Every day, anaesthetists switch consciousness off and on again in millions of people, reliably, with drugs as different from each other as propofol and the inert noble gas xenon — and nobody can say precisely how. Into that gap, in the mid-1990s, stepped the most ambitious answer ever proposed: a mathematician and an anaesthesiologist arguing that consciousness is a sequence of quantum-gravitational events inside the protein scaffolding of neurons. The theory has been attacked for thirty years and defended for thirty years, and both sides usually treat it as one thing. It is not. It is at least four independent claims, bolted together so that each borrows credibility from the others. This piece unbolts them. Since the authors’ 2014 review, one of the four has been hit hard by an experiment under an Italian mountain; one has gained real support from animal studies; and the support it gained turns out not to require any quantum physics at all.
Current Conditions
Primary Route — One Theory, Four Grades
Start with the mystery, because it is real and it needs no physics beyond the ordinary kind. An anaesthetist can take a conscious adult, administer a gas, and within a minute or two have a person who feels nothing, remembers nothing, and responds to nothing — then reverse it on schedule. It is done millions of times a year with a safety record most engineering disciplines would envy. And the field cannot tell you, at the molecular level, why it works. The oldest clue is more than a century old: the potency of an anaesthetic tracks its solubility in oil, which pointed first to cell membranes and later to hydrophobic pockets inside proteins. Specific receptor targets have been found for specific drugs. But the agents are too chemically varied — a noble gas that bonds with nothing sits in the same functional category as complex synthetic molecules — for any single receptor story to feel complete. Something common to all of them switches off experience. What, exactly, is still open.
That openness is the fairest possible entry point to the theory this piece grades, because it is the one place where the theory’s authors started from a genuine problem rather than a philosophical preference. Stuart Hameroff is an anaesthesiologist. His interest in microtubules — the hollow protein cylinders that form the internal scaffolding of every neuron — began with the observation that anaesthetic gases bind in the same kinds of oily, electron-rich pockets that line the tubulin proteins they are built from. Roger Penrose arrived from the other direction, with an argument from mathematical logic and a proposal about quantum mechanics and gravity. In the mid-1990s the two ideas met, and the result was Orch OR: orchestrated objective reduction.
The Theory, Stated Fairly
In its 2014 form, the claim runs as follows. Inside neurons, tubulin subunits of microtubules can exist in superpositions of two states — originally described as shape changes, later as electric dipoles, and in the 2014 review as possibly magnetic spin dipoles. These superpositions are shielded from the surrounding environment long enough to evolve as a quantum computation. They then terminate not randomly but by a specific physical rule, the Diósi–Penrose criterion: a superposition of two mass distributions is unstable, and collapses on average in a time of about ℏ divided by the gravitational self-energy of the difference between them. Bigger displaced mass, faster collapse. Each such orchestrated collapse is, the theory proposes, a moment of conscious experience; the selected outcome feeds back into when the neuron fires. At the bottom of the stack sits the claim that started it: that human understanding is non-computable, so consciousness must rest on physics that no algorithm can simulate — and quantum collapse is the only candidate gap in current physics.
It is an extraordinary edifice, and its defenders are right about one thing that critics often skip: it is quantitative. It makes numbers. The 2014 review calculates that roughly twenty billion tubulins held in coherent superposition would collapse in 25 milliseconds — one cycle of the 40-hertz gamma rhythm that correlates with awareness — and lists twenty testable predictions published in 1998, with a status report on each. Few theories of consciousness expose themselves that much. The question is what happens when you check the numbers one claim at a time.
The Scalpel
The physicist Henry Stapp, himself sympathetic to quantum approaches to mind, supplied the right instrument years ago. He divides the proposal into parts and notes something that transforms the whole debate: the collapse rule and the quantum computer are different claims with different requirements. A gravitational collapse rule is not threatened by the brain being warm and noisy. Interaction with the environment actually helps it along — it sharpens the alternatives the rule chooses between. Only the quantum-computation claim needs the delicate isolation that thermal noise destroys. Max Tegmark’s famous 2000 calculation, which put decoherence in microtubules at around 10−13 seconds, therefore strikes one component and misses the other. Because the authors sell the components as a unit, every hit on any part looks like a hit on everything, and every defence of any part looks like a defence of everything. Separate them, and each can be graded on its own evidence. There are four.
Claim One: Gödel — the Foundation That Failed First
Penrose’s original argument is that a mathematician can see the truth of statements that no consistent formal system containing arithmetic can prove, and therefore that human understanding cannot be the output of any algorithm. It is the reason the theory needs exotic physics at all: if understanding were computable, ordinary neurons would do. The argument has been examined at length by logicians, and the verdict has been close to unanimous that it does not go through — broadly, because it requires the human mathematician to know that his own reasoning is consistent, which is precisely what Gödel’s results suggest no sufficiently rich system can establish from inside. Stapp’s summary is blunt: the Gödel part cannot be regarded as established. Grade: a foundational premise that the relevant discipline rejected, still present in the structure because everything above it was built on it.
