Animal bait and switch

You may be familiar with this scenario: A child starts annoying another child, and they start bickering. To keep things from escalating, an adult distracts the children by focusing their attention on something else, and the conflict ends.

Male mice who are being picked on by other male mice use the same trick when female mice are nearby, according to new work by University of Delaware neuroscientist Josh Neunuebel and his research group published in the journal PLOS Biology in October.

Neunuebel, an associate professor in the Department of Psychological and Brain Sciences, and his team study how communication shapes social behaviors, and what neural circuits process this information.

His research team found that when a male mouse chases another male and a female is close by, the mouse that is being chased will run to the female, interact with her for a second, and then dart away as soon as the male mouse who started the chase interacts with the female, distracting the aggressor and ending the conflict. The researchers called the behavior a "bait and switch."

Neunuebel and his group recorded five hours of video of two male and two female mice interacting, "mice just being mice." They used a machine learning tool to help sort the mountains of data present in each recording and extract distinct behaviors. The team then focused on aggressive behaviors in males and saw an interesting sequence of events that led to the discovery of the bait and switch.

"It was a really pretty striking and robust pattern that we were seeing," Neunuebel said. "Aggressed animals would go up to a female, briefly interact and then dart off, and the other animal, the aggressor, would spend more time with the female."

The number of fights that happened after the male mice went their separate ways dropped, supporting the idea the tactic is used to de-escalate the confrontation.

"When animals fight, there's always a potential cost," Neunuebel said. "There's a chance of getting hurt, so when possible, fighting is best avoided. This makes distraction a powerful mechanism for avoiding conflicts and unwanted situations."

But the behavior comes at a cost too, Neunuebel said. While the aggressed mouse avoids further conflict, and possibly getting hurt, he gives up the opportunity to interact with the female, which is important when it comes to reproduction.

Even so, those costs can outweigh the risks of getting hurt in a fight.

The findings open the doors to questions about how neural circuits process information about the behavior. Research shows that humans and mice share about 90% of the same genes. Understanding the neurobiology of mice may give researchers valuable insights into the mechanics of human communication and what happens when those systems don't work correctly.

The project also shows how machine learning is becoming an indispensable tool in behavioral research. Social interactions can be hard to study objectively, especially with large volumes of data. Neunuebel calls artificial intelligence a "staple" in his lab, one that allows researchers to make sense and understand huge amounts of information.

"It's transforming the field," Neunuebel said. "It's allowing us to ask questions in ways that are much more naturalistic and identify meaningful information."

UD is leading an effort in the Mid-Atlantic region to build a new position on research teams - the research software engineer (RSE) - who understands the science and advanced computational skills in AI and machine learning.

Human oversight is always part of his team's process, Neunuebel said. Machine learning is "a powerful tool to help look for patterns, but it's critical that you verify that the algorithms are working correctly," he said.

In earlier work Neunuebel found mice make specific sounds when they do specific actions. As a next step in this research, he and his team may look at the sounds the female mice make during the bait and switch to further understand how they are contributing to the social dynamics.

Support for the work comes from the National Institutes of Mental Health, the National Institutes of Health, the University of Delaware Research Foundation and Delaware's General University Research Program.