A startup named Kernel came out of stealth mode yesterday and revealed its ambitious mission: to develop a ready-for-the-clinic brain prosthetic to help people with memory problems. The broad target market includes people with Alzheimer’s and other forms of dementia, as well as those who have suffered a stroke or traumatic brain injury.
If the company succeeds, surgeons will one day implant Kernel’s tiny device in their patients’ brains—specifically in the brain region called the hippocampus. There, the device’s electrodes will electrically stimulate certain neurons to help them do their job—turning incoming information about the world into long-term memories.
Kernel’s device will be based on a research effort led by Ted Berger, director of the Center for Neural Engineering at the University of Southern California. Berger tells IEEE Spectrum that his experiments with rats and primates make him confident that “it’s really time” for a clinical device. “We’re testing it in humans now, and getting good initial results,” he says. “We’re going to go forward with the goal of commercializing this prosthesis.”
Berger’s pioneering work on memory prosthetics was featured in an IEEE Spectrum article reporting on attempts to end all physical, emotional, and intellectual disabilities.
In Berger’s approach, electrodes in the hippocampus first record electrical signals from certain neurons as they learn something new and encode the memory. These electrical signals are the result of neurons “firing” in specific patterns. Berger studied how electrical signals associated with learning are translated into signals associated with storing that information in long-term memory. Then his lab built mathematical models that take any input (learning) signal, and produce the proper output (memory) signal.
An implanted memory prosthetic would have electrodes to record signals during learning, a microprocessor to do the computations, and electrodes that stimulate neurons to encode the information as a memory.
For people who have difficulty forming lasting memories on their own, the prosthetic would provide a boost. “We take these memory codes, enhance them, and put them back into the brain,” Berger says. “If we can do that consistently, then we’ll be ready to go.”
Prior research on memory prosthetics by both Berger and other neural engineers has received funding from DARPA, which also aims to develop a clinical device within the decade.
But the money behind Kernel, the new startup, comes largely from tech entrepreneur Bryan Johnson, who sold his payments company to PayPal for US $800 million in 2013. Johnson then started a venture fund called the OS Fund, which aims to “rewrite the operating systems of life” for the benefit of mankind. USC’s Berger says that Johnson “isn’t just trying to make his next $800 million” with Kernel (which Johnson will run as the company’s CEO). “He thinks the next big challenge for the human race is how to improve our brains,” says Berger.
The funding will support more human trials, which are conducted with hospitalized epilepsy patients who have temporary electrodes placed in their brains as part of their regular treatment. In human tests so far, the researchers have recorded from the hippocampus while the patients do memory tests, and have also been able to electrically stimulate the hippocampus to enhance the patients’ memorizing abilities.
Many fundamental questions about the science of memory formation remain to be answered, making it all the more remarkable that Kernel’s founders are already aiming to build a clinical device. For example, is there a common code for memories? If two people memorize the same list of words, do their electrical signals match, or do they each use a unique pattern of signals to encode the memory?
Berger says that in rats, the researchers did detect a “significant common code,” but that they couldn’t find one in their primate experiments. However, he adds, they studied far fewer primates, so they didn’t have as big a dataset to analyze. As for humans, “even if there is a generalized memory code, it’s going to be tough to find it using the tools we have right now,” Berger says.
The problem is that humans have way more neurons than rats; we have about 86 billion, while rats have about 200 million. So electrodes placed in the human hippocampus will record from a much smaller percentage of the neurons there. “Our information will be biased based on the neurons we’re able to record from,” Berger says. One of Kernel’s goals will therefore be to develop implants with denser arrays of electrodes that can record from more neurons.
If Kernel succeeds in turning basic science into an actual product, it wouldn’t be the first brain implant to hit the market. Implants are already used in a technique called deep brain stimulation (DBS), which has become an accepted therapy for Parkinson’s disease and an experimental treatment for depression and many other neuropsychological disorders. And in 2013, regulators approved the first brain implant for epilepsy; it monitors the brain for oncoming seizures and tries to prevent them.
Welcome to the age of the bionic brain.