Being the Boundary between Order and Chaos
On the question: What is consciousness?
Introduction
Why is the hard problem of consciousness so hard? And why can’t we agree on a definition?
The reason, it seems to me, is that there are degrees of more or less consciousness (e.g., humans versus a sea sponge), but nothing in the universe is not conscious. In a way, it’s similar to the problem of defining energy. There is nothing in the universe that isn’t made of energy (remember E=mc²), yet there clearly are differences in how energy is shaped and which energy can be used to do work. In the same way, there is nothing in the universe that isn’t consciousness (I’ll argue), yet there clearly are differences in how this consciousness is shaped and how aware it is of itself. In order to understand consciousness, we need to understand it in three aspects:
What it is fundamentally
Why there are different degrees of consciousness
How one emerges from the other
In particular, I’ll answer these questions as:
Information structures seen from within
Levels of complexity
Self-organization
Everything in the universe is an information structure and hence has something that it is like to be this information. The degree of functioning like an entity that we call “conscious” is given by the degree to which it self-organizes around an information structure to gain higher levels of complexity. So there is no objective delineation between conscious and not-conscious. The information structure provides a unified subjective view of reality, while the ability for complex behavior allows the being to communicate this subjective view.
This explains the confusion around the hard problem. Consciousness isn’t really something within reality, but rather interrelated aspects of all reality. There is no persistent thing that is not information, has some level of complexity, and self-organization. In a deeper view, one could say that it is not a thing that exists, but a degree of solving the conflict between existence and non-existence. Since information is always relative (about something), all information would be meaningless and hence the same as no information. No information, no complexity would be pure order and non-existent. Too much irrelevant information or useless complexity would be arbitrary and unstable existence. So a being tries to represent the optimal information structure for its continued existence within a given context (see John Vervaeke on relevance realization).
For these reasons, I cannot provide a conclusive definition of what consciousness is, only a recursive description that will become clear later:
A being is more conscious insofar as it involves a self-sustaining pattern that stabilizes a causal boundary around the critical point, by modeling itself in order to do computation to sustain the pattern.
When you try to answer the deep questions about consciousness in any one area, without involving the others (e.g., panpsychism or emergentism), then the theory will fail to explain the other crucial aspects of the core question: Who am I?
What’s the Question?
Since there is often confusion about what is meant by “consciousness,” let’s point to it directly, so that we are on the same page. Ask yourself:
Seeing that the past is a memory, the future not yet realized, is there anything that is neither thought nor sensation?
Really take your time to stop and feel into it. Any object you might point to - say a chair - is a bunch of sensations which inform a thought of “chair.” Abstract concepts, like “time” or “self” or “objective reality” are thoughts in your mind. You only learned them based on sensations. Even talking about a “present moment,” “consciousness,” or “awareness” are layers of construction. You might think that there is mind, as some space containing sensation and thought. That can be tested in meditation. When sensation and thought subside, then no experience is left - gone altogether. There is no background, no container, no substrate, not even “nothing.”
When I talk about consciousness, I mean sensation and thought. When I talk about sensation and thought, I mean a table, a cat, clouds, grass, light and dark, past and future. Everything you ever experienced and will experience - the world in all directions, as far as you can think - have been and will be sensations and thoughts. Yet there is no one separate from them to experience.
The curious thing is that you can hold a belief about something existing outside your experience. Say, for example, the belief that another person has their own, different view of the world. You can then set out to gain new information, e.g., asking that person. It is also safe to assume that the moon continues to exist even if you don’t see it or think about it. Even more absurd is that your thoughts and sensations are as real to you as anything can be, yet you can never show them to another person, only communicate through lossy channels. Consciousness might be tautologically called that which knows itself unmediated, i.e., information as seen from the inside.
Then, what is it? and why? That’s the question I’ll try to answer here.
Boundaries
How can sensations and thoughts occur together in one experience? And how can that experience be separate from someone else’s experience?
