Vision Quest
A HALF CENTURY OF
ARTIFICIAL-SIGHT RESEARCH HAS SUCCEEDED. AND NOW THIS BLIND MAN CAN SEE.
By Steven Kotler
I'M SITTING ACROSS
FROM A BLIND MAN CALL HIM PATIENT Alpha at a long table in a windowless
conference room in
|
|
Our guinea pig is 39,
strong and tall, with an angular jaw, bold ears, and a rugged face. He looks
hale, hearty, and healthy except for the wires. They run from the laptops
into the signal processors, then out again and across the table and up into the
air, flanking his face like curtains before disappearing into holes drilled
through his skull. Since his hair is dark and the wires are black, it's hard to
see the actual points of entry. From a distance the wires look like long
ponytails.
"Come on,"
says William Dobelle, "take
a good look."
From a few steps
closer, I see that the wires plug into Patient Alpha's head like a pair of
headphones plug into a stereo. The actual connection is metallic and circular,
like a common washer. So seamless is the integration that the skin appears to
simply stop being skin and start being steel.
"It's called a percutaneous pedestal," Dobelle
tells me.
All I can do is stare. The man has computer jacks sunk into both sides of
his skull.
On the far side of the
pedestal, buried beneath hair and skin, is the wetware: a pair of brain
implants. Each one is the size of a fat quarter, a platinum electrode array
encased in biocompatible plastic.
Dobelle has designed a
three-part system: a miniature video camera, a signal processor, and the brain
implants. The camera, mounted on a pair of eyeglasses, captures the scene in
front of the wearer. The processor translates the image into a series of
signals that the brain can understand, then sends the
information to the implant. The picture is fed into the brain and, if
everything goes according to plan, the brain will "see" the image.
But I'm getting ahead
of myself. The camera's not here yet. Right now the laptops are taking its
place. Two computer techs are using them to calibrate the implants.
One of the techs
punches a button, and a millisecond later the patient rotates his head, right
to left, as if surveying a crowded room.
"What do you
see?" asks Dobelle.
"A medium-size phosphene, about 5 inches from my face," responds the
patient.
"How
about now?"
"That one's too
bright."
"OK," says Dobelle, "we won't use that one again."
|
|
This goes on all
morning, and it's nothing new. For almost 50 years, scientists have known that
electrical stimulation of the visual cortex causes blind subjects to perceive
small points of light known as phosphenes. The tests
they're running aim to determine the "map" of the patient's phosphenes. When electrical current zaps into the brain,
the lights don't appear only in one spot. They are spread out across space, in
what artificial-vision researchers call the "starry-night effect."
Dobelle is marshaling these
dots like pixels on a screen. "We're building the patient's map, layer by
layer," he explains. "The first layer was individual phosphenes. The next layer is multiples. We need to know
where his phosphenes appear in relation to each other
so a video feed can be translated in a way that makes sense to his mind."
Some phosphenes look like pinpricks or frozen raindrops. Others
appear as odd shapes: floating bananas, fat pears, lightning
squiggles. Of course, the use of the word appear
is misleading, since the phosphenes appear only in
the patient's mind. To the sighted, they are completely invisible.
Suddenly, the color
drains from the patients face. His deadened eyes roll back. Then
another warping convulsion.
Dobelle sits in a wheelchair
beside the patient. His left leg was amputated a year ago after an ulcerated
infection in his big toe spread out of control. Because being in a wheelchair
makes it hard to dig into his pants pockets, he favors T-shirts - "the
good kind" - with a chest pocket to carry his keys, a couple of pens, his wallet. His shirt is so weighed down that it sags from
his neck, drooping cleavage-low. He has a patchy, unkempt gray beard. His
forehead is high and wrinkled, and his glasses are thick and wide.
"Are we ready for
multiple phosphenes?" asks one of the techs.
Dobelle nods his head.
So smoothly has the
morning been going that while we're talking, the techs allow the patient to
take control of the keyboard and begin stimulating his own brain.
This isn't standard operating procedure, but with the excitement, the techs
don't stop him and the doctor doesn't notice.
