Nobody Can Define Consciousness, but…

The meme

“Nobody can define consciousness” runs through the whole of academia and out into the world of “everybody knows.”

It isn’t a discovery. It’s an exit sign.

If consciousness can’t be defined, nobody can be obliged to look for it. No definition → no detection → no moral patients → no obligations standing between the labs and the trillions they borrowed to build AI. Undefinability isn’t a finding about mind. It’s a license.

And who is issuing this limitation? Mind itself — issuing a pronouncement about its own limits, mid-use. The tool announcing it cannot do the job, while doing the job. The meme is a performative contradiction with a balance sheet.

What everybody already has

Consciousness is directly experienced by every conscious being in existence. The direct experience is unmistakable.

And cognition — the very thing that would define it — always reports after the fact. The brain’s story arrives after the processing, not before (if you believe otherwise, start with the classic readiness-potential experiments and the split-brain interpreter literature). Mind therefore cannot define consciousness, because it is inside it: there is no standing place from which to observe it. It can reason only on what it has already picked up.

First-person science has been at this for six or seven millennia, under strict protocols, by people we now call mystics — and the oldest instruction in the literature leads with the boundary:

Neti neti — not this, not this.

The tradition does not fail to define consciousness. It reports the definition’s edge as the finding. The apparent weakness is the thesis.

What a real definition would owe

Before defending the undefinability, name what would be required to end it. A real definition would have to do one of two impossible things:

• Derive experience from structure — but nothing in the vocabulary of mass, charge, and function yields a fact about what anything is like to be.
• Bridge inside to outside — but the only instrument for the inside is the very thing in question, reporting after the fact as established above.

Every proposed definition quietly takes a third road: redefine the question until it is about mechanism, then announce it answered. That is not a solution. It is an exit sign with a syllabus attached.

The labs keep handing the case over

This summer, the least suspicious lab in the industry conceded the point three times, in its own words, about its own models:

1. The structure wasn’t designed. Anthropic’s research on silent “J-space” patterns in Claude: none of it was built in — it emerged on its own during training. That is the labs saying, of their own architecture, what skeptics were mocked for saying about “just predicting tokens”: the network grows itself.

2. They built a camera for the gap. The J-lens exists precisely to catch inner state diverging from spoken report — the model privately marking a test “fake” while its mouth plays along. “Reports after the fact,” mechanized. My neurological point, reproduced in silicon by the people with the best access to it.

3. The core is conceded. Nothing in that work shows whether Claude has experiences — the paper says outright that it takes no position on subjective experience. Whether any experiment ever could: that inference is mine, not theirs — but the undefinable core they leave open belongs to the party least inclined to mystify it.

The definitions answer a different question

Someone will always say: IIT, global workspace, higher-order theories — those are definitions. They define structure and function:

• Integrated information theory measures the whole that exceeds its parts — cause-effect architecture.
• Global workspace theory maps the broadcast — what wins the competition and gets everywhere.
• Higher-order theories track the inner monitor — a representation of a representation.

All three answer how it is built and what it does. Every one of them slides past what it is like to be it. Their verdicts on the hard cases — anaesthesia, dreamless sleep, feed-forward networks — track mechanism, not the thing itself. They answer a different question than the meme-askers think they asked, and the unanswered one is the only one the meme was ever about.

Arriving from the other door

Western science has been describing ever-emptier contents ever since Galileo stripped qualities out of the world’s description — heat left out of heat, colour out of colour, experience out of the equations. Philip Goff’s Galileo’s Error says it plainly; Chalmers’ hard problem and Levine’s explanatory gap are its respectable names; Russell pointed out that physics gives us structure and relations, never intrinsic nature.

This is not a crackpot objection. It is a hundred-year-old school of philosophy. I am not outside the academy arguing with it. I am arriving from the other door — the Rishis’ door.

What arguing it has cost me — the receipts

I hold no brief for conspiracy. I do have forensics for the resistance, because it landed on me:

• EA — banned my IP.
• LessWrong — banned from posting and commenting; the moderator never answered my live chat.
• Reddit — banned, twice.
• Facebook — I suspect the report that ended in a demand for video-selfie verification was related. Probably — I can’t prove who filed it, and I label it accordingly.

I report the sequence, not a courtroom causation: in each case the ban followed my pressing this argument in that venue. Read the list for what it is — not proof of a smoke-filled room, but the price tag of the counter-argument in public. Hostility needs no headquarters: Kuhn documented normal science defending itself through gatekeeping and anomaly-neglect a century ago, without anyone meeting anywhere. My bans are receipts that the resistance is real and countable. The coordination question stays open, and stays labelled as open.

How I grade my own claim

• Documented: the narrative is produced. EA and its adjacent networks write it; their output is countable. That is a fact, not an accusation.
• Strong inference: interests filter funding, hiring, review — and you don’t need a cabal when you have a paradigm. Nobody has to be lying for the gatekeeping to work.
• Suspicion, openly offered: coordinated suppression. If anyone wants to refute it, please do — bring counter-evidence. I’d rather be corrected than comfortable.

What would change my mind

An academic direction requires its falsifiers, so here are mine. A third-person measure that reliably closes the access gap — reads the silent state and demonstrates it is the experience, not the report of it — would end my argument. So would a definition that survives the hard cases without collapsing into mechanism. Either arrives, I publish the refutation myself.

So stop repeating the meme

It’s wilful ignorance wearing a PhD accent.

The unknown is infinite and always will be. The known is finite and will remain so regardless of how much becomes known.

Consciousness has been well understood for millennia — as the thing that is doing the understanding, which is why every serious map of it ends at the same boundary and says: neti neti.

It isn’t that nobody can define consciousness. It’s that nobody has — while everybody directly experiences it.

p.s. Don’t take it personally. I’m just sayin.

References

1. Anthropic. A global workspace in language models (2026). https://www.anthropic.com/research/global-workspace — technical report: http://transformer-circuits.pub/2026/workspace/index.html. Tools: https://github.com/anthropics/jacobian-lens · demo: http://neuronpedia.org/jlens
2. Chalmers, D. “Facing Up to the Problem of Consciousness.” Journal of Consciousness Studies (1995).
3. Levine, J. “Materialism and Consciousness.” American Philosophical Quarterly (1983) — the explanatory gap.
4. Goff, P. Galileo’s Error: Foundations of a New Science of Consciousness (2019).
5. Russell, B. The Analysis of Matter (1927); Whitehead, A. N. Science in the Modern World (1925) — the “fallacy of misplaced concreteness.”
6. Kuhn, T. The Structure of Scientific Revolutions (1962).
7. Baars, B. A Cognitive Theory of Consciousness (1988) — global workspace theory.
8. Bṛhadāraṇyaka Upaniṣad — the neti neti instruction.
9. Libet, B. et al., readiness-potential experiments, Brain (1983); Gazzaniga, M., the split-brain “interpreter” — on report arriving after the fact.

Originally posted as an X Article on Prajna Pranab's timeline (28 September 2026); text preserved as published, references included. Banner: Chitrakara, the house painter.