The contents of this post:

  • I argue that you can create exotic experiences at will via mere analytical thought. I think you can easily enough learn the weirdest types of synesthesia. Without any substances.
  • This post touches on the binding problem and (a little bit) on the solvability of the hard problem of consciousness.
  • This post describes unusual possible properties of experience. You're free to disagree with those properties, but they are new ideas.
  • I try to give examples of my idea.

My arguments are mostly based on my models. The arguments repeat the same idea 9 times from 9 different angles.

1. Experience is prediction

A simple argument that experiences can be learned:

  1. Each human thought carries at least some experience.
  2. A new type of thoughts = a new type of experience.
  3. "Predictions" are a very sharp and repeatable and verifiable type of thoughts.
  4. Learning a new type of prediction = learning a new type of sharp, repeatable and verifiable experience.

Imagine you can't "experience" sentences, you experience only particular words. You can reason about a sentence only after studying every particular word. I come along and say "hey, I can teach you to predict words in a sentence before you read them".

If you learned this - you learned to experience sentences. Even though your experience of sentences can still be weird and not equivalent to the normal experience.

2. Experiences = situations

Every important situation/context affects, at least a little bit, the way you experience the smallest things. So, in order to create a new type of an experience (A)... you just need to experience A in a new important situation/context! It's that easy.

Smells trigger memories, the perception of time is 100% tied to situations, music easily connects to situations (take any popular tune and people are going to describe memories and elaborate associations: HOME - Resonance, Mareux - The Perfect Girl, Donkey Kong Country 2 - Stickerbush Symphony, Chrono Trigger - Secret of the Forest)... 

You can't separate the perception of people from situations and relationships. Sadly, traumatic events can turn the simplest experiences into triggers. Liminal spaces is the most popular/relatable/developed meme about subjective experience.

I know, "experience is tied to situations/contexts" sounds extremely trivial. Of course it is!

But it's really strange, actually. Why even the simplest experiences link to complex situations? Is there any "pure" experience? What all those situations ultimately link to? Only to each other? How do we gain our first experiences, before we can experience situations? Or do first experiences come simultaneously with situations? Here's an unusual recursive idea:

An experience A is the average of the most important situations where A is experienced.

But you can learn to associate different averages with A, making your experience "multimodal". (Remember this, it'll be important later. But now we'll approach the same idea from other angles.)

3. Fundamental experiences

Some experiences are "different". And some experiences are 100% different on a fundamental level.

Imagine seeing "red" color and "blue" color. (Or hearing a sound and seeing a color.) Those are very simple and very different experiences.

Now imagine seeing a big blue square and a smaller blue square. Each object is a unique experience, but those experiences don't feel very different.

Now imagine seeing a red circle and a blue square. Those objects are very different experiences. But it's hard to feel the essence of the difference, because the overall difference is overshadowed by the smaller differences between specific features (color and shape).

...

That was the set up. Now the punchline argument:

Each experience is fundamental and simple.

If you can pay attention to that experience. But your attention depends on the importance of the experience. You can't pay attention to experiences which feel unimportant.

So, in some sense you experience billions of the most crazy experiences. You just can't pay attention to them.

The importance of an experience can depend on your ability to predict this experience or "derive" it from other important experiences.

If you can't predict/derive that a square should be red, then you can't feel "a square + red" as a fundamental and simple experience.

But you can feel, for example, "time + a circle" or "time + a line" as simple and fundamental. Simply because those associations are important enough. You don't even notice how crazy they are. However, in different possible worlds you could also experience "egg-shaped time" or "time with a chocolate taste". If only there was enough context to make it important. (Related: Ideasthesia.)

So, you are essentially experience-omnipotent. You can conjure any impossible experience on a whim. You are just not used to acknowledging it. Partially for a good reason (you care about reality and other people). Partially for a bad reason.

For example, you probably never tried to experience other people in a simple way. Because you weren't interested enough in their personalities or were too busy judging their actions. So you can experience such concepts as "a prick" or "a Karen" on the gut level, but can barely experience the uniqueness of your friend's personality on the same gut level. Or maybe you got interested in substances and "forgot" that experience of other people is much more fascinating than your substance experiences.

