Melodic Identity
In my previous attempts to explain the nature and origin of music, I have identified the concept of Melodic Identity - the idea that, for “strong” melodies, the listener’s brain has a strong sense of which melody they are listening to.
This can be described in terms of qualia, ie a strong melody has a strong qualia, a feeling of being a thing where the feeling of that thing is quite distinct from the feeling of being some other thing. Similar to how the colour “yellow”, for example, has a qualia of yellowness, which is subjectively quite distinct from the feeling of redness associated with the colour red, even though they are both colours.
But what is a qualia, and can we give a pragmatic description of what a qualia is that is not tied to our own subjective experience of what it is like to have the subjective experience of that particular qualia?
Qualia – a working hypothesis
My working hypothesis is that qualia are those subjective perceptions that are likely to be immediately relevant to conscious decision-making.
For example, there are many situations where seeing something of a particular colour (or seeing something be a different colour to what it normally is) is likely to be highly relevant.
We can note that in our modern technological society we have the ability to add almost any chosen colour as a coating on almost any manufactured object, and we can potentially be saturated with colours which don’t really mean anything.
However, in less technological times, a lot of our environment consisted of things that were green or brown, and the observation of a different or unexpected colour was much more likely to be directly relevant to dealing with a situation.
Applying the working hypothesis to music
If we assume that this working hypothesis applies to melodies, we have an immediate difficulty: it’s not obvious why the identity of a particular melody should matter that much. If we listen to music for enjoyment, or to achieve an altered state of mind, it doesn’t seem like it matters that much to know which melody we are listening to. (Of course if you are the one performing a musical item, it does generally matter to know which item you are performing, but that is a secondary consideration, and it could be considered part of the general technical issues involved in performing music that depends on the learned skills of doing that – it’s not essential that all music listeners experience qualia from different musical items just so that performers more easily know which is which.)
I propose a solution to this difficulty, which is that melodies have strong identities, because at some earlier stage in the evolution of music, the ancestor of music – so-called protomusic – was a type of language where specific melodic identities communicated specific meanings.
These meanings were culturally assigned, similar to how the meanings of words are culturally assigned in word-based language. And these meanings had to be learned by each child growing up in a society, just like the meanings of words have to be learned, presumably by observing how other individuals around them used the protomusic to express meaning.
The unevolution of the meaningfulness of melodic identity
Of course in the modern world, any tendency to learn specific meanings of specific melodies is very limited.
Even in cases where a melody is usually accompanied by words, ie the lyrics of a song, the corresponding association is mostly determined by the repeated occurrence of those words.
In other words, one could write new lyrics for a particular melody, and provided that the new lyrics were consistent with the emotional feel of the melody, then those new lyrics would be acceptable to music listeners, and the fact that previously the song had only been heard with the old lyrics would not cause any issue.
Also, in the case of music without lyrics, nobody ever determines a specific meaning that the music is intended to have.
Musicality
Identity is one property of a melody.
But our primary criterion for listening to music is not whether or not it has a strong identity, but rather how “good” the music is.
In principle one can string a random series of notes together and call it a melody, but in practice no one will want to listen to it, because it won’t be very musical.
Indeed the problem of composing music can be defined as the problem of finding melodies which are as musical as possible.
This implies the existence of a musicality function, which determines the quality or “strength” of an item of music as a function of the audio content of the music.
Composition then becomes the problem of maximising the value of the musicality function.
But what actually does this musicality represent or mean?
And what is the actual formula for calculating it?
As far as we can tell, musicality has no meaning other than telling us whether something is good music or not. And music consists of anything that has a high level of musicality. Which is all very circular.
The implication is that musicality does not have any external meaning.
Which raises the question of why such a thing exists at all.
However, if we assume that melodic identity is something that matters, or mattered in the prehistoric past, then we can observe that:
- A melody is a local optimum of the musicality function
- The musicality function is defined so that a suitable set of local optimums can be determined and assigned culturally-determined meanings.
In other words, musicality is defined so that melodies can exist. And they don’t just exist, they are stable because those hearing them are motivated to maximise musicality, which means that any item which is a local optimum will not move away from that position of local optimality.
Activity Patterns
One thing I have observed previously about musicality is that some aspects of music, such as pitch scales and nested regular beat, can be explained by assuming optimisation of a function which is based on the occurrence of certain types of activity patterns within cortical maps in brain areas that process sound, and also that the observation of these activity patterns is plausibly carried out by certain glial cells.
On the one hand, when I originally discovered this explanation, I was unable to determine what the occurrence of these activity patterns actually represented.
But, under the current hypothesis, we can see that the activity patterns do not have to represent any significant fact about the real world, rather they can define a complex function the optimisation of which generates a suitable set of local optimums, where those local optimums determine a set of symbols that can be the basis of a communication system.
Also, the musicality function is determined by the same patterns occurring in different cortical maps, the implementation is consistent across those different cortical maps, and therefore the amount of actual mutation required to make it happen is not so great.
