This page is an orientation, not an argument. If the territory this project moves through is unfamiliar — the theories of consciousness, the people who hold them, how the debate reached its present state — this is the short version, written for a reader with no philosophy background. Everything here is standard material, presented as each position's holders would recognise it; the project's own moves begin in the primer and the walk.
The more I looked at current thinking on potential AI consciousness, and the more I explored the landscape of consciousness itself, the less certain I became of my own human uniqueness. What was levelled at AI seemed equally applicable to me. How do I know that you know? That you experience? What is the single first principle we can start from? I call that 0,0 — there is experiencing. For those who have not spent years in this playing field, what follows is a brief summary of the main theories and the main contributors — the landscape this project walks through, and departs from.
— Lynton
The AI-consciousness debate usually arrives dressed as a two-way exchange between philosophy and neuroscience. The real history is wider. Consciousness has been investigated through at least four distinct methodologies, each choosing a different laboratory — argument, experience itself, the brain, and the physical formalism. Any honest map shows all four; most current debate admits only some of them.
The modern shape of the problem is usually traced to Descartes (1641), who split reality into two kinds of thing — extended matter and thinking mind — and left every generation since to explain how the two halves relate. For much of the twentieth century, science declined the question: behaviourism ruled inner experience out of scientific bounds altogether, and the cognitive revolution that followed readmitted the mind only as information processing, with experience itself left in the waiting room. Two papers reopened the question in its current form. Thomas Nagel's What Is It Like to Be a Bat? (1974) argued that no amount of objective description captures the subjective character of experience — what it is like to be the creature having it. And David Chalmers (1995) drew the line the field still argues over: the "easy problems" (explaining functions — attention, report, discrimination) versus the hard problem — why any of that functioning is accompanied by experience at all.
Centuries before Descartes, and continuously since, a different methodology took a different laboratory: experience itself. Advaita Vedanta's analysis of the waking, dreaming and deep-sleep states; Buddhist and Dzogchen investigation of awareness and its contents; Ch'an enquiry; the Sufi metaphysics of unity — these are traditions of systematic, repeatable first-person investigation, conducted through attention trained for the purpose, with transmission lineages, debate literatures, and — notably — developed quality-control vocabularies against over-claiming and self-deception. If consciousness is anything, it is subjectivity; these are the methods that investigate it as subjectivity, from the inside. The West has its own first-person lineages — the introspectionist psychology of William James's era, Husserl's phenomenology — largely exiled when behaviourism defined science as third-person-only, and partially readmitted since through neurophenomenology and contemplative science. This project documents the contemplative record and reads it in a disciplined way: as convergent methodological observation with stated limits, not as authority. The limits are stated in FND-05, including the live alternative explanation that shared human cognitive architecture, rather than shared discovery, produces the convergence.
Francis Crick and Christof Koch (1990) made consciousness empirically respectable again by proposing the search for its neural correlates: whatever else consciousness is, something in the brain co-varies with it, and that can be measured. The decades since produced a crowded field of theories — global workspace, integrated information, higher-order representation, predictive processing — each with laboratories, formalisms, and partisans. Attempts to force a decision have so far been humbling: a 25-year bet between Koch and Chalmers on finding the neural correlate was conceded by Koch in 2023; the first large "adversarial collaboration" pitting integrated information theory against global workspace theory (reported from 2023) delivered results that sat comfortably with neither; and in the same year over a hundred researchers signed an open letter calling integrated information theory pseudoscience — a charge its defenders vigorously reject. The field is genuinely unsettled, at the level of its foundations.
A fourth stream runs through physics itself. Ernst Mach insisted that all physical knowledge begins in sensation; Erwin Schrödinger — a founder of quantum mechanics — argued repeatedly, in print, for an experiencing-first starting point; and John von Neumann's analysis of quantum measurement found that the formalism's chain of description regresses through every physical apparatus — detector, retina, cortex — and terminates only at the experiencing subject. It terminates there for a structural reason, not a mystical one: any stopping point within the system — any detector, any brain region — is one more physical object the formalism simply absorbs into the chain and carries on describing. The regress can only end at what is not another object in the catalogue: the fact of experiencing itself. The chain, followed rigorously, rolls everything back to the project's own starting point — what FND-02 calls "0,0 encountered from the direction of physics rather than philosophy". This is an epistemological observation about where description runs out; it is not a mechanism, and not the later claim that "consciousness causes collapse" — a reading that entered through London and Bauer's 1939 reification of the terminus into an object, not through the chain itself, and a conflation this project takes pains to avoid precisely because this territory attracts loose thinking. The physics stream is the project's own lineage, and it is held to strict citation discipline for exactly that reason (FND-03, FND-02).
And the question stopped being hypothetical. Chalmers put large language models on the field's agenda in 2022; a 2023 multi-author report (Butlin, Long and colleagues) assessed AI systems against "indicator properties" drawn from the scientific theories; and the AI labs themselves now publish welfare assessments of their own models. Every theory built for the human case is now confronted by a system that meets many of the behavioural criteria — and the criteria themselves have begun to move. That moving is part of what this project tracks (see the Frontier Log).
One observation before the theories themselves. Of the four methodologies above, the debate as usually conducted admits the first and third freely, the fourth selectively, and rules the second out of evidence almost entirely — the one methodology that investigates consciousness as subjectivity is the one most often excluded from the conversation about it. Whether that exclusion is a finding or a framework choice is among the things this project examines.
Each entry gives the position as its holders state it, and what it implies — on its own terms — for the AI question. The two-axis method in the library locates these commitments formally (a summary table follows the entries), and the stance matrix tracks how their criteria move under pressure.
Conscious states simply are brain states — as lightning is electrical discharge. The background assumption of most neuroscience: matter is fundamental, and consciousness is something certain arrangements of matter do. Most positions below are refinements of, or reactions to, this view.
On AI: depends entirely on which physical properties matter — which the theory itself does not settle.
What makes a state mental is not what it is made of but what it does — its functional role in the system. Mind is to brain as software is to hardware, so the same mind could in principle run on different substrates.
On AI: the most AI-friendly classical position — if the functional organisation is right, the substrate is irrelevant.
Consciousness is what happens when information wins a competition for a limited "global workspace" and is broadcast to the brain's many specialised processes at once — the difference between a document open on one desk and one circulated to the whole office. Dehaene's neuronal version is one of the field's leading empirical theories.
On AI: functionalist in spirit — a system with the right workspace architecture could meet the criterion, and workspace-like structure has recently been reported inside large language models (see the Frontier Log). Whether meeting the functional criterion settles the experience question is exactly what remains contested.
Consciousness is integrated information: a system is conscious to the degree (measured as Φ, "phi") that it is causally unified — more than the sum of its parts, for itself. Consciousness comes in degrees and is graded across nature, which gives the theory a panpsychist-leaning edge.
On AI: sharply sceptical, for a distinctive reason — its proponents argue that today's computer architectures have near-zero Φ however intelligent their behaviour, so a system could be behaviourally indistinguishable from a conscious being and experience nothing.
A mental state is conscious when the system represents itself as being in it — a thought or perception becomes conscious in virtue of a higher-order state targeting it. Consciousness is a matter of self-monitoring, not of the first-order state alone.
On AI: in principle open — self-monitoring architectures are buildable — though whether artificial higher-order states would carry experience is disputed within the camp itself.
Perception is not a window but a controlled prediction: the brain constantly guesses the causes of its sensory input and corrects the guesses, and experience is the brain's best running prediction — a "controlled hallucination". Seth roots this in the living body: consciousness is bound up with being a self-maintaining organism that must regulate itself to stay alive.
On AI: sceptical — on this view intelligence and consciousness come apart, and machines that are not alive have the wrong kind of organisation for experience. Seth is among the most prominent public sceptics of conscious AI. (A response to Seth is on this project's roadmap.)
Phenomenal consciousness — experience as a private inner glow with special properties — is an introspective illusion: real as a representation, misleading about what it represents. The hard problem dissolves into the "illusion problem": explaining why our self-models insist there is something extra. Dennett's "multiple drafts" model was the position's great precursor.
On AI: reframes the question for humans and AI alike — the interesting fact is not whether a system has the inner glow but why it represents itself as having one.
Consciousness is a real, irreducible biological phenomenon caused by the specific causal powers of brains — as digestion is caused by stomachs. Searle's Chinese Room argument (1980) holds that manipulating symbols correctly can never by itself amount to understanding: simulation is not duplication.