If you can't experience other people on the level of simple experiences (like sounds, shapes, colors), then there's something stopping you.

4. Meta-experiences

Imagine that each experience is associated with an output of a function, e.g. "color(X) = blue" (says that the X pixel of an image is blue).

  1. You can experience "color(X) = blue". This is "direct qualia".
  2. You can experience any important algorithm/prediction which yields the "color(X) = blue" output. This is "meta qualia". Yes, this is a very strange recursive axiom, so give it a thought.

By "to experience" I mean "to experience as a fundamental and simple experience" or "to experience as a singular experience".

Imagine a blue bird (e.g. bluebird) and a blue sky. Those objects convey different types of "blueness" because they're associated with different predictions about the "color(X) = blue" output. With the sky the prediction is "whenever I look, I'll see blueness" and with the bird the prediction is "if I look at a specific spot, I'll see blueness". You can notice this if you're sensitive to "meta qualia". (And you can become sensitive if you want to.)

You can keep creating new types of "blueness" if you find new important algorithms/predictions which use the "color(X) = blue" output.

"color(X) = blue" is the average of all important algorithms which yield this output.

You don't experience "color(X) = blue", you experience the algorithms which use it.

And you can learn to associate different averages with "color(X) = blue", making your experience multimodal.

Experiences are fractals

So, the idea above is that (meta) qualia is somewhat like a fractal: "color(X) = blue" is both a single experience and an infinity of situations where it's experienced. It's both an output and every algorithm which makes this output. It's both a branch and the entire tree.

And "color(X) = blue" can feel different in every algorithm, in every context (multimodality). It's as if the output of an algorithm encoded the memory about the entire algorithm.

I know, this sounds really weird and like magic. But I think that's cool because it's an original paradoxical idea.

  • And it's not as far-fetched as you might think, there's a similar idea in deep learning: Iterated Distillation and Amplification (and here). You can take an algorithm which uses a function (A) and try to squish the entire algorithm into the original A.
  • All those ideas about equivalence also could "solve" the open-question argument and the paradox of analysis. Instead of qualia those paradoxes/questions talk about "meaning", but you can treat "meaning" as an experience and analyze the topic using my idea.

5. Glued experiences

You can say there are three types of experience:

  1. Simple experience. Based on a single feature.
  2. Combined experience. Based on many features. But those features don't feel too connected.
  3. Glued experience. Merges many features into a single simple feeling.

Examples of all three:

  1. "Being long". This experience is based on a single feature (length).
  2. "Being long and blue". This experience is based on two features (length and color). But their combination feels a bit arbitrary.
  3. "Being stretched". This experience is based on at least two features (length and the object's normal proportions). But their combination feels simple and meaningful. "Being fat" is another example.

(1) A combination of features becomes "glued" when it's important enough. (Abstract and frequently encountered features are often important.) (2) If you add two glued experiences together - you'll get a new glued experience (e.g. imagine something "fat and stretched": a pretty distinct sight). Because a combination of two important things is important or noteworthy. (3) If you add an important feature to a glued experience, you'll get another glued experience (e.g. imagine a stretched sphere: "being a sphere" is an important abstract feature). Because of the same reason: important + important = important or noteworthy.

Stretched, fat, stretched + fat, stretched + spherical (mostly by DALL-E)
More stretched and fat qualia: DALL-E, Miyazaki's Laputa, Jacek Yerka

In this model it's easy enough to look for and create new fundamental experiences. You just need to explore cool properties of things and care about them.

You may be asking yourself: "how do I know that glued experiences exist? how do I know that they're related to qualia?" Consider this:

  1. "Some combinations of features are more important than others" is an objectively true idea. A statistical fact of sorts.
  2. This fact has to be related to qualia at least in some way. And in what way can it be related?
  3. The idea about glued experiences is one of the simplest ideas.

I think my idea is obvious in retrospect. Like something you knew all along, but couldn't notice.

Look at the end of the post for more interesting examples.

