Sensorimotor behaviors often rely on model-driven control. In vertebrates, such as primates, even simple reaching gestures are constructed through the use of internal models of the body (forward and inverse models) and the world. How such models are implemented at the neuronal level remains a major unsolved question in systems neuroscience. Here I explore this question in a novel context, relying on the sophisticated sensorimotor behavior and compact neuroanatomy of insects to both complement and contrast the extensive work done in vertebrates. Using dragonfly prey capture as a model system, I will discuss four key results. First, high resolution three-dimensional free flight kinematic data demonstrates that interception guidance steering is based inverse models that implement a predictive light path. Second, in-flight steering of the head relies on forward and inverse models to stabilize the prey image against self-motion on and prey-motion. Third, both body and head steering use internal states as well as visual feedback and can proceed with their kinematics in the absence of vision. Finally, I will show that the underlying neuroanatomy is consistent with this behavior being driven by a series of parallel internal models, and I will discuss the anatomical elements and connectivity that we hypothesize form building blocks of the visually-driven portions of the forward and inverse models. If correct, these data imply a neuronal implementation of internal models substantially more distributed and implicit than the discretized and explicit representations we observe at the level of behavior.
Anthony Leonardo received the B.S. degree in cognitive science from Carnegie Mellon University, Pittsburgh, PA, and the Ph.D. degree in computation and neural systems from the California Institute of Technology, Pasadena, in 2002. He then completed postdoctoral work at Bell Labs and Harvard University. Since 2008, he has been a Group Leader at the Janelia Farm Research Campus of the Howard Hughes Medical Institute, where his lab explores the principles underlying neural information processing. As part of his B.S. degree work, he worked on problems in artificial intelligence. His current research focuses on the computations neural circuits implementation, how robust they are, and how different circuits are linked into systems to produce complex behaviors. These questions are pursued in the context of prey capture in salamanders and dragonflies. Dr. Leonardo received the Lindsley Prize in Behavioral Neuroscience, the Capranica Prize in Neuroethology, a Helen Hay Whitney Fellowship, and the Burroughs-Welcome Career Award in the Biological Sciences.