Making waves: connectivity-based models of locomotor circuits

It is well-established that rhythmic patterns involved in coordinated movements are not generated explicitly in the brain but instead through dedicated circuits in the spinal cord. We focus on the circuit responsible for locomotion in aquatic vertebrates, which is characterized by a wave of activity traveling down the spine displaying (1) left-right alternation, (2) variable speed control, and (3) constant phase difference between adjacent segments at all speeds. While classical computational models describe the circuit as a chain of left-right oscillators, they fail to explain fully explain (2)-(3) and do not incorporate recent experimental observations such as (i) intermediate-range longitudinal connections, (ii) differing connectivity patterns among different genetically defined cell types, and (iii) speed-specific recruitment of dedicated neural populations at different locomotion speeds. Here, we propose a series of rate-based neural network models of increasing complexity, revealing which aspects of connectivity (i)-(iii) are responsible for each of the desired properties (1)-(3). Importantly, the desired patterns emerge entirely from population dynamics and connectivity motifs, challenging the requirement of complex single neuron properties present in older models and instead suggesting that cell type-specific connectivity patterns may be a fundamental driver of coordinated locomotor behavior.