6. Abstract experiences

Here's another paradoxical idea:

There are "abstract" experiences which can be realized by different "specific" experiences. For example "abstract happiness", which can be different for different people (or different for the same person).

And there are history-dependent abstract experiences. For example your identity: no matter what person you become, you still count as "you". In different worlds very different people are "you". To determine who counts as "you" we need to look at history.

If you can experience different versions of experience A (multimodality), A itself counts as an abstract experience.

The concept of "objects" which can give you different experiences (e.g. a cube which can be looked at from different perspectives) is about abstract experiences too.

Does it make sense to call abstract experiences "experiences"? I think yes. (Even though we could call them "concepts".) Just because it makes sense to humans to talk about abstract and "shared" experiences (happiness, sadness, love, friendship and so on). Just because it's extremely important to human values. And this way we get a connection with the idea about "multimodal experiences" (meta-experiences). Also, abstract experiences directly affect specific experiences: if you're aware that you experience something familiar, it affects your experience.

The set of any distinct experiences can become an "abstract experience". If this set is important enough. Therefore it's easy enough to seek/create new "abstract experiences".

Ethics and consciousness

How can we judge/assume similarities between experiences? How can we make sets of experiences and reason about them? Maybe you could call this "meta binding problem". This meta problem asks you what creates experiences which nobody actually experiences, such as "abstract happiness". This problem is also important for ethics. Is "aggregate happiness" a meaningful concept? What is the ethical status of two people having a (nearly) identical experience? What is the ethical status of a single second of conscious experience? What is the meaning of experience which you forget? What is "identity" and what is death (teleportation paradox)? What is "society" and what is a good and just society?

"Seeing red" and "remembering seeing red a second ago" and "imagining red" and "seeing red and thinking "I'm seeing red"" (having a higher awareness) are all different experiences. Without the concept of "abstract experience" you can't connect them. So, in some way "abstract" experiences are more important than "specific" ones. The idea of "abstract experience" is the default ontology of your mind. It's kind of required, even, for your consciousness to exist and be coherent. You can reason about consciousness without this concept, but consciousness needs this concept internally in order to function properly.

7. Experience = probability

The next bunch of ideas:

Experiences behave somewhat like probability. I.e. they have to add up to a limited amount.

Experiences behave somewhat like simple agents. An experience wants to get more of a limited resource. The "limited resource" can be your attention or the amount of features to which the experience corresponds.

The way experiences split resources can change them.

What are the reasons for this? (1) Experiences are just combined features. It's at least approximately true. Probably our mind avoids (or makes less important) experiences which combine mostly the same features. It wouldn't be economic otherwise. (2) Bouba/kiki effect and Halo effect. Two experiences split different associations between each other equally, as opposite poles. Experiences tend to group together. (3) There exist "proportion values". I think they correspond to laws of experience. More on that below.

What is a "proportion value"? Imagine this statement: "To me Duchamp's Fountain is art as long as at least 50% of art is not like this." This is a proportion value. I speculate that such values are reflected in experience:

Many experience Duchamp's Fountain as art, including me. However, if most of art were like this, then I would (a) stop experiencing it as art or (b) my perception of art itself would change.

So, if you want to create a new experience, you can just change the proportion of things in an important set.

Some places feel very different when they are not filled with people. This is an example of a "proportion" which affects people's experience. See "dead malls" and "liminal space aesthetic".

By the way, you can say that proportional values introduce a new type of probability. There's probability for predicting the world, probability for describing the world (fuzzy logic, fuzzy set membership)... and probability which is a combination of both (illustrated by proportional values).

Trades between experiences

Distinct experiences don't want to be similar (kind of like people: it's not fun to be 95% unoriginal). So, they have to merge or become more different. However, there's a third possibility:

Similar experiences can split into different "groups". So that those experiences are not activated together most of the time. It's like similar people living in different universes, planets, cities, homes...

Experiences fight for resources and for places in different groups.

