Michael Kahana, a University of Pennsylvania psychologist, has been studying memory for over 30 years: how it works, and what’s going on when it doesn’t.
He’s not just fascinated with memory loss caused by traumatic brain injury — which affects more than 5 million people in this country — or the nearly 7 million Americans with Alzheimer’s. His research has also focused on the memory lapses that impact everyone, regardless of their cognitive health.
“We all have bad memories sometimes,” Kahana told The Post. “It fluctuates across the day, and it can fluctuate from moment to moment. That’s just how our brain circuits work. Once I realized that, then the question was, how do I get my brain to always be in its good mode?”
Kahana’s investigation into memory culminated with a landmark study, published last January, in which he and a team of researchers used computer interventions on a group of 47 epilepsy patients, delivering a pulse of electricity directly to the brain just as a memory lapse was about to occur. They did this via electrodes that had been implanted directly into the patients’ brains as part of their epilepsy treatment.
These electrodes — between 100 and 200 per person — are able to recognize brain signals when a patient is trying to remember something, and send a precisely-timed zap of electricity to the lateral temporal cortex, the part of the brain used for storing and processing memories.
The results were better than even Kahana could’ve hoped, with the brain stimulation leading to a 28% improvement in recall. While he remains cautiously optimistic, he can’t hold back his enthusiasm.
“I think that we are at the threshold of a new era in human neuroscience and human neurotherapeutics,” he said.
Kahana isn’t the only one exploring the possibilities of brain-computer interfaces. Across the country, scientists are developing brain-computer interfaces (BCIs) that can be used to treat everything from memory loss to speech disabilities to paralysis.
Just last year, patients in a Stanford Medicine study were so amazed at their memory improvements after a 90-day treatment with brain implants that a few of them refused to have the devices turned off.
And in August, Neuralink, the neurotechnology startup owned by Elon Musk, announced plans to insert a BCI — designed to give paralyzed patients the ability to use digital devices by thinking alone — into a second human test subject.
Noland Arbaugh, a 30-year-old Arizona man paralyzed from the neck down after a diving accident eight years ago, received the first Neuralink implant in January of this year. In a March livestream on X, Arbaugh demonstrated how he can use his thoughts to control a computer cursor to play games and email. In May, it was announced that the device had unexpectedly begun to detach from Arbaugh’s skull, but that the problem was fixed.
Musk has predicted there will be hundreds of people with Neuralinks within a few years and “millions within 10 years.”
In August, researchers at Switzerland’s Ecole Polytechnique Federale de Lausanne unveiled a brain that converts thought-to-text with 91% accuracy and is even smaller than Neuralink’s chip.
Progress is happening by such leaps and bounds that the FDA will hold a workshop later this month about clinical outcomes assessments for BCIs.
“If preliminary results are replicated, we could be years, not decades, away from some kind of meaningful assistive technology for individuals with severe disease and disability,” said Anna Wexler, a Perelman School of Medicine professor who studies the ethical, legal and social issues surrounding emerging technology.
When we think of computers helping patients with ALS (formerly known as Lou Gehrig’s Disease) to speak, the first person who comes to mind is usually Stephen Hawking, the acclaimed theoretical physicist who spoke with a microprocessing computer powered by Intel. While he could communicate, his voice sounded metallic, like a robot in a science fiction movie.
But for 45-year-old Casey Harrell, who lost his ability to speak due to ALS, a brain-computer interface called the BrainGate2 has given him back his voice — his actual voice.
It’s given Harrell the ability to communicate with his 5-year-old daughter.
“She hadn’t had the ability to communicate very much with me for about two years … ” Harrell told Scientific American in an August 2024 story. “I can help her mother to parent her. I can have a deeper relationship with her and tell her what I am thinking. I can simply tell her how much I love her.”
David Brandman, a UC Davis neurosurgeon who helped develop the brain chip, said the BCI interprets brain signals which are then reproduced by a voice assistant software.
“The system is about 97% accurate, and allows him to say words from a 125,000-word dictionary,” Brandman told The Post. “Using artificial intelligence, we have also recreated the sound of his voice so that the text can be spoken aloud by the computer to sound like him before he was diagnosed with ALS.”
For memory, the challenges get a little murkier. A person’s memory ebbs and flows, and the problem isn’t always consistent. It isn’t always about trying to provide general improvement of memory performance, according to Brent Roeder, Ph.D., but “to improve memory performance for specific important or urgent information, such as, ‘Did I take my medication this morning?’”
Roeder, a research fellow in the department of translational neuroscience at Wake Forest University School of Medicine, studies how to replicate individual codes within hippocampal activity for specific memory information.
He and his fellow researchers accomplished this with a “memory prosthesis,” an electrode inserted into the brain that interacts with the hippocampus, making neural recordings when a patient performs a specific memory task. “Once these unique memory codes were created, we used them to stimulate during the memory task to determine if we could increase the patient’s memory performance,” Roeder says.
In other words, they encoded memories for future reference, creating Post-It notes to remind the brain what it had forgotten.
As they discovered, it helped patients recall very specific information. It didn’t just improve their memory overall — although it did do that, with memory boosts anywhere from 11% to 54% — but specifically memory lapses that interfere with daily life, like forgetting where they put their car keys or if they turned off the stove.
One advantage to this type of approach is that it isn’t limited to a specific condition, said Roeder: “The hope is that once it is ready for clinical use, it will be able to be used to treat any condition that impairs memory function, from traumatic brain injury to dementia and Alzheimer’s.”
As exciting as the research is, there’s still the question of how this technology will be used. Or, as Wexler put it, “the blurring of lines between BCIs for treatment and enhancement.”
“If an implanted BCI allowed people to type at the same speed as we can type with our fingers or dictate with our voices, I doubt most people would be interested,” Wexler said. “But if it can make a really significant or measurable improvement — something that has not been demonstrated — that’s when things will get interesting.”
That seems to be what Musk is counting on. In a July 10 video posted to X, he asserted that the long-term goal of Neuralink is to “give people superpowers” and provide functionality “far greater than a normal human.”
But scientists like Roeder don’t share those ambitions. “The focus of our research has always been to restore function that has been impaired due to disease or injury,” he told the Post. “We feel that giving someone back what they’ve lost is a superpower.“
Just getting the technology to the point where it becomes widely available will be no small feat. It does, after all, involve brain surgery. As Tom Oxley, chief executive at brain-interface startup Synchron said during a 2022 TED talk, “The brain doesn’t really like having needles put into it.”
Kahana agrees that this is a hurdle. “We can’t modulate your brain with a raygun from far away,” he said. “So for this to work, you do have to get into the brain.” But, he adds, it’s becoming safer and safer to do so. “So much has changed in the last few years. The imaging is better, the electrodes are small. When the time comes, I wouldn’t hesitate to have this procedure done on myself.”
He co-founded Nia Therapeutics to help commercialize the brain implants, with funding from the Defense Advanced Research Projects Agency, part of an effort to help veterans with brain injuries. But it’s also personal for him.
“I have a son who can’t speak, he can’t say a discernible word. He uses a device to communicate, which as you can imagine is incredibly awkward. Searching through menus to find the right word. He knows what he wants to say, but how do you translate that brain pattern into spoken language?” Kahana explained. “You and I are doing it so easily, we take it for granted. But if someone could develop a technology to decode those brain signals, well… that would really be something.”
