By Michelle Jaffee
A new study offers insights into measuring social dominance in mice, a field of research that aims to develop new therapies for autism and other neuropsychiatric disorders with social dysfunctions. Published Nov. 23 in eNeuro, the findings stem from a combination of behavioral tasks and machine learning using two genetically distinct mouse models, providing data intended to boost future soundness of studies.
Led by MBI researcher Nancy Padilla-Coreano, Ph.D., a UF assistant professor of neuroscience, the research team ran four distinct tests to measure behavior and calculate social rank. Three of those tests (urine marking, the tube dominance test, and observation of aggression) are commonly used in dominance research, while the team also added a fourth novel, trial-based reward competition designed by Padilla-Coreano to use machine learning to assess subtle differences in behavior beyond just wins and losses.
We used deep-learning to track body parts of multiple mice interacting in a competition task. We then used AI for unsupervised clustering of the poses to produce unique behavioral states, or “clusters,” to compare behavior and show how distinct groups behave differently.
Nancy Padilla-Coreano, Ph.D.
The results showed that in one of the two mouse types, dominance scores from the well-established and commonly used tube test did not correlate with scores from the other three tests, highlighting the importance of selecting a test valid for the mouse model being used, Padilla-Coreano said.
“With the novel data we provide, researchers will have information they need to choose tests that are the best fit for the facet of dominance behavior they hope to measure,” she said. “We also hope this project encourages other researchers to incorporate machine learning tools, which adds an additional, rich layer of data when incorporated into traditional behavioral assays, both within and beyond social behaviors.”
The study, Padilla-Coreano added, provides insight into the variability of dominance tests and offers guidance to researchers studying how social rank affects the brain and behavior.
“In order to capture the nuances of dominance and produce high-quality science about social rank, researchers must be mindful of their test selections, reproducibility in behavioral research, and accurate measures of dominance,” she said.