The Spectrum of Proof
Every frequency has been proven biologically active — except the one worth $1.9 trillion a year
In December 2025, the FDA granted premarket approval to a medical device that treats major depression by passing a direct current of 2 milliamps through electrodes placed on the foreheadi. The therapeutic electric field in the brain cortex is 0.3 to 1.0 volts per meter. At this intensity, neurons change their firing patterns, neurotransmitter release is modulated, and symptoms of treatment-resistant depression improve in randomized controlled trials.
The same month, the International Commission on Non-Ionizing Radiation Protectioni maintained its position that non-thermal electromagnetic effects below established exposure limits do not occur in humans.
These two statements cannot both be true.
The device — Flow Neuroscience's tDCS headseti — is not an anomaly. It is the latest entry in a catalogue of 24 or more FDA-approved device categories whose clinical efficacy depends entirely on non-thermal electromagnetic biological effects. Each approval required what the FDA calls 'substantial evidence of effectiveness': clinical trials demonstrating that electromagnetic energy produces a measurable biological response without heating tissue.
The catalogue spans the entire electromagnetic spectrum. At the lowest frequencies, pulsed electromagnetic field (PEMF) bone growth stimulators have held FDA approval since 1979i. They accelerate fracture healing at field strengths of 1 to 100 hertz — frequencies so low they overlap with Earth's geomagnetic field. TENS units for pain management hold over 12,000 individual 510(k) clearances. Deep brain stimulation, approved in 1997, uses implanted electrodes to deliver continuous electrical pulses that control Parkinson's tremors.
Moving up the spectrum: repetitive transcranial magnetic stimulation (rTMS) for depression, cleared in 2008, induces currents in the brain without any physical contact. Vagus nerve stimulation treats epilepsy and depression. Cochlear implants convert sound into electrical signals that the auditory nerve interprets as hearing. None of these involve thermal heating. All require proof that electromagnetic energy changes biology.
Then comes the intermediate frequency range — 1 kilohertz to 1 megahertz — and the proof becomes impossible to ignore.
In 2011, the FDA approved Optune, a device manufactured by Novocure that treats glioblastoma brain cancer by applying alternating electric fields at 200 kilohertzi. The fields — called Tumor Treating Fields, or TTFields — disrupt cell division by interfering with the molecular machinery of mitosis. They do not heat the tissue. The mechanism is purely electromagnetic: the alternating field exerts forces on charged and polarized molecules in dividing cells, disrupting the formation of the mitotic spindle and causing abnormal chromosome segregation.
Novocure received a second FDA approval in 2015 for mesothelioma, and a third in 2026. The EF-14 phase III trial demonstrated a median survival increase from 16 to 20.9 monthsi — a result significant enough to change the standard of care for the deadliest brain cancer.
The Novocure patent — US 7,016,725i — contains a sentence that should appear in every discussion of electromagnetic safety. In describing which cells are most susceptible to disruption by intermediate-frequency fields, the patent states that 'cells in the ovaries or testicles may be sensitive to the electric fields.' The patent identifies a cell size–frequency relationship: the optimal disruption frequency is inversely proportional to cell diameter. For spermatogonia — the precursor cells of sperm, approximately 12 micrometers in diameter — the predicted resonant frequency is approximately 310 kilohertz.
This is the same frequency range produced by the switch-mode power supplies inside LED light bulbs: 20 to 200 kilohertz.
At the high end of the spectrum, the biological activity of electromagnetic radiation becomes so obvious that we have a special word for it: vision. The human retina responds to individual photons carrying approximately 10⁻¹⁹ joules of energyi. Photobiomodulation therapy — red and near-infrared light applied to tissue — holds FDA clearance for wound healing and pain management. Blue light phototherapy for neonatal jaundice has been standard care for decades. Ultraviolet phototherapy treats psoriasis and eczema.
Taken together, these 24+ device categories establish a single, inescapable fact: non-thermal electromagnetic biological effects are real, clinically proven, and regulatory-validated at every frequency from DC to ultraviolet light.
Every frequency except one range.
The gap sits between 300 megahertz and 6 gigahertz — the frequencies used by mobile telecommunications. This is not a gap in biology. It is a gap in acknowledgment.
The global telecommunications industry generates approximately $1.9 trillion in annual revenue. No regulatory body has approved a therapeutic device operating specifically in this band for non-thermal biological effects — not because such effects have been disproven, but because no manufacturer has submitted one for approval. The absence of a device approval is not evidence of absence of effect.
The safety standards for this frequency range — ICNIRP guidelinesi and IEEE C95.1 — are based on a thermal model: they protect against tissue heating, nothing else. The specific absorption rate (SAR) limits were established on the assumption that if tissue is not heated, no biological effect occurs. This assumption is contradicted by every device approval in the catalogue.
Consider the arithmetic of field strength. The FDA approved tDCS at a therapeutic cortical field strength of 0.3 to 1.0 V/m. Independent measurements of urban ambient radiofrequency fields in European cities range from 0.67 to 1.51 V/m. These are the same order of magnitude.
If 0.3 V/m of direct current is biologically active enough to treat major depression — active enough for the FDA to grant premarket approval based on clinical trial data — then 0.67 V/m of radiofrequency energy cannot be assumed biologically inert. The burden of proof has been inverted: it is no longer the responsibility of critics to prove that ambient EMF has biological effects. That proof already exists in the FDA's own device approval database. The responsibility now falls on safety standard-setters to explain why therapeutic-level fields are biologically active inside a clinic but biologically inert outside it.
They cannot, because the fields do not know they have left the building.
There is a deeper principle at work. The ion channels that mediate electromagnetic sensitivity in biological tissue are among the most evolutionarily conserved molecular structures on Earthi. Voltage-gated calcium channels, potassium channels, and sodium channels are found in organisms from bacteria to humans. Their basic architecture has been preserved for approximately three billion years.
These channels evolved in an electromagnetic environment that contained exactly two signals: Earth's steady geomagnetic field (approximately 25 to 65 microtesla) and the Schumann resonances (approximately 7.83 hertz and harmonics), generated by global lightning activity. For 3.8 billion years, there were no radiofrequencies. No intermediate frequencies. Nothing above a few hundred hertz.
Evolution builds filters for signals that recur. Organisms evolved sophisticated mechanisms to use the geomagnetic field for navigation and the Schumann resonances for circadian synchronization. But they built no filter for electromagnetic frequencies that nature never produced — because there was nothing to filter against.
This is the evolutionary calibration principle: biological electromagnetic sensitivity is calibrated to the natural environment. When the environment changes faster than evolution can respond, organisms are exposed to signals they have no mechanism to reject. The human electromagnetic environment has changed more in the last 130 years than in the preceding 3.8 billion. Mobile telephony — the dominant source of personal RF exposure — has existed for approximately 40 years. LED lighting — the dominant source of intermediate-frequency exposure — has been widespread for approximately 15 years.
Evolution operates on timescales of thousands of generations. Forty years is not a generation.
The therapeutic device paradox is not a conspiracy theory. It is a logical observation about the internal consistency of regulatory frameworks. The FDA and ICNIRP are not adversaries; they are two regulatory bodies whose positions are mutually incompatible on a question of basic physics.
The resolution is simple: either non-thermal electromagnetic biological effects exist (in which case safety standards must account for them) or they do not (in which case 24+ device categories are approved on a false premise and should be withdrawn). There is no third option in which electromagnetic fields are biologically active when applied by a physician but biologically inert when applied by a cell tower.
No one is proposing to withdraw tDCS, TTFields, bone growth stimulators, cochlear implants, or photobiomodulation devices. The clinical evidence is overwhelming. These devices work.
Which means the safety standards are incomplete.