CDS Lecture Series

ABSTRACT

Eric Tytell
Neural Control of Locomotion Lab
Department of Biology
University of Maryland

Feedback loops in lamprey swimming
In fishes, locomotion is a result of a complex network of distributed feedback loops. The basic neural pattern is produced by a network of spinal interneurons, called a central pattern generator (CPG). The CPG’s rhythm can be modulated both by higher control centers in the brain and by sensory input from the body’s movement. The body’s motion, in turn, is determined by a feedback between muscular forces and fluid forces. Also, fishes can sense fluid motion through the lateral line system, which may have an impact on the descending control from the brain. Finally, all of these feedback loops have multiple inputs and outputs, distributed throughout a fish’s body. This distributed, nested control network has important implications for how fishes swim. In this talk, three examples are given. First, the spatially distributed pattern of sensory input results in an apparent discontinuity in muscle activation phase along the body. Open loop stimulation shows that muscle near the tail tends to be active at an appropriate phase for transmission of force to the fluid, while muscle near the head is active at a phase in which it absorbs energy, which is not appropriate for swimming. Distributed mechanical feedback due to the continuity of the body may serve to smooth out this apparent discontinuity to produce effective muscle activation patterns. Second, sensory input has a generally excitatory effect on the CPG rhythm, which increases its frequency. Thus, the isolated CPG tends to have a much slower rhythm than actual swimming. Third, the lateral line sense of the fluid motion around the animal may help to modulate the swimming kinematics as the flow changes. Tests on lampreys with pharmacologically disabled lateral line systems show that several swimming parameters, including the body wavelength, do not change appropriately as the animal swims faster. Finally, a new technique for studying these interacting feedback loops is described. In this technique, termed a biomechanical dynamic clamp, the output of the CPG from a living neuronal preparation is measured, then the biomechanical effect of that output is estimated using a real-time computer simulation, and the resulting motion is then fed back to the CPG as a bending input. This technique allows a controlled exploration of the roles of neural and mechanical feedback loops. As an example, the effect of mechanical resonance in stabilizing the CPG output is investigated. This work is being done with Avis Cohen in the lab for Neural Control of Locomotion.


Back to CDS Lecture Series
Back to Intelligent Servosystems Laboratory