Voting: The wisdom of ants

If the nest of the tiny (about 3mm long) eusocial ant Leptothorax Albipennis is damaged, then the inhabitants will collectively locate new nest site candidates and choose among them before moving the colony there and building a new nest. Their decision-making process is described and modeled in

N.R.Franks, S.C.Pratt, E.B. Mallon, N.F.Britton, D.J.T.Sumpter: Information flow, opinion polling and collective intelligence in house-hunting social insects, Philosophical Transactions of the Royal Society of London B357, 1427 (2002) 1567-1583.

The ants' process is very similar to the (more heavily studied and better understood) ingenious range-voting process employed by honeybees, but with the following differences:

1. The ants walk at millimeters per second while the bees fly at 3 meters/second. That allows the bees to explore an area of about 150 sq km seeking the best nest site. Meanwhile the ants only explore an area of a few square meters. The ants make their decision in 2-5 hours, while the bees take 5-350 hours.
2. Most scout ants inspect more than one candidate site, enabling a direct comparison. But most scout bees only inspect one.
3. Bees have a "dance" sign language that allows them to inform other bees of both the location and the numerical desirability of a possible nest site. However, ants do not have sign language and can only inform other ants about locations by requesting that they follow (or physically carrying them) there, and apparently cannot directly indicate a numerical desirability level at all, only a "yes" or "no" ala approval voting.

In view of the last fact, it might at first seem impossible for ants to do range voting at all. However, ants can recruit other ants to inspect a site with different probabilities and with different delays between recruitments. (And they do.) That still enables the ants, in a statistical sense averaged over many recruitment acts, to convey continuum "site quality" information as a continuously variable "exponential growth rate." Eventually, the faction representing the site with the greatest average "perceived quality" value, will dominate.

The idea (same as the honeybees use) is for each "faction" supporting some site-candidate to recruit new members, where each faction grows exponentially at different growth rates proportional to the average (up to statistical noise) of some measure of its site's quality.

More about social insects and their evolution

National Geographic July 2013 article on this

This whole investigation is described in more detail in my paper #96 here.

Return to main page