This post originally appeared on MIT Technology Review
“Allí,” says Bernardeta Gómez in her native Spanish, pointing to a large black line running across a white sheet of cardboard propped at arm’s length in front of her. “There.”
It isn’t exactly an impressive feat for a 57-year-old woman—except that Gómez is blind. And she’s been that way for over a decade. When she was 42, toxic optic neuropathy destroyed the bundles of nerves that connect Gómez’s eyes to her brain, rendering her totally without sight. She’s unable even to detect light.
But after 16 years of darkness, Gómez was given a six-month window during which she could see a very low-resolution semblance of the world represented by glowing white-yellow dots and shapes. This was possible thanks to a modified pair of glasses, blacked out and fitted with a tiny camera. The contraption is hooked up to a computer that processes a live video feed, turning it into electronic signals. A cable suspended from the ceiling links the system to a port embedded in the back of Gómez’s skull that is wired to a 100-electrode implant in the visual cortex in the rear of her brain.
Bernardeta Gómez wearing the glasses with the cameras. Unfortunately, she no longer has the brain implant, which is still a temporary device.
Using this, Gómez identified ceiling lights, letters, basic shapes printed on paper, and people. She even played a simple Pac-Man–like computer game piped directly into her brain. Four days a week for the duration of the experiment, Gómez was led to a lab by her sighted husband and hooked into the system.
Gómez’s first moment of sight, at the end of 2018, was the culmination of decades of research by Eduardo Fernandez, director of neuroengineering at the University of Miguel Hernandez, in Elche, Spain. His goal: to return sight to as many as possible of the 36 million blind people worldwide who wish to see again. Fernandez’s approach is particularly exciting because it bypasses the eye and optical nerves.
Much of the earlier research attempts to restore vision by creating an artificial eye or retina. It worked, but the vast majority of blind people, like Gómez, have damage to the nerve system connecting the retina to the back of the brain. An artificial eye won’t solve their blindness. That’s why in 2015, the company Second Sight, which received approval to sell an artificial retina in Europe in 2011—and in the US in 2013—for a rare disease called retinitis pigmentosa, switched two decades of work away from the retina to the cortex. (Second Sight says slightly more than 350 people are using its Argus II retinal implant.)
During a recent visit I made to palm-studded Elche, Fernandez told me that advances in implant technology, and a more refined understanding of the human visual system, have given him the confidence to go straight to the brain. “The information in the nervous system is the same information that’s in an electrical device,” he says
Restoring sight by feeding signals directly to the brain is ambitious. But the underlying principles have been used in human-electronic implants in mainstream medicine for decades. “Right now,” Fernandez explains, “we have many electric devices interacting with the human body. One of them is the pacemaker. And in the sensory system we have the cochlear implant.”
This latter device is the hearing version of the prosthesis Fernandez built for Gómez: an external microphone and processing system that transmits a digital signal to an implant in the inner ear. The implant’s electrodes send pulses of current into nearby nerves that the brain interprets as sound. The cochlear implant, which was first installed in a patient in 1961, lets over half a million people around the globe have conversations as a normal part of everyday life.
“Berna was our first patient, but over the next couple of years we will install implants in five more blind people,” says Fernandez, who calls Gómez by her first name. “We had done similar experiments in animals, but a cat or a monkey can’t explain what it’s seeing.”
Her experiment took courage. It required brain surgery on an otherwise healthy body—always a risky procedure—to install the implant. And then again to remove it six months later, since the prosthesis isn’t approved for longer-term use.
Seizures and phosphenes
I hear Gómez before I see her. Hers is the voice of a woman about a decade younger than her age. Her words are measured, her cadence is perfectly smooth, and her tone is warm, confident, and steady.
When I finally see her in the lab, I notice Gómez knows the layout of the space so well she barely needs help navigating the small hallway and its attached rooms. When I walk over to greet her, Gómez’s face is initially pointing in the wrong direction until I say hi. When I reach out to shake her hand, her husband guides her hand into mine.
Gómez is here for a brain MRI to see how things look half a year after having her implant removed (they look good). She’s also here to meet a potential second patient who is in town, and in the room during my visit. At one point during this meeting, as Fernandez explains how the hardware connects to the skull, Gómez interrupts the discussion, tilts forward, and places the prospect’s hand on the back of her head, where a metal outlet used to be. Today there’s virtually no evidence of the port. The implant surgery was so uneventful, she says, that she came to the lab the very next day to get plugged in and start the experiments. She’s had no problems or pain since.
Gómez was lucky. The long history of experiments leading to her successful implant has a checkered past. In 1929, a German neurologist named Otfrid Foerster discovered that he could elicit a white dot in the vision of a patient if he stuck an electrode into the visual cortex of the brain while doing surgery. He dubbed the phenomenon a phosphene. Scientists and sci-fi authors have since imagined the potential for a camera-to-computer-to-brain visual prosthesis. Some researchers even built rudimentary systems.
In the early 2000s, the hypothetical became a reality when an eccentric biomedical researcher named William Dobelle installed such a prosthesis in the head of an experimental patient.
In 2002, the writer Steven Kotler recalled with horror watching Dobelle crank up the electricity and a patient fall to the floor writhing in a seizure. The cause was too much stimulation with too much current—something, it turns out, brains don’t like. Dobelle’s patients also had problems with infections. Yet Dobelle marketed his bulky device as nearly ready for day-to-day use, complete with a promotional video of a blind man driving slowly and unsteadily in a closed parking lot. When Dobelle died in 2004, so did his prosthesis.
