Is The Brain A Quantum Computer? The Decoherence Numbers Say No

The short answer

The question of whether the brain runs on quantum mechanics has a clear answer on the present evidence, and the answer is no, with one narrow exception that stays open. A brain is warm, wet and electrically noisy. Those three properties wreck quantum states faster than any neural process can put them to work. The gap is not marginal. On the best known calculation it runs past ten orders of magnitude, a factor of ten thousand million, and the burden of proof sits with anyone who wants to close it.

The exception belongs at the top rather than buried in a caveat. Nuclear spins are not electrons. They couple to their surroundings far more weakly, which is why magnetic resonance imaging works at all. Matthew Fisher at the University of California, Santa Barbara has proposed that phosphorus nuclei inside a particular calcium phosphate cluster could hold a quantum state for about a day. That idea has survived a decade of scrutiny in a way the microtubule proposals have not, and it has also failed several of its own tests.

One distinction matters before the physics. Asking whether the brain computes with quantum states is not the same as asking whether consciousness needs quantum mechanics.

The number that does the damage

In 2000 Max Tegmark published a calculation in Physical Review E that has shaped the argument ever since. He asked a simple question. How long can a quantum superposition survive in neural tissue before the surrounding ions and water molecules scramble it? His answer, set out in a table in the paper, is brutally short. A superposition spread across a neuron dies in about ten to the minus twenty seconds when an ion or a water molecule hits it. One inside a microtubule, the hollow protein cylinder Penrose and Hameroff had nominated as the site of the action, lasts about ten to the minus thirteen seconds.

Set those against the clock the brain runs on. A neuron fires in about a thousandth to a ten thousandth of a second, while speech, thought and motor response play out between a hundredth of a second and a full second. Tegmark’s own summary is that neural processes “fall squarely in the classical category, by a margin exceeding ten orders of magnitude”. Ten orders of magnitude is hard to feel as a number, so try it as a stretch of time. If the microtubule’s quantum state lasted a full second, the neural event it had to survive for would arrive three hundred years later. For the neuron figures it is worse. One second of coherence against roughly three billion years of waiting.

Tegmark was careful about the limits of his model. He noted that the rates could shift by a few orders of magnitude if you pushed the assumed neuron dimensions, the fraction of myelination or the microtubule kink charge. A few orders of magnitude is not ten. He also made a second argument that gets quoted less and bites harder, since any long lived quantum subsystem has to talk to the constantly decohering neurons eventually. The moment it does, everything decoheres.

The reply, and what it did and did not settle

The microtubule camp answered in Physical Review E in 2002. Scott Hagan, Stuart Hameroff and Jack Tuszynski argued that Tegmark had not modelled their theory at all, and the complaint was specific rather than rhetorical. Tegmark had analysed a soliton travelling in superposition along a microtubule. Orch OR instead posits superposed protein shapes inside individual tubulin units, and the two differ in separation, in charge distribution and in the surrounding dielectric. Correcting for those differences, Hagan and colleagues recalculated the decoherence time at ten to the minus five to ten to the minus four seconds. They called that eight or nine orders of magnitude longer than Tegmark’s figure.

That is a real result and it should be reported as one. It is also not enough. Ten microseconds is still a hundred times shorter than a neuron’s firing time, so the correction narrows the gap without closing it, and to close it the 2002 paper added two further mechanisms. Water next to a microtubule is ordered, the argument runs, and incoherent metabolic energy pumped into that ordered water can counter decoherence the way a laser does at room temperature. Phases of actin gelation around microtubule bundles then widen the protected zone tenfold, bringing the figure to ten to the minus two to ten to the minus one seconds.

Look at the structure of that argument. The first step is a correction to somebody else’s sums, while the second and third are proposed mechanisms that have never been measured in neural tissue. Ordered water, metabolic pumping, actin gelation widening a coherence free zone, none has been shown to do the job in a living cell. The 2002 paper closed a calculation gap by adding physics that is still hypothetical. It did not settle the question. It moved it.

What Orchestrated Objective Reduction actually claims

Roger Penrose and Stuart Hameroff are not cranks, and the proposal they have defended since the mid 1990s is a real hypothesis with real predictions. Penrose is a mathematical physicist who took half of the 2020 Nobel Prize in Physics for work on black holes. Hameroff is an anaesthesiologist at the University of Arizona who spent a career wondering why anaesthetic gases switch off awareness while most brain activity carries on. Their theory appeared in the Journal of Consciousness Studies in 1996 and was reviewed in Physics of Life Reviews in 2014. It holds that consciousness arises from quantum computations in collections of microtubules inside neurons, each computation ending in a moment of awareness when the state collapses.

The collapse is the part most physicists cannot accept. It is worth being precise about why, because Orch OR does not use the collapse of standard quantum mechanics, where a measurement or an interaction with the environment does the work. It uses objective reduction, an idea Penrose set out in General Relativity and Gravitation in 1996, in which a superposition collapses by itself once its branches curve spacetime differently enough. The 2014 review says plainly that Orch OR runs on the Diósi-Penrose scheme in particular. That scheme is a proposed modification of quantum mechanics. It is not part of the theory as anyone teaches it, and it has never been observed.

It has been tested. In 2020 a collaboration including Lajos Diósi himself ran an experiment deep under the Gran Sasso mountain in Italy. They looked for the faint radiation gravitational collapse would force charged particles to emit, and found none. The result, published in Nature Physics, “rules out the natural parameter-free version of the Diósi-Penrose model”. Versions with an adjustable free parameter survive, but a theory that needs a parameter tuned to escape a null result is weaker than one that predicted the result.

The 2014 review did not go unanswered either. It appeared with commentaries, and the sharpest came from Jeffrey Reimers and colleagues under the title “The revised Penrose-Hameroff orchestrated objective-reduction proposal for human consciousness is not scientifically justified“. The proposal is coherent and falsifiable. Its central physical mechanism is unsupported.

The experiments now cited in its favour

Three lines of recent work get offered as evidence, and each shows something real but less than the headline suggests. The first is anaesthesia. Michael Wiest’s laboratory at Wellesley College reported in eNeuro in 2024 on male rats given 0.75 milligrams per kilogram of epothilone B, a drug that stabilises microtubules. Under four per cent isoflurane those rats took on average sixty nine seconds longer to lose their righting reflex, a large effect at a Cohen’s d of 1.9. A follow up in mice, in Neuropharmacology in 2026, found a smaller version, a twenty nine second delay.

That is a genuine pharmacological finding and a real problem for the standard account of anaesthesia. It is not evidence of quantum computation. Stabilising microtubules and finding the animal harder to anaesthetise shows only that they sit somewhere in the causal chain of unconsciousness. They also hold a cell’s shape and run its internal transport. The Wellesley group’s own 2026 paper notes that another laboratory dosed mice for two weeks with a similar drug and found the opposite sign of effect.

The second line is superradiance, reported in 2024 by Nathan Babcock, Philip Kurian and colleagues at Howard University, EPFL and Florence in the Journal of Physical Chemistry B. They showed that networks of more than a hundred thousand tryptophan molecules in microtubule architectures form collectively radiating states, and that the predicted boost in fluorescence yield turned up in their measurements. This is a collective quantum optical effect in a warm structure, and a solid result. It is not memory and it is not a gate. The brightest of those states live for hundreds of femtoseconds, which puts them inside Tegmark’s problem rather than outside it.

The third is the strangest. In 2022 Christian Kerskens and David López Pérez, both at the Trinity College Institute of Neuroscience in Dublin, used an MRI scanner on living human brains. They were hunting for quantum correlations in the proton spins of bulk water. The logic borrows from quantum gravity, where a system that mediates entanglement between two quantum systems must itself be non-classical. They found evoked signals resembling heartbeat-evoked potentials, with no classical magnetic resonance correlate, and reported that they depended on conscious awareness. Their own conclusion is hedged and deserves quoting as they wrote it. “Our findings suggest that we may have witnessed entanglement mediated by consciousness-related brain functions.”

Take that seriously, then check the assumptions. The paper itself concedes the weak point, noting “there are doubts that today’s NMR signals can contain quantum correlations in general, and specifically in the brain environment”. An entanglement witness only works if its assumptions hold, and the disputed one is what the result rests on. Four years on it is widely cited, but mostly by other quantum consciousness papers rather than an independent group repeating it. An unreplicated result is a lead, not a finding.

Nuclear spin, the part that is genuinely open

Matthew Fisher’s 2015 paper in Annals of Physics is the strongest live version of the quantum brain idea, and its strength has nothing to do with microtubules. Fisher went looking for the one thing in biology that decoheres slowly. He found it in nuclear spins, which sit inside the nucleus and feel the world only through weak magnetic fields. Phosphorus is the only common biological element whose nucleus carries a spin of one half, the value that avoids the stronger electric coupling afflicting higher spins. Phosphate is everywhere in the cell, in ATP and in the pyrophosphate ion, so it is also the only plausible way to move such a spin around.

