a small subset of all possible music. After all, you only have to compose some music that's good to be considered a create composer.
In other words, to compose an item of music, the composer has to develop intuitions about that item of music, ignoring the problem that the musical item doesn't exist yet, because it hasn't been composed. This seems impossibly circular. But we can escape the circularity by assuming that it is sufficient to develop intuitions about musical items which are very similar to the item we are trying to compose.
Which leads to the concept of incremental composition, as follows:
- Pick a starting point. This can be some sequence of notes and chords which "sound nice".
- Experiment with minor variations. "Muck around" (or to use a more technical term, improvise).
This process of variation and experimentation will not lead immediately to wonderful new music. But it will give your brain information about how the musicality of an item varies as a function of how the item varies. The plan is that you will eventually accumulate a large enough set of intuitions about particular musical ideas to get to the point of being able to compose good quality original music.
Music can be composed on any instrument, but some instruments are better than others for composition. What I currently use is a Yamaha PSR-620 electronic keyboard. This is not the most recent model, but it has touch sensitivity and a reasonable range of good quality sounds. My experience is that timbre is important, and that it is difficult to compose tunes on an instrument that does not "sound nice". Also some tunes work better with some timbres, so the more timbres you can experiment with, the more opportunities you will give yourself to create new music.
A keyboard allows easy combination of melody and harmony, optionally with different instrument sounds for each. However, keyboards aren't suitable for all music, particularly electric guitar music with lots of "bendy" notes, for which basically you need an electric guitar.
You might consider using computer software to compose music, entering notes with a mouse or keyboard. However, to exploit your brain's natural ability to develop intuitions about things, it is desirable to experiment and perform with an instrument that allows a real-time "hands on" approach to creating music, and which gives you instant feedback about the decisions that you make. Typing in notes, and then clicking the "Play" button, and then hearing the result, is not quite instant enough.
Not all music gets composed at the keyboard. I often find myself humming a tune that has "popped" into my head.
However these tunes do not come completely out of nowhere: they are almost always related to tunes I have been playing previously. Often they are based on a chord sequence that I am already familiar with, although I might actually have to sit down and play the tune on a keyboard to realise that this is the case (more on chord sequences below).
I also find that the tendency for music to appear "out of nowhere" is a function of my current emotional or mental state. A tune can unexpectedly "pop" into my head when:
- I feel suddenly sad about something.
- I feel suddenly happy about something.
- I am concentrating on something very intellectual (e.g. a serious maths book).
- I have just been listening to music. (Sometimes I am playing music on my keyboard, then I decide to finish, so I turn the keyboard off, and I go to do something else like housework. But then I find myself humming a little tune, which is not a tune that I have just been playing, and I have to go right back to the keyboard to play the tune I just started humming.)
Some people talk about composing music to "express" their emotions or their feelings about something. However, people usually only "express" themselves this way if they have previous experience creating music. One way to explain the relationship between music and emotion is to observe that musicality acts on emotion, amplifying its effect. As a consequence, a person in an emotional state is more sensitive to the musicality of music, and this enhances their ability to compose new music.
Music that suddenly appears can also disappear. If you are serious about composing, you need to capture all these tunes that seem to come out of nowhere. Depending on my circumstances at the time, I do one of the following:
- Record the tune, using a voice recorder (which can be a cellphone or an MP3 player with recording capability).
- If I'm near my keyboard I go straight to it, and play the tune. Ideally I would still record it, but usually I don't bother, because I find that once I've played a tune a few times, I can remember it in the future (if it's good enough to be worth remembering).
Recording a tune can be tricky, especially in public places. It can be difficult to find a place which is both quiet and private. Also, you might find when you get home that your recorded singing was not sufficiently in tune to be recognisable when you play it back. (One strategy I use is to record both a hummed version and a whistled version of a new tune – typically I can make out the notes of the melody better in the whistled version.)
Humming a tune is one way to perform music to yourself. But rhythm is an important component of music, one which can be difficult to express by humming, and it is helpful if you have some way to accompany yourself with a suitable "percussion" instrument. This can be something as simple as tapping your fingers on a table.
One technique I find to be useful in expressing rhythm is "internal beat-boxing". This is like normal beat-boxing, in that you are using your mouth to create rhythm, but it only has to be loud enough for you to hear it yourself. For example, you can tap your tongue inside your mouth, and it is quite easy to do this while also humming the notes of a tune.
Chord sequences are an important part of most modern popular music, and also of many older types of music including folk music and classical music.
On the one hand, a chord sequence by itself has little or no musicality. Just playing a chord sequence may seem to give a musical result, but this always turns out to depend on the listener hearing an individual sequence of notes (i.e. at least one note from each chord). On the other hand, there is often an intense feeling of musicality when a chord change occurs.
If follows that the musicality of chord changes must be a function of both the melody and the chords.
A complete theory of chord sequences is as elusive as a complete theory of music. But even a partial theoretical understanding may be enough to help us develop intuitions about the musicality of chord changes.
Change of Frame of Reference
Naively we might suppose that the "feeling" of a chord change is a simple function of the contrast between old and new chord, and to some extent this is true. However, if this was all, then every change from chord X to chord Y would have exactly the same degree of musicality.
My theory suggests that a chord determines a local "frame of reference" for the portion of melody occurring within that chord. The model for a cortical map responding to chords is only slightly different from the model of a cortical map responding to a scale and a home chord, and the latter two more obviously determine a frame of reference (see chapter 10 of my book for a full discussion). When the chord changes, then it would seem desirable for the brain to record information about the change in frame of reference (otherwise the record of the melody is piecewise disconnected). To give the melody a role in how the brain encodes information about this change in frame of reference, we can assume that the brain records information about the change in terms of the change in "coordinates" of the melody, i.e. what the coordinates were for the portion of the melody within the old chord, and what they would be if the same fragment of melody had been played over the new chord.
Given that the musicality appears to occur at the moment of the chord change, this suggests that the musicality of the chord change is only a function of the coordinates of the portion of melody played over the old chord, and does not depend on the portion of melody within the new chord (except perhaps for the first note if it occurs simultaneously with the chord change). I am assuming here that there is no significant anticipatory component of this aspect of music perception.
This analysis is incomplete, in that it does not say anything specific about how the information about a chord change is represented in a cortical map (or how this representation corresponds to musicality). However, the hypothesis that chord change musicality is determined by just three things (old chord, new chord, portion of melody in old chord) can be helpful in developing an intuition about how they determine it. In particular, one can consciously pay attention to the old chord, the melody in the old chord, and the new chord, and to the relationship of these three things to the musicality of the chord change. Simply paying attention to the things that matter (and ignoring the things that don't matter) can significantly help the brain to develop accurate intuitions about how to perform a task better.
Other More Global Aspects of Musicality
It might seem from this analysis of chord changes that musicality is determined very locally, and that a tune is just a sequence of mini-melodies, one within each chord. However there are other aspects of musicality which are determined more globally. For example, tempo, rhythm, scale and progression are determined globally. Another intrinsically global aspect of musicality is structure. Among other things, the requirement for structure constrains the different mini-melodies to be similar to each other in a recognisable manner, and disallows the "piecewise" composition of music.
So, for example, to develop an intuition about melodies based on a chord sequence X, Y, Z and W, you have to develop an intuition about the simultaneous optimisation of:
- Musicality of chord change X to Y, as a function of X, Y and melody within X.
- Musicality of chord change Y to Z, as a function of Y, Z and melody within Y.
- Musicality of chord change Z to W, as a function of Z, W and melody within Z.
- Musicality of structural relationship between melodies within X, Y, Z and W.
(To further complicate things, the structural relationship may actually be between phrases, where each phrase starts within one chord and finishes within the next chord.)
Music can be seen as composed of various "components", where each component optimizes a particular aspect of musicality. Examples of components include:
- Chord sequence
- Nested regular beat (i.e. time signature)
- Melodic contour
If the role of each component in optimising a particular aspect of musicality was independent of every other component, then a totally "mix and match" approach to musical composition would be possible. Since this doesn't happen in practice, we conclude that there is a degree of interaction between these components, such that the presence of one component in a tune alters the ideal form of all the other components.
Nevertheless, the interaction is sufficiently small that the components are recognisable as such. It can be useful to think of the components as being fuzzy, fuzzy in the sense that the ideal form of each component is partly a function of its circumstances.
The concept of the fuzzy component allows for compositional techniques that go beyond a purely incremental approach. Instead of just altering one tune bit by bit until an improved result is achieved, we can pull "components" from different tunes, and put them together. The results of these recombinations will require further improvement, but they will be good starting points for creating new tunes. To some extent, once we start paying attention to the musicality of our own creations, our minds will perform these recombinations subconsciously, and this may account for some of the "pop-into-my-head" experiences that composers experience.
For some examples of my own attempts to compose music intuitively, see My Compositions.