Harvester ants use their own internet for hive-mind decision making

Harvester ant colonies are able to decide how many foragers they need to send out of the nest by using a protocol system that's eerily similar to the one IT professionals use to determine how much bandwidth is available on the internet. This so-called "anternet" apparently proves that ant decision-making and behavior is regulated by a sophisticated set of algorithms that have evolved over the course of millions of years — algorithms that we ourselves have only recently discovered.

The research was conducted by Stanford ant biologist Deborah Gordon and computer scientist Balaji Prabhakar, whose findings were recently published in PLOS. Prabhakar was brought on board to see if he could find any correlations between ant behavior and computer science. After some preliminary observations he quickly realized that the ants were essentially using the same algorithm that describes TCP, the Transmission Control Protocol. TCP is used to discover how much bandwidth is available for transferring a file. But instead of bandwidth, the ants are using their own version of TCP to determine how much food is available based on the reports coming back from foraging ants.

TCP was an important breakthrough that allowed information technologists to take the internet from a few dozen nodes to the billions currently in use today. It works by transferring a file from a source to the destination in a series of packets. The source is continually informed by the destination as to when the packets have arrived. So, if it's taking too long, there must not be much bandwidth available, and the source will re-adjust by decreasing the rate of packet transmission — what's referred to as throttling.

Harvester ants use their own internet for hive-mind decision makingS

As it turns out, the ants are doing exactly the same thing — but they're not looking to throttle data, they're looking to control the release of forager ants.

A forager harvester ant will only return to the colony after it finds some food. Thus, if there's lots of food in the immediate environment, the number of ants returning with positive food messages will be high; but if there's little to no food around, very few ants will be transmitting their acknowledgement messages back to the colony. So, with a returning ant frequency established, the colony can adjust the forager throttle accordingly. When there's more food available, the hive mind releases more ants given the confirmed presence of food.

It's worth noting that harvester ants do not use pheromone trails, and that they transmit information to each other using their antennae.

In the study, the researchers did not specify why the ants did not increase their foraging activities in consideration of trace amounts of food. One possible explanation is that the colony trusts the data coming back from the foragers, and that sending more foragers would simply be a waste of time and energy.

As for testing this theory, Prabhaker wrote an algorithm to predict the foraging behavior of ants depending on the amount of food available. It turned out that his ant-algorithm was almost identical to TCP. Moreover, he discovered that ants follow two other aspects of TCP, namely slow start phases (where a source sends out a large wave of packets at the beginning of a transmission to assess bandwidth), and a transmission time-out (where the source will stop sending out data packets (or ants) when a data transfer is broken or disrupted). In the latter case, ants will also stop sending out ants if, after 20 minutes, no foragers have returned to the nest.

Fascinatingly, the team speculates that, had IT developers discovered this ant algorithm during the 1970s, they very well might have used it to develop TCP. Moreover, they believe the ants have more to offer, and strongly suspect they could still teach us about the design of network systems. The researchers contend that each ant may be very basic in terms of its capacities, but that the collective is capable of forming incredibly complex and sophisticated tasks.

The entire report can be read at PLOS.

Image via Andrey Pavlov/Shutterstock.com. Inset image via.