8. Experience is language

I think it's convenient to split thoughts (mental states) into two types:

  • Subjective experience (qualia), undefinable properties. Deals with relationships which "exist" in the world of your senses. Like when you see or imagine a cup standing on a table: you can say "the cup on the table" is a relationship which objectively exists in the real world or on your brain's "drawing board". Let's call it a "real relationship".
  • "Language", objective connections between tokens ("words") which can be easily defined. Deals with relationships which do and don't "exist" in the world of your senses. Like "I belong to the class of humans": you can't see, hear, smell or touch this relationship. Let's call it an "unreal relationship".

Properties of experience itself (e.g. "pain is unpleasant") can be called "hyperreal relationships". Qualia deals with them, language doesn't.

There's stuff beyond language and stuff beyond qualia.

However, there are some strange facts and possibilities:

  1. Maybe "linguistic qualia" (qualia created by unreal relationships) exists. If it doesn't exist, how can you be aware of your thoughts? (Which Mental States Possess Qualia?) Also, you can be a Boltzman brain which has experienced only a second of qualia. Most of your experiences can be illusions, fake memories.
  2. Even if "linguistic qualia" doesn't exist, it seems like some very important qualia can be reached only via unreal relationships. Think about "love", "empathy", "respect". Such feelings are based on unreal relationships with other people.
  3. "Relationships" don't actually exist in the real world, all relationships are "unreal". And how can you distinguish real and hyperreal relationships? Given the possibility of solipsism/"cogito ergo sum" situation (only hyperreal exists).

I think qualia is similar to / hard to distinguish from / entangled with language. I think the 2nd point is evidence that you can and already did create new experiences via language.

Experience is truth

I think experience relates to truth in the four following ways:

  • You can experience truth. (Sensitivity)
  • You can reveal truth by changing your experience. (Exploration)
  • You can make something indistinguishable from truth by changing your experience. (Desire)
  • You can modify those statements by changing your experience. (Meta)

So, I think that (1) there's no hard separation between those four ways, (2) you can't always distinguish them even if they're distinct, (3) you may need to use one way in order to get the other (e.g. have strong desires about truth in order to be sensitive to what's actually true).

Do infinities exist? Do numbers exist? Do probabilities exist? Do moral facts exist? Do identities exist? Do natural kinds exist? Do high-level properties exist? What epistemology is the best? Do epistemologies exist? How much of your experience exist? Studying those questions you may notice that sometimes it's impossible to tell apart truth and experience. Truth is entangled with experience and experience is entangled with desire, intentionality.

9. Experience, reality and language

Definition: by "language" and "words" I mean any word of any language, any signal of any communication system.

Let's define the worst (most undefinable) and simultaneously the best (for "qualia maximalists") version of qualia. I think subjective experience has 8 following properties: (see my arguments in this post)

  1. Any real thing can be turned into an experience. (e.g. any physical or mathematical structure)
  2. Any experience can represent a real thing. This idea is indistinguishable from truth. (If you apply the same idea to words and languages, you get something like mathematical universe hypothesis or other ontological maximalism.)
  3. Any experience can have any objective meaning (correspond to anything real).
  4. Any experience can combine any amount of experiences of any kind. Put it another way, any experience can have any property. (E.g. any experience can feel "fundamental" or "simple".) Within the limits of logical possibility.
  5. Any set of experiences can be an "abstract experience".
  6. You can have a process in which experiences "create and destroy other experiences". This idea is indistinguishable from truth.
  7. Any experience is always indistinguishable from the immediate reality. Unlike words: if you read the definition of a "unicorn" (a non self-descriptive word), you understand that you're not encountering a unicorn. Even if you think that unicorns exist because they're definable.
  8. 99% of experiences are indistinguishable from words (concepts and memories) from the inside. Therefore, any crazy experience you can describe is possible. Any crazy experience you can desire is possible. If you could believe in having an experience, you could have it for real. This idea is indistinguishable from truth. And 100% of experiences are indistinguishable from words from the outside.

First 5 properties are properties of language (just replace "an experience" with "a word"). Last 3 properties are unique to subjective experience.

I think subjective experience has the properties of language. Experience is the bare essence of any possible language. Experience is what would happen if language were a physical phenomena. If we lived inside of a language.

