Saturday, September 20, 2014

Bees and Ants on How to Make Decisions

Bees and Ants on How to Make Decisions


When it comes to making decisions, bees and ants can act against stereotype


By Robert M. Sapolsky in the Wall Street Journal

It can be challenging to choose the best way to solve a problem, to find the right tool for a job. Work intuitively or analytically? Alone or in a group? If in a group, through top-down or bottom-up decision-making? And how to figure out if it's actually a task you should just avoid?

Scientists study decision-making; consultants flourish advising on the subject. Remarkably, insects navigate similar challenges—often with great success and in ways that undermine our stereotypes of their species.

Social insects excel at what we've come to call the "wisdom of the crowd," in which a group of moderately informed individuals is more accurate than a lone "expert." Suppose two bees each discover a different food source. As research beginning almost a century ago has shown, each bee then returns to the hive and "dances," communicating the direction and distance of the food; when a bee in the hive encounters a dancer, she investigates that source. How does the hive figure out which is the better resource?

Suppose site B has twice the food as site A, and as a result, the scout from that source dances for twice as long. The other bees in the hive will encounter the site B scout dancer twice as often as the other bee; soon, twice as many bees investigate and return from site B as from site A, and they dance for twice as long, too. This results in four times as many bees checking out B than A, then eight times as many, then…everyone. No bee investigates both sites, yet the better site is chosen.

Work by Takao Sasaki and Stephen Pratt of Arizona State University and colleagues, published in the Proceedings of the National Academy of Sciences, explores something similar in ants. Given two nests, how does an ant colony determine which is preferable (in this case, which has a darker interior)? A scout returns, and the better the nest, the sooner she leads a second ant to it, causing that exponential shift of colony preference for that site. Ants sometimes move randomly, so this system can amplify a wrong choice; still, it works much better than chance.

To test the limits of this method, the authors had groups of ants and single ants pick nests. The choice was either easy (major lighting differences between the two nests) or difficult (subtler distinctions).

For difficult comparisons, ant groups were more effective than singletons in selecting the best nest in a set time. But critically, for easy tasks, it was the other way around. For one thing, by the time a lone ant had made the obvious choice, the group was still forming subcommittees to write the Environmental Impact Statement. And for easy choices, a single ant is likely to be more accurate, since it isn't vulnerable to random fluctuations amplifying the wrong choice.

So even ants struggle with finding the right tool for the task.

Insects demonstrate another type of wisdom. Publishing in the same journal, Clint Perry and Andrew Barron of Macquarie University studied honeybees using two vials of fluid, each attached to a symbol that a bee had to learn to recognize. The vial near the correct symbol contains sugar water—yum. The vial near the wrong symbol contains quinine—yech. By choosing, the bee could be rewarded or punished. Or it could opt out and not choose.

When the contrast was easy—say, between a circle and triangle—bees typically chose, and with great success. But for difficult choices—say, a circle versus an oval—the bees took longer to choose. More important, they frequently opted out, thereby greatly improving their success rate. They knew when to hold 'em and when to fold 'em.

What to make of this, besides the fact that these are impressive feats for organisms with roughly one neuron for every 100,000 we have? First, even ants need some rugged individualism. Second, judicious bees aren't drawn like bees to honey if the odds of succeeding aren't great.


No comments: