Meeting Abstract
Context dependence, redundancy, multifunctionality, and evolutionary history can obscure a function description of sensory feedback. Given the complexity of neural and mechanical systems, how do we identify common strategies for sensory feedback? One fruitful approach has been careful, systematic characterization of each successive component in a neuromechanical cascade. However there are pitfalls in such interpretation due to the context dependence. As an alternative, we can explicitly analyze and design experimental manipulations that alter the feedback dynamics of a behavior. This approach affords strong, quantitative predictions especially for understanding how the role of sensory feedback changes from one context to another. Here I will focus on dynamic flower handling in hawk moths, a behavior where they must hover in mid air, cast back and forth up to 14 times a second to track flower movement, and do so in exceptionally dim light. Using this system identification approach we have been able determine how redundancy and sufficiency play a role in the integration of mechanosense and vision. We have learned how multiple species tune sensory processing to their preferred ecological contexts and how moths adjust sensing to be robust to changing conditions (unsteady environments, or large mass changes). We can frequently capture these shifts in just one or a few parameters showing that neuromechanical systems can manifest “simple” dynamics, even if the implementation is potentially complex and the underlying neural and mechanical interactions are non-linear. This feedback system identification approach generates descriptions of sensory feedback strategies analogous to the template mechanics models ubiquitous in locomotion –low dimensional targets that are realized, frequently in multiple different ways, in the full complexity of neural systems.
