Fish can adjust their sensitivity to the actions of others – such as fleeing due to a false alarm – in order to reduce the risk of overreacting to misinformation, according to a new study.

When wild coral reef fish swim alone or in small groups they are extra-sensitive to their surroundings and are highly attuned to the actions of others. For example, when fish around them startle, they’re more likely to flee themselves. But in large, dense schools, where fish around them are more likely to startle or dart for no reason, individuals are more willing to take risks and tune down their sensitivity to social cues, making them less likely to flee when a neighboring fish does.

Such decision-making mechanisms may be preserved in other animals, including humans.

The behavior doesn’t necessarily discriminate between true threats and misinformation, rather, it adjusts sensitivity in a way that will lower the probability of responding to a false alarm.

Analysis of footage from underwater camera observatories found that individuals in schools of wild foraging fish flee for shelter even when no predator or threat is present about once every 8 minutes.

For the study, “Wild Animals Suppress the Spread of Socially-Transmitted Misinformation,” published March 28 in the Proceedings of the National Academy of Sciences, researchers used new computer vision tools, machine learning and computational modeling to analyze underwater recordings of startle events among foraging coral reef fish in Mo’orea, French Polynesia.   

“When we looked at the features of the model that matched observed behavior, we found that it adjusts the sensitivity of individuals to signals produced by others, based on the past history of what they’ve been seeing,” said senior author Andrew Hein, assistant professor of computational biology in the College of Agriculture and Life Sciences. Ashkaan Fahimipour, an assistant professor of biological sciences at Florida Atlantic University, is the study’s first author.

When there is a lot of visual motion, individuals appear to lower their sensitivity to it, and when there is very little visual motion, they increase their sensitivity, Hein said. “So they seem to be dynamically adjusting the sensitivity,” he added.

While there is a big focus in politics these days on disinformation, where individuals deliberately create misinformation to deceive others, other forms of misinformation are more subtle, inadvertent and innocent. For example, the sight of a crowded restaurant may suggest the food is good, but a meal there might prove otherwise. Or, while standing on a busy street corner, other pedestrians begin to cross the street, and you follow, only to find the light is still red and a truck is approaching, prompting you to step back.

In the study, the researchers used coral reef fish as a model system. The fish feed in dangerous areas where there are predators, prompting them to be edgy and skittish, even when predators are not present. Using computer vision technology to analyze footage of the fish from underwater observatories in coral reefs, the researchers precisely tracked the actions of every individual in a frame and reconstructed what each individual saw and the decisions they made.

In the past, scientists have proposed several ideas about how misinformed behaviors might spread from one individual to another within human and animal groups. These hypotheses predicted that misinformed responses should spread widely through fish groups; but real world data showed that such misinformation cascades rarely occur, Hein said.

“Mechanisms for adjusting sensitivity are actually crucial if you’re going to maintain control over your behavior,” Hein said. The new model was based on well-studied properties of fish neural circuits that control these behaviors.

“When we took our new model, and we applied it to our data, we found that it explains the data beautifully, and appears to be very consistent with what the animals are actually doing,” Hein said.

While more study is needed, the paper suggests that the need to cope with misinformation may have driven the evolution of how brains process information, Hein said.

“Because of its simplicity, and the ease with which it can be implemented in the nervous system, we believe this form of dynamic control of sensitivity may be widespread in biological systems, and may have evolved as a simple but robust way of coping with misinformation,” he said.

The study was funded by the National Science Foundation.