Projects
Neural Dynamics in Social Behavior
PI: David J Anderson (Division of Biology and Biological Engineering)
SASE: Nestor Coria, Scholar
Our understanding of social behavior has been revolutionized by the ability to observe the rich dynamics of neural activity in the brains of freely interacting animals under naturalistic conditions. By studying these patterns of neural activity, the Anderson laboratory at Caltech has uncovered signals deep within the hypothalamus of the mouse brain that correlate with aggression, mating, and defensive behaviors. To unlock the relationship between neural activity and behavior, the lab currently uses a suite of mathematical tools. More recently, the lab has also pioneered the use of machine learning (ML)-enabled dynamical systems models that allow the dissection of neural activity during social behavior in an unbiased, unsupervised manner. Application of these ML models have allowed the identification of new computations performed by the brain during aggression and mating. In order to support the scientific work being done, the Schmidt Academy collaborated with the Anderson Lab to develop two new tools to perform the tasks of behavioral analysis and for analyzing neural dynamics.
The tools developed for these tasks were Batik, which uses next generation deep learning architectures for end-end analysis of behavior videos, automating common lab tasks with state-of-the-art accuracy and Onam, a software suite with a user-friendly interface that allows experimental neuroscientists to perform reproducible complex population analysis and dynamical system modeling techniques. Both tools were written in Python, allowing for future techniques which involve ML to be more easily implemented.
Through this collaboration with the Anderson Lab, the Schmidt Academy has helped accelerate the pace of scientific discovery within the lab, providing easier access to techniques and analysis methods which were previously more difficult to set up.