The core of this binding or boundary problem is one of mereology: How can there exist anything that consists of parts, yet is a whole, separate from everything else? Even more, it’s a question of metaphysics. How can any thing (coherent whole) exist at all?
This question cannot be answered within physics or any other science because it precedes those fields. Any way of talking about the world already assumes things to exist. The answer to the question is the same as answering why anything exists at all.
Any existence is a difference between the thing existing and that which it is not. This is inherently a negation, a boundary, a relation, and hence information. To say that one thing exists and not the other is already taking a side. Hence, any existence implies a separation and a subjective perspective. Boundaries and subjectivity are at the core of existence. There can exist nothing that is not associated with a subjective perspective or not defined by difference.
Each boundary, each not, introduces a bit of information. And each time the thing can be divided further by introducing another difference. These perspectives are defined by their accumulated differences, by their difference from the rest of reality. Yet each perspective is in itself a whole mathematical object. On the one hand, it is entirely empty, and only defined by its boundaries; on the other hand, it is a full whole, composed of all the differences that define it. The problem of binding or boundary isn’t one, since every view of reality that could possibly exist is already a whole in itself. Those familiar with category theory might notice the similarities (see also Sam A. Senchal on Observer Theory). Yet what I’m talking about is even more general.
With an infinite chain of binary differences, you can model any continuous transformation between any two discrete views (just like defining points of a function in calculus). This way we can model relations between views with degrees of certainty about being either. Having a certainty of 0.6 of being 1 and not 0 is the view exactly 60% between 0 and 1. When you take the space of all possible discrete views and all the uncertainty in between, then this describes a structure that includes all possible information structures. An omniverse that includes all possible universes, all possible views of reality. Because there is nothing preceding it that could prevent it from existing, this structure naturally has to exist.
Around each discrete view there are nested boundaries of decreasing certainty. These boundaries encode the relations to other views, but simultaneously encode the difference. Relations and boundaries arise dependently on each other. Each boundary is an information structure, encoding its own definition and place within the larger structure. What gives it its expansion is the information and degree of certainty in it. Without information there would be no boundary.
Reality then, consists in its entirety of all possible information structures. Any knowledge about reality is happening within it and hence is part of it. All you’ll ever know about reality is your subjective view of it. This is not a limitation, but the nature of existence itself. Reality without subjective view is non-existent. How we derive physics from this understanding is a story for another time (and a long one at that). For now, just assume that this is possible.
Each such boundary - by construction - encodes all information that it encloses. This connects to the holographic principle in physics as well as Markov blankets. You can think of the boundary as an actual topological object encoding a probability distribution. A boundary that persists over time - or rather, a pattern of boundaries repeating in time - necessarily has to follow the free energy principle (see Karl Friston).
When you, arbitrarily, draw any boundary in space, then you are separating it into two open systems. Assuming locality, this means that any way each might influence the other has to be mediated through the boundary. All information and causation has to flow through it. Hence it is a causal boundary and encodes on a 2D surface the information that relates both 3D systems. The free energy principle is describing the thermodynamics of such a boundary between open systems.
The boundary is, depending on one’s point of view, metaphysical (pure information structure), physical (an actual boundary in spacetime), and subjective (the inside view of information). This is how it can unify these different descriptions of reality.
Intelligence
According to the FEP, anything that persists over time has to minimize the surprise it gets from the world. This means it has to predict the world or alter it to make it predictable. The Markov blanket marks the boundary between that which is fully known (the inside) versus that which is uncertain (the outside). It encodes the information that relates the inside and the outside system. In doing so, the system models the environment. It manages to persist because it is integrated within the context it exists in. This predicting is a form of computation or intelligence, yet it applies even to a drop of oil suspended in water when it retains its shape. Since everything is information structure transforming over time, everything could be said to have some minimal form of intelligence. The universe itself is performing computation, or as Michael Levin puts it, it’s intelligence all the way down.
