The most basic perception that comes from
listening to music is simply that it is music that we are listening to.
When we listen to music we perceive the "musicality" or "musicalness" of the music.
The concept of musicalness as a variable quantity reflects
the observation that we perceive some music to be more musical (i.e.
better, or stronger) than other music.
Another consequence of perceiving music is its emotional effect.
However it is generally found that the intensity of the emotional effect of
music is proportional
to its musicalness, which suggests that the brain calculates musicalness first,
and then uses that calculated information to alter its emotional response
accordingly. Yet another consequence of perceiving music is pleasure,
but the extent of musical pleasure is also a function of musicalness.
So we can continue under the assumption that the primary output of
the music processing sub-system is musicalness.
But what is the meaning of "musicalness"? If the brain contains a system for
calculating the musicalness of music, then the value that it calculates must have
some meaning, otherwise why even bother to calculate it?
Does music have any meaning? We can try to answer this question by
considering where music comes from. Music comes from those who compose it, and from those
who perform it. Why do composers and performers make music? In practice they
may have various goals, but it seems likely that, in many cases,
particularly with regard to the commercial music that most of us listen to, the main goal is
to make music that is as musical as possible.
However, if musicalness is a calculated property of music, and music is just something
that has been created to be as musical as possible, then our understanding of
the relationship between music and musicalness is too circular.
But what if musicalness was a perceived property of something else?
That Which Is Like Music But Which Is Not Music
If musicalness was a property of something else, then we might expect that something
else to be similar in at least some respects to music. There is only one natural
phenomenon that I know of which bears any major resemblance to music, and that phenomenon is speech.
Speech is certainly like music in several respects. It has recognisable "melody" and "rhythm",
and it is spoken using one of the world's most popular musical instruments - the human voice.
So we can tentatively conclude that the brain has a "musicalness-calculator", which calculates
the musicalness of speech.
To sum up the argument so far:
- The brain is an information processing system.
- Music is information, and some part of our brain processes that information.
- The output that comes from processing music information is "musicalness".
- The brain sub-system that calculates "musicalness" serves some useful purpose.
- Music exists only because it is contrived to be musical.
- As an explanation of what music is, this is too circular,
so the purpose of the music processing sub-system must relate to the
perception of something else.
- The only thing like music which is not music is speech, so
the input that the music processing sub-system is designed to process must be speech.
Now normal speech is not in general very musical, and this
might seem to be a fatal flaw in the theory. But we have already observed that the creators
of music try to make music as musical as it can possibly be. This is consistent with the
assumption that music is a super-stimulus for the perception of musicalness.
The musicalness of speech may be many times less than the musicalness of music, but that
does not rule out the possibility that the major purpose of the musicalness-calculator
is to calculate the musicalness of speech. Indeed many perceptual systems in humans and animals
are specifically "designed" (by evolution and natural selection) to perceive things in
the world around us that are only just perceptible, because sometimes it matters to
know something if it is at all possible to know it.
If we accept the perception of musicality as being an aspect of speech perception, then
two major questions follow:
- What is the meaning of musicalness? (Which I asked before, and have not yet answered.)
- How do we explain those features of music which do not seem to have any parallel in speech,
such as scales, the regularity of musical time and the occurrence of harmony?
Ultimately we must answer the first question, if the theory is to be convincing. But we
will find it more productive to answer the second question first.
A general answer to the second question is that
a super-stimulus for a perception may be qualitatively different from the normal stimulus
for that perception. I will illustrate this general principle by considering individual
musical features, starting with harmony.
The problem with harmony, in any theory that seeks to link our perception of music with
our perception of speech, is that music contains simultaneous pitch values, whereas
normal speech does not. It seems unlikely that the brain would devote significant resources
to the perception of pitch relationships between the speech melodies of two different
speakers speaking simultaneously. Indeed our most common experience of listening to two
speakers is that the best we can do most of the time is filter one out and listen to the other.
A common form of harmony, in many popular forms of music, is the occurrence of chords.
And a curious fact about chords is that we can perceive the quality of chords if the notes
of the chords are played simultaneously, and we can also perceive the quality of chords
if the notes are played sequentially. We can strum a chord on a guitar, or, we can play
the notes of the chord one by one. So if there is a cortical map which responds to the harmonic
relationships between notes in chords, then this cortical map is capable of
responding both to relationships between simultaneous pitch values and to relationships
between sequential pitch values. It is entirely possible that the purpose of this map
is to respond to the harmonic relationships between pitch values
occurring within a single speech melody, even though the super-stimulus for the same cortical
map consists of multiple simultaneous pitch values.
Why would the brain have such a cortical map? Why would it care about the harmonic
relationships between pitch values occurring at different times in a speech melody?
One of the major properties of our perception of melody
is pitch translation invariance, which basically means that if we transpose a musical
item into a different key, it is still essential the same music. This invariance also applies to
speech melody: the melodic aspects of speech, including
intonation, pitch accent or lexical melody, are preserved by the operation of
adding a constant interval to the pitch contours of speech.
This allows speakers with different pitch ranges to speak the "same" speech melodies.
To achieve pitch translation invariant perception of a melody, it is necessary to
perceive the relationships between different pitch values within the melody. Harmonic
relationships are based on simple fractional ratios of frequencies, and these ratios
are intrinsically pitch translation invariant. (Note that the term "translation" makes
sense if we consider pitch to represent log frequency, in which case multiplication
of a frequency by a ratio can be considered to be an additive operation.) This gives
us a plausible explanation of why the brain might be interested in detecting harmonic
relationships between different pitch values in a speech melody.