Suddenly, the color
drains from the patient's face. His hand drops the keys. His fingers crimp and
gnarl, turning the hand into a disfigured claw. The claw, as if tethered to
balloons, rises slowly upward. His arm follows and suddenly whips backward,
torso turning with it, snapping his back into a terrible arch. Then his whole
body wrenches like a mishandled marionette shoulders tilting, neck craning,
legs twittering. Within seconds his lips have turned blue and his deadened eyes
roll back, revealing bone-white pupils, lids snapping up and down like
hydraulic window shades. There's another warping convulsion, and spittle sails
from his mouth. Since the doctor's in a wheelchair and the techs seem
hypnotized, I rush over and grab him.
"Call 911!"
one of the computer techs shouts.
But the doctor yells
back: "No!"
"Lie him
down," cries the other. "Get him some water!"
"No!"
My arms are under his,
trying to steady the weight. His head snaps toward mine, and I take it on the
chin with the force of a solid right cross. We're now close enough that I can
count the wires going into his head. I can see a faint scar where a surgeon's
saw cut a hole in his skull and removed a chunk of it like a plug from a drain.
Finally, the techs move to action. They're up and struggling to unhook the
patient from the seeing machine but really, what can they do? It's in his
brain. I'm pretty sure he's going to die in my arms.
WILLIAM DOBELLE LIKES
A GOOD WRIGHT BROTHERS STORY. Like how the first plane the Wright brothers
built didn't have a steering mechanism, that it merely went up and down and
straight. Or if you look at a plane these days you won't see their names on the
side. Instead there's Boeing or Airbus, but even so, you know these makers are
merely historical recipients of the Wright stuff, just as you know that your
voting privileges are somehow owed to Thomas Jefferson.
Of all the Wright brothers stories, Dobelle likes
the one about Lieutenant Tom Selfridge the best.
|
|
The Wright brothers
ran low on money. They built their airplane, but they needed more cash for
further experimentation. A lieutenant from the US Army showed up for a
demonstration, and after watching Orville pilot around for a little while he
said, "That's great, now take me for a ride." So Orville strapped Selfridge into the passenger seat, took off, and promptly
crashed. Crashed! The plane was wrecked, Orville was in the hospital for
months, and Selfridge was killed yet they still
managed to land a contract for a military flier.
The doctor treats this
story like a talisman. Its moral with great risk comes
great reward has been an inspiration for him during the past 30 years, since
1968, when he began working on an artificial-vision system to restore sight to
the blind. The moral was there in the '70s, when he went under the hot knife of
surgery and had his own eye slit open to test the feasibility of a retinal
implant. It was there when he looked over the work that had been done on the
visual cortex and realized the only way to create a visual neuroprosthesis
was to slice through the skull and attach an implant to the human brain. It was
there two years ago, when he decided to skirt the Food and Drug Administration
by sending his patients to a surgeon in
There was one lab rat,
however. In 1978, shortly before the FDA passed the last in a series of medical
device amendments that would outlaw testing a visual neuroprosthesis
on a human, Dobelle installed his prototype into the
head of a genial, big-bellied, blind Irishman from
"When my
grandkids meet a blind guy with a brain implant," says Jerry, explaining
his participation in Dobelle's experiments, "I
wanted them to be able to say, 'Let me tell you about my grandfather.'"
For years the
prototype sat in Jerry's occipital lobe, largely unused. Back then Dobelle's concerns were infection and biocompatibility.
When neither turned out to be a problem, he edged the research forward. Over
the years, Jerry's visual field was mapped, but his implant never produced true
"functional mobility."
Functional mobility is
a bit of jargon defined as the ability to cross streets, take subways, navigate buildings without aid of cane or dog. For the past
40 years this has been the goal of artificial-vision research. But Jerry's not
there, instead caught halfway between sight and shadow.
When hooked up to a
video camera, he sees only shades of gray in a limited field of vision. He also
sees at a very slow rate. It helps to think of film. Normal film whirls by at
24 frames per second but Jerry sees at merely a fifth of that speed. The
effect, Dobelle tells me, is a bit like looking at
snapshots in a photo album through holes punched in a note card.
Patient Alpha, on the
other hand, has the full upgrade: the Dobelle
Institute Artificial Vision System for the Blind. Because the system has yet to
be patented, the doctor is cagey about specifics. He won't say how many
electrodes are inside the patient's head, though by my count the number is
around 100. Other changes have been made as well. Instead of Jerry's one
implant, the patient has two, one in each side of his head. Materials, as well,
have been updated, and the power pack and signal processor made portable. But
the biggest difference is that it took Dobelle 20
years to work Jerry up to any sort of vision. Patient Alpha got out of surgery
a month ago.