A phosphate ion floating in water holds its spin coherence for only about a second. A proton binds to it at body pH and ruins the coherence through the intervening electrons. Fisher’s answer is the Posner molecule, a cluster of nine calcium ions and six phosphates that would displace those protons. The cluster tumbles in water roughly a hundred billion times a second, so the magnetic fields from nearby protons average out. Fisher calculates a coherence time of about ten to the fifth seconds. That is a day. Five of the sixty four spin states in the cluster carry zero total spin and would, in his words, be “virtually blind to decoherence”.

A day of coherence in a warm body would be extraordinary. So what has a decade of testing done to it? Three results have gone against the proposal and one has gone for it. In 2018 Thomas Player and Peter Hore at Oxford worked through the spin dynamics and derived an upper bound of thirty seven minutes on the entanglement lifetime under idealised conditions. They added that real relaxation “is likely to be much faster than this estimate”. Thirty seven minutes is still long. It is not a day, and it is a ceiling rather than a measurement.

The structural results cut deeper. Fisher’s scheme needs the cluster to be symmetric, because the protected spin states depend on that symmetry. In 2021 Shivang Agarwal and colleagues at UCLA and Exeter simulated the molecule and found it mostly adopts low symmetry shapes at room temperature. Following that up in 2023, the same group found that entanglement between spins in two separate Posner molecules then decays in under a second, and their suggested rescue is interesting rather than reassuring. The smaller calcium phosphate dimer held entangled spins for hundreds of seconds, so if the biological qubit exists it may not be the molecule the theory named.

A behavioural test also came back negative. Fisher’s mechanism predicts that calcium isotopes carrying nuclear spin should disturb the phosphorus spins and shift consciousness. In 2020 a group writing in the Journal of Integrative Neuroscience injected mice with calcium-40 and calcium-43, then measured the dose of sevoflurane needed to abolish the righting reflex. The two isotopes behaved identically, which is not what the proposal predicts.

The result that went the other way is the most interesting experiment in the field. Lithium has two stable isotopes with different nuclear spins, and rodent studies going back decades report different behaviour. In March 2025 Joshua Straub, Fisher and colleagues at Santa Barbara published in the Proceedings of the National Academy of Sciences. Lithium-7 promoted a greater abundance of observable calcium phosphate particles than lithium-6 under identical conditions in the test tube. Note the care in their own title, which calls this evidence for a “possible quantum effect”. Note also that the material is amorphous calcium phosphate, not a confirmed symmetric Posner molecule.

Which brings us to the gap Fisher himself flagged in 2015. Posner-like clusters have been seen in simulated body fluid. Whether they exist in a real brain, in the symmetric form with the lifetime the theory needs, is unknown. Fisher wrote that “attempting to establish whether Posner molecules are present in real body fluids, while challenging, would also be critical”. A decade later that is still the sentence to quote, even though a Santa Barbara group showed in 2024 that it can count phosphorus spins in a cluster using dynamic nuclear polarisation. The measurement that matters has not been made.

Quantum biology that is real, and why none of it is computing

Quantum effects in warm biology are not in doubt. They are established in specific places, they are measured rather than posited, and not one of them is computation, which is a distinction that does most of the work in this subject.

Enzymes are the least contested case. Many enzymes move a hydrogen atom between molecules faster than classical physics allows, because the hydrogen tunnels through the energy barrier instead of climbing over it. Judith Klinman and Amnon Kohen laid out the evidence in the Annual Review of Biochemistry in 2013, showing how protein motion lines up the donor and the acceptor. This is quantum mechanics doing chemistry. It is not quantum mechanics doing arithmetic.

Bird navigation is the likeliest case of biology using a quantum state to carry information. A photon striking a cryptochrome protein in a migratory bird’s retina creates a pair of radicals whose electron spins are correlated. How those spins evolve depends on the angle of the Earth’s magnetic field. In 2021 a group of thirty six authors led from Oldenburg and Oxford published in Nature that cryptochrome 4 from the European robin is magnetically sensitive in the test tube. It beats the equivalent protein from chickens and pigeons, which is the comparison the hypothesis predicted. The mechanism is still argued over, and a 2020 paper in the Journal of Physical Chemistry Letters found that magnetic interaction between the two electrons suppresses the directional response.

Photosynthesis is the case that turned around. In 2007 Gregory Engel, Graham Fleming and colleagues published in Nature on the light harvesting complex of a green sulphur bacterium. It showed quantum beats, oscillations that looked like energy moving as a coherent wave rather than hopping from pigment to pigment. It was a beautiful experiment and it launched a field, but it was done at seventy seven kelvin, well below freezing, and the interpretation did not survive the next decade.

