Cortical Sculpting of a Rhythmic Motor Program

Abstract

The standard model of motor cortex maintains that firing rates evolve over time based on the local connectivity of neurons within the cortical network and on external inputs to the network. The resulting activity patterns influence spinal circuitry through descending projections, and ultimately determine neuromotor output. This model has been highly successful in explaining cortical control of discrete, voluntary movements such as reaching. Yet it fails to explain cases in which cortical commands for voluntary action must be integrated with commands generated by independent subcortical centers. Our recent work has focused on coordination between neural population dynamics in motor cortex and the spinal central pattern generator (CPG) in mice performing a voluntary gait modification task. To enable animals to traverse an obstacle, motor cortex must integrate inputs conveying the barrier’s proximity and the phase of the locomotor CPG, then compute the appropriate commands to modify the forelimb trajectory. By combining large-scale neural recording in freely-moving animals, behavioral and optogenetic manipulations, and computational approaches, we identify the population-level representations motor cortex uses to implement this computation, and propose a simple algorithm that generates the appropriate output through phase-dependent gating. Our results reveal a regime of cortical function in which higher brain areas do not directly specify motor output, but must sculpt an ongoing, spinally-generated motor program to flexibly control behavior in a complex and changing environment.