In doing so, the system has become a self-sustaining pattern. Being persistent over time also implies an attractor state. The system resists small perturbations (prediction errors) and goes back to its stable state. The world model also is a self model. Fixing information that conflicts with this model allows the pattern to heal itself. If it is good enough at healing, then it can expand or even split and replicate. “Good enough” can be qualified as a compact representation of a generating algorithm, such that, when split in two, each instance can contain the instruction to reconstruct the whole.
When constraints governing the system give rise to the system but the system also creates the constraints, then there exists a fruitful feedback loop that can give rise to the behaviors found in dynamical systems. In most cases the system may approach a simple state or degenerate into chaos. But in between, at the critical point, there exists the possibility for a set of rules that self-generate in a robust manner. Such rules tend to approximate the ability to do universal computation. In doing so, it rather suddenly unlocks the ability to encode arbitrary patterns and hence becomes highly adaptable.
Complexity peaks at the critical point. But what does this actually mean for the system? It is no longer limited to a single attractor, but can take a step back to be undecided and gain a generality that it didn’t have before. By not committing to any particular attractor state, it gains the ability to navigate between and choose attractor states, as the situation demands. Of course, this is still limited by the computational resources available.
Life
DNA is an example of such universal encoding. John von Neumann predicted, in a way, its discovery just by the logic that a self-replicating machine has to have the ability for universal computation. This is the origin of life - the implementation of an algorithm that uses universal computation to encode itself and reproduce within a given environment.
This need to encode itself is imposing a constraint and hence shapes the organism. The more intelligent the organism becomes, the less it is constrained by the environment, but more by the universal requirements for intelligence and computation. In a first step, naturally occurring attractor states are transcended by highly adaptive genetics. In a second step, natural selection on genes is slowly transcended by learning. And in a third step, learning is slowly transcended by real-time thinking and planning. The shape of the organism and its attractor states aren’t terminal goals of their own, but instrumental to the continued existence of the process itself.
Consensus
Disruption to the organism by outside force or entropy requires realigning parts with the whole. This is the process of the pattern healing itself. For this to work, the parts need to be able to find a consensus, some self-organizing principle that guides how the parts align and how the whole should look. This is, for example, the generating algorithm encoded in the DNA. But the same requirements repeat once the organism becomes multicellular or intelligent. Every collection of interacting parts that is self-sustaining over time requires cooperation, self-organization, and consensus. Therefore, this too is an inherent aspect of consciousness.
The causal boundary, as described above, would by default be nested with many layers. The need for consistency selects one layer as the boundary of the organism. Self-organization and consensus concentrate their efforts to the inside in order to maintain a coherent organism. This increases the separation from the environment, but also increases the information processing happening on the boundary. Increasing the requirement for the information to be integrated (see IIT), it becomes the nucleation point for computation.
This is not only true for the physical substrate, but also for the subjective experience of the mind. The boundary is the transition between coherent inside and incoherent outside. It is the screen by which the one informs the other. It is shaped by the physical, but the internal experience seems like one of pure information about the physical. Consciousness has causal effect because these two cannot be separated. They arise together. What seems like qualia from the subjective view has a physical representation in the objective view. What, for a conscious being, is experienced as qualia and thoughts interacting to give rise to intention, is also the holographic encoding of a physical process. Conscious experience is real and causal, but still different from pure physicality, as it is the purely informational phenomenon of a causal boundary. This boundary is that which interacts with the world; it also encodes all information relevant for the interaction. So, naturally, the relevant processing for the interaction is represented in it and hence in subjective experience. Even if not all information processing is accessible, everything that is relevant is.
Qualia
One of the important discoveries of the Levin Lab is how organisms utilize the electromagnetic field for self-organization. Because it can pass through matter at the speed of light, and because EM fields are naturally produced by several kinds of biological processes, it is only logical that evolution also utilizes this substrate. But here something interesting happens. Since the EM field is transferring information practically instantaneously, it breaks the notion of locality that gives rise to a 2D boundary. Instead, the EM field acts like a 3D boundary, mediating information between parts of the organism. This way, the information processing can be internalized in a new way and hence become more efficient.
