The Pattern Beneath The Form

There is a moment that composers sometimes describe before a piece has become music. No key has been chosen, no tempo marked, and perhaps not even a single note has been written, yet something about the piece already seems present. It may feel heavy or light, restless or still, open or compressed. The composer doesn't necessarily hear the finished music. There is simply enough of a direction that some musical choices begin to feel more appropriate than others.

Other kinds of artists describe similar experiences. A painter may have a sense of temperature or pressure before choosing a color. A choreographer may feel the character of a movement before counting its rhythm. A writer can know that a sentence is wrong before knowing which words would make it right.

I've become interested in that interval between recognition and form.

It's tempting to say that all of these artists are encountering the same underlying thing and then translating it into different media. I used to make that claim much more confidently. There may indeed be useful structure shared among these experiences, but resemblance across modalities doesn't establish that we've discovered a single object underneath them.

Something interesting is still happening.

We can investigate that without deciding too early what it is.

When sunlight becomes music

Suppose someone says that a piece of music feels like sunlight.

Most of us don't respond by pointing out that sunlight is electromagnetic radiation while sound consists of pressure waves moving through a medium. We understand that the speaker isn't claiming physical equivalence. Somehow a relationship has been proposed between experiences that arrive through very different modalities.

Ask several people what makes music feel like sunlight, however, and the apparent simplicity begins to break apart. One person may hear warmth. Another may hear brightness. Someone else may associate sunlight with openness, morning, safety, exposure, summer, glare or even discomfort. Their musical choices may differ accordingly.

That variation is useful because it tells us something about what the comparison actually does.

“Sunlight” doesn't contain a musical instruction waiting to be decoded. It provides a field of possible relationships. The composer selects some properties as important, maps them into musical choices, listens to the result and decides whether the mapping preserved whatever mattered about the original intuition.

Someone else can make a different mapping and produce something equally recognizable.

The interesting object may therefore be less like a hidden code and more like a transformation.

What survives the translation?

Consider a simpler cross-modal mapping:

warm → amber → low strings → slow movement → rounded language

There's an intuitive coherence to that sequence for many people. We can probably imagine the room, music, movement and prose belonging together.

That doesn't tell us why.

Some of the association may come from our bodies. Some may come from culture, language, artistic convention or personal history. Some relationships may recur across many people while others vary sharply between populations or individuals. Several causes can participate at once.

Instead of immediately asking whether we've found the underlying pattern, we can ask a more useful question:

What are we preserving as we move from one form to another?

Perhaps warm becomes amber because we're preserving an association with heat. It becomes low strings because we're preserving weight rather than brightness. Slow movement preserves calm. Rounded language preserves softness.

Another person might preserve different properties and produce:

warm → yellow → acoustic guitar → loose movement → conversational language

The outputs differ because the transformations preserved different things.

Neither needs to be the correct translation of warmth.

Now we have something we can study.

The artist as translator

Artists do this constantly, although translation may itself be an imperfect word. Translation suggests that there is a stable message on one side and an equivalent representation waiting on the other. Creative work is usually messier.

A composer begins with an impression and makes a musical choice. Hearing that choice changes the impression. A second choice responds to the first. Something unexpected appears, and the emerging piece begins pushing back against the intention that started it.

The artist isn't simply extracting a completed work from a hidden source.

The source changes through enactment.

This helps explain why artists can begin with something they can't articulate and still recognize when a particular expression doesn't fit. They don't necessarily possess a secret specification of the finished work. They have enough constraints to reject some possibilities and continue exploring others.

The work becomes through that exchange.

Compression without essence

This also gives us another way to think about artistic compression.

A few notes can imply a much larger emotional landscape. Three colors can establish the character of a visual world. A movement phrase can tell us something about a character before the character speaks. A sentence can establish the pressure under which an entire story will operate.

The small form doesn't literally contain everything we later experience in it. It constrains what can plausibly grow around it.

That distinction matters.

If I describe a character as warm, grounded, unpretentious, comfortable with ordinary pleasures, I can use those properties to guide music, costume, movement, dialogue, architecture or food. Each medium forces different decisions, and each decision may reveal that my original description was incomplete.

The abstraction is useful because it travels.

It isn't necessarily truer than the forms it helps produce.

In fact, the forms can teach us something the abstraction missed.

A pattern needs somewhere to be wrong

This is where the exercise becomes more interesting than simply finding poetic correspondences among colors, sounds and movements.

Suppose we describe a quality across several modalities:

Sound: sustained high tone with bell-like overtones

Color: deep indigo

Movement: slow rotation

Language: a question that gradually answers itself

At first glance, these may feel related. We can ask people to describe the relationship, generate additional forms from the set, remove one modality and predict it from the others, or deliberately substitute something that seems not to belong.

