This research addresses a fundamental question in biology and neuroscience: how do animals process sensory information for the control of locomotor behaviors? Behaviors can be described as a sensorimotor loop: sensing (sensori-) governs action (-motor), action changes the environment, and these changes are perceived via sensing. Animal behavior arises from a concert of sensory, computational, and mechanical systems. Often, these mechanisms are studied independently (and often isolated from the context of the behavior) and the behavioral model is constructed from knowledge of the constituents, a bottom-up synthesis. Complementary to this approach, we model behavior at the level of the sensorimotor loop (the task-level) and subsequently generate hypotheses as to the mechanistic constituents. These top-down models serve to constrain permissible mechanisms and identify necessary neural computations.;We design an assay of experiments and frequency-domain analyses to identify task-level behavioral models, specifically for image-stabilization behaviors. Image-stabilization describes a broad class of behaviors in which animals modulate movement to fixate a sensory signal. In this dissertation, we study analogous behaviors in two species: refuge-tracking in weakly electric knifefish and stripe-fixation in fruit flies.;Glass knifefish swim forward and backward to maintain their position relative to a moving refuge. Fish were recorded performing refuge-tracking behavior for sinusoidal (predictable) and sum-of-sines (pseudo-random) refuge trajectories. System identification reveals a notable nonlinearity in the behavior; the frequency response functions (FRFs) generated from predictable and pseudo-random experiments are categorically different. The data support the hypothesis that fish generate an internal dynamical model of the stimulus motion, hence enabling improved tracking of predictable trajectories (relative to unpredictable ones) despite similar or reduced motor cost.;Fruit flies adeptly coordinate flight maneuvers to seek, avoid, or otherwise interact with salient objects in their environment. In the laboratory, tethered flies modulate yaw torque to steer towards a dark vertical visual stimulus. This stripe-fixation behavior is robust and repeatable; in series of experiments, flies stabilize moving stripes oscillating over a range of frequencies. We parameterize this FRF description to hypothesize a Proportional-Integral-Derivative (PID) control model for the fixation behavior. We demonstrate that our hypothesized PID model provides a parsimonious explanation for several previously reported phenomena.
展开▼
