Invasive electrodes have substantially higher bitrate, but surgical risk confines them to medically necessary settings

What it is

Electrodes on or in cortex sit near neuronal populations and skip skull filtering, so bitrate can sit well above the scalp. The exchange is craniotomy, infection, package failure, and chronic foreign-body reaction. That capability is therefore pinned by medical necessity: use is discussable when the gain in communication or motor replacement outweighs surgical harm, not as an upgraded consumer headset.

Why it happens

Local field potentials and multiunit firing supply millimetre to sub-millimetre spatial scale; decodable dimensions rise, and continuous movement plus faster selection become physically reachable. Cost lives on the implant chain: stroke and infection from surgery itself, glial scarring around probes, mechanical fatigue of leads or wireless modules, and revision surgery to explant or replace. SNR can fall over months to years as glia encapsulate the device; upkeep is not “swap a cap.” Ethically, unimpaired people or reversible conditions almost never see benefit that covers the harm; locked-in or severe paralysis flips the comparison. Regulation placing these devices on a medical-device path is itself an institutional statement of the bound.

Studying it

Clinical BCI trials report post-implant decode rate and adverse events (infection, device failure, seizures). Factors: array type (cortical surface / microelectrode), implant site, decode task. Outcomes: bitrate or cursor throughput, months of device availability, serious adverse events. Comparisons with scalp systems must count surgery and care, not only peak rate. Ns are small and selection-biased (often people already in a neurosurgical path); extrapolation should stay cautious.

Where it stops holding

Semi-invasive routes (subdural grids, endovascular electrodes) try the middle; risk and bandwidth both sit between scalp and penetrating probes, and long-term data remain thin. Animal lifespan and immune response are not adult-years. If packaging and wireless power someday drive infection extremely low, the medical-necessity bar may move down; that is unpaid engineering, not a premise for current consumer products. Marketing a scalp headset with bitrate plots from invasive papers welds the two bounds together.

Applying it

• Do not promise continuous control or typing rates from invasive papers in consumer or office products.
• For clinical products, write surgical informing, a revision plan, and a non-invasive fallback on failure into the main flow, not an appendix.
• When explaining capability to non-clinical audiences, give bandwidth and irreversible harm together; do not only show the cursor video.
• Verify that any comparison table has a column “craniotomy / implant required”; rate comparisons without that column must not support a product claim.

Related

• Same group: C9.07.1 Non-invasive EEG can reliably separate only a few coarse intents, not continuous fine control · C9.07.3 Current BCIs fit discrete choice, not replacing continuous pointing or text input · C9.07.4 Capability bounds move with signal-processing progress; they are not fixed physical limits
• Adjacent: C9.02 Brain-Computer Interfaces · C9.02.3 Currently suited as a replacement channel rather than an enhancement
• Search: invasive BCI · surgical risk · electrocorticography