🧬 biology

What a radical-pair quantum reservoir would require: readable capacity, no quantum advantage, and a nuclear-register criterion

This paper demonstrates that while a cryptochrome radical pair functions as a readable quantum reservoir with measurable information capacity, it offers no quantum advantage over classical networks and is constrained by chemical turnover rates rather than spin coherence lifetimes, leading to a new, experimentally testable criterion for biological quantum computation.

Original authors: Hikaru Wakaura, Taiki Tanimae

Published 2026-10-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Hikaru Wakaura, Taiki Tanimae

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

For decades, a quiet question has lingered at the edge of biology and physics: could the human brain, or perhaps the nervous systems of other animals, rely on the strange rules of quantum mechanics to think? We know that certain birds navigate using the Earth's magnetic field, a feat scientists have traced to a specific chemical reaction inside their eyes. This reaction involves a pair of electrons that become "entangled," meaning their states are linked in a way that defies classical physics, allowing them to sense magnetic directions. This discovery sparked a bolder idea: if nature can use this delicate quantum link for sensing, could it also use it for processing information, effectively turning a molecule into a tiny computer? The prevailing skepticism has always been that the brain is too warm and wet for such fragile quantum states to survive; they should collapse into ordinary chaos almost instantly.

A new study by researchers at the Quantum Integrated Research Institute in Tokyo takes this debate and turns it into a concrete, testable experiment. Instead of asking if a molecule can perform complex quantum calculations like a futuristic computer, they asked if it could function as a "reservoir." Imagine a reservoir not as a lake, but as a complex, swirling tank of water. You pour a patterned stream of water in at the top, and the turbulence inside mixes it. You don't need to understand the exact path of every drop; you just need to look at the water coming out the bottom to see if the original pattern is still recognizable. In this new view, the molecule is the tank, the magnetic field or a chemical signal is the input stream, and the chemical products it spits out are the readout. The researchers built a detailed simulation of this process, treating the electron pair and its surrounding atomic nuclei as a system that takes in information, holds it for a moment, and then releases it.

The team simulated a specific type of molecule found in nature, known as cryptochrome, which hosts a pair of electrons that dance between two energy states. They drove this system not by resetting a single particle, as previous theories had done, but by mimicking a real chemical cycle: a fresh pair of electrons is born, they interact with the molecule's atomic nuclei, and then they recombine to form a product. The key question was whether the atomic nuclei, which act as a memory bank, could hold onto the information long enough to be useful. The results were surprising and humbling. The simulation showed that the molecule could indeed store information, but it did not offer a "quantum advantage." When the researchers compared the molecule's performance to a standard, non-quantum computer network of the same size, the classical network performed better, exceeding the molecule's capacity by nearly half to more than double. The quantum nature of the electrons did not make the system smarter; in fact, the system worked best because of the way the chemical cycle refreshed itself, not because of any magical quantum speed.

Perhaps the most significant finding concerns the timing. For years, critics argued that these molecules could not compute because the electron pair only lasts for a microsecond—a time so short it is useless for brain activity, which happens in milliseconds. The new study overturns this objection by changing the clock. The researchers realized that the electrons are consumed and replaced in every cycle, but the atomic nuclei are not. The nuclei survive the reaction and remain in the product molecule, carrying the memory forward to the next cycle. This means the "clock" of the computer is not the fleeting life of the electron pair, but the much slower speed at which the chemical reaction repeats itself. If the reaction happens every ten to one hundred milliseconds, the memory horizon stretches into the range of brain rhythms, potentially making biological quantum computing theoretically possible.

However, a new, stricter condition emerges from this possibility. For the memory to work, the same atomic nuclei must be reused in every single cycle. The simulation revealed that the memory is stored not in the delicate quantum phase of the atoms, but in their basic population, a more robust form of information. Yet, even this robustness has a limit. The study calculates that for the memory to survive the pause between chemical cycles, the specific atomic nuclei acting as the memory must be able to reorient themselves incredibly fast—about two and a half times faster than the protein molecule they are attached to. If they move as slowly as the protein, the memory fades in milliseconds, falling short of the neural timescale. Furthermore, the study identifies that only hydrogen atoms (protons) can serve as this memory; the nitrogen atoms in the molecule relax too quickly to hold any information.

The paper concludes that while the idea of a quantum brain is not dead, it is far more constrained than once hoped. The molecule does not compute better than a simple classical network, and its ability to function as a memory device depends entirely on a specific chemical criterion: the same proton must be reused across cycles, and it must move with a speed that current measurements suggest is unlikely for a protein in a cell. The researchers have not proven that this happens in the brain, nor have they ruled it out entirely. Instead, they have replaced a vague, untestable dream with a sharp, measurable requirement. They have defined exactly what must be true for this system to work: a specific type of atom, moving at a specific speed, in a specific chemical environment. This shifts the question from "Is it possible?" to "Can we measure this specific motion in a living cell?" It is a move from speculation to a concrete challenge for experimental chemistry.

