2015/05/28

The Hardworking Ant?

Last week we talked about ants as an example of hard work, so I thought we might look at this article in the Boston Review on ants again.

Network interactions, and the uses of downtime
Among harvester ants—the ants I know best—the important interactions are brief antennal contacts. An ant uses the rate at which it meets other ants to decide what to do. If you have ever watched ants closely, you have seen them touch antennae. When a harvester ant moves from tasks inside the nest to tasks outside, its odor changes, so an ant’s hydrocarbons identify its current task as well as its colony. To test how brief antennal contact influences ant behavior, my colleague Michael Greene and I presented ants with little glass beads coated with the odor of ants who are performing a particular task. Some of the beads smelled like patrollers, the first ants to go out of the nest each morning and travel around the colony’s foraging area. The safe return of the patrollers, at a rate of about ten ants per second, stimulates the first foragers to go out to search for food. When foragers meet beads bearing the hydrocarbons of patrollers, at the correct rate, they leave the nest. This experiment shows that an ant’s rate of brief antennal contact influences what the ant does next.

And what an ant does next may not be much at all. Contrary to another of our beloved myths about ants, told by Aesop, Homer, and the writer of Proverbs 6:6, many ants don’t work very hard. In a large harvester-ant colony, about a third of the ants at any time are hanging around doing nothing. As Mark Twain put it, this “will be a disappointment for the Sunday schools.” Because colony behavior is regulated by a network of interactions, inactivity might have its uses. Idle ants may act as a buffer to dampen the interaction rate when it gets too high. My colleagues and I have found that ants will move around to adjust their interaction rate—either they seek each other out when there are few ants, or they avoid each other when crowded. Sometimes interactions create positive feedback, as when ants go out to forage in response to interactions with foragers bringing food back to the nest. But eventually this could lead ants to search for food when there is none left. The colony may need some inert ants, unlikely to be stimulated by interactions, to buffer the network.

In this quote, it's suggested that the use of the ant downtime of ⅓ of the colony is to buffer the signal ratio of the network. Might there be other uses for this downtime, like for the health of the network or health of the colony, or any other purpose?


Ant colony as a set of network interactions.
A real ant colony is … more like an office that communicates by meaningless text messaging in which each worker’s task is determined by how many messages she just received. The colony has no central purpose. Each ant responds to the rate of her brief encounters with other ants and has no sense of the condition or the goals of the whole colony.

For ants, only the structure of the network matters. For us, the content is crucial. We care about what the emails say; the ants care only about how often they get them. As we move through the networks that shape our lives, we constantly produce a narrative about what is happening and why. We may be wrong about what we think is going on, but it is vitally important that we think we know.
Our stories about ants always have morals about how people ought to behave: soldiers should die for their country; we should conserve resources and plan for the future; a dutiful factory worker should cheerfully perform his or her appointed task. These morals come from stories about ants that are not true.

Real ants do not offer lessons in behavior. They do, however, provide insight about the dynamics of networks. Ants can show us how the rhythm of local interactions creates patterns in the behavior and development of large groups. There are no morals to be taken from the ants, but there is much to learn about systems without central control.

Can we distinguish 'behavior' from 'network interactions'?
At what point are people individuals, and at what point are they network interaction nodes?
Is society more than a network? What distinguishes 'society' from 'network'?



Do hierarchy and central control always appear together in a system?
Understanding how ant colonies actually function means that we have to abandon explanations based on central control. This takes us into difficult and unfamiliar terrain. We are deeply attached to the idea that any system of interacting agents must be organized through hierarchy. Our metaphors for describing the behavior of such systems are permeated with notions of a chain of command. ...
Ant colonies, like genes, work without … programming. No ant understands what needs to be done or what its actions mean for the welfare of the colony. An ant colony has no teams of workers dedicated to fighting or foraging. Although it is still commonly believed that each ant is assigned a task for life, ant biologists now know that ants move from one task to another. How does an ant decide which task to do and when to do it? We all know that where there is a picnic, there will be ants. So what determines which ants go to the picnic, and how many show up?

Colonies are regulated by networks of interaction. Ants respond only to their immediate surroundings and to their interactions with the other ants nearby. What matters is the rhythm of interactions, not their meaning. Ants respond to the pattern and rate of their encounters with each other, as well as to the smells they perceive in the world, such as the picnic sandwiches.

In what ways are people like this description of ants?
In what ways are they different?