Then we watch what happens.

Perhaps many observers independently reconstruct something similar. Perhaps the agreement disappears when we cross cultures. Perhaps one of the modalities turns out to be doing most of the work. Perhaps an AI model reproduces the mapping because those associations occur frequently in its training material. Perhaps two people agree on the feeling while choosing completely different manifestations of it.

Every result tells us something.

None requires us to announce beforehand that the four expressions are secretly the same thing.

Pattern Grammar

This is one reason I've been experimenting with what I call Pattern Grammar.

Take a finished work and compress it, not by summarizing what happens, but by choosing relationships that seem important to preserve. Remove the names, setting, medium and surface details while retaining some combination of movement, tension, sequence, recurrence, contrast, uncertainty or tone.

Then give that compressed grammar to another generator.

That generator might be another person. It might be an AI model. It might even be yourself months later after you've forgotten some of the original.

The regenerated work will not reconstruct the source. Too much has been discarded for that. Instead, it shows what can grow from the properties we chose to preserve.

That difference turns compression into an experiment.

If the new work strongly resembles the original in ways we expected, the grammar may have captured something important. If it resembles the original in ways we didn't expect, we have another question. If it wanders somewhere completely different, perhaps the grammar discarded something essential to the effect we were trying to preserve.

Or perhaps the new work shows us that something we thought was essential wasn't.

The artifact gets a turn in the argument.

Moving across modalities

The same experiment can be performed across media.

Begin with a short piece of music and ask what you want to preserve when translating it into an image. Then turn the image into movement, the movement into prose, and the prose back into music.

Don't ask whether the final music matches the original note for note. It shouldn't.

Ask what survived.

Maybe the original tension survives while the tempo disappears. Maybe visual symmetry becomes rhythmic repetition. Maybe something that felt melancholy in music becomes peaceful in color. Somewhere along the path, an apparently stable property may split into several possible interpretations.

That is not necessarily translation failure.

It may be the most informative part of the exercise.

Every medium forces the pattern through different constraints, and those constraints expose choices that were invisible while we remained inside one form.

Intuition and structure

None of this diminishes artistic intuition. If anything, it gives intuition more interesting work to do.

An artist doesn't need a formal grammar in order to create. People have been navigating these relationships for as long as we've made music, images, stories and movement. Formalization becomes useful when we want to compare mappings, communicate an intuition to someone else, generate variations deliberately or discover where two people who thought they understood one another actually made different choices.

The formal structure shouldn't become an authority over the intuition that produced it.

If the map says indigo and the painting needs yellow, use yellow.

Then ask why.

Perhaps the original mapping was poor. Perhaps context changed the relationship. Perhaps the exception reveals another variable we weren't representing. Perhaps nothing general can be learned from that particular choice.

Reality gets another turn.

What AI changes

Artificial intelligence makes these experiments unusually easy because a model can regenerate forms rapidly under changing constraints. We can give several models the same description and compare their outputs. We can vary one property while holding others relatively stable. We can ask for mappings across language, visual composition, sound descriptions, movement and spatial organization.

This creates a large experimental surface.

It doesn't turn the model into an instrument that detects hidden universal patterns.

Models have their own histories. Their outputs reflect training data, architecture, prompts and the cultural associations embedded in the material from which they learned. If several models independently associate a concept with the same color, that might reveal something interesting about the concept, or it might reveal something interesting about their training distributions.

The distinction is exactly what makes the experiment worth doing.

We can disturb the mapping and see what remains.

The shape before the form

So I would no longer say that the composer is accessing a rule that existed before the music, waiting to be expressed.

Something less complete and more interesting may be happening.

Before the notes, there may be a set of pressures, associations, expectations and half-formed relationships that constrain what comes next without determining it. The first notes give those pressures a form, and once the form exists, it changes the space of possibilities for everything that follows.

The composer responds.

The piece responds back.

Painters, writers and choreographers may recognize versions of the same process without requiring us to conclude that every medium is connected to a single hidden language underneath experience.

Perhaps there is deeper structure there. Recurring mappings across people, cultures and modalities could give us reasons to investigate that possibility.

We don't need to settle it before beginning.

For now, the interesting observation is already within reach. We can feel something before we know how to express it. We can carry parts of that feeling into another form. The new form preserves some relationships, loses others and occasionally gives us something we didn't know was present when we began.

Then we can look back at the thing we thought we were translating and understand it differently.

The shape before the form may not be a hidden object waiting to be decoded.

It may be the beginning of a conversation with the form that comes next.