Claim Two: Gravitational Collapse — Hit Under a Mountain
The Diósi–Penrose rule is not a crank idea. It is a serious proposal for the deepest open problem in quantum mechanics — why definite outcomes appear at all — and it has one enormous virtue: it predicts things. A collapse driven by gravity should, as a side effect, jiggle the particles involved, and jiggled charged particles radiate. In 2020 a team including Diósi himself put a germanium detector, wrapped in lead, 1.4 kilometres under the Gran Sasso massif to shield it from cosmic rays, and looked for that radiation. It was not there at the level the natural version of the model requires. The result rules out the parameter-free form of the model — the version in which the collapse time follows from the formula with no knobs to turn.
The fair statement of what survives is narrower than either side usually admits. Modified versions with an adjustable spatial scale remain alive, and theorists have since proposed forms that claim to evade the bound altogether. But the price of survival is a free parameter, and the whole appeal of the Orch OR numbers — twenty billion tubulins, twenty-five milliseconds — rested on the formula having none. A theory whose central clock now needs to be tuned has lost the property that made its predictions impressive. Grade: the most rigorous component of the theory, and the one that took the hardest experimental hit, which is not a contradiction — it is what happens to claims precise enough to be tested.
Claim Three: Microtubules and Anaesthesia — Real Support, Wrong Conclusion Drawn
Here the theory’s anaesthetic roots pay off. If gases act partly by binding microtubules, then stabilising the microtubules with a drug should make an animal harder to anaesthetise. In 2024 a Wellesley group tested exactly that. Rats given a single dose of epothilone B, a microtubule stabiliser that crosses into the brain, took an average of 69 seconds longer to lose their righting reflex under isoflurane, and the delay could not be explained by tolerance from repeated exposure. The effect size was large. The authors have since reported the same pattern in mice. This is a genuine, pre-registered-in-spirit prediction of the theory’s authors, tested by a laboratory willing to be wrong, and it came out in the predicted direction.
Now read what it actually shows. It shows that microtubules are among the functional targets through which an anaesthetic produces unconsciousness. It does not show that anything quantum happens there. A drug that stiffens a scaffold could delay unconsciousness through any number of classical routes — by occupying or distorting the binding pocket the gas would otherwise use, by altering transport along the tubules, by changing how neurons integrate their inputs. The experiment is equally at home in a world with no quantum biology whatsoever. Grade: the best new evidence the theory has received, supporting the claim that needs no new physics, and leaving the claims that do exactly where they were.
Claim Four: Warm Quantum Computation — the Evidence Has One Address
This is the component every physicist flinches at, and the defence has two lines. First, the decoherence calculations: Tegmark’s 10−13 seconds was challenged by Hagan, Hameroff and Tuszynski, who argued he had modelled the wrong kind of superposition and recalculated to 10−4 to 10−3 seconds. Stapp, reviewing the exchange, judged that the corrections still left a gap of a factor of around ten thousand, with no detailed account of how living tissue would close it. Second, the experimental line: resonant conductance measured in single microtubules at specific gigahertz, megahertz and kilohertz frequencies by Anirban Bandyopadhyay’s group in Japan, which the 2014 review calls clear evidence for coherent microtubule quantum states. The weakness is not that the measurements are wrong; it is where they come from. In the 2014 review, the strongest supporting claims about microtubule coherence rest on that single laboratory, several of them cited as personal communications. A resonance is also not the same thing as a quantum computation, any more than a ringing bell is a calculator. Grade: not refuted, not shown, and resting on a base too narrow to carry what is placed on it.
Claim Five, and the One the Others Exist to Carry
Even if all four held, the claim that makes Orch OR a theory of consciousness rather than of cell biology would remain a separate step: that each orchestrated collapse is a moment of experience, and that unorchestrated collapses everywhere in the universe are flickers of ‘proto-consciousness.’ No experiment in the file tests this, and it is difficult to see what experiment could. It is an identity claim, and identity claims are the part of any theory of mind that data reaches last.
The Number That Moved
One more tool, because it generalises. Defending the theory, Stapp points out that its core calculation lands in a plausible range: the number of tubulins required for a half-second conscious moment comes out at around one percent of the brain’s supply, when it could have come out billions of times too large or too small. A result that could have been absurd and wasn’t looks like evidence. But set that figure beside the authors’ own 2014 review. There, the same half-second moment requires roughly a billion tubulins — about one neuron’s worth. The one-percent figure reappears in the 2014 paper too, but attached to a different version of the theory, the new ‘beat frequency’ scheme, for a collapse time of ten nanoseconds. The number shifted by around nine orders of magnitude across versions, driven by a modelling choice: whether the displaced mass is the whole smoothed-out protein, its atomic nuclei, or its individual nucleons. A plausible ballpark that can be moved nine orders of magnitude by an assumption is not a confirmation. It is a dial.
Four claims bolted together, each borrowing credibility from the others. Hit one and it looks like an attack on all; defend one and it looks like a defence of all. The only honest move is the one nobody in the argument wants to make: grade them separately.
Three Honesties
First: the theory deserves credit it rarely gets. It committed to numbers, published predictions, and put one of its load-bearing components within reach of a decisive experiment. That component took a heavy hit, and it could only take that hit because it was precise. Most rival theories of consciousness have never exposed themselves to anything comparable. A failed prediction is an achievement a vague theory cannot claim.
