Which is smarter: a swarm of brainless mini-robots with clockwork guts, or a colony of ravenous, half-blind Argentine ants?
If you answered mindless robots, you’re right — but just barely.
Researchers studying the problem-solving abilities of foraging ants enlisted the aid of 10 sugar-cube-sized robots to determine whether the real-life insects had to put any thought into deciding which direction they should go when they came to a fork in the road or an obstacle in their path. The answer to that question is important for the understanding of how large communities of organisms interact and coordinate their behavior.
The Argentine ant was selected for the study because it’s among the world’s most successful invasive species. When it gains a foothold in new lands, such as California, Florida, southern Europe, Australia and Southeast Asia, it out-competes local ants and can sever links in the larger food chain.
“These guys are a real problem; they’ve caused a lot of trouble,” said Simon Garnier, who studies animal behavior at the New Jersey Institute of Technology and was lead author of the study published Thursday in PLOS Computational Biology. The ants, which measure about an eighth of an inch long and have very poor vision, are native to South America.
Certain species of ants can travel farther than the length of two football fields to find food, and then tote morsels back to their nest. The paths they take can be extremely complicated, and Argentine ants deposit pheromones along the way to serve as guideposts for their trailing comrades.
The behavior of individual foragers can have drastic consequences for the entire group. A series of wrong turns by one or several workers can transform an otherwise successful picnic raid into a catastrophe: Wayward ants can accidentally lock their supply network into a closed loop, causing the group to march in a fruitless spiral until they drop from exhaustion.
Scientists at the New Jersey institute and the Research Center on Animal Cognition, in Toulouse, France, hypothesized that the ants’ foraging success was a result of a scripted set of instinctive behaviors, and not of calculations made by individual ants. Using grant money from the French government, the researchers tested their hunch by setting up a competition between real ants and a squad of micro-robots designed at the Federal Institute of Technology in Lausanne, Switzerland.
In the live-animal experiment, a colony of 500 worker ants was starved for a couple of days and then set free in a maze carved into a plastic board. Researchers placed a cotton ball soaked in a sugar solution at the opposite end of the maze and observed as the ants went into a frenzied search for food before returning to their nest.
The robot experiment took a lot longer to set up and conduct.
Each robot comes equipped with two Swatch watch motors and four tiny wheels. (Their top speed is a blistering 8 feet per minute, about four times faster than real ants.) The robo-ants communicate with light instead of pheromones, so they sport light sensors instead of antennae.
The electronic critters were programmed to move randomly, but in the same general direction — just like real ants.
The robot ants were released into a cardboard maze with infrared light beacons to simulate their nest and their food source. As they wheeled down passageways, an overhead projector beamed blue circles onto the pathway behind them, as if they had left a pheromone marker for their buddy robots behind them. When the robots encountered an intersection, they were programmed to take the route that deviated least from their general direction of travel. However, if they encountered a blue circle of light, they followed that instead. (The projected light circles gradually faded in intensity, just as real pheromone deposits evaporate and lose strength.)
After running the contest between ant and machine many times, their rates of success and overall routes were very similar, although the robots tended to use shorter routes, the researchers found. Also, when the robots bumbled their way into closed loops, they were more likely to break free.
The research team concluded that “a complex cognitive process is not necessary to explain the ants’ behavior.”
Though it might appear that the robots were somewhat more efficient, or “smarter,” Garnier said it wasn’t exactly a fair comparison. With hundreds of ants in the maze at once, traffic jams would cause the insects to disperse in different directions. “If we had performed the experiment with 500 robots, we probably would have run into the same problems,” he said.
Although the study’s methods were novel, its conclusions were “not very surprising,” said Doug Yanega, a senior scientist at UC Riverside’s Entomology Research Museum. Computer simulations by animal behaviorist Nigel Franks have provided similar insights into ant behavior, he said.