← PortfolioThe body and the brain

Science Corporation

Biohybrid brain-computer interfaces

Team

Max Hodakfounder

Founded

2021

Invested

2026

Links

Website

The problem

How do you wire into a brain without wrecking the part you touch?

Click the left tissue to push in another needle and watch what it costs. Click the device on the right to seed another neuron and watch its fibres grow in.On the left, needles are pushed into the brain. Each one cuts through the cells in its path, support cells gather and wrap it, and the signal at its tip fades. On the right, living neurons sit in the device and their own fibres grow down to the cells already there. Whether those links carry information is still an open question.An illustration, not real data.

How brain interfaces listen and talk

A brain-computer interface, or BCI, measures brain activity and turns it into something useful, most often control of a computer or a robotic limb. Options run from EEG caps on the scalp, through electrodes on the brain's surface, to pushed into the tissue. Going inside buys accuracy.

Invasive BCIs aim to pick up , the electrical spikes a neuron fires, from single cells or small groups near each electrode. The Utah array is a bed of 100 silicon needles that record only from their tips. Talking back to neurons can be done with electric current, or with light: makes chosen cells light-sensitive by giving them light-sensitive ion channels.

The eye is the same problem on a smaller stage. In age-related macular degeneration, the that turn light into electrical signals die, while the other retinal cells stay intact. An implant under the retina can stand in for the lost cells and hand its signal to the ones still working. caused by the disease is the leading cause of irreversible blindness and affects more than 5 million people worldwide.

Further reading Brain–computer interface (Wikipedia)Microelectrode array (Wikipedia)Optogenetics (Wikipedia)Macular Degeneration (Science Corporation)Retinal implant (Wikipedia)Subretinal Photovoltaic Implant to Restore Vision in Geographic Atrophy Due to AMD (New England Journal of Medicine (via Europe PMC))

Why it is hard

i.
Every probe is a wound
The brain has almost no spare room. Its neurons sit in a dense tangle of connections, so anything placed inside destroys some tissue, however fine. Destroying 10,000 cells to record from 1,000 can be worth it after a serious injury, but it scales badly. More channels have meant more damage.
ii.
The body fights back
Within hours of implantation, support cells called astrocytes and microglia gather around the device and start inflammation. Over time they wrap it in a sheath tens of micrometres thick, which insulates the electrodes and raises their impedance. Long-term implants tend to lose nearby neurons, collect and see working electrodes drop off, like a microphone slowly wrapped in felt.
iii.
Cells are fragile guests
Putting living neurons into a device brings its own trouble. Graft cells have to be compatible with the patient's immune system or they are rejected. Growing them from the patient's own cells would take many months and cost over a million dollars per patient, and making stem cells that hide from the immune system is a major undertaking. Then the neurons must survive low oxygen and blood sugar shocks while maturing inside live electronics.
iv.
Seeing in big pixels
A retinal implant's sharpness is capped by its pixels. With 100 μm pixels, the best acuity matches about 20/420, which is enough to read large letters. Patients report that faces are hard to make out.

Further reading Biohybrid (Science Corporation)Biohybrid neural interfaces: an old idea enabling a completely new space of possibilities (Science Corporation)Microelectrode array (Wikipedia)Maximizing the fidelity of a photovoltaic subretinal prosthesis for human patients (Journal of Neural Engineering (via Europe PMC))Simulation of prosthetic vision with the PRIMA system and enhancement of face representation (Journal of NeuroEngineering and Rehabilitation (via Europe PMC))

What Science Corporation is after

Science Corporation wants a connection to the brain with far more bandwidth than today's devices, without the damage that comes from putting wires in. Its idea turns the usual approach around: it integrates neurons into the electronics, rather than electronics into the brain.

Nearer term, it wants to restore some central vision to people whose photoreceptors have died, with a retinal implant called PRIMA.