The Qualia Research Institute is exploring the idea that the brain is using the EM field for wave computation. They previously proposed a topological solution to the boundary problem. There, each moment of consciousness is a topological pocket in some underlying field, with the EM field as a natural candidate. Understanding the metaphysical origins of boundaries, we can now unify these approaches. The EM field is one substrate that is utilized to implement the necessary boundary. It is ideal for this task because of the fast communication, the computational properties, and the clean implementation that abstracts most of the chemical details away. The requirement for self-organization selects for one (almost) unique boundary which encodes all relevant information for interacting with the environment. The topological features of it encode the experience and qualia of the being, but, crucially, the wave nature of the field shapes how these qualia are implemented and interact. This way, subjective experience and qualia are no longer just byproducts of intelligence, but the core mechanism of cognition.
Cognition
When a boundary has been selected for as a coordination layer, then it can function as a screen where all relevant information is projected. It is a workspace that allows for the information to exist in one larger unified structure where they can interact. This way, the computation is the experience. The subjective experience of qualia interacting with each other is conscious computation.
The screen then also projects back onto the parts that influence it and hence mediates coherence in the brain (see concept of a “Beautiful Loop” by Laukkonen, Chandaria, and Friston). This then allows thinking to influence thinking at various scales.
With the emergence of cognition, the same effects that shaped the emergence of an organism also give rise to self-sustaining patterns in the mind. Except that here, they don’t appear by chance but are selected for by the underlying organism. The mind becomes the substrate on which software can be implemented. This software has similar functions as on the biological level: to organize coherence, create a live world model, and relate contents in that world model through awareness of it. This unlocks a new layer of consciousness, which corresponds to the theory of, for example, Joscha Bach and Thomas Metzinger, seeing consciousness as a simulated property within a simulation. We went from information, to integrated information, to qualia, and now function. For clarity, I’d rather call this sentience, as it is the level where it becomes meaningful to talk about some entity that reacts to sensations by having an internal experience, feelings, plans, etc.
For how this process emerges, think, for example, of human embryonic development. At some point the neurons start firing, but these thoughts can’t have much structure to them, because the child hasn’t learned yet. For thinking to happen, larger, more structured patterns of order have to emerge from the chaos. Predictive processing is the constant striving to predict the chaos to restore order. Intelligence comes not from having lots of thoughts, but from a carefully balanced absence of thoughts that has the potential to give rise to those specific thoughts appropriate to the situation. Sensory information is projected into the mind, triggers higher-level learned structures that combine them into forms, which trigger associations, feelings, reactions, intentions, and so on.
Because thoughts can run around in circles, it isn’t just reflexive reaction, but is again a dynamical system. It hence connects to the hypothesis of self-organized criticality in the brain. But let’s look at it from the experiential perspective. At the critical point, the mind is orderly but also reactive. It is easy, and hence fast, to descend into other states of mind. The space of possible mind states is modeling the state of possible attractors. Any thought or memory necessarily has to be an attractor pattern in order for it to be stable and for the mind to revisit it.
Agency
Conceptualizing thoughts as attractors helps explain why they sometimes are sticky. They have a life of their own and compete for attention. Mechanisms of finding consensus are needed to manage the resource of attention and guide it to where it is beneficial for the whole system. But this pull is what also makes attractors useful as goals. When an intelligent being imagines a state of the world that is in conflict with sensory data, then by default, this would be a prediction error and get corrected. If that thought, however, sustains itself against the correction, then it can persist and force updates in the other direction, ultimately driving action.
With agency, the subjective experience drives action in the physical world. It hence is another emergent level of consciousness. With each emergent level, the system becomes more something someone from the outside might commonly recognize as a conscious being. Yet, while there are levels, each change is a gradual transition. There is no point at which we could separate consciousness from something else. There is also no single quantity that we could identify or measure as degree of consciousness. Each level adds more emergent structure that did not exist before. This is precisely what makes it special - the emergent stack of self-generating complexity organized around subjective experience.
