An interactive computational exploration of fractal electromagnetic resonance in biological nanostructures
In 2020 and 2022, Anirban Bandyopadhyay and colleagues published findings reporting fractal electromagnetic resonance in single microtubule nanowires, with a self-similar "triplet of triplet" pattern repeating across four frequency decades (Hz, kHz, MHz, GHz). They also reported that cytoskeletal filaments fire ~250μs before the membrane ionic spike — challenging the classical Hodgkin-Huxley model of neural signaling.
These are extraordinary claims. Extraordinary claims deserve rigorous computational scrutiny — not dismissal, but not uncritical acceptance either.
This simulator was built to ask: Are these patterns mathematically plausible? Are they internally consistent? Could simpler structures produce the same effects? Not to prove or disprove, but to sharpen the questions that matter before (or alongside) expensive laboratory experiments.
The simulator has five interactive panels and two operating modes. In Visual mode, it provides intuitive animated illustrations of the phenomena. In Computed mode, it runs real numerical physics engines and displays quantitative results.
The triplet-of-triplet pattern across Hz, kHz, MHz, and GHz scales. Click any peak to zoom into its fractal sub-structure.
13-protofilament chiral cylinder with resonance mode visualization. Sweep driving frequency to find resonances.
Filament-first firing sequence showing the ~250μs lead time over the membrane ionic spike.
Optical vortex ring visualization of angular momentum states emitted by the internal clock assembly.
7 testable hypotheses with computational engines, honest revised verdicts, and 3 meta-analysis tools that interrogate the simulator itself.
Each hypothesis was tested computationally and assigned an honest verdict using a revised taxonomy that distinguishes between independent evidence, model self-consistency, and unvalidated claims.
| Hypothesis | Verdict | Summary | |
|---|---|---|---|
| H1 | Fractal Coherent Amplification | Plausible | 36 coherent modes give ~36x amplification vs random ~6x. But a regular lattice actually beats the fractal pattern. |
| H2 | Chirality Creates Triplets | Falsified | Helical modes do not cluster into the triplet pattern. Chirality alone does not explain the observed structure. |
| H3 | Boundary Dominates Bulk | Consistent | Sub-linear length scaling confirmed, but this is tautological — the model is confirming its own boundary assumptions. |
| H4 | Filament-First Temporal Ordering | Consistent | Fast oscillator leads slow oscillator as expected. Also tautological — built into the coupled oscillator structure. |
| H5 | Pitch Angle Optimality | Inconclusive | Peak near 12° (the actual MT pitch), but the result depends on the αcritical parameter. Weakest robustness (55%). |
| H6 | Thermal Noise as Fuel | Plausible | Stochastic resonance present in the model (SNR peaks at intermediate noise). But noise parameters are tunable, not empirically constrained. |
| H7 | Scale Invariance Predicts Schumann | Unvalidated | 4/5 Schumann harmonics matched, but Monte Carlo null test gives p=0.179. Not statistically significant. |
The fractal resonance pattern described by Bandyopadhyay is mathematically real but not uniquely special. Simpler ordered structures (like a regular lattice) can produce equal or greater coherent amplification. The specific triplet-of-triplet arrangement is not magic — order itself is what matters.
The strongest finding is stochastic resonance: the well-established physical phenomenon where noise actually improves signal detection in nonlinear systems. This mechanism is physically grounded, experimentally verified in other biological systems, and present in our model. If microtubules exploit stochastic resonance, it would be remarkable but not unprecedented.
The energy budget analysis provides a hard constraint: active MHz oscillation in microtubules is energetically implausible (requiring 16x a neuron's total power budget). Any valid mechanism must be passive or resonant, driven by ambient thermal noise rather than ATP hydrolysis.
The Schumann resonance alignment (H7) — perhaps the most provocative claim — does not survive statistical scrutiny. Random fractal patterns match Schumann harmonics almost as well (p=0.179).
Each hypothesis explained without jargon, along with what we tested and what it means.
Imagine 36 tuning forks. If they vibrate in random directions, they partly cancel out (like a crowd all talking at once — you get about 6x louder). If arranged in a specific pattern, they reinforce each other (like a choir singing in harmony — you get 36x louder). Our test found the fractal pattern does amplify the signal, but a simpler regular arrangement actually works slightly better. The nesting structure matters, but the specific fractal pattern is not uniquely optimal.
What we computed: Berry phase accumulation across 36 modes. Coherent sum = 36x, random average = 6x. Regular lattice = 22.9, fractal = 17.7.
Microtubules are twisted (chiral), like a spiral staircase. The question: does this twist create the triplet resonance pattern? We compared a twisted tube to a straight one. The twist changes which frequencies resonate, but it does not make them cluster into groups of three. Whatever creates the triplet pattern, it is not chirality alone.
What we computed: Cylindrical standing wave modes with and without helical pitch angle. Chiral modes shifted but did not cluster into triplets.
Does the amplification happen at the surface of the tube or throughout its interior? If it is a surface effect, making the tube longer should not help much (the surface-to-volume ratio decreases). Our model shows sub-linear scaling — consistent with a boundary effect. But this is the model confirming its own design, not independent evidence.
What we computed: Amplification factor vs. tube length. Scaling was sub-linear. Caveat: Tautological test — the boundary assumption was built in.
Classical neuroscience says the membrane fires first, then the cytoskeleton responds. Bandyopadhyay's team measured the opposite: filaments fire 250 microseconds before the membrane. We modeled this as a fast oscillator (filament) coupled to a slow one (membrane). The fast one does lead — but that is simply what coupled oscillators do when one is faster. The model confirms its own structure, not the biological claim.
What we computed: RK4 integration of coupled ODEs. Fast oscillator phase-leads slow oscillator. Caveat: Expected from the math, not an independent prediction.
Microtubules have a specific twist angle (~12 degrees). Is this angle special? We swept from 0 to 45 degrees and found a peak near 12 — but the peak position depends on a free parameter we can tune. With different parameter choices, the peak moves. This is the least robust result (55% stability under perturbation).
What we computed: Berry phase resonance condition and coupling efficiency vs. pitch angle from 0° to 45°. Peak near 12° but sensitive to αcritical.
Usually, noise destroys signals. But in certain nonlinear systems, a specific amount of noise actually makes the signal stronger — a well-known phenomenon called stochastic resonance. We found it in our model: too little noise and the signal is weak, too much and it drowns, but at an intermediate level, the signal-to-noise ratio peaks. This is real physics, experimentally verified in other biological systems. The caveat: we can tune the noise parameters to get this result.
What we computed: SNR vs. noise amplitude for 36 modes in a bistable potential (Euler-Maruyama integration, 5000 trials). SNR peaks at intermediate noise.
If the fractal pattern truly extends to all scales, it should predict Earth's Schumann resonances (electromagnetic standing waves in the atmosphere at 7.83 Hz, 14.3 Hz, etc.). Our extrapolation matched 4 out of 5 harmonics. Sounds impressive — until we ran 10,000 random fractal patterns and found that 17.9% of them match just as well. This is not statistically significant (p=0.179). The apparent alignment may be coincidence.
What we computed: 10,000 Monte Carlo trials with random triplet patterns. Observed 4/5 matches vs. random mean of 2.56±1.01 matches. p=0.179.
This simulator was built as a collaboration between human scientific curiosity and AI computational capability. It uses Claude Opus 4.6 (Anthropic) as both a development partner and a reasoning engine for framing, testing, and honestly interpreting computational results.
Feedback, corrections, and contributions are welcome. If you are an experimentalist with access to microtubule spectroscopy, we would be particularly interested in your assessment of the predicted resonance frequencies.