EUGENE, Ore.--At first glance, the worm known as Caenorhabditis elegans seems humble enough. Unlike its disease-causing and crop-killing cousins in the nematode family, C. elegans--which reaches at most a millimeter in length--is a peaceable resident of temperate soils. Its body is clear to the point of transparency. It rarely lives more than 18 days, and it mates with itself before it passes on.
But to researchers at the intersection of biology, chemistry, physics and engineering, C. elegans is invaluable. Scientists at the University of Oregon--funded in part by the Navy--are utilizing C. elegans's brain wiring to run an electronic robot that could one day be a model for a cheap, artificial eel that can locate explosive mines at sea.
Their eel would be built with a computerized brain that allows it to think, sniff and move as efficiently as C. elegans does. Much like the fruit fly, Drosophila melanogaster, C. elegans became a scientific star because it is both physiologi cally simple and has a quick reproductive cycle.
Scientists were able to map C. elegans's synapses--the message-bearing connections within its nervous system--because it has a mere 302 neurons, or nerve cells. By comparison, humans have 1 trillion. Despite its tiny brain, C. elegans is pretty smart for its world. If one adjusts for the nematode's smaller size, C. elegans can actually handle about 1,000 times as much information per second as an Intel Pentium processor can.
That comparison, conceived and jotted down by University of Oregon neuroscientist Shawn R. Lockery a few years ago, "supported the idea that there are really important engineering secrets hidden in an animal system," Lockery said.
Thinking about artificial intelligence that way represents a sharp break from the past. Historically, scientists have tried to fashion electronic systems into approximations of animal behaviors. By contrast, Lockery and his team are trying to hard-wire the principles of animal brains into the instructions that run electronic robots.
"We don't call it artificial intelligence--we call it biological intelligence," said Joel L. Davis, a program officer with the Office of Naval Research, which has funded Lockery's work. "Our goal is to look at the animal kingdom for behaviors or capabilities that we can reverse-engineer into devices that solve real-world Navy problems."
Several other federal agencies, including the Defense Advanced Research Projects Agency (DARPA) and the National Institutes of Health, have followed the Navy's lead on biological intelligence, Davis said.
The Navy and DARPA have funded efforts by Joseph Ayres to pry into the minds of lobsters and lampreys. Lobsters are famed for their skill at moving through rocky, underwater surfaces buffeted by heavy water currents. Ayres, of Northeastern University's marine biology station in Nahant, Mass., has studied the simple "pattern generator" in the lobster's neural networks that governs how each leg moves.
The goal, Davis says, is to use a synthetic copy of the lobster's pattern generator to drive an artificial lobster that could one day be a prototype for an autonomous undersea vehicle. Early versions of the simulated lobster body and legs are complete and are scheduled for testing this summer, Davis says.
Similarly, the lamprey--a relatively primitive sea creature--might eventually provide scientists with clues about mimicking the movements of fish. Ayres is studying how the lamprey's brain controls its sine-wave-like movements, but the project is still in an early stage. "We'd like to create a device where the movement of the body mimics the flex of a fish," Davis said.
The focus of Lockery's lab is a process known as chemotaxis--the method animals use to follow smells or tastes. A blindfolded man finding his way toward a just-baked apple pie uses chemotaxis to decide which direction to move. Similarly, nematodes use chemotaxis to find bacterial food sources by following the odors of their favorite bacteria's chemical byproducts. Chemotaxis, Lockery says, "is arguably the most widespread form of goal-directed behavior--that is, intelligent behavior--in the animal world."
To figure out how nematodes practiced chemotaxis, Lockery and lab mates Thomas Morse and Jonathan Pierce constructed chemical gradients for their worms to wander through. What they found is that nematodes forged ahead as long as they were finding equal- or higher-strength odors that they liked. As soon as the odor began to abate even slightly, the worms spun around. If their new direction offered stronger odors, they continued in that direction. But if the odors proved to be weaker still, the nematode kept spinning until it found a direction that offered more of what it wanted. Once Lockery and his lab mates understood that process, they wrote computer instructions that mimicked the behavior and installed them in a $350 makeshift robot made of Lego tiles, model airplane parts, a light sensor and the plastic canister from a gumball-machine prize. The foot-long robot moves exactly as a nematode would under a microscope: It meanders haltingly, but within a minute or two, it always winds up at the brightest light source in the room.
Eventually, Lockery would like to adapt that sensing system to track the minuscule plumes of waterborne chemicals that leach from underwater mines. To do that, they would embed a microchip containing the nematode's chemotaxis instructions into an eel-like robot fitted with a chemical sensor.
The little Lego robot--though it's only "phase one, if not phase zero" of Lockery's project--has already showcased how eons of evolution come up with ingenious solutions. For instance, Lockery unexpectedly discovered that C. elegans and its robot progeny use an elegant method to right themselves once they have hit an obstacle. When the robot hits a chair or a bookcase, it doesn't remain stuck for long, even though the light level at that spot, to the naked eye at least, seems constant. In reality, the light level is always fluctuating slightly, so before long it usually drops low enough for the brain to conclude that it's time to turn around.
Building a real-life mine detector from Lockery's research "is possible," said Anne Hart, a neuroscientist who studies C. elegans at Harvard Medical School and Massachusetts General Hospital. "But either way, we're going to learn a lot. The fundamental mechanisms behind chemosensation in C. elegans seem to be the same as for humans. And it's a lot easier to learn those principles from a worm."
This article first appeared in The Washington Post.
NEXT STORY: House chair probes DOJ enviro affiliations