This chapter tells the story of a hypothesis that failed, what replaced it, and why the failure made the theory stronger.
Every molecule in your body is vibrating. At thirty-seven degrees Celsius, thermal energy shakes proteins, rattles water, jostles every bond. That jostling, on a quantum scale, is enough to tear the whole system apart. For decades, the assumption was simple: biology is too warm and too wet for quantum coherence. Photosynthesis complicated things. Plants and bacteria move energy through protein scaffolds using quantum coherence, and the scaffolds are structured, membrane-shielded, and relatively quiet. Finding quantum effects there still surprised everyone, but the environment at least gave coherence a real glimmer of hope. The brain is not a chloroplast. It runs hotter, soaks in saltwater, fires electrical signals, and burns through metabolic fuel at a rate that would kill a plant cell. If quantum coherence barely survives in a photosynthetic membrane, asking it to operate in a neuron is asking an ice cube to survive a foundry.
The original version of Level B proposed that consciousness exploited Environment-Assisted Quantum Transport (ENAQT) in microtubule tryptophan networks. The argument was seductive. In photosynthesis, the FMO protein complex achieves optimal energy transport at intermediate noise. Microtubules contain vast tryptophan networks with similar architecture. And ENAQT would explain why decoherence is a feature rather than a bug.
Two numbers determine whether ENAQT works in a given system. 1) how fast the quantum signal hops between sites, and 2) how fast the environment scrambles it. If hopping dominates, coherence wins and the system behaves quantum-mechanically. If scrambling dominates, decoherence wins and the system behaves classically. The ENAQT sweet spot is in between. In microtubule tryptophan networks, the hopping rate (γ ≈ 0.2–2 ps⁻¹) and the dephasing rate (κ ≈ 17–20 ps⁻¹ at physiological temperature) give γ/κ ≈ 0.01–0.1 — the intermediate regime where ENAQT should operate.
The prediction was specific: quantum probability sculpting via excitonic transport through tryptophan chromophore networks at the ENAQT optimum.
We tested it. The results were unambiguous.
We ran three different simulation approaches: a phenomenological model, a physically-derived quantum dynamics calculation, and an evolutionary optimization over geometry parameters. The excitonic hypothesis does not work at physiological temperature.
The ENAQT sweet spot does exist in the microtubule geometry. The problem is where it falls. The optimal noise level for quantum transport enhancement is about 1,700 times quieter than the actual operating conditions inside a living cell. The system is simply too warm. At body temperature, thermal energy is nearly four times the strength of the quantum coupling between tryptophan molecules. It is like trying to hear a whisper at a rock concert. The quantum coherent dynamics are drowned out.
We tested every network size we could. Eight sites, thirteen, twenty, twenty-six. The quantum advantage over classical transport was 0.18% at every size. It did not scale. There was no hint that larger networks would help.
We also tested conformational tunneling as an alternative mechanism. The quantum rate was a million billion times smaller than the classical rate. Not viable.
The temptation after a negative result is to explain it away: adjust parameters, invoke shielding mechanisms, propose exotic environments where the effect might survive. We did not do this, because a theory that accommodates any result predicts nothing. The full simulation methodology, parameter choices, and sensitivity analyses are reported in [Malloy, "ENAQT and Radical Pair Spin Coherence in Microtubule Tryptophan Networks].
Instead, we asked: is there a different quantum mechanism in microtubules that operates at a timescale and energy scale where thermal energy does not dominate?
There is. And it was hiding in plain sight.
Picture two electrons spinning like tiny tops. If they spin in opposite directions — one clockwise, one counterclockwise — the pair is in what physicists call the singlet state, and one chemical reaction happens. If they spin together, the pair is in the triplet state, and a different reaction happens. The quantum part is that the pair doesn't commit to either configuration. It exists as both simultaneously, and the ratio between them shifts depending on the magnetic environment. Which products form depends on which state the pair is in when the reaction occurs.
