Roll Your Own Octarine


Topics: Origin of feeling, different categories of feelings/qualia, genre as a model for the feeling of self, art skill/process as analogous to self.

Epistemic status:


If you were someone else instead of yourself, I think most would say that it would feel different to be that person, even though, in any given moment where you’re someone else, the combinations of emotions you feel could be the same as any you’d felt when you were you. If so, why would it feel like something to be someone? Why would being different people feel different?


Imagine setting a swarm of machine-learning-machines1 with random heuristics for movement on a grid. You’re selecting each new generation based on the ones that stop on top of a red sticker you put in the top-right of the grid. There are two ways for them to succeed: either they learn to move rightwards and upwards, or they learn to learn that they have to stop on top of the sticker.

Similarly, every adaption of humans are either 1. implicit knowledge of the environment or 2. tools for extracting and storing knowledge from the environment.

Note that these aren’t seperate categories in evolution’s eyes- that we’re given tools to learn from the environment encodes the knowledge that learning from the environment is useful. The feeling of curiosity embodies this (encouraging knowledge-gathering because knowledge-gathering is useful.)

So if curiosity- and other emotions- count as knowledge instead of tools for gathering knowledge, what does count as a tool? The obvious contrast to emotions, reason comes to mind, but further thought places sight in this category as well. But even then there are categories unaccounted for- where would physical pain or taste fit into this model? Like emotions but unlike sight they “feel like something”, but they otherwise seem easy to group with sight as being senses.

Let’s return to curiosity- it “feels like something” because it needs to encourage a certain action. Pain and taste, then, are the same way (encouraging avoiding/fixing harm or encouraging eating certain things and not others). The difference here is that curiosity and other emotions depend on context, while pain and taste don’t, just as sight doesn’t.

But then what of reason?

Certainly reason isn’t a sense, like the things that we’ve considered direct to be, and so this seems like the only place it can go. Can we be more convincing?

To put the diagram into more words, reason is like sight, but not reliant on the external environment. Is this true? Consider that though reason as an inner monologue may be the popular depiction, visual thinking is common, and so we here we have our process, guided/affected/controlled by the self instead of the external environment, which is more or less exactly like sight.

Don’t get me wrong- obviously seeing something in the external environment can make us think about it. What matters is how far downstream it is in the chain of environment → senses → self → feelings/thoughts.

And finally the self shows up. Can anything useful be said about it?

Perception and Action

Firstly, though I originally considered labelling thoughts/feelings (the non-sensory category) as contextual, I discarded it because that’s only true by one interpretation of the word- it’s not that senses lack a “context component”, but that this component is the external environment. From that point of view, self can be seen as an internal instead of an external environment. “It feels like something to be me” may then be similar to something like “it feels like something to live in a city”. It seems to me as if this involves trying to filter out both the self and any particular experiences- if f(s, E) outputs an experience given the self and the set of variables that is the environment, then “what does it feel like to live in a city” might be asking for description of the difference between the image of f and the image of f when one of the values in E is given. Alternatively, just call it f(E) and say that one of the values in E is s. “What it feels like to be me” is then the same procedure.

Secondly, recall the idea that every cognitive feature is either implicit knowledge of the environment or a tool for extracting (and storing) knowledge of the environment. How feelings feel is constant (presumably anger is anger to anyone like red is presumably red to anyone), so they’re implicit knowledge of (what motivational tools will be needed in) the environment, but if it feels different to be different people then selfhood must have something to do with extracting/storing knowledge of the environment. Thinking in that vein, certainly memory (stored knowedge) does seem to have something to do with the self. But it’s not a 1:1 correspondance- we don’t hurriedly remember/compare our entire life histories to the current situation whenever we act. If our experiences shape us to a significant degree then it seems like the information gleaned from them must be compressed into instinct, much like or the same as how we don’t have to consciously think about precisely how to walk when we walk unless something unusual happens.

Genre and Art-making

So we have something like “what it feels like to live in the city”, which seems something to do with perception and aesthetic, and we have something like walking, which is something like skill or procedural memory. Metaphors will be used to illustrate these two mental processes more clearly.

Firstly, genre. Fairy tales, murder mysteries, cyberpunk- all different genres with different feelings to them. Genre is more than just a label, providing a sense of how to understand the material, eg., a superhero movie and an “artsy” movie are meant to be approached differently. (To support this, consider firstly that in general beliefs about what’s expected change perception, and secondly and more specifically, that genre subversion is interesting because genre inviting us to perceive the work a certain way sets us up for thrilling lostness.) Placing the work our mental map of genre, that is, determining how to perceive a work, is based on previous experience and knowledge of genre. Then, similarly, how a person perceives the world is based on previous experience and knowledge, and like with genre, different perceptual approaches feel different.

Secondly, an apparent companion metaphor for the skill part of selfhood might be improv. Acting exists despite people being no more able to consciously simulate an entire psychology than we could calculate the physics of walking.

In this way it’s instinctive the way being ourselves is instinctive. Can we connect this with genre? Well, we can imagine the actors given a prompt to improvise scenes/stories of a certain genre. To follow our metaphor back to being about selfhood, this means that our instinctual reactions are influenced by perceived success criteria.

But maybe this metaphor is unnecessarily convoluted as an aid for understanding, or perhaps it seems that by talking about people pretending to be someone else we’ve snuck in some analogue of selfhood and so muddied the data of the thought experiment. One might also protest that one is certainly not just a creature of instinct but also of reason, and this metaphor doesn’t seem to lend itself to clarity on that point.

Here I would point at the example of skill in visual art. Drawing is a process that involves reason/is consciously guided. One can choose what to draw and adjust how to draw the drawing while drawing it. Drawing isn’t a process that takes place fully in the conscious or subconscious mind, but is an interplay across both. Art skill is like a complex tool which provides feedback when used and develops differently depending on how it’s used/the training data it’s given. (Of course, if the heavy lifting is done unconsciously, one could flip perception of tool and agent and say that conscious involvement is just a tool a that art skill uses to “see” and adjust to criteria for success that it wouldn’t have known about otherwise.)

Looking further into using this metaphor, artists view art differently than art novices. To quote the linked abstract: “Artists may extract visual information from paintings based on high-level features … such as textures and composition of colors”. This seems to be related to the idea of genre sense as being/changing perception- there is some mental model/map to fit lower-level features to.

A possible point where this paper doesn’t seem to map to anything is that it compares perception between two groups instead of between individuals even within a group. I don’t think this makes it inapplicable, however. Consider that different artists have different styles. An artist laying down a line in a drawing can be broken down into the artist on some level perceiving where the line should be and then placing it, and so if it seperates artists and novices/laymen and by extension their art, then it seems plausible that variation in style is also linked to differences in perception.

New Feelings

The color red doesn’t exist in the world; only photons do. We can’t consciously process what each rod and cone receives as input- in some sense we might as well be hallucinating in total darkness. Evolution extrapolated constant laws from training data and then served us the “imaginary” higher-level feature: red.

Knowing art principles alone doesn’t make an artist; art principles are largely aids for learning and troubleshooting. Knowing art principles allows one to train one’s instincts on the relevant parts of data until imaginary higher-level features can be perceived.

Next up?: More on perception and action.

Additional links: Thou Art Godshatter

  1. Technical term.