The Auditory Pathway as a General-Purpose Data Channel: capability scope, physical limits, and a dual-use analysis published alongside

Hearing technology treats the auditory pathway as a repair target. It is also a high-bandwidth sensory input the brain parses exceptionally well, already reached by implanted transducers in routine clinical use, and those transducers already accept arbitrary digital audio. This preprint examines what follows if that pathway is treated as a general-purpose input channel, and publishes the misuse analysis with the capability rather than after it.

The hardware is not a proposal. An active transcutaneous mastoid implant — a piezoelectric actuator on an osseointegrated screw, with the external processor held magnetically across intact skin — is a regulated commercial device with multicentre investigation, systematic review and meta-analysis behind it. The question is therefore not whether such hardware can be built but what it is permitted to carry and for whom.

The binding constraint is not bandwidth but encoding and learning, which puts the first work on non-implanted hardware: if an encoding does not work through a headset, implanting it will not rescue it. A section proposes interleaved frequency-band encoding of a learned spatial code for single-sided deafness, with the spectral budget computed and the cost to music perception stated rather than assumed away.

The constraint that should govern near-term work is a risk asymmetry stated in both directions. Implanting hardware in a person with normal hearing accepts surgical risk for convenience, where a deaf person's alternative is not hearing — but a large elective market is also how a surgical safety envelope gets characterised and funded, and small populations historically inherit technology they could not have paid to make safe. The document states the asymmetry and declines to resolve it, because that judgement belongs to the people who would live with the device.

The dual-use analysis traces nearly the whole risk profile to one property: the device cannot be taken off. Governance requirements are stated as things a regulator could require — a hardware output ceiling not defeatable in firmware, a physical mute independent of software, a funded explantation pathway as a condition of approval, and separate regulatory classes for restorative and elective use. Risk categories and remedies are given deliberately without implementation detail for any misuse pathway, and that omission is stated on the page.

Preprint, not peer reviewed. Prior-art searching was targeted and conducted at abstract and extract level; 11 claims are tagged sourced and 6 unverified.