Thursday, July 23, 2009

Scientific American - Mindless Collectives Better at Rational Decision-Making Than Brainy Individuals

I have no doubt that Howard Bloom would simply nod his head in agreement.

Mindless Collectives Better at Rational Decision-Making Than Brainy Individuals

New experiments show how ant colonies don't fall prey to irrational choices as humans sometimes do

By Charles Q. Choi

CROWD SOURCING: A colony of the common acorn ant (Temnothorax curvispinosus) in a one-inch-wide artificial nest. Stephen Pratt

Humans often make irrational choices when faced with challenging decisions. Ant colonies, however, can make perfectly rational selections when confronted by tough dilemmas. This isn't because lone ants are especially knowledgeable—they're not. Instead, when ants are grouped together, a kind of "wisdom of the crowds" avoids the kind of mistakes that individuals can make, new research shows.

In terms of evolutionary biology, animals strive to maximize their fitness. Still, actions that seem counterproductive and irrational occur not only in human societies, but also all over the animal kingdom. For instance, when honeybees and hummingbirds have two equally tempting choices of nectar, a third alternative inferior to both can sway them to prefer one of the initial two options over the other. The animals apparently compare the inferior choice against the originals and conclude that one of the originals is better, even though nothing about them has changed.

Such irrationality can lead to deep insight, because "finding what makes the system fail can give a clue about how it works," explains Stephen Pratt, a behavioral ecologist at Arizona State University's School of Life Sciences in Tempe. Of special interest to Pratt is how groups of animals such as ant colonies make collective decisions. "We can even think of a colony as an analogue for a nervous system—by understanding how decisions emerge from interactions among ants in a colony, we may learn something about how decisions emerge from the interactions among neurons in a brain," he says.

To see if collectives behave rationally, Pratt and his student Susan Edwards investigated a common acorn ant of eastern North America, Temnothorax curvispinosus, which is tiny—a colony of 50 to 200 such ants can make its home inside a single nutshell. When their nest is damaged beyond repair, the ants choose their new home en masse. Scouts look for potential nests, and if enough of them close in on the same area, they then carry nest mates over.

The researchers made two artificial nests as potential homes. Nest A had a larger, less defensible entrance but a dark interior that suggested strong, thick walls, whereas nest B had a smaller entrance (more defensible) but a bright interior (weaker walls). As expected, when the researchers ran 26 ant colonies past these nests, the insects split roughly equally on the nests.
Then they provided inferior "decoy" nests to spur irrational choices. For instance, if they presented a decoy that was similar to nest B yet had an even brighter interior, the ants might irrationally prefer nest B over nest A, if past results with humans and animals are any guide.

Surprisingly, the decoys had no effect on the colonies—they always made rational decisions.

"All minds, both collective and individual, have limited capacity—they have to use shortcuts and rules of thumb to solve difficult decision problems, and those shortcuts are expected to sometimes cause mistakes," Pratt says. "The ant colonies, however, were unfazed by a challenge that often elicits such mistakes in other animals."

So what makes these colonies so rational? Surprisingly, the very ignorance of the lone ants might be key. Instead of making comparisons between choices that can sway humans toward irrational decisions, individual ants typically know of just one option, which prevents them from making potentially misleading comparisons. Although the researchers expected the ants to behave irrationally, "we accidentally got a different insight about a possible advantage of collective cognition," says Pratt, whose findings appear today online in the Proceedings of the Royal Society B: Biological Sciences.

"We thought our brains and those of most animals could not always decide rationally because it was impossible to do so," notes behavioral ecologist Anna Dornhaus of the University of Arizona in Tucson, who did not participate in this study. "The ants show us that…there is a way to construct a decision-making system" that doesn't make irrational choices. The question then, she observes, is why our brains have evolved to act irrationally on occasion.

Reasons could be historical—"evolution constructed the brain this way and can't easily change it now," Dornhaus speculates. "Or it might not be that relevant or that big a cost to make irrational decisions in these cases, possibly because they are rare. Or, maybe there is another reason why what looks like an irrational decision is actually a good thing."

Pratt notes these findings "underscore a nonintuitive point—getting lots of information about a problem may not help decision making if you have only limited computational capacity to process it. You might do better with a strategically limited set of information. The trick, of course, is knowing what information to use and what to exclude."

No comments: