Using weakly electric fish to study sensing, localization and mapping

Through natural selection, animals evolved the abilities required to survive in a complex dynamic world: they accurately perceive their surroundings, map their environment and localize themselves and others within it, develop flexible policies to achieve various goals, and translate these policies into motor actions. The natural mechanisms executing these tasks outperform currently existing artificial systems. They are implemented in hierarchical networks of closed circuits composed of complex non-linear dynamic elements – neurons and synapses. In this talk, I will show how frameworks and tools of engineering are used to investigate these mechanisms in the weakly electric fish, an animal that provides unparalleled access to the study of sensing, learning, and control in three-dimensional space. A novel method I developed enables continuous, long-term recording of motor, sensory and neuronal signals in freely swimming fish. I used this method to study a brain circuit implementing an internal forward model involved in Adaptive Noise Cancelation of incoming sensory signals. I have found that such cancelation is critical to the animal’s ability to perceive and that its implementation requires both motor command and sensory feedback signals. I will discuss future directions related to how the brain constructs an internal map of the environment and localize within it. Insights generated about these processes in natural systems will be used to advance artificial sensing and autonomy in terrestrial, aquatic, or aerial systems.

Zoom Meeting