Tropical army ants are known to form devastating armadas that march through forests eating other insects and small creatures. When ants march, they deposit chemical trails of pheromones that others behind them sense and follow. But occasionally one of these ants will walk in a circle, which prompts a subset ofants behind them to follow until they all collapse and die. 

“They end up in this cycle where they’re reinforcing the same pattern over and over again,” said Andrew Hein, associate professor of computational biology in the College of Agriculture and Life Sciences. “This is a physical instance of an echo chamber, where the animals involved actually die.” 

Hein is senior author of a new study that examines how a few very general constraints can lead to echo chambers – self-reinforcing feedback loops – in groups of biological agents such as birds in a flock, cells in tissue, or ants in a swarm, and explores ways to control them.  

The paper, published Sept. 22 in the Proceedings of the National Academy of Sciences, uses mathematical models to test parameters that increase and decrease the likelihood of an echo chamberforming. 

In humans, the phenomenon occurs in groups when individuals send and receive similar messages and lack external sources of differing information. Many biological systems where individuals collectively make decisions – such as with fish in a school that dart away from suspected predator – suffer from two key constraints, according to the paper.  

First, an individual often only observes a neighbor’s discrete actions and doesn’t know all the internal information that led that neighbor to behave in a certain way. And second, individuals have limited attention at any moment and can only attend to the actions of a few neighbors, as opposed to many of them at once. 

The study reveals that these constraints can lead individuals to become extremely sensitive to the messages they receive from others and they encourage conditions for sending and receiving very similarmessages back and forth in a group, which then creates an echo chamber.  

“It happens when people are telling each other the same thing, but there’s also an implication that this group has become unresponsive to what’s actually happening in the world around them,” Hein said.  

  In the study, Hein and colleagues started with computer models studied by other researchers in the past where agents in a network shared the actual data that caused them to form a particular opinion or take a particular action.  

“It’s as if I express my opinion to you, but instead of telling you what my opinion is, I tell you everything that I ever learned that caused me to form that opinion,” Hein said.  

The model also assumed that any agent in their network can pay attention to all of its neighbors at once. With these two assumptions in place, the models did not form echo chambers.  

“One of the reasons why we found these assumptions interesting is that they appear to be violated in every system we could think of,” Hein said, meaning individuals rarely have all the information that lead to an action or opinion nor can they pay attention to many things at once. 

The researchers then tinkered with these assumptions and removed one or both of them to create more realistic constraints. “The behavior of the group of agents as a whole could become just totally dysfunctional and the form that this dysfunction takes is really an echo chamber,” Hein said. “Individuals can reach a consensus decision but it becomes totally decoupled from the state of the world around them.” 

But some factors can prevent an echo chamber. For example, individuals who stand at a group’s periphery may be connected to those at the center but are also influenced by others and can pick up their own information from the environment and can trigger change in the whole group. 

“It’s really the free spirit, the individual who has stayed at the edge and doesn’t have many social influences who are the first ones to change,” Hein said. “They can set off a cascade.” 

Another factor that prevents echo chambers comes from individuals who recognize that theirdecisions are wrong, and who start ignoring the group. 

“If some of the individuals in the network, even just a small fraction of them, do that some of the time, it will destroy an echo chamber,” Hein said.  

Ling-Wei Kong, a Schmidt AI postdoctoral fellow in Hein’s lab, is the paper’s first author. Naomi Ehrich Leonard, professor of mechanical and aerospace engineering at Princeton University, is a co-author.  

The study was funded by the National Science Foundation and Schmidt Sciences, L.L.C.