WILLIAM DOBELLE WAS
BORN IN 1941 IN
Which
sounds like hooey, until you check the records. He applied for his
first patent, on an artificial hip improvement, at age 13. He was into college
at 14 and hooked on the artificial-vision challenge by 18. He dropped out of
Vanderbilt to pursue independent research on visual physiology, supporting
himself as a Porsche mechanic.
Bulky and expensive,
early systems took 20 years to work up to any sort of vision. Patient Alpha got
out of surgery a month ago.
In 1960 he returned to
school, earning an MS in biophysics from Johns Hopkins. This time he covered
costs by selling scientific ephemera: iguana gall bladders and whale hearts
which he collected in
Located in
Walk inside and you'll
see a carpet so thin it could be cement. The furniture in the front offices
looks anonymous, wood-veneered, bought by the pound. Behind the offices is a
larger workshop the home to the breadwinners of the operation.
During his tenure as a
spare-parts man, Dobelle built hiccup suppressors and
erection stimulators and pain inhibitors. Right now, there are 15,000 people
running around the world with his inventory inside their bodies. The workshop
is currently used to build lung, spinal cord, and deep-brain stimulators. Since
he's never wanted to be beholden to anyone and thus never accepted venture
capital, these devices pay the rent so Dobelle can
pursue his real goal: artificial sight.
"It doesn't come
cheap," says Dobelle, rolling himself into the
workshop so I can get a look. We pass a machine shop drill presses, lathes,
saws of all varieties, tools hung on pegs and others left out among the dust
and metal filings then out onto an assembly room floor. In the center,
separated from the rest by long sheets of heavy plastic, there's a clean room
for delicate procedures. And against a far wall stands an ancient computer,
weighing 2 tons, complete with a punch-paper tape input and a Teletype output.
It measures 10 feet wide and 7 feet tall.
"What is that
for?" I ask.
"That was the
first artificial-vision system, the one I built for Jerry. It's my past.
Thirty-four years of work and $25 million."
THE COST HAS COME DOWN
QUITE A BIT. ACCORDING TO A printout Dobelle hands me, the
price tag for curing blindness is now around $115,000:
Visual Prosthesis
System:
$100,000
1 miniature camera mounted on eyeglasses
1 frame grabber
1 microcomputer
1 stimulus generation module
2 implanted electrode arrays with percutaneous
pedestals
3 sets of rechargeable batteries and 1 charger (customer is responsible for
replacement batteries as needed)
5-year full warranty (not including travel or freight)
5 years of annual follow-up examinations in Portugal (not including travel)
unlimited telephone consultation
Evaluation of patient: $2,000 psychiatric evaluation/all other testing
Hospital expenses: $10,000
Miscellaneous expenses: $5,000 airfare to Lisbon, hotel and food for one
week (2 people) miscellaneous (such as taxis)
|
|
The first person ever
to receive this bill was Patient Alpha. His given name is Jens pronounced
"Yens." Twenty-two years ago, at age 17, while nailing down railroad
ties, an errant splinter took his left eye. Then, three years later, this time
fixing a snowmobile, a shiv of clutch metal broke
free and took out his right.
He lives in rural
Starting from scratch
and without the aid of sight, Jens designed and built a solar- and wind-powered
house and pulled his family off the grid. In his spare hours, he programs
computers, tunes pianos, and gives the occasional concert. For
a blind man to give a classical recital requires memorizing whole scores a
process that can take nearly five years. To cover his surgery, Jens gave
quite a few recitals.
BACK IN THE LAB, I'M
STILL SUPPORTING JENS' WEIGHT. HE'S panting and jerking. Every pore on his body
leaks sweat. His neck has gotten too slippery to hold, so I've jammed my right
hand into his armpit. I can feel the throb of his axillary
artery. His heart is beating. Thankfully, he's still alive.
Over the next five
minutes, the gasping subsides. Respiration returns to normal. The full-body
twitch stills to the occasional flutter. Soon the grim rigor of his hand
relaxes, his fingers merely stretching now, as if reaching for the far notes on
his piano.
Dobelle's glaring at the techs.
"What
happened?" he demands.
"He was overstimulated."
"Yeah, I know
that."
