Central pattern generators (CPGs) are neural networks to produce a rich multiplicity of rhythmic activity types like walking, breathing and swim locomotion. Basis principles of the underlying mechanisms of rhythm generation in CPGs remain yet insufficiently understood. Interactive pairing experimental and modeling studies have proven to be vital to unlocking insights into operational and dynamical principles of CPGs and support the consensus that the most of essential structural and functional elements in vertebrate and invertebrate nervous systems are shared.

We have developed a family of highly-detailed, biologically plausible CPG models using the extensive data intracellularly recorded from constituent interneurons of the swim CPG of the sea slug {\it Melibe leonina}. We also have deduced fundamental properties needed for the devised Hodgkin-Huxley type neuronal models with specific slow-fast dynamics to become qualitatively and quantitatively similar to biological CPG interneurons and their responses to parameter and external perturbations. We have studied the onset and robustness of rhythmogenesis of network bursting the CPG circuits comprised of tonic spiking interneurons coupled with mixed inhibitory/excitatory, slow chemical synapses. We have shown that the mathematical CPG model can be reduced functionally from an 8-cell circuit to a 4-cell one using the calibration of timing and weights of synaptic coupling between CPG core interneurons.

We demonstrate that the developed mathematical network meets all the experimental fact-checks obtained for the biological Melibe swim CPG from a variety of state-of-the-art experimental studies including dynamic-clamp recordings, external pulses perturbations as well as from its forced behaviors under applications of neuro-blockers such as curare and TTX.

Our model and developed mathematical approaches and computational methodology allow for laying down theoretical foundations necessary for devising new detailed and phenomenological models of neural circuits and for making testable predictions of dynamics of rhythmic neural networks from diverse species.