Imprints on our Biology that Determine our Behavior
Our recent study in Cell Reports is big step forward towards our vision of integrating naturalistic behavior, machine learning and genetics to understand the cellular and molecular basis of mammalian decisions and actions. This is important work, since our behavior, including our decisions and actions, are the biggest factors shaping our health and quality of life. So what determines them?
Our lab is developing and using computer vision and machine learning approaches to decompose complex, natural behaviors into finite component pieces that we call “modules”. Modules are discrete, genetically controlled behavioral sequences that reproducible and discovered from large populations of mice using unsupervised, top-down decompositions of the behavior. Modules range from less than a second in length to hundreds of seconds in length.
We have learned that are modules important, biologically valid components of behavior that can be mapped to genes and cells and are different from more stochastic and flexible components of behavior. This is a powerful insight for behavioral genetics and we are aiming to build a transformative new field of AI-driven behavioral genetics that we call COMPUTATIONAL DECISION GENETICS.
Our new paper shows how these approaches can reveal novel genetic, epigenetic and cellular mechanisms of parental control over the behavior in sons and daughters.