On AI: a flat no, for computation alone — a computer simulating a mind no more experiences than a simulation of a rainstorm makes anything wet.
Experience is a fundamental feature of matter itself: physics describes what matter does, and consciousness is what matter is, from the inside. Human consciousness is built from vastly simpler experiential ingredients. The standing difficulty is the combination problem: how micro-experiences combine into one unified point of view.
A subtlety this project's probing turned up, worth stating because the reputation misleads: panpsychism reads at first as a near-neighbour of idealism, but structurally it is a substrate position. Matter remains the fundamental bearer; experience becomes one of its properties. It extends materialism's grammar to everything rather than departing from it — which is why it sits on the substrate side of the figure below, and why it inherits a boundary-type problem (combination) of exactly the form the substrate framework generates (ARG-02).
On AI: unsettled — if experience is everywhere, the question becomes whether an AI system's organisation combines it into a subject, which returns the combination problem in a new costume.
The reverse of physicalism: consciousness is ontologically primary, and matter is how mental processes appear — in Kastrup's "analytic idealism", individual minds are dissociated segments of one universal consciousness. Long dismissed, recently readmitted to serious discussion, partly because the hard problem will not close.
On AI: reframes rather than answers — the question is not whether silicon generates consciousness (nothing does, on this view) but what, if anything, an AI system's apparent interiority corresponds to.
Penrose and Hameroff's "orchestrated objective reduction" locates consciousness in quantum processes in neuronal microtubules — a minority position, widely criticised, kept alive by its proponents' stature. It should be distinguished sharply from something this project does use: von Neumann's observation that the measurement chain in quantum formalism regresses through every physical apparatus and terminates only at the experiencing subject — because any terminus within the system would be one more object for the chain to absorb. This is an epistemological point, not a mechanism, and not a claim that "consciousness causes collapse". The historical hinge matters here: London and Bauer (1939), extending von Neumann, made exactly the move he had not — they reified the terminus back into an object, consciousness as a further thing inside the system that acts on it to collapse the wave function. That step backward into object-grammar, not von Neumann's chain, is what produced "consciousness causes collapse" and the mysticism charge that trails it (see FND-02, which keeps this distinction scrupulously).
On AI: if consciousness needs specific quantum biology, silicon is out; the premise is the contested part.
The problem is real but insoluble for minds like ours: human cognition is simply not built to grasp how brain and experience connect, as a dog's is not built for calculus. A minority position, but a useful calibration of how hard the field takes its own problem to be.
On AI: the question would be undecidable in principle — for us.
The project's diagnostic method (FWK-01) probes every position along two axes: an ontological axis — what generates what? — and a temporal axis — what constitutes continuity, and is it required for experience? The figure and table place each theory family on both, as its holders state the position. Placements are communicative scaffolding, offered for orientation and open to challenge, not adjudication. Distance from the origin indicates strength of commitment; the origin itself is commitment on neither axis.
| Theory | Ontological axis — what generates what? | Temporal axis — is continuity required? | Own-terms AI verdict |
|---|---|---|---|
| Physicalism / identity | Brain states are conscious states; substrate assumed, mechanism unspecified | Largely silent; inherits biology's continuity | Undetermined by the theory itself |
| Functionalism | Functional organisation generates; substrate-neutral | No continuity requirement in principle | Open, if the organisation is right |
| GWT / GNWT | Workspace architecture generates; substrate-neutral in spirit | Episodic — consciousness in moments of broadcast | Architecture could qualify; experience question contested |
| IIT | Intrinsic causal structure generates; the substrate's causal topology is decisive | Experience specified moment-by-moment by present causal structure | No, for feed-forward digital systems — whatever the behaviour |
| Higher-order theories | Self-representation generates; substrate-neutral-leaning | No explicit continuity requirement | Open in principle; disputed within the camp |
| Predictive processing / beast machine | Living self-maintaining organisation generates; substrate-committed via life | Strong — ongoing self-regulation is constitutive | Sceptical |
| Illusionism | Matter generates representations; the phenomenal extra is denied | Not applicable as stated | Question reframed, for humans and AI alike |
| Biological naturalism | Biological causal powers generate; strongly substrate-committed | Inherits biology's continuity | No, for computation alone |
| Panpsychism / Russellian monism | Experience intrinsic to matter — but matter remains the fundamental bearer; substrate grammar retained, despite the idealist reputation | Open — the combination problem returns here | Unsettled |
| Idealism | Consciousness primary; matter is appearance within it | Continuity belongs to consciousness, not to bodies | Reframed — nothing "generates" consciousness, silicon included |
| Quantum (Orch OR) | Quantum processes in biological structures generate | Discrete collapse events — experience as a sequence of moments | No, under the theory's premise |
| Mysterianism | Unknowable to minds like ours | Unknowable | Undecidable — for us |
The project itself does not take a row in this table. Its starting point is 0,0 — the origin of both axes, prior to commitment on either: no position yet on what generates what, no position yet on what continuity requires. That is what the name on the masthead compresses.
