Meeting Abstract

40.5  Friday, Jan. 4  Determining Neuromechanical Control Architecture Using Kinematic Phase Response to Perturbations REVZEN, S.*; BERNS, M.S.; KODITSCHEK, D.E.; FULL, R.J.; U. Cal. Berkeley; U. Cal. Berkeley; U. Penn. Philadelphia; U. Cal. Berkeley shrevz@berkeley.edu

We define several neuromechanical control architectures that represent rhythmic motion. The first class is a mass-spring system interacting with the environment whose motions are triggered by specific events. The second class is a mass-spring system driven by the feedforward signal of a CPG-like clock. The third class is a mass-spring system coupled to a clock, but with proprioceptive feedback that tracks a trajectory without altering the dynamics of the clock. The fourth class is similar to the third, but allows feedback to modulate the clock dynamics. We propose that a battery of perturbations to an animal can provide outcomes that allow identification of an architecture. To define these architectures, we selected a vertical hopping model that has received analytical treatment in both robotics and biomechanics, because it is a simple model that captures the essential phase and frequency responses of a neural pattern generator coupled to a mechanical oscillator. We assume that kinematically derived measurements of mechanical phase reflect the coupled internal neural clock phase and can be used to capture aspects of the various motor systems� phase response curves during rhythmic behavior. We subjected the four models to three perturbations that include a bump, a step, and an incline. In the first class phase and frequency change continuously. In the second, frequency is preserved by the clock and phase exhibits several discrete values. Both phase and frequency are preserved by the third class, whereas the fourth is similar in outcomes to the first, but at much longer time scales. Experiments on polypedal running animals will reveal the true empirical power of these architectural hypotheses.