Computational modeling of hindlimb locomotor pattern generation of the cat

Nearly a century has passed since the discovery of an intrinsic factor within the spinal cord of cats capable of coordinating motor activity in a pattern conducive for locomotion. The neuronal architecture of the intrinsic factor remains poorly understood. Theoretical models attempting to explain the neuronal mechanisms governing the central pattern generation for locomotion have focused on half-centre oscillators using reciprocal inhibition to generate patterned alternation of activity to antagonistic muscles. However, evidence supporting the existence of half-centre oscillators for ipsilateral coordination of motor activity is scant.;In this dissertation, I present an alternative mechanism for coordinating the ipsilateral efferent activity associated with hindlimb locomotion of the cat. Using the topographical arrangement of the hindlimb motor pools, I demonstrate that a series of waves of activity propagating rostrocaudally through the lumbosacral enlargement generates efferent motor patterns with a high correlation to the motor patterns observed during normal stepping.;To evaluate whether the discrepancies between the normal locomotor patterns and the wave-generated locomotor patterns were functionally significant, I developed a computational model of the hindlimb musculoskeletal system that exhibited stable stepping behavior when the muscles were actuated in the patterns of normal locomotion. When actuated by the wave-generated patterns, the musculoskeletal model also exhibited stable stepping indicating that the discrepancies between the patterns were not substantial enough to impair the ability to step.;To explore the neuronal mechanisms governing the rostrocaudal propagation of activity and rhythm generation, I developed computational models of the lumbosacral spinal network with patterns of synaptic connectivity based on experimental observations. When the components of the network were modeled as simple population rate code neurons, a brief stimulus to the rostral, dorsal region of the network generated a rostrocaudally propagating wave of activity. When the components of the network were modeled as more realistic Hodgkin-Huxley neurons, rhythmic bursting that propagated rostrocaudally could occasionally be observed. However, the sensitivity to parameter variation, the high frequency of bursting and the high velocity of propagation make interpretations of these results inconclusive.