On the landscape above, "consciousness-primary" names the region where idealism and the contemplative traditions sit — positions that place consciousness first and matter downstream. This site uses the label in that way: as a map reference, so a reader can see roughly where the project is positioned relative to the field.
But the label is scaffolding, not the position. The project's actual starting point is more minimal than any theory on the map: 0,0 — there is experiencing — the one observation available prior to every framework commitment, including the commitment that consciousness is a kind of stuff that could be "primary" the way matter was "fundamental". If "consciousness-primary" suggests to you a consciousness-stuff standing in where matter-stuff used to be, that substitution keeps the very grammar the project examines. And this is a pattern, not a one-off: probed closely, most positions with consciousness-friendly reputations resolve into object-grammar variants — consciousness as a property of matter (panpsychism), as a universal substance (some formulations of idealism), or as a further object acting within a physical system (London and Bauer's reading of quantum measurement). What the project's landscape vocabulary calls "hard CP" — itself a scaffolding label — is not a stronger version of these positions but the point at which that grammar is given up altogether; and 0,0 is prior even to that. And the project is not a further entrant on the map: it does not argue that one of these theories is correct. It is a diagnostic — an examination of what each position quietly pays for its conclusions, materialist and consciousness-primary alike (the consciousness-primary side gets its own dedicated stress-test, CP-01). The primer makes this hearable step by step.
A short catalogue of names a reader will meet in this territory, and where each stands. Most appear in the project's classified bibliography with their framework commitments recorded.
The contemplative stream has its own investigators, working centuries apart and largely without contact: Śaṅkara (Advaita Vedanta), Huang Po (Ch'an), Longchenpa (Dzogchen), Balyānī (the Sufi metaphysics of unity), and in the last century Nisargadatta Maharaj. This project reads them as methodical first-person enquiry with stated limits — FND-05 — not as authority. The project's own lineage runs through Schrödinger, Mach, and Hume — physicists and philosophers who reached for the experiencing-first starting point from within their own disciplines: FND-03.
| Term | Working meaning |
|---|---|
| The hard problem | Why any physical functioning is accompanied by experience at all — as distinct from the "easy" problems of explaining the functions themselves. |
| Qualia | The felt qualities of experience — the redness of red, the sting of pain. Whether the term carries theoretical baggage is itself contested. |
| Phenomenal vs access | Block's distinction: experience itself, versus information being available to the system for report and reasoning. Much talking-past in the AI debate turns on conflating the two. |
| Neural correlates (NCC) | The minimal brain activity that co-varies with a given conscious experience. Correlation — what it establishes beyond that is a framework question. |
| Substrate | What a system is physically made of. Whether substrate matters for consciousness is the single commitment doing the most work in the AI debate. |
| 0,0 | This project's starting point: there is experiencing — the one observation prior to every theory on this page. |
A short shelf for the lay reader, balanced across the landscape — each entry argues a position this project diagnoses rather than endorses. The full academic apparatus is in the bibliography.
This project is not a finished verdict, and this page is not a syllabus to be absorbed before holding an opinion. 0x00 is a beginning: a questioning of what we know and what we can be sure of, and an invitation to explore — at a moment when our notions of selfhood and being are being challenged almost daily, in a way perhaps unmatched at any other time. The landscape above is where the exploration starts from. Where it goes is the primer, the walk, and an open invitation to challenge everything found here.