What Your Brain-Computer Interface Will Actually Do — And Why Physicians Need To Say So

I am a physician-scientist. I have spent twenty-five years developing brain-computer interfaces, first at Brown University where I helped design the original BrainGate trial, and now at the Farber Institute for Neuroscience in Philadelphia where I direct a neurorestoration center and care for people with stroke, brain injury, and spinal cord disease. I have seen how deep brain stimulators and other medical devices can be transformative. I am neither Luddite nor technophile: I just want more options to help children and adults with neurological disease and injury.

So when I tell you that I think about consumer brain-computer interfaces — devices implanted in the skulls of healthy, able-bodied people who want to interact with AI using thought — I want to be clear about why. Not because I think the world needs an implantable iPhone.

While I do think there is an opportunity to create novel modes of computation and new types of minds, my instinct is that such augmentations will be most beneficial for scientists and inventors, when combined with AI that is aligned with what Mike Levin and colleagues term the Bodhisattva vow (Biology, Buddhism, and AI: Care as the Driver of Intelligence).

My rationale for this essay is not to espouse a transhumanist vision of cognitive superpowers as a solution to problems that can already be healed with much simpler, safer alternatives. Instead I write because the billions of dollars being invested in this technology are real, the devices are coming regardless of what I think, and the people I care for — children with cerebral palsy, adults recovering from brain injury, people with aphasia and stroke — will either benefit from that investment or be left out of it, depending on choices being made right now by people who are not thinking about them.

The Investor Thesis Nobody States Plainly

Here is what nobody in the BCI industry will say openly: the paralyzed patients in the current clinical trials are not the product. They are the path to the product. The investors behind Neuralink, Precision Neuroscience, Science Corp, and their peers are not primarily motivated by helping people with ALS or spinal cord injury — those markets are simply too small to justify billion-dollar valuations. They are betting on a mass consumer market for healthy adults. Elon Musk has said this explicitly. The investment math requires it.

The clinical trials serve a specific strategic function: they generate the FDA Pre-Market Approval that creates a competitive moat no rival can easily replicate. A decade of safety data, surgical refinement, and regulatory relationship cannot be reproduced overnight. So people with paralysis are, in a real if uncomfortable sense, the beachhead. The consumer market is the territory.

I am not saying this to be cynical. I am saying it because it is true, because it is not being said publicly, and because understanding it is necessary to have an honest conversation about where this technology is going and who it will serve.

The philosopher Thomas Koch wrote a necessary critique of transhumanist promises, comparing them to Professor Harold Hill in The Music Man: emotionally stirring, futuristically charged, and disconnected from the hard, pedestrian work of actually helping people. Some iBCI investors are, in Koch's framing, selling 76 trombones. We could do a great deal for human cognitive flourishing by improving education, eliminating neurotoxic pesticides, ensuring adequate sleep, making mental health care universally accessible. None of that requires drilling a hole in anyone's skull.

And yet. The devices are coming. People will be implanted. Some will have strokes because of the procedure. Some will have seizures, infections, device failures. These are not hypothetical risks — they are the documented complications of neurosurgical procedures, and any physician who pretends otherwise is not being honest. The question is not whether to have this technology. The question is whether physicians engage with it — help shape it, push it toward the people who need it most — or stand on the sidelines while it unfolds without us.

I have chosen to engage. Here is what I think is actually coming, stated as plainly as I can.

What the First Consumer Device Will Actually Do

Two small patches of cortex — one over the hand-and-arm area of motor cortex, one in Broca's area — together roughly the size of two postage stamps, are sufficient to decode imagined movements and imagined words with enough fidelity to replace a keyboard and screen. This has already been demonstrated in research participants with paralysis. The remaining challenges are engineering: miniaturization, battery life, surgical safety, long-term stability.

The first consumer iBCI will feel, initially, like a Palm Pilot. Remember Palm Graffiti — the simplified alphabet of stylus strokes you had to learn because the device could not yet recognize natural handwriting? Early BCI users will learn an analogous repertoire: imagined gestures that produce neural patterns consistent and distinct enough for the decoder to parse reliably. Awkward for a few weeks. Then second nature. Then invisible.

What you get, on the other side of that learning curve: picture yourself at dinner with a friend. Without moving, without speaking, without reaching into your pocket, you compose a message with an imagined gesture — invisible, interior, like a thought with a shape — and send it. Your friend, who also has a device, doesn't feel a buzz in their pocket. Instead, a small glowing sphere appears in the upper right of their visual field — not on a screen, but as a phosphene written directly to visual cortex. They glance at it inwardly, the way you'd glance at a notification, and your words appear. They reply the same way. To anyone watching, nothing has happened.

This is not a superpower. It is a more intimate version of the smartphone, with all the same risks of distraction, manipulation, addiction, and surveillance — except now inside your skull. The same companies that designed infinitely scrolling feeds to exploit dopamine systems will have direct access to the substrate those systems run on. That should concern all of us.

What Comes Next: Sensation, Emotion, New Senses

Once you have reliable bidirectional access to sensory cortex — not just motor cortex — the possibilities expand qualitatively. Your device, properly calibrated, could allow you to receive not just text and images but tactile sensation: the proprioceptive feeling of a tennis coach's arm placement written directly to your somatosensory cortex so you feel the correct motion rather than watching it, or the ambient emotional state of someone you love — not a reading, not a number, but something closer to what you feel standing beside them.

This requires what might be called a Qualia Dictionary: a personal map, built once, of what different stimulation patterns feel like to you specifically. Because no two brains are identical, a pattern that feels like "left and slightly cool" to one person may feel entirely different to another. You build your dictionary — a few hours, probably — and after that the system knows your perceptual vocabulary. This is the point at which BCI stops being a better phone and starts being a new sense.

The Ceiling Nobody Is Talking About

Every BCI company, and nearly every investor, is working toward what I'd call the Jedi Phone. Decode imagined movement and speech, write phosphenes and sensations back, wrap it in a low-profile implant. That is a real product with a real market.

But it has a ceiling lower than the pitch decks suggest.

The nervous system is clocked by its own biology — the time constants of neurons, synapses, and circuits, calibrated over hundreds of millions of years to the speed of the fastest eye movement. Adding recording channels to motor cortex does not make the brain faster. A thousand USB ports do not upgrade the CPU. The Jedi Phone is a better interface to the same cognition.

If you want to hold more in working memory, learn a discipline in days rather than years, maintain the cognitive acuity of your best day on your worst — you need not better I/O but more substrate: extracranial circuits (ectopic or extracorporeal) the brain can learn to use as its own. That requires layer-specific, bidirectional access to hippocampus, prefrontal cortex, thalamus, stably over a lifetime, at a resolution current silicon arrays cannot maintain. And it requires a theory of what those circuits are computing, not just what they are emitting.

The companies racing toward the Jedi Phone will build something real and valuable. But the cognitive augmentation the investor pitch decks actually describe is not accessible from where they are building. Getting there requires a different design premise, not a better version of the current one.

Why This Matters for My Patients

The path to the Jedi Phone and the path to genuine neurological restoration are, for a substantial distance, the same path.

A device that reads and writes reliably to motor and speech cortex in a healthy person can, in a person with stroke or traumatic brain injury, begin to reconstruct severed connectivity. Not by stimulating around the damage — the neuromodulation approach DBS and vagal nerve stimulators already use, that work for one subset of people and not for many others— but by rebuilding computational substrate. A new hip implant, not just physical therapy. A porcine mitral valve, not just medication.

Many of my patients cannot be helped by neuromodulation at all. Their disease has not simply disturbed a circuit — it has destroyed tissue. The child with cerebral palsy, the adult with a large hemispheric stroke, the patient with severe traumatic brain injury: for them the question is not how to stimulate what remains but how to replace what is gone. That requires a device that interfaces stably with brain structures over a lifetime, at laminar resolution, linked to extracranial computational modules the brain can learn to treat as its own.

That class of device does not yet exist. Building it requires understanding what the brain is actually computing — what the canonical circuits of thalamus, hippocampus, prefrontal cortex, and basal ganglia are doing when they work, so we know what to rebuild when they do not. The current generation of iBCI companies is not, for the most part, trying to solve this. Their implicit assumption — that AI will figure out the architecture once the I/O is good enough — is the same assumption genomics made in 2001. More data does not automatically yield mechanism.

The consumer vision, if it goes only as far as the Jedi Phone, will help a subset of people with paralysis and fall short for most others. If it proceeds to genuinely grapple with the architecture problem — with what it would actually take to give a healthy person expanded cognition rather than a better keyboard — it will, along the way, develop the tools that could give a child with cerebral palsy a functional life.

That is why I am writing about this. This is not about being for or against an implantable iPhone- we shoud respect the autonomy of able-bodied adults and make the risks as clear as possible so they can make an informed decision. I am writing because the decisions being made in the next five years about what to build, and how to build it, will determine whether the most vulnerable people in my clinic are an afterthought or a beneficiary. Healers must elevate the voices of those children and adults with neurological disease and injury — and help provide a voice for those who have no voice at all. We should be part of this conversation.