If you want to explain subjective experience and solve the hard problem of consciousness in the most general case, you want to deny some of the properties above. But those properties are somewhat equivalent and desirable. And if the properties above are true enough, you can easily learn exotic experiences. Because language and experience are similar: you can just make up a new word/experience!

Conclusion

If the arguments above are somewhat true, it means you can create experiences via analytical thoughts. Including very exotic experiences.

Now I just want to share examples of "glued experiences" I promised and some general opinions.

Part 2

Two aspects of a theory

I think there are two aspects of a theory:

  1. Describing dynamics of a process.
  2. Putting a process into a new context.

I think a good theory requires both aspects. In this post I tried to (2) put the concept of "experience" into unexpected contexts and (1) give ideas about possible dynamics of experience:

  • How experience develops based on important predictions/algorithms.
  • How experience can "glue" features.
  • How experience can be probability-like.

I described unusual possible properties of experience. You're free to disagree with those properties, but they are new ideas.

Importance of experience

I think those are three main reasons why subjective experience is important:

  1. Qualia creates all meaning and value. (Including experience-independent meaning and value.) Without qualia all information would be just a meaningless noise. And understanding qualia is also important for understanding cognition.
  2. Some qualia are related to experiencing other people. And other people is the most important thing in the world.
  3. Qualia is not only about experience. It's also about understanding a large and unexplored part of reality. Qualia shows that we can experience tons of real-world "feature clusters" which we can't even begin to conceptualize analytically. For example, you can't define how a dog looks, but you can easily experience it.

And there's the 4th reason I believe in: I think analytical intelligence and experience are ultimately equivalent. They don't have to go together (maybe), but they work according to the same underlying principles.

I think for people interested in studying experience one danger of psychedelics is this: psychedelics can make you forget that experience is also about real properties of real things. And about things which are not often thought of as "experiences", e.g. about your perception of personalities of your friends.

Examples of "glued experiences"

Here are example of "glued experiences" which should be new to you.

I'm going to talk about paintings by Jacek Yerka. I'm going to interpret those paintings as real places (e.g. 3D videogame levels).

Those paintings show "green places". Some of them are fat and long and bloated (the lower picture). Other have flat surfaces and thin towers (mostly the upper picture). All have smooth ground (no mountains).

If you find those properties important enough (for places), then you should experience them as "glued". I.e. experience them as qualia.

Also, 99% of the places are literally green. The default assumption is that color is arbitrary, but you can empirically find that it's not arbitrary. Your associations with the color may convey information. For example, if you think about a green lawn, you may realize that most of those places are flat or smooth (lack mountains). Or maybe you think about mustiness and swamps and other green spooky themes.

Another factor: in the real world, different places are going to have different lighting. Therefore, the same color is going to look different in different places. So, pay attention to the literal color!

Those paintings show "blue places". Blue places are focused on depth/volume.

  • In places 1-2 volume is created by a border and a pit.
  • In places 3-4 volume is created by an "open" border + water/cosmos.
  • In places 4-6 volume is created by an abstract space itself (underwater or cosmos).
  • In place 7 volume is created by a small-scale border + a small-scale river.

Volume/depth created by a place is an obviously important property (imagine standing in a plain field; then imagine standing in a giant cube). "Absence/presence of giant pits", "abstract spaces", "different size categories" - are all important concepts. Important for places.

Therefore, at least some combinations of those properties should create qualia.

Places 1-5
Places 6-16

Those paintings show "red places". Red places are focused on height of human-made (human-inhabited) structures. By the way, a single place can have multiple colors.

  • Places 2-5 have a human-made tower in the center.
  • Places 6-11 have brute human-inhabited towers in the center. Place 3 can be interpreted as a tower (it's narrow enough).
  • Places 12-13 have solid-ish surfaces standing on human-made "towers".
  • Place 14 is a very small human-made "tower". Place 15 is just similar.
  • Place 16 is a giant wide human-inhabited "tower".

"Being human-made" is an important property, since we analyze places. "Focusing on height", "having a solid surface" are likely important properties for places too. Different size categories are important for analyzing places. Therefore, at least some combinations of those features should give a "glued experience".