Digital Communication
If this hypothesis about melodic identity is correct, then we can see it as the first case of a digital communication system with a set of specific symbols with assignable meanings.
It is digital in the sense that the set of distinct “symbols” corresponds to the set of distinct strong melodies that are known to the members of the society that knows them.
Here we are talking about members of prehistoric tribes that lived in relatively small groups, so the total number of melodies that they knew would not be very large – perhaps only a few dozen, or a few hundred at the very most.
In modern word-based language, the set of distinct “symbols” corresponds to the number of distinct phonemes, ie consonants and vowels.
Also, in word-based language, the minimal unit of meaning generally consists of a small sequence of phonemes that constitutes a word, or in some cases, part of a word.
And these meanings can be used to construct larger meanings by putting those words or parts of words into sequences that constitute sentences.
Whereas for the system of protomusical communication, the unit of meaning would just be the individual “symbols” (ie melodies), and also there was probably no process for assembling individual symbols into more complex utterances.
So compared to modern word-based language, this hypothesized system of protomusical communication was very limited and constrained in what it could communicate.
When we look at comparisons between word-based language and music, we can be tempted to identify individual notes in a melody as minimal components of a digital communication system.
However there isn’t anything like the rules of word-based language grammar that is known to apply to music.
And my hypothesis does not treat notes as individual components of a digital system.
Rather the existence of notes from pitch scales and regular nested beat is a consequence of the optimisation of the musicality function, because for certain of the cortical maps processing audio information, maximisation of the activity patterns relating to the musicality function will always require the melodies to have pitches from pitch scales and to have rhythm defined within nested regular beats.
At the same time, I have previously proposed the hypothesis that word-based language originally evolved as an enhancement to protomusic – where words initially evolved as additions to protomusical melodies in order to add extra details to the information being communicated by the protomusic.
For word-based language embedded within a melody, the individual notes typically correspond to syllables, so there is some correspondence between the notes of the melody and the occurrence of digital symbols of the word-based language.
In a sense the regularity of pitch and rhythm in protomusic actually provided a scaffold for word-based language to exist and be embedded in the protomusical melodies.
The Pleasure of Music
There are various ways in which this hypothesis explains our experience of music as a pleasure.
The first is that if music is determined is determined by the optimisation of a “musicality” function, then the process of optimisation generally involves finding the thing that the listener most “likes”. And discovering something that you “like” is more or less what pleasure is.
To put it another way, listeners to protomusic had to like hearing strong melodies in order to be motivated to discover the locally optimal solutions to the musicality function required to create and preserve the melodies that constituted the protomusical communication system for their tribe.
In my previous article https://blog.thinkinghard.com/blog/music-has-a-meaning-but-we-never-find-out-what-it-is I have proposed the hypothesis that music evolved from protomusic when the ability to determine and assign meanings to individual melodies was lost.
So there was a three-step process:
- Discover new melodies
- Get the feeling that a new melody should have a meaning
- Determine what that meaning should be (from observation of how others used the melody, or possibly in the case of a new melody that you just thought of, make up something plausible)
All these 3 steps involved some pleasure. But what got lost was step 3, and therefore the musical listener got stuck on step 2, and as a result each item of music had the ability to cause almost unlimited pleasure from repeated occurrences. (I say “almost”, because the best item of music can suffer from too much repetition – although even the effects of excessive repetition can be recovered from after giving it a bit of a rest, ie we still want to listen to many of our “favourite” tunes that we first heard years ago.)
The Emotion of Music
In this article I haven’t actually said much about the emotional aspect of music.
On the one hand a musical item is not just characterised by how musical it is, it also has an associated emotional feel.
The emotional feel of a musical item necessarily constrains any meaning that might be associated with it, which, for example, constrains the set of possible lyrics that might “work” with a given melody.
In the case of protomusical items, we would assume similarly that any culturally assigned meaning of a protomusical item would have to be consistent with it’s emotional feel.
On the other hand, I hypothesise that musical items are the result of the optimisation of a musicality function where that function is in effect a single-dimensional number.
But emotion cannot be reduced to a single number.
The implication is that the emotional feel and the “musicality” are two orthogonal aspects of a musical item.
I have proposed that the optimisation of the musicality function determines the existence of musical items as local optimums, to provide “digital” symbols to which meanings can be assigned.
However this could be completely separate from how emotional feel is determined.
For an evolutionary point of view, one of those two things had to come first, and my best guess is that it was the communication of emotion that came first.
So we can suppose the existence of a “pre-protomusic” which had emotional feel, where that emotional feel was expressed in a “analog” nature, and where musicality and the associated digitality was not yet a thing.
So the pre-protomusic started as a non-digital system of communication of emotions, and then musicality was added to support the digitalisation of the protomusical melodies so that they could be assigned discrete meanings, but where those meanings still had to be consistent with the emotional feel of the protomusical item in each case.