Ants make tough choices better when working in groups, study says
Our society often touts teamwork, but when faced with an easy task, groups may actually perform worse than individuals – at least when the group is made up of Temnothorax rugatulus ants.
A study published Monday in the Proceedings of the National Academy of Sciences found that T. rugatulus ant colonies could consider two competing nesting sites and make a good collective decision about which was the better option if the difference between the two was small. When the difference between the nests was obvious, however, the “group advantage” disappeared and colonies were more likely to choose the inferior site.
When T. rugatulus colonies in coniferous forests choose a potential nest, they dispatch about 200 scouts to scope out the territory. If a scout thinks a nest will make a good home, it releases a chemical signal that attracts a single colony member. If the recruit likes the site, it also will release a chemical signal and repeat the process. The number of recruits then steadily rises until it passes a threshold, or quorum. That’s “when things really take off,” and the recruits carry the other ants piggyback to the site, said study author Stephen Pratt, a behavioral ecologist at Arizona State University.
Rather than weighing multiple options, each scout ant assesses only one site and gives it a thumbs up or thumbs down. A collective decision emerges from competing recruitment efforts at each site in a positive feedback loop -- the better the site, the more likely recruits are to rally other colony members, and the faster the colony hits a quorum.
Earlier studies on collective decision making hadn’t compared group with individual performance, Pratt said. To help test whether “the group advantage” really did exist, he and his colleagues presented the same problem to both isolated ants and colonies, making them choose between two nests -- one dimly lit and one brightly lit.
The researchers maintained the same brightness within the dimly lit nest throughout the study but varied the brightness of the comparison nest with each round of testing so that the lighting differences between the two ranged from subtle to dramatic. Since darker is better for Temnothorax ants, they expected the insects to prefer the dimly lit site, Pratt said.
When the lighting differences were small — and therefore trickier to detect — colonies chose the dimmer nest more often than did individual members. But when the lighting differences were large and easily detectable, colonies fared worse than single ants.
The research team observed that when individual ants decided on a nest, they crawled back and forth to each site before settling on a new home. When it wasn’t clear-cut which nest was dimmer, they spent several minutes weighing the two options. Sometimes the ants “ended up making up their minds before gathering enough information” because the decision making process took so long, often leading them to choose incorrectly, Pratt said.
Colonies, however, decided on nesting sites much faster because none of the ants had to compare both sites. Their track record was impressive, but not perfect — when deciding between two slightly different nests, the ants chose incorrectly in a few cases.
Using a mathematical model, the team found that this cost in accuracy was “not as bad as that of a single ant making that decision,” Pratt said. When faced with a tough choice, the ants were “better off deciding collectively and accepting a small chance” that the competing recruitment efforts would lead them to the wrong choice.
Speed came at a higher price when the choice between nests was simple. In this case, the slight chance of a colony deciding incorrectly was higher than it was for individuals. Although isolated ants still had to make multiple trips to the two abodes, the process didn’t take so long that they ended up settling on a site before they had enough information. Even if individuals took longer than colonies to decide, they tended to be more accurate.
Although careful deliberation paid off in lab tests, speed often matters more than accuracy in the wild, even for easy decisions, Pratt said.
“If your nest is destroyed and you’re exposed, it’s important you find a new nest fast,” he said. “That probably makes up in large part for occasional loss of accuracy.”
Some “general comparisons” can be made between decision-making in ants and other social species, including humans, Pratt said. Humans tend to be more heavily influenced by social information, so their interactions are often much more complex, he said.
In humans, “There’s all this calculus that goes into interpersonal interactions rather than interpreting infromation,” said Anita Woolley, an organizational behavior researcher at Carnegie Mellon University who was not part of the study.
Ant colonies, which are typically decentralized, often make poor decisions by chance, she said. In contrast, hierarchical human groups “sometimes go awry because a high status member favors a decision that isn’t optimal,” Woolley said. Even if other group members have decided on an alternative, “because of social dynamics nobody speaks up, so the group makes the wrong decision.”
Pratt is currently collaborating with engineers at the University of Pennsylvania to create groups of tiny robots modeled after T. rugatulus ant colonies and other social species. He hopes to apply his latest findings to programming the robots to solve similar problems, including in dangerous search and rescue missions.