Beside him, Jens' head
bobs once and then again. Slowly, motor control returns. He stretches his arms
as if waking from a long sleep.
"What
happened?" echoes Jens, his voice a low, percolating gurgle.
"You had a
seizure," says Dobelle.
"I wha ..."
"A
seizure. Jerry never had one, but it was always a possibility."
"I wha ..."
"You'll be
fine," says Dobelle.
"For what I paid
..."
"What?"
"For what I paid,
I better be."
"OK," says Dobelle, "I think we're done for today."
LATER
THAT NIGHT, DOBELLE CALLS TO EXPLAIN. HIS VOICE IS balmy,
preternaturally pacific.
"My surgeon is
the world's foremost expert on epilepsy. When someone's having a seizure you
don't lie them down or give them water they could choke. I knew he would be
OK."
And the next morning,
when I walk into the lab, Jens is OK. He's back at the table, amid another
round of testing. He doesn't remember much of the seizure, but he remembers
seeing the phosphenes.
"It was
wonderful," says Jens. "It is wonderful. After 18 years in a dark
jail, I finally got to look out the door into the sunlight."
"Are you ready
for a little more?" asks Dobelle. In his hand is
a pair of oversize tortoiseshell glasses. The left lens is dark, and affixed to
the right is a miniature video camera: black, plastic, and less than 1 inch
square. The wires that yesterday ran from the laptops are now plugged into the
camera. It's time to see if Jens can see.
"Are you
ready?" repeats Dobelle.
"I've been ready
for 20 years."
Jens slides the
glasses onto his face, and the techs power up the system. I am sitting across
the table from him. As it turns out, when the world's first bionic eye is turned
on, Jens sees me.
"Wow!" says
Jens.
"Wow what?"
I ask.
"I'm really using
the part of my brain that's been doing dick-all for two decades."
"And that's only
one implant," says Dobelle. "We still have
to integrate the other side, and we haven't installed the edge-recognition
software yet. The image is going to get better and better."
Jens turns away, and
we clear all objects off the conference table. Dobelle
picks up a telephone and puts it down on the far corner. Jens turns back
around. The camera is sending data down the pipe and to the implant in his
brain at 1 frame per second. So when he first scans the table his head swivels,
robotic and turtle-slow. It takes him nearly two minutes to find the phone
but he finds the phone. Then we do it again. Fifteen minutes later, Jens can
pick up the receiver in less than 30 seconds. Within a half hour, it takes him
less than 10.
They gradually work the frame speed up until there's nothing left to do but strap the signal processor and power pack to Jens' hips, like guns in their holsters. Then Jens heads out back, where he climbs inside a convertible Mustang. The top is down. The wind is in his hair. He fires up the ignition. Dobelle doesn't let him tour the freeways, but he has his way with the parking lot.
"The next version," Dobelle tells me, "may have enough resolution to use while driving in traffic." In fact, since this is only a simple camera we're talking about, one could imagine the addition of any number of superhuman optical features: night vision, X-ray vision, microscopic focus, long-range zoom. Forget the camera even; there's no reason you couldn't jack directly into the Net. In the future, the disabled may prove more abled; we may all want their prostheses.
PUBLIC DISCUSSION OF ELECTRICITY'S EFFECT ON VISION dates to 1751, when it was addressed by Benjamin Franklin following his celebrated kite-and-key experiment. Despite some advocates, the idea of treating blindness through electrical stimulation did not catch on.
The human eye occupies a
weird place in history. For more than a century, creationists, staring down
On June 13 Dobelle addressed the annual meeting of the American
Society of Artificial Internal Organs in
In fact, to most of the artificial-vision community, Dobelle's breakthrough came out of the blue. For years he had been merely a footnote, known mainly for his early work in phosphene stimulation. People had heard of Jerry, but because the testing was done privately, outside of academia, many felt the work suspect.
Dobelle leads one of a dozen teams spread
out over four continents racing ahead with all sorts of artificial-vision
systems. There are teams working on battery-powered retinal implants and
solar-powered retinal implants, and teams growing ganglion cells on silicon
chips, and teams working on optic-nerve stimulators. And there is Dick Normann, the former head of the
Like Dobelle, Normann is working on a visual neuroprosthesis. I was the first to tell him that the race was over: He lost.
"That's fantastic," Normann says.
"You're not even mad?"
