Project Summary/Abstract There is a fundamental gap in our understanding of how complex, multi-action, behavioral variability is generated by neural circuits in the brain. Understanding this process is crucial to understanding how many neurological diseases cause restricted and inappropriate selection of actions. The goal of this proposal is to understand how populations of neurons control the behavioral variability in action selection for complex behaviors. We will utilize the fruit fly to investigate this question because despite having a much simpler brain than humans, the fruit fly performs complex behaviors with high variability. We have found that activating just 24 neurons in the fruit fly brain causes variable expression of four distinct aggressive actions: stopping, wing elevation, leg extension, and lunging. The relative simplicity of this system is ideal for pin-pointing the contribution of different sources of variability: cell-autonomous stochasticity in action potential generation, variability at a network-level or variability in muscles or other neurons affected by the 24 neurons. To test these sources for variability, the project will consist of three main aims. The first is to determine whether the actions are discrete states of behavior or lie within a continuum by using advances in machine learning and neuroethology to automate the identification of the aggressive behavioral actions and the transitions between these actions. The second is to test the hypothesis that each action is caused by neurons with similar neuroanatomical connections using neuroimaging, genetic manipulation, and controlled behavioral experiments. The third is to use functional imaging, in vivo electrophysiological recordings, and computational modeling to determine the neural basis for the variability of behavior. By combining theory, experimentation, and modeling, we will be able to provide a systems level explanation of the logic behind why animals perform different actions despite consistent levels of neuronal activation.