Technical Summary: "What a radical-pair quantum reservoir would require"

Problem Statement
For decades, it has been conjectured that the electron–nuclear spin coherence underlying the radical-pair mechanism (RPM) of magnetoreception might support biological information processing, potentially enabling a "quantum brain." While the RPM is experimentally established as a sensor for weak magnetic fields (e.g., geomagnetic fields), its capacity to function as a computational reservoir has remained a qualitative hypothesis. Previous proposals often relied on abstract quantum channel models, assumed error correction, or identified the memory horizon with the short lifetime of the radical pair ( $\sim \mu$ s), leading to the conclusion that biological computation is impossible on neural timescales. This paper seeks to quantitatively test the radical-pair architecture as a reservoir computer, driven and read out under chemically consistent conditions.

Methodology
The authors model a cryptochrome-type separated radical pair (flavin–tryptophan) as a quantum reservoir computer. Unlike previous studies that reset a single electron or rely on magnetic resonance readouts, this study employs a chemically faithful protocol:

Drive: The reservoir is driven by the "birth" of a new spin-correlated electron pair in each turnover cycle, where the singlet character encodes the input. The nuclear spins (the register) are retained from the previous cycle.
Dynamics: The system evolves under the Liouville-space master equation, explicitly including electronic decoherence and spin-selective recombination.
Readout: Readouts are restricted to observables accessible to downstream chemical reactions, specifically product yields and nuclear polarisation (Chemically Induced Dynamic Nuclear Polarisation, CIDNP), rather than spectroscopic observables.
Baselines: Performance is measured via Information Processing Capacity (IPC) and Memory Capacity (MC), compared against matched classical echo-state networks (ESNs) and incoherent limits.
Clock Definition: The memory horizon is defined not by the radical-pair lifetime ( $\tau$ ) but by the turnover interval ( $T_{turnover}$ ), as the nuclear register survives recombination.

Key Results

Readable Capacity without Spectroscopy: The reservoir is readable by ordinary chemistry. Using a CIDNP readout (nuclear polarisation carried by the product), the system achieves an out-of-sample information-processing capacity of 4.7 against a memoryless floor of 2.0 (an excess of 2.64). Time-resolved kinetics also clear their respective floors. This demonstrates that the reservoir's output is accessible to biological downstream reactions without requiring EPR spectroscopy.
No Quantum Advantage: When compared to classical reservoirs matched by readout dimension (channel-for-channel), the quantum reservoir shows no advantage. At the biologically relevant operating point, a classical ESN exceeds the five-spin quantum reservoir by 49% to 123%. Even when matched by physical units (5 spins vs. 5 nodes), the quantum advantage is negligible (within statistical scatter).
The Clock is the Turnover, Not the Lifetime: The memory horizon is determined by the turnover interval ( $T_{turnover}$ ), not the radical-pair lifetime ( $\tau \sim 1 \mu$ s). Because the nuclear register survives recombination, the system can theoretically operate on millisecond-to-second timescales if the turnover rate allows.
The Nuclear-Register Criterion: The binding constraint for biological computation is not temporal but chemical: the same nuclear register must be reused across successive turnovers.
- Relaxation Mechanism: The memory is carried by nuclear populations (specifically $z$ -populations), not coherences. Therefore, the limiting factor is the nuclear spin-lattice relaxation time ( $T_{1n}$ ), not the dephasing time ( $T_{2n}$ ).
- Nucleus Specificity: The flavin $^{14}$ N nucleus is quadrupolar and relaxes in microseconds, wiping the register every cycle. The register must therefore be carried by protons ( $^1$ H).
- Reorientation Constraint: At the geomagnetic field ( $\sim 50 \mu$ T), a proton bound to a rigid protein relaxes in $\sim 2.4$ ms (due to extreme narrowing). To reach the neural band (10 ms – 1 s), the register's dipolar vector must reorient approximately 2.5 times faster than the intact protein ( $\tau_{eff} \lesssim 6$ ns).
- Drive Mechanism: The drive cannot be light-driven (photochemical) due to insufficient photon flux in a cellular environment; it must be catalytic.

Significance and Claims
The paper does not claim to prove that the "quantum brain" hypothesis is true. Instead, it replaces a qualitative conjecture with a falsifiable chemical criterion.

Refutation of Previous Objections: The paper overturns the standard "timescale objection" (that $\mu$ s lifetimes preclude computation) by showing the clock is the turnover interval. However, it replaces this with a stricter chemical constraint: the nuclear register must survive the pause between turnovers.
Modest Claim: The authors state that the "quantum brain" proposal survives only if a specific, narrow set of conditions is met: a catalytic radical pair must regenerate on the same molecule, the register must be proton-based, and the molecular reorientation must be fast enough to extend $T_{1n}$ beyond the turnover interval.
Testability: The central claim is that this criterion is testable in vitro using synthetic radical pairs and relaxation dispersion kinetics, without needing to measure parameters in living neurons. The paper concludes that while the radical pair can function as a reservoir, the specific conditions required for it to function as a biological reservoir in the neural band are highly restrictive and currently unverified.

Conclusion
The study demonstrates that a cryptochrome radical pair can function as a quantum reservoir with chemically accessible readouts, but it offers no quantum advantage over classical networks. The viability of this system as a biological computer hinges entirely on the chemical criterion of nuclear register reuse, specifically requiring a proton-based register with rapid reorientation dynamics ( $\tau_{eff} \lesssim 6$ ns) to overcome relaxation limits at the geomagnetic field.

Technical Summary: "What a radical-pair quantum reservoir would require"

More like this