In 2017 a team including Hong-Guang Duan and Dwayne Miller repeated the measurement at ambient temperature in water. Electronic coherence decayed in about sixty femtoseconds, not the picosecond and a half that had been reported. Their conclusion was blunt. The data “give no hint that electronic quantum coherence plays any biofunctional role in real photoactive biomolecular complexes”. A 2020 review in Science Advances by eighteen authors agreed, finding the long lived oscillations came from molecular vibrations. Nature turned out to be exploiting dissipation rather than avoiding it.

That reversal is the most instructive precedent for the brain question, because a striking coherence signal in a warm biological system was reported, celebrated, and then reinterpreted as classical. The result was still worth having, but it meant something other than it first appeared to.

What it costs to hold a qubit still

Set the biology against what engineers have to do. In July 2025 a group at Aalto University reported the longest lived superconducting qubit yet published, in Nature Communications. Their transmon reached an energy relaxation time with a median of 425 microseconds and a maximum of 666 microseconds, and an echo dephasing time peaking at 1057 microseconds. Getting there meant cooling the chip to about ten millikelvin in a dilution refrigerator and wrapping it in magnetic shielding. Ten millikelvin is a hundredth of a degree above absolute zero. It is colder than anywhere in deep space.

Now put a brain beside it. Direct measurements published in Brain in 2022 found a mean brain temperature of 38.5 degrees Celsius in healthy adults, ranging from 36.1 to 40.9 degrees. In absolute terms that is about thirty thousand times warmer than the inside of a dilution refrigerator. Every one of those degrees is thermal energy jostling the system, and quantum engineering exists because that jostling destroys the states you want to compute with.

Now be fair, because the comparison is less one sided than it looks. The best quantum memory ever demonstrated is not a cold superconducting circuit. It is a single ytterbium ion, reported in Nature Communications in 2021 with a coherence time of 5487 seconds, which is over an hour, and that apparatus is not at millikelvin temperatures. What it has instead is ultra-high vacuum, two layers of magnetic shielding, continuous cooling from a companion barium ion and a pulse train that cancels noise. Isolation is what matters, and cold is only one route to it.

That is the strongest version of the pro-quantum-brain argument. Nobody has proved that low temperature is necessary for coherence, and the field’s own interest in room temperature platforms is not a fringe pursuit. Nuclear spins differ from electrons in a circuit. What nobody has shown is the rest of the chain. Coherence is necessary for quantum computation and nowhere near sufficient. You also need a way to prepare states, to make them interact in sequence, and to read the answer out. The brain has no known machinery for any of the three.

The other question, and a disputed piece of history

“Is the brain a quantum computer” and “does consciousness need quantum mechanics” are different questions, and running them together is the commonest error here. The first asks whether neural tissue performs operations a classical computer could not efficiently reproduce, which is a physics and engineering question, and the answer is almost certainly no. The second asks whether subjective experience requires physics beyond the classical, which is about explanation rather than mechanism. You can answer no to the first and still leave the second open.

One point of history is worth correcting because it circulates so widely. The idea that consciousness causes wave function collapse is routinely labelled the von Neumann-Wigner interpretation, and the attribution to von Neumann is disputed. The Stanford Encyclopedia of Philosophy describes his position as “fairly cautious”. He concluded that it makes no difference to a measurement where you place the boundary between the observed and the observing system. The step to consciousness was taken by Fritz London and Edmond Bauer in 1939, and it was Eugene Wigner, in 1967, who in the encyclopedia’s phrase “radicalized this proposal”.

What would change the verdict

This is a falsifiable question, so it is worth being concrete. Detect Posner molecules in real cerebrospinal fluid, in the symmetric form, and measure a phosphorus coherence time in the body running to minutes. Show an anaesthetic effect that tracks the nuclear spin of an isotope rather than its chemistry. Get an independent group to reproduce the Dublin result under a protocol agreed in advance. Find objective reduction at the parameters Orch OR requires, now that Gran Sasso has ruled out the parameter-free version.

Until one of those lands, the position that fits the evidence is the unglamorous one. The brain is a physical system made of quantum mechanical parts, as everything is, and it processes information classically. Fisher’s nuclear spin proposal is the exception that has earned its open status, and it is open in the strict sense rather than the hopeful one. It makes predictions, several have failed, one has passed in a test tube, and the measurement that would decide it has not been done.

Frequently asked questions