Scales and Regular Beats
Two more features of music not found in normal speech
are musical scales and regular beat. I consider them together,
because we can construct plausible models of cortical
maps which respond to these features of music, such that there is a strong analogy
between the patterns of activity which occur in these cortical maps when responding to music
and which do not occur when responding to normal speech. This analogy suggests a deep
connection between the roles of time and pitch, two major aspects of music
which seem at first glance to be qualitatively distinct from each other.
For scales, we can start by assuming that the scale in which a musical item
is played is a perceived property of that item. How would this property be perceived?
The scale consists of the set of notes which occur, with the order of occurrence
factored out of the perception. This suggests the existence of neurons which respond
somewhat persistently to the recent occurrence of pitch values. If we consider
these neurons to exist in a tonotopic map (which means that pitch is correlated
with position), then their response to music will be such that:
- regions of the map corresponding
to the pitch values of the scale will be active;
- regions corresponding to in-between
pitch values will not be active;
- this pattern of activity will be constant for the duration of the tune.
Now let us consider regular beats. If a tune is in, for instance, 4/4 time, and contains
quarter notes, eighth notes and sixteenth notes, then there will be five perceptible regular beats,
corresponding to once per bar, twice per bar, four times per bar, eight times per bar and sixteen times per
bar. If there are neurons that respond to regular beats, and they exist in a cortical map
where position is correlated with the beat period, then our piece of music in 4/4 time
will cause a pattern of activity in that cortical map with the following characteristics:
- there will be five active regions;
- the regions between the active regions will mostly be inactive;
- this pattern of activity will be constant for the duration of the tune.
Constant Activity Patterns
The constant activity patterns that I have just described
will occur in these cortical maps when listening to music, but they
will not occur when listening to normal speech. They will not occur in the map that persistently
responds to recent pitch values, because speech melody is continuous, and therefore all
neurons corresponding to the pitch range of the speech melody will be active, and there will be no gaps.
They will not occur in the map that responds to regular beats, because the rhythms of speech
are irregular, and the spectrum of irregular rhythms will be continuous rather than discrete.
If similar activity patterns exist in different cortical maps that respond to music,
and if we remember that we are assuming that music has been constructed specifically to
maximise "musicalness", then we might wonder if there is some relationship between
constant activity patterns and perceived musicalness. Whatever musicalness is, it appears to be something
that is maximised by the occurrence of constant activity patterns in cortical maps
involved in speech perception, where the activity patterns include multiple zones of
activity and inactivity.
Speaker's Mental State
Unfortunately the development of the theory becomes somewhat more speculative at this point,
but I will persist nonetheless.
If musicalness is a perceived property of speech, then it must be something that tells the
listener useful information about the speaker and the speech that they are speaking.
Perhaps information about the patterns
of activity in the cortical maps of the speaker tells the listener something useful
about the mental state of the speaker, and this information is used
within the listener's brain to adjust their emotional response to the content of the speech.
(This would explain the emotional effect of music, especially given that music is assumed
to be a super-stimulus.)
In order for the listener to perceive patterns of neural activity in the speaker's
brain, there would have to be some relationship between neural activity patterns in the speaker's
brain and neural activity patterns in the listener's brain, in a way which preserves
the geometric nature of those patterns, at least to a sufficient
extent that the patterns can be perceived. This implies some form of "neural mirroring".
The mirroring does not necessarily have to be very accurate - it just has to be
accurate enough that some observation can be made of the patterns of activity in the
speaker's brain. Furthermore, the perception can be performed by combining perceptions
made across as many different cortical maps as possible.
We have already observed that musicalness is
a very subtle property of normal speech (compared to its unsubtleness as a property of
music), so this is consistent with a perceptual process that operates at the very
edge of what is feasible.
At the beginning of this article I gave the example of the colour-processing sub-system
in the brain, which occupies a specific region within the occipital cortex.
The theory of constant activity patterns suggests a more spread out location for the
music processing sub-system. Indeed it implies that musicalness is determined according to patterns of activity
in at least two (and probably many more) cortical maps, and those cortical maps have
purposes other than the calculation of musicalness. Thus musicalness appears to be
a secondary perception calculated within those cortical maps. In as much as it relates
to actual physical patterns of neural activity, it seems plausible that there
exists a specific type of neuron which is specialised to respond to the activity patterns
which represent musicalness. These "musicalness neurons" would have their own special
anatomical form, they would be spread throughout various cortical maps that play
a role in the perception of speech, and their outputs would presumably have some connection
to those parts of the brain that process emotional responses to speech.
Identifying such a population of neurons would be
a strong confirmation of the constant activity patterns theory.
Music science has experienced a revival over the last twenty years.
and the Brain
is a recent Scientific American article about scientific
research into music, which could be regarded as giving a good indication of the current
"state of the art" in music science. In the third paragraph of that article we get an admission
that scientists do not know what are the selective pressures that led to the evolution of our ability to
respond to music. Or to put it more bluntly, scientists still don't have a clue as
to what music actually is.
The theory of "constant activity patterns"
comes as close as any other theory to providing a plausible answer to the question of what music is.
The theory unifies the roles of pitch and time, it provides a neural model of what "musicalness" is,
and it even suggests a plausible evolutionary motivation for our ability to perceive musicalness:
the perception of the musicalness of speech represents the perception of something about the mental state of the
speaker which helps to determine an optimal emotional response (by the listener) to the content of that speech.