Those paintings show "white places". White places are focused on something small and singled out (mostly the upper picture) or on human-made structures which are self-contained, not mixed with any land (mostly the lower picture). Those properties are obviously important, so they should give a "glued experience". Given the complexity of the world, you should have some qualia to detect such a simple property of such an important concept ("a place"). However, every place realizes its "color" in a unique way. White color is connected to red color.

Grey places. They are about human-made flat surfaces or something like that.

Cyan / "cosmic color" places. They are either (a) self-contained places which can be split into a lot of parts and/or (b) small places on the sidelines, near the edges.

Yellow places. I'm not sure. One idea I have: yellow places are focused on something in the center of a field or forest or other landscape. I can't always explain the difference with the white color.

Do those combinations of properties give you any unusual experiences? Feel free to comment and disprove my ideas. But at the same time try to understand why/how each property can be important.

New Comment
6 comments, sorted by Click to highlight new comments since:

mmmm... sort of, but not all experiences can be reached by the thoughts your particular network of cells can have. eg, an exotic experience you can't have: you can't tell your skin to drastically change shape; the thought cannot be encoded into the neurons you have. despite that your skin can probably be told how to change shape electrically, see michael levin's work.

I agree, there are physical limits. But this doesn't really matter for the post.

The post claims there are a lot of crazy experiences which are within reach. Not even counting altered states of consciousness. And the post studies the question in general, explores what experiences are possible in principle (for any possible being). Not everything detected by a body gives a unique qualia. Or it's not obvious if this is true or false. (Which Mental States Possess Qualia?)

you're merely dissociated from your skin cells' detailed qualia; they're still in the business of having qualia about what shape to take.

Why do you think so? That's not a popular position. The post contains arguments for something similar, but I suppose you believe that for different reasons.

hmm. Rambling mode, activate!

I already believed something akin to panpsychism, but with a weighting function towards higher interaction processes of some specific kind I'm not sure I know how to mathematically identify, but which relates closely to integrated information, signal to noise ratio, compressibility, agency. I suspect I'm looking for some relationship between those, and of course some of those need additional parameters that an ideal form of the concept selector wouldn't. it has to accurately match my brain-internal qualia, while also having strong reason to believe it generalizes. Because intelligence arises from a series of simple objectives, a fundamental component of which is usefulness per watt ie a signal to noise ratio towards an agentic target, we can expect that anything that contains optimization will contain some magnitude of consciousness or agency, by generalizing from the fact that our neural substrate seems to.

The activation maps of artificial neural networks look like the maps of qualia would in my brain; because the internal steps of the behavior share structure and we share the same physics, I was already granting artificial neural networks "qualia" (but not "personhood", which I believe arises from counterfactual maximum agency or so).

when I saw michael levin's talks on bioelectricity, I came to the conclusion that non-neural cell-level patterns also are qualia, and are simply not integrated with the main qualia network. something approximately like, qualia is the feeling of increasing signal to noise ratio by distilling locally intentional, intelligent behavior out of our surroundings and into the networks of interaction between neurons' weights, synapses, neurotransmitter types in dna. our skin (and entire body map) does a similar job but distilling only the local shape of skin.

evo devo talks about the development from a relatively static perspective, but levin's work shows that it is in fact highly internally communicative and dynamic, which means there's a lot more distillation and noise-rejection occurring at runtime; noise rejection, being the basis of intelligence in our own brains in ways that we can observe fairly clearly from neuroscience, gives me significant confidence that our entire body is comparable to neural cellular automata in the way large transformer networks are very vaguely comparable to the brain - probably more vague of a comparison than the NCA one, honestly!

Thank you for sharing your perspective, even though most of it was too technical to understand for me.

I already believed something akin to panpsychism, but with a weighting function towards higher interaction processes of some specific kind I'm not sure I know how to mathematically identify, but which relates closely to integrated information, signal to noise ratio, compressibility, agency.

As I understand, this is your expectations about properties of qualia: some unknown process + integrated information + signal/noise, compressibility, agency.

I believe my post should give you new information (if the post is true) and new ideas.