Stop asking what consciousness is

The interesting question is not what consciousness is. That question is old, honorable and still unanswered. The interesting question is the one an engineer asks when facing an unknown machine that happens to be himself.

Can I inspect and operate my own cognitive architecture?

That is a different move. It shifts the field away from the why, from consciousness science, toward something you could call self-system engineering: the what, the how, the places where you can actually draw a boundary. And that field is surprisingly open. There are excellent individual maps. What is missing is the integrated operator's map that shows which systems run inside you, what feeds them, what they emit and where they interlock.

That is exactly what we attempt here. Not as a theory of mind, but as version 0.1 of a map you sharpen later.

We are not building on a theory. There is none.

This has to be said first, because everything after it depends on it. Science currently holds no generally accepted theory of consciousness. That is not a failure, that is the state of the art.

In 2025 the global workspace theory was tested in an unusually large, preregistered head to head against integrated information theory, with 256 people and three measurement methods at once. The result was precisely no clear winner: central predictions on both sides were challenged. A 2026 review still runs half a dozen theories side by side, global workspace, integrated information, recurrent processing, predictive processing and others.

From that follows the building rule of this map: we claim no truth about the brain. We draw functional boundaries that are useful, and we write next to every field how sure we are. Well established, plausible theory, or open. That third column is not the weakness of the map, it is its backbone.

No clean server diagram

The seductive picture is this: the body has servers and subservers, cleanly wired, with clear responsibilities. It is not like that. The brain looks more like a distributed system that grew over time, never rebuilt from scratch, only rebuilt further and further:

• strongly specialized modules, but with fuzzy edges
• redundant signal paths, several routes to the same target
• recurrent feedback everywhere, hardly any pure one way streets
• parallel processing on several timescales at once
• a few more central points of integration, but no main processor
• and surprisingly few clearly drawn, software like boundaries

That is exactly what makes the question interesting in the first place. Functional interfaces do not have to coincide with biological boundaries. A module on the map is not a claim about a region in the head. It is a claim about a function that can be described, measured and perhaps operated. The map lives in the functional layer, not in the anatomy.

The loop you are running in

Roughly and deliberately simplified, the processing steps hang together like this, from perception through to action and back into the world. Almost every step is a science of its own.

1. Perception. Sensory input plus prior knowledge are assembled into objects and events.
2. Predictive model. The system constantly guesses what comes next and measures the error against reality.
3. Salience. Novelty, danger, reward and body signals decide what matters at all.
4. Attention. Scarce processing capacity is allocated to what matters, often with several claims in conflict.
5. Context construction. Out of working memory, memory, body and environment comes what is relevant right now.
6. Valuation. States get labeled good, bad, valuable or costly. Emotion sits here.
7. Motivation and goals. Needs and values become desired future states and their priority.
8. Executive control. Plan, inhibit, switch, decide. The deliberate steering above the automatisms.
9. Action selection. Competing options and habits resolve into one chosen action.
10. Motor and autonomic output. The action is realized in the body and changes the world.

The real trick is the feedback. The output changes the world, the world delivers fresh sensory data, and that runs straight back into step one. You are not a receiver that takes in images. You are a simulator that runs and lets itself be corrected by sensory data.

The body is part of the simulator

The old order goes: body, then brain, then consciousness. Closer to current research is a closed loop. The brain predicts body states, senses them and regulates them at the same time. Interoception, the sensing of the inner signals from heart, breathing, gut, hormones and the immune system, is not mere body telemetry. Recent work treats it as central to emotion, motivation and the bodily self, and describes large networks for it rather than a single interoception center.

The loop closes like this: from the body to perception, to the model, to prediction, to valuation, to decision, to action and back into the body, then round again. What you experience as mood, need, motivation or bodily self could largely be the felt edge of this control loop. Not an add on to thinking, but its underground.

What is mappable, and what is not

The most common mistake is to treat three completely different questions as one. Kept apart, they look like this.

Problem A: cognitive architecture

How does a human process information? Memory, attention, prediction, decision. Here research is fairly far along, this maps well.

Problem B: self architecture

How does the system build the model of the self? Body ownership, agency, autobiographical self, metacognition. Understood in parts.

Problem C: phenomenal experience

Why does any of these processes feel like anything at all? Red, pain, fear, music. Fundamentally unresolved.