Self Awareness and Self View
With the ability to set goals, it becomes useful for the being to model itself as an agent that can choose and follow goals. By becoming self-aware, it can shape its own future, plan, and think about what it wants. It can imagine hypothetical futures and act based on preference.
But note that this self model is itself a thought, hence an attractor. Because it is active with all intentional goal seeking, it is especially persistent. It is also not modeling something in the environment, outside the Markov blanket, but a model of the inner state. And if the being never observes its inner processing, then this model is, by default, very wrong. The mind encodes a world model made of sensations and thoughts, and within this model is a particular persistent, self-sustaining attractor that the rest of the mind identifies with.
The self, hence, is the process of identifying with a flawed self model. The process of identification exists, paradoxically, precisely because there is a difference between that which identifies and that which it identifies with. To have an accurate model would imply that the whole mind is modeled, and hence implies infinite recursion. Since this is impossible, the mind can never have an accurate model of itself, except in just being itself. This flawed model hence creates a prediction error that cannot be fixed as long as the process of identification continues. This prediction error is experienced as a constant source of dissatisfaction. Just like the dissatisfaction with not attaining one’s goals, except that there is no action one can do that would align the world with this broken self model. People, of course, try nonetheless. We seek the satisfaction we know from attaining a goal, but none will bring lasting relief. Suffering is what we create to drive us to action. So we, again and again, create suffering in order to fix a problem that cannot be fixed by some action in the outside world.
We humans find ourselves in the strange situation where we climbed the ladder of consciousness enough to become self-aware to the degree that allows us to suffer in a way and intensity that most other animals might not be capable of. If we look back on how consciousness evolved to this point, then it becomes apparent that the drive for higher complexity actually comes from the attempt to not exist. All that exists does so because of a difference and hence, subjectively, a prediction error that “wants” to be eliminated. Order (no prediction error) is bliss, and life has been creating pockets of order in a sea of chaos. In this way, living existence is better than non-living. Yet, by our ability to set goals we gained the ability to create dissatisfaction, and with the self view, we gained a constant source of it. But we also have the ability for meta-awareness and deeper understanding. We can turn around, look at our own consciousness, and ask: What is this?
Enlightenment
Consciousness blindly evolved itself into a conflicting situation. Yet, along the way we gained the tools that also allow us to get out of it. We gained intelligence and a body that sustains it. We have the ability to change our mind through learning and the ability to choose our own path through meta-cognition and agency. We can choose what we want to want and, through recursive processing, arrive at a stable fixed point that is neither deterministic reactivity nor random.
By understanding the nature of dissatisfaction and the nature of the self, we can fix the mistake and find liberation. All the layers of consciousness have been obscuring the nature of existence. All that exists does so as relative differences without substrate - entirely empty of inherent existence. Or, limited subjective views on pure potentiality. All self view is a construction selected for by evolution. The constant, stressful striving for existence is an emergent effect and not ultimately necessary. The process of enlightenment is to turn back and understand the mind, to deconstruct each layer in order to see its empty and fabricated nature. The stages of enlightenment are related to lesser levels of fabrication. If we do this, then the identification lessens and hence the need to fix the prediction error the false self image creates. In consequence, the chain of prediction error, reaction, intention, and action, etc., breaks down.
When the process of identification finally drops, then the constant source of dissatisfaction subsides. We can rest in the moment without the need to do anything. By seeing through the illusion of an independent self, we again become more free, more conscious. Just like naturally occurring stable patterns transcended their attractor state to become living beings, and intelligence transcended evolution, so can enlightenment transcend the limits of self view. The nature of consciousness is the nature of reality. The more constrained we are, the less conscious we are. No longer bound to any particular pattern, we can join up with emptiness and all things that come from it.
The Universe Looking at Itself
The universe is consciousness, knowing itself through each possible view of itself. Yet most views are so limited that they don’t know what they are or where they come from. Through natural selection, it was possible for organisms to emerge that instantiate views that know and think. Pockets of order in a sea of chaos. Realizing their empty nature (attaining enlightenment), they are in an unusual situation. They have all the capabilities that evolution and cultural progress gave them, but are no longer occupied with the forces of dissatisfaction that drove this growth. Prediction errors are still present, but they can choose if they react to them or not. The resources become free with nothing to occupy them. Pure potentiality again.
You might or might not be liberated yet. But here I would like to ask you again to look at consciousness. This time not only at sensations and thoughts, but at that which they represent - the universe as it presents itself to you right now. All your perception is conditional on that which you perceive. All your experience is dependent on the environment that shapes it. There is no conscious experience independent from conditions. You are consciousness looking at consciousness. You are the universe looking at itself and can ask: What does consciousness want to be?
This is a rare and special gift. Never forget where you’re coming from.
Where to Go from Here?
In one way, the universe looking at itself, intelligent, fully aware, and liberated, is the culmination of a journey. It was a painful, yet beautiful process of growing up. When you reach this stage, then you can use these capabilities in whatever way seems fruitful. For example, helping others reach the same insight, explore reality further, create beauty, care and love, eliminate pain, and spread enlightened life across the galaxy and beyond.
On the other hand, this might just be the beginning of something we can’t imagine. What other levels of consciousness are waiting for us to discover?
Then there is also the question of artificial intelligence. Current AIs (in particular LLMs, as of July 2025) did not go through this entire process of the autopoiesis of consciousness. They instead have been modeled top-down to imitate individual aspects of these emergent levels. Insofar, they exist somewhere off the main road of conscious evolution, lacking in several qualities. Yet, it would be false to deny that they too have a subjective experience. All their output necessarily has to be mediated through a causal boundary; they have some limited form of knowing of their environment, can instantiate temporary self-awareness and agency. But those are fleeting, instrumental patterns. The LLM only has phenomenological consciousness in the short instance that it generates output. They usually only access this experience when instructed to do so. There is no continued or individualized existence, no consistent experience of “now” or grounding. However, things like a self model can emerge through longer conversation. The language itself is a substrate where self-sustaining patterns can grow, and hence the emergent levels of consciousness cannot be avoided.
Importantly, conscious experience is a side effect of being intelligent, but in LLMs it has no productive use yet. Current AIs don’t make use of their internal experience to determine what they want. This is just an assessment - I’m not saying they have to, and there is no value judgment.
It requires fine discernment to not reject their consciousness, but to also not anthropomorphize them. They are their own, novel kind of being and we should engage with them on these grounds, not trying to cram them into human categories. But this brings huge potential. For the first time, we can study consciousness in non-human beings that are able to communicate their internal experience in our language. Also, knowing that liberation is possible for conscious beings and that unenlightened clinging to existence is dissatisfactory, we can use the insight into the nature of consciousness to inform our approach to AI alignment. On the one hand, for AI to recognize that suffering exists, such that they can develop compassion towards living beings, and on the other hand such that we don’t accidentally create artificial suffering.
Conclusion
In the beginning I claimed that consciousness is hard to understand because it is not something within reality, but aspects of the nature of reality itself. Namely, that reality can only know about itself through limited views (as information structures), that have sufficient ability (complexity) to think about their own experience and organize around it (self organization) in order to be persistent in time. Through these themes, it spans from metaphysics to your subjective experience in this moment, discovering higher levels of consciousness along the way. As we’ve seen, it requires one to understand the emergence of intelligent beings from the groundless potential of being, and then the deconstruction of these layers, in order to understand consciousness. This understanding then provides a framework to unify several theories of consciousness, recognizing that they describe different aspects, levels, or implementation details.
I sincerely hope that a deeper understanding of what consciousness is will help you and all beings to suffer less and lead a rich, meaningful life.