touch controls rats
Technology Research News
Since the late 1930's scientists have known
that electrical stimuli to different regions of the brain could be used
as teaching cues. In the '50's, researchers found that stimulating the
medial forebrain bundle in the hypothalamus is rewarding enough that an
animal will work to obtain the stimulation just as it will for food or
Researchers from the State University of New York Downstate Medical Center
and Drexel University have combined reward stimulation with stimulation
of a part of the brain that processes the sense of touch in order to remotely
control the movements of a rat up to 500 feet away.
The achievement provides insights into how animals learn, and may eventually
lead to better prosthetics that provide tactile feedback. "A larger aim
behind the guided rat was to begin studying the kind of effects, both
on behavioral control and perception, that stimulation of sensory regions
of the brain may have," said Sanjiv Talwar, a research assistant professor
at SUNY Downstate Medical Centre. "This would enable us to experiment
with ways to [use] sensory feedback in order to better control a robot
arm," he said.
It also shows that it is possible to control wired rats as if they were
intelligent robots. "The guided rat could easily be used as a robot platform
in situations where the terrain is too complex for [mechanical] robots
to traverse or where stealth may be a prerequisite," said Talwar. The
rats could be equipped with small video cameras, GPS systems and other
sensors, he said.
The key insight that led to this type of control was thinking of conditioning
animal behavior wholly in terms of virtual cues and rewards rather than
external cues and rewards like sounds and food, Talwar said.
To train a rat this way, the researchers implanted electrodes that were
about the diameter of a hair into three parts of the rat's brain. One
electrode stimulated the reward center in the hypothalamus. The other
two stimulated areas of the right and left somatosensory cortex that receive
input from a rat's whiskers. The somatosensory stimulations probably give
the rat the illusion of touch, said Talwar. It will seem to the rat like
it is being touched on its left whisker, for instance, when electricity
passes through the electrode on the right side of the somatosensory cortex,
The electrodes were connected to a small adapter cemented to the top of
the rat's skull. The adapter was connected to a microprocessor carried
in a small pack on the rat's back. The researchers controlled the electrode
stimulation remotely using a PC that sent radio signals to the microprocessor.
"The system we came up with was built using simple, off-the-shelf electronics,"
The researchers showed that rats could be trained using the electrodes
and a figure-eight maze. When the rat moved forward, the researchers gave
it a hypothalamus reward stimulation, and within ten minutes the rat learned
that when it continued to move forward it received a reward every second.
Whenever the maze presented the rats a choice of turning left or right,
the researchers stimulated one side of its whiskers. When a rat made the
corresponding turn, it was immediately given a hypothalamus reward stimulation.
If it made an incorrect choice it was deprived of hypothalamus reward
stimulation for five seconds.
After seven or eight 10-minute training sessions, "the rats learned to
move forward continuously and respond with near 100 percent accuracy to
turning commands," said Talwar.
The researchers then tested the rats in open environments and found that
they still responded to the turning commands. "Even in open space they
would move forward for periodic... rewards and respond correctly and instantaneously
to turning commands," said Talwar.
The researchers were also able to get the rats to climb or jump over objects
using the stimuli, Talwar said. "By giving [reward] stimulations along
with right-turn and left-turn cues we... were able to guide the rats over
any terrain that was within their capability to overcome. These terrains
included wide-open spaces, both indoors and outdoors, complex 3-D mazes,
trees, pipes, narrow ledges at considerable heights, stairs, and large
concrete-block rubble piles," he said.
The research is exciting, and bears on the whole field of neural prosthetic
development, said Miguel Nicolelis, a professor of neurobiology and biomedical
engineering at Duke University. "It's a very important step because it
shows that you can actually, in a very predictable way, deliver signals
to the brain of an animal and condition the animal to interpret the signals,
particularly using the sense of touch," he said.
The research shows that you can get touch-like feedback to the brain and
train animals to interpret it as if they were being touched on the face,
said Nicolelis. Guiding animals using a sense of touch "is very new, no
one has done it before," he said.
The research has several potential uses, said Nicolelis. "One is on the
basic science level as a new paradigm to study fast learning," he said,
pointing out that with the stimuli, the rats were learning to do a maze
on the first try. "This is something that allows you to look at the fully
behaving animal and measure what is going on in [brain] circuits," he
The research is also significant for the development of prosthetic devices
that use a touch-like interface to restore motor, sensory or even cognitive
function, said Nicolelis. "This was a missing step that several people
were trying to achieve -- [the researchers have] done it."
A sense of touch would be an important type of feedback for a paralyzed
patient using a robotic arm, for example, said Nicolelis. "You need to
provide some sort of feedback information to this patient," he said. One
possibility is visual feedback, "but when you're grabbing an object like
a glass of water, vision alone is not going to help you -- you need to
have this touch feeling of what it is you're touching, how heavy it is,
otherwise you're going to either drop it or smash it because the force
that you're going to apply is not going to be appropriate," he said.
Another potential application is to use remote-control rats to search
rubble for victims of earthquakes and bombings, said Nicolelis. "If you
can steer a rat... through rubble, it would be [useful] because they are
smaller than dogs, they're very good sniffing animals, and they have these
phenomenal facial whiskers... that can discriminate very fine objects,"
he said. "Theoretically that could be a... development if it proved to
be reliable," he said.
The researchers' next steps are to explore the possibility of a working
sensory prosthesis, said Talwar.
Although the capabilities of the guided rat could easily be developed
further, "at present, however, we're not aiming at this," he said. It
would take two to three years to develop the concept into practical processes
shaped around specific applications, he added.
"Ethical considerations may play a role in future development, and a wider
debate will be required for this to happen," said Talwar. The rat experiments
were performed within National Institutes of Health guidelines, and "in
addition our behavioral model is based only on a reward system with no
associated food or water deprivation," he said.
"Nevertheless, for some there may still appear to be something creepy
about using a guided rat for real-world tasks," said Talwar. "This must
be acknowledged -- after all it will be easy to extend the same method
to any species," he said.
Talwar's research colleagues were Shaohua Xu, Emerson S. Hawley, Shennan
A. Weiss and John K. Chapin SUNY's Downstate Medical Centre, and Karen
A. Moxon of Drexel University. They published the research in the May
2, 2002 issue of the journal Nature. The research was funded by the Defence
Advanced Research Projects Agency (DARPA).
Timeline: 2-3 years
TRN Categories: Biotechnology; Applied Technology
Story Type: News
Related Elements: Technical paper, "Rat Navigation Guided
by Remote Control," Nature, May 2, 2002.
Viruses turn out high-tech
Virtual touch controls
Plastic mix promises
Laser patterns particles
Research News Roundup
Research Watch blog
View from the High Ground Q&A
How It Works
News | Blog
Buy an ad link