Problem C is the famous hard problem, the explanatory gap. This map does not solve it, and no map does today. The trick is to keep C cleanly apart from A and B, so the big open question does not poison the many answerable ones. You can learn a great deal about your attention without knowing why attention feels like anything.

The self is not a module, it is a stack

You instinctively look for the one place where the self sits. It probably does not exist. Closer to the truth: several self models, stacked, each on a different timescale, from the second to the lifetime. From the bottom up:

• L0, homeostatic organism. Keep this organism alive.
• L1, bodily self. This body is me.
• L2, acting self. I cause this action.
• L3, metacognitive self. What am I thinking and knowing right now?
• L4, social self. Who am I in relation to others?
• L5, narrative self. Who am I over the years?
• At the very top, open: subjective experience. Why does all of this feel like anything?

The open point at the very top is deliberate. What towers over the whole stack, subjective experience, is the point where the functional map ends and the hard problem begins.

Thirty systems, honestly labeled

No claim to biological precision. A first functional cut, every entry with a core function and an estimate of how sure the state of knowledge is. The last column is the most important one, it keeps known, hypothesized and unknown honestly apart.

#	System	Core function	Conscious?	State
01	Homeostasis	keep the body within survivable limits	mostly no	well established
02	Interoception	sense the inner state of the body	partly	well established
03	Exteroception	sense the outside world through the senses	partly	well established
04	Proprioception	track the body's position and movement	partly	well established
05	Perception	assemble signals into objects and events	partly	well established
06	Predictive model	generate expectations, measure the error	mostly no	strong theory
07	Salience	decide what matters right now	mostly no	well established
08	Attention	allocate processing capacity selectively	partly	well established
09	Working memory	hold a few items briefly active	partly	well established
10	Long-term memory	store knowledge and experience	retrieval partly	well established
11	Memory retrieval	reconstruct relevant memories	partly	well established
12	Context construction	build the current mental context	partly	functional model
13	Valuation	tag states with value and cost	partly	well established
14	Emotion	coordinate whole organism states	partly	well established
15	Motivation and drives	energize and direct behavior	partly	well established
16	Goal system	hold desired future states	partly	well established
17	Action selection	decide between competing actions	partly	well established
18	Executive control	plan, inhibit, switch	partly	well established
19	Habit system	run automated routines	mostly no	well established
20	Motor system	realize actions in the body	partly	well established
21	Global access	make information available to many systems	strongly linked to consciousness	contested theory
22	Metacognition	judge your own thinking	often yes	well established
23	Agency model	model that I cause this action	partly	well studied
24	Bodily self	model the body as mine	partly	well studied
25	Narrative self	form a coherent story of the self	yes	partly understood
26	Language system	structure meaning symbolically	partly	well researched
27	Social modeling	model other agents	partly	well researched
28	Norms and culture	internalize social rules	partly	fragmented
29	Learning system	change the models from experience	mostly no	well researched
30	Phenomenal experience	it feels like something	by definition	fundamentally unresolved

How to read the state column: well established means a solid empirical base. Theory or contested means a plausible but open model. Fundamentally open means the point where the hard problem begins.

From science to operation

Science asks: what is consciousness? This map asks something else and more practical: can I observe attention, memory retrieval, emotional state, body state, goals, prediction and the running loops, and what happens when I intervene?

Observe. Understand. Intervene. Observe the consequence.

That is the engineer's loop, applied to your own system. It forces no theory of consciousness, it only needs an honest functional map and the willingness to keep known, hypothesized and unknown apart. A complete, practically operable map of the human being does not exist yet. Many excellent partial maps, yes. The operator's map, no.

The next step is version 0.2: turn these thirty systems into a more serious map. For every module the same columns, carried through hard:

• inputs and outputs
• state and timescale
• feedback loops and body interfaces
• failure modes
• conscious or unconscious
• modulable and measurable?
• known, hypothesized or unknown

That turns a collection of thoughts into a research program. Not thinking about consciousness in more abstract terms, but actually building the map, field by field, and staying honest about where the big question mark stands.

This stays a functional cut, not an anatomical or philosophical final word. The state of consciousness research is open at the level of theory. This map shows that rather than hiding it. Version 0.1, a starting point to build on.