How The “Anternet” Succeeds by Showing Restraint

Some people show more self-control than others. If a marshmallow is placed in front of me, I will probably eat it. You might exercise more restraint than me, or you might not. Either way, your actions will depend on thousands of neurons in your brain. It would make little sense to describe any one of these neurons as showing restraint. They’re just interacting with each other in simple ways, and restraint emerges from these interactions.

You can see the same emergent behaviours in an ant colony. The red harvester ant mainly eats seeds. Some colonies go out searching for seeds no matter the weather but others hold back on dry days when they risk death by dehydration. Deborah Gordon from Stanford University has found that these restrained colonies are actually more successful as those that forage come rain or shine. And they even pass on their reserved behaviour to the next generation.

But it’s not that each individual ant is showing lots of self-control, any more than a single neuron does. The worker isn’t not weighing up how much food the colony already has. It’s not making active decisions based on the weather.

Here’s all that happens: If workers bump into others who have returned to the nest with food, they’re more likely to go out on their own foraging trips. Every colony has its own bar for the number of interactions it takes to make a worker leave to find food. That’s it.

These rules, which Gordon discovered last year, are similar to protocols that control traffic on the Internet—the Anternet, as she calls it. They allow the harvesters to tune their behaviour to their environment without any conscious knowledge. If there’s lots of food around, more foragers will return with seeds, stimulating more nest-bound workers to venture out. If seeds are scarce, fewer ants leave the nest. The size of the foraging parties automatically change to fit the amount of food around.

Harvesters live in deserts and lose water whenever they leave the nest. But they can only gain water by finding seeds. “Colonies must spend water to get water,” explains Gordon. Colonies balance their water budget differently. In some, workers are slightly and consistently less likely to respond to incoming foragers when the humidity is low. “This adds up, year after year, to less foraging activity by that colony,” says Gordon. On some days, not a single ant will leave these nests.

Each ant is behaving like a neuron in a brain, part of a literal hive-mind. It’s just going about its business and interacting with its colony-mates according to extremely simple rules. From these connections, restraint emerges. It’s a collective behaviour that happens at the level of the colony.

What effect does this have in the long-term? That’s a difficult question since harvester colonies can live for 25 years, and produce daughter colonies for 20 of those. To understand how their foraging differences pay off in the long run, you’d need to study a group of harvesters for decades.

That’s exactly what Gordon has done. Since she was a graduate student in 1985, she has almost single-handedly kept an annual census of around 300 harvester colonies in an area of New Mexico (here’s her TED talk on the work). Her records showed that the colonies that hunker down on dry days are just as likely to survive as those that head out all the time. Even though they may stay indoors up to two-thirds of the time, they can find enough seeds on good days and lose fewer workers in the process. They even seem to produce more daughter colonies. “In much of foraging theory, it’s assumed that ‘bigger is better’ – the more food collected, the more successful the forager,” says Gordon. “These results show the opposite.”

Iain Couzin from Princeton University, who studies collective behaviour, calls the study a “remarkable feat” and is particularly impressed that the collective foraging strategies seem to be heritable. Daughter colonies forage in very similar way to their parents. “Due to the relatively large distance between parents and offspring, it is unlikely that such synchronization is based on cultural transmission of behaviour,” he says.

For decades, scientists have been studying how animal swarms, from locusts to fish to birds, move as one, and even think as one. (You can read more about this in my Wired feature on the science of swarms.) But recently, they’ve begun to look at not just how collective behaviours work, but how they evolve.

Couzin, for example, showed that shoals can evolve to fool predators, even if the prey animals don’t know they’re in danger. Another group showed that parasites can make shrimps gather in shoals so they’re more likely to be eaten.

Gordon’s study is slightly different – it’s not about how collective behaviour arose, but how it continues to evolve. “It’s the first study of how natural selection is acting on collective behaviour in a natural population,” she says.

More on collective behaviour:

• Heart of the Swarm – the Amazing Science of Shoals, Flocks, Hives and Brains
• Parasites Make Their Hosts Sociable So They Get Eaten
• The Real Wisdom of the Crowds
• What are you looking at? People follow each other’s gazes, but without a tipping point
• To work out why fish swim together, tempt a predator with virtual prey
• March of the locusts – individuals start moving to avoid cannibals