Frontiers in Systems Neuroscience (Nov 2021)
A Model of Predictive Postural Control Against Floor Tilting in Rats
Abstract
Humans and animals learn the internal model of bodies and environments from their experience and stabilize posture against disturbances based on the predicted future states according to the internal model. We evaluated the mechanism of predictive control during standing, by using rats to construct a novel experimental system and comparing their behaviors with a mathematical model. In the experiments, rats (n = 6) that were standing upright using their hindlimbs were given a sensory input of light, after a certain period, the floor under them tilted backward. Initially, this disturbance induced a large postural response, including backward rotation of the center-of-mass angle and hindlimb segments. However, the rats gradually adjusted to the disturbance after experiencing 70 sequential trials, and a reduction in the amplitude of postural response was noted. We simulated the postural control of the rats under disturbance using an inverted pendulum model and model predictive control (MPC). MPC is a control method for predicting the future state using an internal model of the control target. It provides control inputs that optimize the predicted future states. Identification of the predictive and physiological parameters so that the simulation corresponds to the experiment, resulted in a value of predictive horizon (0.96 s) close to the interval time in the experiment (0.9–1.15 s). These results suggest that the rats predict posture dynamics under disturbance based on the timing of the sensory input and that the central nervous system provides plasticity mechanisms to acquire the internal model for MPC.
Keywords
