Building Embodied, Integrated Models of the Nervous System

How can neural circuit models become testable theories of how the brain generates behavior? I argue that computational models should move beyond simplified tasks to be tested in embodied, naturalistic settings. I will present my work translating biological circuit structure into trainable neural networks, such as a spinal–brainstem controller for a virtual rat, and I will discuss how this approach can be extended to other neural subsystems, including visuomotor and basal ganglia circuits.