Matt Angle’s claim might have sounded eccentric before: for years, he insisted that the key to solving one of neuroscience’s most intractable challenges lay in a 1960s-era technology invented in the tiny nation of Moldova.
It’s a lot harder to dismiss Angle’s approach now. Today, the U.S. Department of Defense selected Angle’s small San Jose-based company, Paradromics Inc., to lead one of six consortia it is backing with $65 million to develop technologies able to record from one million individual neurons inside a human brain simultaneously.
Recording from large numbers of neurons is essential if engineers are ever to create a seamless, high-throughput data link between the human brain and computers, including to restore lost senses.
That goal has been in the news a lot lately. In April, Elon Musk announced he was backing Neuralink, a $100 million company working on a brain-computer interface. Facebook followed up by saying that it had started work on a thought-to-text device to let people silently compose e-mails or posts.
The announcements generated worldwide headlines but also skepticism, since neither Musk nor Facebook disclosed how they’d pull off such feats (see “With Neuralink, Elon Musk Promises Human-to-Human Telepathy. Don’t Believe It.”).
Now the federal contracts, handed out by DARPA, offer a peek into what cutting-edge technologies could make a “brain modem” really possible. They include flexible circuits that can be layered onto the brain, sand-size wireless “neurograins,” and holographic microscopes able to observe thousands of neurons at once. Two other projects aim at restoring vision with light-emitting diodes covering the brain’s visual cortex.
Paradromics’s haul is as much as $18 million, but the money comes with a “moonshot”-like list of requirements—the implant should be not much bigger than a nickel, must record from one million neurons, and must also be able to send signal back into the brain.
“We are trying to find the sweet spot—and I think we have found it—between being at that cutting edge and getting as much information out at one time, but at the same time not being so far out that you can’t implement it,” says Angle.
Since learning to use metal electrodes to listen in on the electrical chatter of a single neuron a century ago, scientists have never managed to simultaneously record from more than a few hundred at once in a living human brain, which has about 80 billion neurons in all.
Angle, 32, says he ran into that problem as a graduate student. He wanted to study the way that odors are represented in the olfactory bulb, a part of your brain right behind your nose. But the effort was stymied by the lack of any way to record from more than a handful of neurons at a time.
That’s when a professor at Howard University, and the father of one of Angle’s old college friends, mentioned an obscure Moldovan company that had developed a way to stretch hot metal and mass-produce coils of extremely thin insulated wires, just 20 microns thick.
The technique, similar to the one that produces fiber-optic strands, is used to create antennas and to make magnetic wires that hotels can sew into towels to prevent customers from stealing them. But Angle and his collaborators—Andreas Schaefer of the Francis Crick Institute and Nick Melosh of Stanford University—realized the materials could let them make electrical contact with large numbers of brain cells at once.
Today, Angle says, his team orders spools of the wire and then bundles strands together in cords that are 10,000 wires thick. One end of the wires can be sharpened, creating a brush-like surface that can penetrate the brain as needles would. Angle says the thickness of the wires is calibrated so that it is strong enough not to buckle as it is pushed into the brain, but thin enough not to cause much damage.
The other ends of the wires are glued together, polished, and then pressed onto a microprocessor with tens of thousands of randomly spaced “landing pads,” some of which bond to the wires. These pads detect the electrical signals conveyed through the wires from the brain so they can be tallied and analyzed. Angle says “connectorizing” so many wires is what’s held such concepts back in the past.
In Paradromics’s case, the eventual objective is a high-density connection to the speech center of the brain that could let the company tap into what words a person is thinking of saying. But if the technology works out, it could also vastly expand the ability of neuroscientists to listen in as large ensembles of neurons generate complex behaviors, knit together sensory stimuli, and even create consciousness itself.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.