Second: the theory has also displayed the habits of a hypothesis under pressure. Between versions, the information carrier moved from tubulin shape changes to electric dipoles to magnetic spins; the relevant timescale moved from tens of milliseconds to ten nanoseconds; tubulin requirements moved accordingly. The 2014 review grades its own twenty predictions and concludes they have “fared rather well” — while listing several as untested or moot. And among the evidence it cites for temporal anomalies in conscious experience is Daryl Bem’s 2011 precognition study, which became a founding case of psychology’s replication crisis. None of this refutes anything. All of it is what the drift of an unfalsified idea looks like.
Third: the anaesthesia mystery does not need this theory to be real. The strongest thing in the whole file is the question Hameroff started from: a chemically incoherent set of drugs reliably switches off the one phenomenon science has no mechanism for, and nobody knows the common step. The microtubule findings of 2024 are a genuine contribution to that question. Whether they lead anywhere near quantum gravity is a separate matter, and treating the two as one is the bundling error this piece exists to undo.
What Actually Transfers
This blog files this as the purest case yet of a pattern it keeps finding: the bundle that travels as a unit because its parts could not survive the journey separately. For a deliberately philosophical approach to consciousness rather than a physical one, see One Subject, Many Windows.
Evidence supports a claim, not a theory. Eight rats resisting anaesthesia were widely reported as support for a quantum theory of consciousness. They support one sentence inside it — that microtubules are an anaesthetic target — and that sentence would be true in a purely classical brain. Whenever a result is said to confirm a big idea, locate the specific claim it bears on and check whether any rival idea predicts the same thing.
Precision is how theories earn the right to be wrong. The Diósi–Penrose collapse law was ruled out in its natural form because it was sharp enough to make a detector click or stay silent. That is a credit to it, not a scandal. Distrust theories that have never been hit more than theories that have been — the former may simply never have said anything.
Watch the dial. A calculation that lands in a satisfying range is persuasive only if it could not easily have landed elsewhere. When the same theory produces one percent of the brain in one version and one neuron in another, the result is being set by the modeller. Ask, for any striking number, which assumption would move it most, and by how much.
Instrument Check — Worth Your Attention
Study — Donadi et al., “Underground test of gravity-related wave function collapse,” Nature Physics, 2020. The experiment that took the parameter-free collapse law off the table: a lead-shielded germanium detector under 1.4 kilometres of Italian rock, looking for radiation that should be there if gravity collapses superpositions in the simplest way. It wasn’t. Note that one of the model’s originators signed the paper that constrained it.
Read — Khan et al., “Microtubule-stabilizer epothilone B delays anesthetic-induced unconsciousness in rats,” eNeuro, 2024. Short, clean, honestly limited: eight animals, one drug, one gas, one proxy for unconsciousness. The finding is real. The interesting exercise is to list how many classical mechanisms would predict the same result before reading the discussion, where the quantum interpretation is offered.
Follow — Hameroff & Penrose, “Consciousness in the universe,” Physics of Life Reviews, 2014, Section 5.7. The authors grading their own twenty predictions, in italics, one by one. Read it as a primary document in how a research programme reports on itself: which predictions it counts as met, which it reclassifies as moot, and which it leaves as untested. Then compare the tone of the conclusion with the contents of the list.
Flight Log — Dispatch From Altitude
Every pilot has had consciousness switched off on purpose, at least a little, in a controlled setting. It is called hypoxia training. You sit in an altitude chamber, the pressure is taken down to the equivalent of a high cruise level, you remove your mask and start working through simple tasks — arithmetic, card sorting, writing your name — and you are told to put the mask back on when you notice something is wrong. The lesson of the exercise is that you do not notice. Handwriting degrades into scrawl, sums go wrong, a colleague has to tap your shoulder, and afterwards you watch the video of yourself cheerfully insisting you feel fine. The regulator publishes a figure for this — the time of useful consciousness — and at typical airliner cruise altitudes after a sudden decompression it is measured in tens of seconds.
What the chamber teaches, and what this piece kept circling, is that consciousness has no fuel gauge. The faculty that would report its own failure is the faculty that is failing. That is why the aviation answer to hypoxia is never “pay attention to how you feel” but a procedure executed by rote before judgement goes: masks on first, then think. It is also why the anaesthetist’s question is so hard. The only instrument that directly observes consciousness is consciousness, and it switches off along with the thing being measured — which is why the rats in the 2024 study were scored not on experience but on whether they could turn themselves upright.
And a small professional footnote on bundling. Checklists exist precisely because the human mind treats a coherent story as evidence that each of its parts has been checked. A cockpit does not accept “the aircraft is configured for landing” as one claim; it asks about gear, flaps, speedbrakes and lights separately, and it asks aloud, because the whole is exactly where an unchecked item hides. The Penrose–Hameroff theory is a beautiful whole. Read it the way a crew reads a landing checklist, one item at a time, and it becomes something more useful than beautiful: a list of four very different claims, two of them tested, and a clear picture of which items are still open.