Font Size: 

In this study, we present a dynamical systems approach to modeling physiological gestures. We use as a test bench the generation of a diverse repertoire of respiratory gestures used during birdsong production by canaries and zebra finches. In our approach, the expiratory respiratory gestures used to generate different syllables are built from basic dynamical units (“sub syllabic gestures”), each one being the response of an excitable system (either a Wilson-Cowan model or a Takens-Bogdanov normal form) to a pulse. The reconstruction process employed a differential evolution algorithm to explore the parameter space, optimizing the alignment between generated gestures and actual motor behaviors. Dimensionality reduction and clustering techniques applied to the model-generated signals revealed that similar birdsong syllables tend to employ coherent sets of subsyllabic gestures across distinct individuals. This finding suggests a structured neural encoding of motor gestures, potentially representing a general principle across species and behaviors. Our results underscore the power of dynamic modeling in unraveling the principles of neural control of complex motor actions, offering insights with implications for neuroscience in general.