That is a radical pair reaction. The key parameter is the spin coherence time: how long the quantum superposition of singlet and triplet persists before the environment destroys it.
In biological radical pair systems, spin coherence times are on the order of microseconds. Not picoseconds. Not femtoseconds. Microseconds. This is 5,000 times longer than excitonic coherence in tryptophan networks [B].
The reason is physical. Spin-spin interactions are vastly weaker than electronic excitation energies. The thermal bath that overwhelms excitonic coherence in picoseconds barely touches spin coherence for microseconds. The same warm, wet, noisy environment that kills one mechanism leaves the other largely intact.
We ran the same kind of simulation on radical pair dynamics that had failed for excitons. The results were categorically different.
Radical pair spin coherence showed a 12.7% quantum-classical yield difference, with coherence persisting for 1.48 microseconds at 310K (body temperature). Trajectory divergence between quantum and classical systems persisted for the entire 10 μs simulation window. Compare this to the excitonic mechanism, which showed 0.18% advantage with coherence gone in 0.25 ps. It was kind of a big deal.
In 2026, Zadeh-Haghighi and colleagues published a result in Science Advances that changes the conversation. They swapped magnesium isotopes in tubulin polymerization experiments under magnetic field. Magnesium-25 has a nuclear spin, a tiny magnet in the nucleus that can nudge electron spins in a radical pair between singlet and triplet states. Magnesium-24 and Magnesium-26 have no nuclear spin. Same element, same chemistry in every other respect. Under magnetic field, Mg-25 altered polymerization rates. Mg-24 and Mg-26 did not (P < 10⁻⁷). If the reaction depended on anything other than spin dynamics, swapping isotopes of the same element would change nothing. The substrate is real.
Other evidence points the same direction. Li and colleagues (2018) tested xenon isotopes as anesthetics in mice. Same gas, same chemical properties. The only difference is whether the nucleus has a spin. The spinful isotopes were weaker anesthetics. If consciousness had nothing to do with spin dynamics, swapping isotopes of the same noble gas would change nothing. Turin and colleagues (2014) showed anesthetics alter electron spin in Drosophila, another line connecting spin to consciousness.
Radical pairs are not new to biology. Birds navigate magnetic fields using them. But those radical pairs are activated by light. The microtubule mechanism runs on reactive oxygen. It works in the dark, inside a neuron, without a photon.
All of this establishes that radical pair chemistry occurs in tubulin. It does not establish that consciousness uses it. The Zadeh-Haghighi result is substrate, not function. The anesthesia results are consistent with radical pair involvement in neural processing, but "radical pairs play a role in anesthetic pharmacology" and "radical pairs in microtubules are the mechanism of consciousness" are separated by several uncontrolled steps. Different anesthetics work through different biochemical pathways, and none has established direct effects on microtubule radical pair chemistry specifically. The bridge from substrate to function is a research program, not an established result.
Twelve point seven percent does not sound like much. How can a 12.7% yield difference in radical pair reactions matter for consciousness?
The same way a compass matters for navigation. A compass does not need to be perfectly accurate. It needs to provide a consistent directional bias that is better than chance. Even a slight bias, a few percent improvement in the probability of choosing a fitness-relevant future, compounds over millions of decisions and billions of years of evolution into an enormous selective advantage.
Natural selection is exquisitely sensitive to small systematic differences. A 1% fitness advantage can fix an allele in a population in a few thousand generations.
Two caveats. First, the 12.7% is from a minimal model (two electron spins with a single nuclear spin and idealized dephasing). Real tryptophan radical pairs in microtubules involve multiple nuclear spins, spin-orbit coupling, and molecular motion. The actual yield difference in biological microtubules is unknown and could be substantially smaller. Second, the fitness advantage is not the yield difference itself but whatever behavioral difference the yield difference produces after transduction through the full chain from radical pair to neural signal. Each step in that chain has its own noise and attenuation (see Chapter 9).
But the 12.7% is the microscale effect. The macroscale effect depends on amplification. And that is where criticality enters.