Has the simple movement of a finger ever generated such
widespread interest and media attention as Christopher Reeve''s
ability to move his left index finger? It certainly provides more
hope that medical researchers will someday be able to reestablish
the mind-body connections severed in spinal cord injuries. An
approach championed by Reeve is stem cell research, the hope
being that stem cells can be used to regenerate or replace the
broken nerve connections. This hope has been bolstered by
promising results in recent animal studies. However, stem cell
research has generated heated political and ethical controversies
when it comes to the use of human embryos.
Although stem cells, if they work, would be the ideal solution,
there are other approaches to improving the quality of life of
victims of spinal cord injury. Over the years, there have been
many studies and some limited progress in using computer
controlled prosthetic devices and robotics to accomplish limb
and hand movement. The computer is typically controlled by
some sort of arrangement involving movement of the patient''s
head or even eye movement. These movements are translated
into activating the computer through mechanical means or via
sensors, for example, sensing the eye movements.
An approach that I find especially fascinating combines studies
of the mind coupled with computer technology. The ultimate
goal here is to bypass the broken connections in the spinal cord
and control limb and digit motion by "thinking" the action into
being. How do you normally move your arms and grasp that cup
of coffee? You tell your brain what you want to do and it sends
out the appropriate electrical impulses to accomplish your
mission. You "think" your hand into picking up the cup. These
repetitive motions have become so automatic that you don''t think
about them consciously but the thinking is there.
Coincidentally, after reading about Reeve in our newspaper, I
found an article in the October issue of Scientific American by
Miguel Nicolelis and John Chapin. The article is titled
"Controlling Robots with the Mind". The brains of humans and
animals have been probed, scanned, dissected and treated to all
kinds of insults over the past century or so. In the process, an
amazing amount of information has been obtained as to the
locations where various functions are handled in the brain.
Chapin and Nicolelis met at Hahnemann University some 14
years ago. At the time, an active field was the recording of the
electrical activities of motor neurons in the brain. For example,
it had been found that the electrical output of a neuron associated
with movement of the arm of a monkey depended on the angle of
the arm to a certain preferred direction.
The prevailing practice then was to measure the electrical output
of a single neuron at a time. Nicolelis and Chapin wanted to
measure many more neurons at the same time. There was a
problem. The electrodes used in those days were made of
stainless steel coated with Teflon and were rigid, more like
needles than wires. The electrodes worked fine - for a few hours.
Cellular substances formed around the tips, fouling up the
measurements. Also, because of the rigidity and sharpness of the
needles, they tended to damage the neurons over time. By
devising microwires that were quite flexible with blunt, rounded
tips, Chapin and Nicolelis were able not only to implant many
electrodes but also to measure the outputs of the neurons over
periods of months, not hours.
They also got together with an electrical engineer, Harvey
Wiggins, and a couple fellows named Woodward and Deadwyler
at Brian Trumbore''s alma mater, Wake Forest, to devise a
"Harvey box" (I presume named after Mr. Wiggins). The
Harvey box contained specially designed electronics to monitor,
sort out and process the electrical activity patterns from many
electrodes. With the Harvey box and the microwires, there was
plenty of time to measure and correlate the electrical patterns of
many-electrode implants in the brains of rats. The researchers
wired up one unnamed rat to measure 46 of its neurons. The rat
had been taught to press a bar, which would cause a drop of
water to be delivered for the thirsty guy to drink. Eventually,
when they processed the data from those 46 neurons, it turned
out that they could come up with a single output that could
predict the path the rat''s forelimb would follow.
Now comes the really neat experiment. The researchers
disconnected the bar from the lever that delivered the water.
Naturally, the rat was frustrated and banged away at the bar to no
avail. Suddenly, when the computer detected the same brain
neuron pattern that the rat normally used when pressing the bar,
the computer flipped the lever and the rat''s thirst was assuaged.
After some time, the rat got smart and realized "Hey, I don''t have
to reach out and press the bar at all. I just have to look at it and
imagine I''m pressing the bar in a certain way." Needless to say,
Nicolelis and Chapin were ecstatic. If a rat could "think" a
computer into quenching his thirst, why not a human?
Of course, it''s a big step from a rat to a human. Nicolelis moved
to Duke University and, still with Chapin''s collaboration, they
decided to work with more human-like owl monkeys and with a
three dimensional robotic arm. They chose the owl monkey
because its brain is smooth in the motor control area and it was
relatively easy to surgically implant the microwire assembly.
They also attached sensors to the monkeys'' wrists to follow the
trajectories when the monkeys reached out to pick up pieces of
fruit from a tray. By correlating their brain patterns with their
motions, the researchers became able to actually predict the
position of a monkey''s arm a few tenths of a second before the
animal moved the arm into that position!
Now it was time for Belle, an owl monkey who loved a game in
which she received a drop of fruit juice by moving a joystick
right or left to match a light flashed on a video screen. Could
Belle, with a 100-microwire implant, control the actions of a
robot arm? Not only one robot arm, but two - one in the next
room, the other in the lab of Mandayam Srinivasan at MIT, 600
miles away and controlled by a different computer! They had
just 3 tenths of a second to process the data and get the output to
the robots to match the motion of the robot arms to that of Belle''s
arm. (The data was transmitted to MIT over the Internet.) Sure
enough, all three arms moved in sync! Belle had managed to
"think" the robots into moving just as she moved her arm.
Next, it was on to Aurora, a macaque monkey. Macaques have a
more complex brain with furrows and convolutions similar to our
own. Aurora learned to move a joystick to place the cursor on a
monitor screen inside a target that would appear on the screen.
She had to do this within less than a half second to get her sip of
juice. She enjoyed the game and could hit the mark 9 times out
of 10. As with the rat, they tricked Aurora, this time by disabling
the joystick about a third of the time. Well, Aurora wasn''t going
to let a rat get ahead of her. She soon learned that she could
dispense with the hand movements. That is, she could "think"
that cursor into the target almost every time. This was
accomplished initially with 92 neurons being monitored. When
the article was written it had been a year and they were still able
to monitor 50-60 neurons. Even with the reduced number of
neurons, Aurora was just as proficient as she was initially.
Aurora''s not finished yet. They''re going to add a gripper that
simulates a grasping hand to the robot arm. Also, they plan to
feed back to Aurora''s hand some kind of information related to
the force exerted by the gripper. This probably will be in the
form of vibrations of different frequencies or intensities sensed
through a patch of some sort attached to the palm of her hand.
With this added input, they hope they can teach her to "think" the
robotic arm and hand into exerting the proper force to pick up
objects like a piece of fruit and bring it back to her without
dropping or squishing it.
What next? How about eliminating the robot completely? Work
has begun on prototype backpack computers and "neurochips" to
be used in the following scenario. The spinal injury victim has a
microwire array of electrodes implanted in his motor cortex with
a neurochip in his skull. The neurochip broadcasts the data (no
wires) from the electrodes to the backpack computer hanging
from his chair. The computer processes the signals and transmits
the processed signal to either the controls for the chair or to a
"motor control chip" implanted in the arm. That motor control
chip then transmits the signals to the nerves in the arm. This is a
real arm, not a robot arm, that picks up that piece of fruit and
carries it back to the person''s mouth.
I''m sure Christopher Reeve, and virtually everyone else, hopes
that the stem cell approach proves successful. If not, I suspect
that Superman might well be among the first to have the courage
and determination to agree to a microwire implant. If not, that
unnamed rat, Belle and Aurora still deserve every accolade for
their pioneering roles in exposing the inner workings and the
potential of that most miraculous of objects, the brain.
Allen F. Bortrum