Potential Benefits of Observing Rats

It revolves around nonverbal communication.

A novel AI technique for monitoring the social interactions of rats may assist scholars in grasping the connection between brain function and social behavior, with potential relevance to human conditions like autism.

The machine-learning methodology was outlined in an article, “Mapping the Landscape of Social Behavior,” recently released in Cell. Bence P. Ölveczky, a professor in organismic and evolutionary biology (OEB) and co-author of the study, mentioned: “We are essentially mapping the social behavior of rats by documenting every aspect of their movement. We observe how they interact with one another, and we see recurring patterns of engagement.

“These animals exhibit personalities that are quite fascinating. In many respects, these variations help us comprehend the foundation of several captivating behavioral phenomena, including sociality,” he added.

Rats are inherently social beings. Similar to humans, they engage with one another in manners that shape their behavior through intricate social dynamics of touching and body language. The researchers assert that these rat interactions closely resemble our own.

Ölveczky provided a real-world illustration: “When individuals enter my laboratory, I might scratch my head a bit, and shortly after, they will do the same, or if I cross my legs, they follow suit. We are subconsciously exchanging signals.”

Although research on rat behavior has been conducted for years, earlier studies depended on direct observation and a limited scope of data.

“The conventional approach entails someone watching countless hours of rat footage and stating, ‘Oh, I believe they touched each other at that moment. I think one was mimicking the other,’” Ölveczky explained.

The recent study provided a comprehensive examination of how these social behaviors are conveyed thanks to advanced technology. Utilizing videos of the interactions, a machine-learning pipeline derived more than 110 million 3D poses that tracked diverse points on the rats’ bodies during movements and interactions. Researchers were then able to visualize how these animals behaved around their peers, encompassing how they learned and evolved through these interactions.

“By employing this methodology, we can supplant the subjective human observer with a highly rigorous and reproducible method for quantifying and identifying specific gestures or even interaction patterns,” stated Ölveczky.

AI also empowered the researchers to “process volumes of data that would take humans an eternity to analyze,” noted Ugne Klibaite, a postdoctoral fellow in the Ölveczky Lab and lead author of the paper.

“Given the advancements in computer vision and deep learning along with the technology available, including cameras and computers, we can obtain high-definition animal movement in 3D,” Klibaite remarked. Now, she added, “we have the opportunity to consider what that could signify.”

This advancement is already paving the way for new inquiries into autism. As a complex condition, autism likely involves environmental factors, according to Ölveczky. However, it also appears evident that there is a genetic element, with certain high-risk genes making an individual more susceptible to autism.

“The pertinent question is how does a mutation or deletion in this gene impact the brain, and how does that result in alterations in social behaviors?”

With backing from the Simons Foundation for Autism Research, which supplied rats with variations of these specific genes, the researchers were equipped to examine how these genetically altered rats interacted socially.

While emphasizing that autism is fundamentally a human condition, the scholars observed some intriguing similarities.

“This is a spectrum disorder, and we observe some of that variability in our various rat models as well,” Ölveczky noted.

Pointing out that children on the autism spectrum frequently socialize differently than those not on the spectrum, he expressed, “We also witness a broad range of diverse variations in social interactions in these rats that depend on which specific gene was deleted.”

Continuing research will delve into these parallels and how they might relate to the modified genes.

“Using this framework, we aim to explore how different regions of the brain respond to social gestures,” Ölveczky added. “Can we probe deeper and accurately identify the circuits responsible for these behavioral differences? And if we succeed in that endeavor, it could very well inspire new therapeutic strategies.”

Moreover, enhancing the study’s value, the data — which includes films of the rats and the extracted movement trajectories — will be made accessible, stated Klibaite, who spearheaded the data collection and behavioral evaluation.

“By sharing this with the community and encouraging engagement with the data, we hope to enable people to contribute to the conversation and develop improved models of how the brain facilitates social behavior.”

---

Financial support for this study was partly provided by the National Institutes of Health.