3/18/96

CONTACT: Stanford University News Service


Workers without bosses: How ants and bees know what to do when

Go to the ant, thou sluggard; consider her ways and be wise: which having no guide, overseer, or ruler, provideth her meat in the summer, and gathereth her food in the harvest. ­ Proverbs 6:6-8

Thousands of years ago, one of the authors of the Old Testament was a close enough observer of nature to notice that ants keep busy without anyone telling them what to do. Modern biologists have long known the same thing ­ that ants, honeybees, termites and other social insects can marshal just the right number of forces to forage for food, repair the nest, care for eggs and battle off intruders, without any central or hierarchical control.

There's no ant boss, termite overseer or bee middle manager shouting, "Hey you! The nest needs urgent repair, so stop collecting food!" says Deborah M. Gordon, assistant professor of biology.

Yet somehow social insects divide up tasks and switch from one task to another when the need arises. Gordon and other biologists would like to know how they do that.

In a progress article in the March 14 issue of the journal Nature, Gordon writes that research on social insects long has shown that the task a worker insect performs partly depends on internal factors, such as the individual's size or age. But in the past decade, studies have shown that the insects also respond to external factors. They choose to rest or rush to work, and they switch tasks rapidly and often, in response to cues from the environment and from the actions of other individuals.

Gordon says that the actions of a colony of ants or bees are like the many specialized cells produced as an embryo develops, or like the firing patterns of neurons in the brain. In each case there is no central headquarters giving orders, and the individual cells do not start out with a predetermined task.

"A single neuron does not think '10' or 'coffee cup,' " she says. "Its function depends on what other neurons are doing at the same time. No single neuron can think, but the brain can think."

Not exactly assembly line workers

In the 1970s and early '80s, most researchers thought that social insects were like super-specialized assembly line workers, with each individual suited to only one task. In some ways this is true, Gordon says: For example, some species of ants come in two sizes, small foragers and giant-sized soldiers. Honey bees move from one task to another as they grow older, and juvenile hormone levels influence this transition.

However, recent research has shown that even worker insects predisposed to do one task sometimes will switch to another if the colony's need is urgent enough. "The data show that some combination of . . . internal and external factors . . . contribute to individual decisions about task performance," Gordon writes. "Eventually we will probably abandon the dichotomy between internal and external causes."

Scientists like Gordon have turned to computer models to test theories about how each worker finds out what task it is supposed to do. The biologists are borrowing ideas from the artificial intelligence community ­ which in turn was inspired by the analogy between real-life ant colonies and some kinds of computational systems.

One model shows how task allocation can work even when all the individuals in a colony are intrinsically the same. Another predicts how quickly a colony can track and respond effectively to a changing environment.

The models are often used together with experiments to see if the colonies of insects act as predicted. In her own work, Gordon studies how ants allocate and re-allocate tasks in response to various challenges.

She says that an insect colony is like a computerized neural network, or like a mammalian brain, in the sense that individuals making simple decisions together do complicated things.

For example, an ant colony gets the right numbers of ants to perform a task like nest repair, and recruits extra ants for emergency repair jobs, yet no single worker can count how many ants are doing repair, and no single worker can decide how many are needed for the job.

How does an individual ant know what to do and when to do it?

"Workers might use some simple rule based on the rate of encounter with others," Gordon says. "Say a forager expects to meet another every 2 seconds, and if she does, she goes on foraging. But if she starts to meet other foragers every 0.5 seconds, she stops foraging.

"If the number of foragers goes up, she will experience a higher interaction rate. Using this simple rule the worker can respond to a change in worker number without having to count anything global, only having to assess the interval between contacts that she experiences."

Gordon's review describes the contributions of the theoretical models and field biology in helping biologists understand how social insects make group decisions.

"The behavioral ecology of social insects is a young field," she writes. "So far only a tiny fraction of social insect species have been studied. Social insects obtain food, build nests and defend their colonies in an astounding variety of ways. We will probably discover the same diversity in the ways that these tasks are regulated."

-30-

By Janet Basu