Further reading Biohybrid (Science Corporation)Macular Degeneration (Science Corporation)

How they go at it

Step 1: Grow the wires
Stem cell-derived neurons are engineered, embedded in the device in the lab, then engrafted onto the brain. The cell bodies stay in the device, and only their axons and dendrites grow out into the brain to form new connections. The bet is that a million embedded neurons could form a billion synapses, which is a very different thing from counting electrodes.
Step 2: Light in, spikes out
The embedded neurons are modified with optogenetics so they fire when lit. MicroLEDs on one side do the writing, and recording electrodes detect when the embedded neurons are triggered by signals coming back from the brain.
Step 3: A first test in mice
In an early feasibility study, light-sensitive neurons were housed in a scaffold of tiny wells on the surface of the cortex. They survived, fired on their own and integrated with the host brain, and mice learned to report when their graft was stimulated. That is information going in, in mice, a long way from a human device.
Step 4: Solar panels under the retina
PRIMA pairs a tiny photovoltaic implant under the retina with glasses that project patterned near-infrared light, carrying both image and power. In an open-label trial with no separate control group, 26 of 32 people with geographic atrophy followed for a year gained a clinically meaningful amount of acuity. There were 26 serious adverse events in 19 participants, mostly in the two months after surgery, and nearly all of those early ones cleared up.

Further reading Biohybrid neural interfaces: an old idea enabling a completely new space of possibilities (Science Corporation)Biohybrid (Science Corporation)Optogenetic stimulation of a cortical biohybrid implant guides goal directed behavior (bioRxiv (via Europe PMC))Subretinal Photovoltaic Implant to Restore Vision in Geographic Atrophy Due to AMD (New England Journal of Medicine (via Europe PMC))Macular Degeneration (Science Corporation)

Still open

Does a graft that wires in actually carry information?
Physicians have treated a patient or two at a time with neural tissue transplants for various diseases. The grafts often seem to survive and integrate, and nothing noticeable typically changes for the patient. Biohybrid interfaces are at a low technology readiness level.
How sharp can prosthetic vision get?
Modelling suggests 3D electrodes could shrink retinal pixels to 20 μm, which would correspond to about 20/80 acuity, five times better than the current device. That still has to be built and tested in people.
Can implanted electrodes be made to last?
Problems with long-term implantation have driven much microelectrode array research, and electrodes still typically degrade over time, though progress has been made.

Further reading Biohybrid neural interfaces: an old idea enabling a completely new space of possibilities (Science Corporation)Maximizing the fidelity of a photovoltaic subretinal prosthesis for human patients (Journal of Neural Engineering (via Europe PMC))Microelectrode array (Wikipedia)

About Science Corporation

Science Corporation is developing invasive brain-computer interface technology using a novel biohybrid approach. Instead of traditional electrodes that damage brain tissue, Science's technology uses highly engineered, stem cell-derived neurons embedded in electronics that are engrafted into the brain, forming new biological connections.

The device sits on top of the brain and contains a silicon honeycomb-like structure with 100,000 "microwells" where individual living neurons grow from the microwell circuitry directly into the brain tissue.

Max Hodak graduated from Duke University's Pratt School of Engineering in 2012. He co-founded Neuralink in 2016 with Elon Musk and served as President through early 2021.

Words used here

microelectrode arrays: Grids of tiny electrodes, often on needles, that record from or stimulate neurons up close.
action potentials: The brief electrical spikes neurons fire to send signals.
optogenetics: A technique that gives chosen neurons light-sensitive proteins so light can switch them on or off.
glial scarring: The build-up of support cells around an implant that walls it off from nearby neurons.
photoreceptors: The retinal cells that turn light into electrical signals.
geographic atrophy: A late stage of macular degeneration in which patches of retina waste away.
biohybrid: A device that combines living cells with electronics, so the cells make the connection to the body.

Sources

How do you wire into a brain without wrecking the part you touch?

Every probe is a wound

The body fights back

Cells are fragile guests

Seeing in big pixels