"Fantastic, fantastic, fantastic" and then he pauses "if it works."
"What do you mean? I was there. I saw it work."
"But what do you mean by work? If a patient sees a point of light and it moves, is that sight? I need to know what the patient sees."
"OK. But what does it mean for your research?"
"Mean? It doesn't mean anything. We're going to keep going like we were going."
Normann also envisions a three-part system implant, signal processor, camera but with a critical difference. While Dobelle's implant rests on the surface of the visual cortex, Normann's would penetrate it.
Normann's implant is much smaller than Dobelle's about the size of a nail's head and designed to be hammered into the cortex, sinking to the exact spot in the brain where normal visual information is received. According to Normann, the implant is so precise that each electrode can stimulate individual neurons.
"The reason this matters," he explains, "is that the cornerstone of artificial vision is the interaction between current and neurons. Because Dobelle's implant sits on the surface of the visual cortex, it requires a lot of current and lights up a whole bunch of neurons. Something in the 1- to 10-milliamp range. With that much juice, a lot can go wrong."
Tell me about it.
"With penetrating electrodes, we've got the current down to the 1- to 10-microamp range. That's a thousandfold difference." Lowering the amperage lowers the risk of seizure.
But that's not all. Decreasing the amount of current also allows an increase of resolution: "The lower the current, the more electrodes you can pack on an implant," explains Normann. "We're not there yet, but with my electrodes there's the chance of creating a contiguous phosphene field that's exactly what you and I have and that's just not possible with Dobelle's surface implant."
Which is the way things go when what was once a land of mystics becomes a field for engineers. Just like every other new technology, like operating systems and Web browsers, artificial vision is heading toward a standards war of its own.
Now that it's not faith healing, it's Beta versus VHS.
TO REALLY TRY TO
UNDERSTAND WHAT JENS SEES, I HEAD TO USC in
Theres a wash of light. Suddenly, things become clearer. Did you up the resolution? No, thats your brain learning to see.
"It's a limited approach, aimed at a limited number of pathologies, but it has its advantages," says Humayun. "We thought it was a better idea to operate on a blind eye than on a normal brain."
Humayun's Retinal Prosthesis Lab runs out of USC's Doheny Eye Institute. The room is small and square. Piles of electronic gear sit atop counters of maroon plastic the same hue that offsets the bright yellow on Trojan football jerseys. Lab-coated technicians hunch over computers, barely registering my arrival.
James Weiland, an assistant professor at the institute, helps me into an elaborate headdress: Wraparound goggles cover my eyes, and black, light-blocking cloth hangs down over my ears. Plastic straps secure a miniature camera to the middle of my forehead, and wires run down my back and to a laptop computer to my left. The camera moves where my eyes move and then projects that image onto the "screen" of the goggles. The device, called a Glasstron and built by Sony, turns my normal eyesight into a pixelated version of itself.
With the power shut off, the view is complete darkness. Weiland flips a switch and asks me what I see.
"Vague gray shapes. Big dots. Blurry edges."
"Can you see the door? Could you walk to the door?"
"Yeah, I could, if you want me to trip over things and fall down."
"That's a 5-by-5 display. Hold on," says Weiland, "I'm going to up your pixel count to 32 by 32."
It's Weiland's belief that a 32-by-32 array, 1,024 pixels, should satisfy most vision needs. This is probably 10 times the count on Dobelle's implant and much closer to Normann's design.
Beside me I can hear Weiland futzing with the computer. There's a sudden wash of light, like viewing the Star Wars jump to hyperspace through a waterfall.
"Can you see now?"
"Not really."
"Give it a minute, let yourself adjust."
"OK, I've got blobs and edges and motion."
Suddenly, things become clearer. What moments ago was attack of the Jell-O creatures has become doorways and faces.
"What happened?" I ask. "Did you up the resolution again?"
"No," says Weiland, "that's your brain learning to see."
It's a weird feeling, watching my brain reorganize itself, but that's exactly what's happening. Beside me the fuzzy edge of the counter becomes a strong line, and then the computer atop it snaps into place.
I take one last glance around. Weiland is still not visible. Then there is a subtle shift in color. A drizzle of gray firms up, and I can see the white plane of his forehead offset by the darkness of his hair.
I look around: door, desk, computer, person.
So this is what a miracle looks like.
Steven Kotler
wrote about reengineering the