Unlike Dobelle, who proclaimed a cure for the blind, Fernandez almost constantly says things lik, “I don’t want to get any hopes u,” and “We hope to have a system people can use, but right now we’re just conducting early experiments.”
But Gómez did in fact see.
Bed of nails
If the basic idea behind Gómez’s sight—plug a camera into a video cable into the brain—is simple, the details are not. Fernandez and his team first had to figure out the camera part. What kind of signal does a human retina produce? To try to answer this question, Fernandez takes human retinas from people who have recently died, hooks the retinas up to electrodes, exposes them to light, and measures what hits the electrodes. (His lab has a close relationship with the local hospital, which sometimes calls in the middle of the night when an organ donor dies. A human retina can be kept alive for only about seven hours.) His team also uses machine learning to match the retina’s electrical output to simple visual inputs, which helps them write software to mimic the process automatically.
The next step is taking this signal and delivering it to the brain. In the prosthesis Fernandez built for Gómez, a cabled connection runs to a common neuro-implant known as a Utah array, which is just smaller than the raised tip on the positive end of a AAA battery. Protruding from the implant are 100 tiny electrode spikes, each about a millimeter tall—together they look like a miniature bed of nails. Each electrode can deliver a current to between one and four neurons. When the implant is inserted, the electrodes pierce the surface of the brain; when it’s removed, 100 tiny droplets of blood form in the holes.
The implanted array has a 100 electrodes and resembles a tiny bed of nails.
Fernandez had to calibrate one electrode at a time, sending it increasingly strong currents until Gómez noted when and where she saw a phosphene. Getting all 100 electrodes dialed in took more than a month.
“The advantage to our approach is that the array’s electrodes protrude into the brain and sit close to the neurons,” Fernandez says. This lets the implant produce sight with a much lower electrical current than was needed in Dobelle’s system, which sharply reduces the risk of seizures.
The big downside to the prosthesis—and the primary reason Gómez couldn’t keep hers beyond six months—is that nobody knows how long the electrodes can last without degrading either the implant or the user’s brain. “The body’s immune system starts to break down the electrodes and surround them with scar tissue, which eventually weakens the signal,” Fernandez says. There’s also the problem of the electrodes flexing as someone moves around. Judging from research in animals and an early look at the array Gómez used, he supposes the current setup could last two to three years, and perhaps up to 10 before it fails. Fernandez hopes a few minor tweaks will extend that to a few decades—a critical prerequisite for a piece of medical hardware that requires invasive brain surgery.
Eventually, the prosthesis, like a cochlear implant, will need to transmit its signal and power wirelessly through the skull to reach the electrodes. But for now, his team has so far left the prosthesis cabled for experiments—providing the most flexibility to keep updating the hardware before settling on a design.
At 10 pixels by 10 pixels, which is roughly the maximum potential resolution Gómez’s implant could render, one may perceive basic shapes like letters, a door frame, or a sidewalk. But the contours of a face, let alone a person, are far more complicated. That’s why Fernandez augmented his system with image recognition software to identify a person in a room and beam a pattern of phosphenes to Gómez’s brain that she learned to recognize.
At 25 by 25 pixels, Fernandez writes in a slide he likes to present, “vision is possible.” And because the Utah array in its current form is so small and requires so little power to run, Fernandez says there’s no technical reason his team couldn’t install four to six on each side of the brain, offering vision at 60 x 60 pixels or higher. Still, nobody knows how much input the human brain can take from such devices without being overwhelmed and displaying the equivalent of TV snow.
What it looks like
Fernandez and his grad student with a prototype camera hooked up to the computer.
Gómez told me she would have kept the implant installed if she had been given the choice and that she’ll be first in line if an updated version is available. When Fernandez is done analyzing her array, Gómez plans to have it framed and hang it on her living room wall.
Back in Fernandez’s lab, he offers to hook me up to a noninvasive device he uses to screen patients.
Sitting in the same leather chair Gómez occupied during last year’s breakthrough experiment, I wait as a neurologist holds a wand with two rings against the side of my head. The device, called a butterfly coil, is connected to a box that excites neurons in the brain with a powerful electromagnetic pulse—a phenomenon called transcranial magnetic stimulation. The first blast feels as if someone is shocking my scalp. My fingers involuntarily curl into my palms. “Look, it worked!” Fernandez says, chuckling. “That was your motor cortex. Now we will try to give you some phosphenes.”
The neurologist repositions the wand and sets the machine for a rapid series of pulses. This time when she fires, I feel an intense zzp-zzp-zzp, as if someone were using the back of my skull as a door knocker. Then, even though my eyes are wide open, I see something: a bright horizontal line flashes across the center of my field of vision, along with two shimmering triangles filled with what looks like TV snow. The vision fades as quickly as it arrived, leaving a brief afterglow.
“This is like what Berna could see,” Fernandez says. Except her “sight” of the world was stable as long as the signal was being transmitted to her brain. She could also turn her head and, with her glasses on, look around the room. What I had seen were merely internal phantoms of an electrically excited brain. Gómez could actually reach out and touch the world she was looking at for the first time in 16 years.