Using high-speed video tracking combined with a machine learning algorithms that allow to automatically classify larval behavior genetic manipulation of sparse neuronal populations and single neurons and neuroanatomy and synaptic connectivity studies, we map the neural circuit elements and pathways underlying the five different actions that occur in response to a mechanical stimulus, the air-puff. In response to an air-puff, larvae perform five different type of escape or avoidance actions: Hunch, Bend, Stop, Back-up and Crawl. By quantifying changes in behavioral probability in the different actions upon neuronal silencing, we identify neuronal elements that could underlie the competitive interaction between the five behaviors. Using neuroanatomy studies, we determine the identity of neurons involved in mechanosensory responses (Masson et al., Plos genetics, 2020) and found neurons that are located downstream of two types of sensory neurons (chordotonal and multidendritic class III neurons) that we previously found were involved in sensing air-puff or their first order partners. Our data revealed new putative second order interneurons and ascending pathways in the mechanosensory network as well as a feedback from the premotor domain. Overall, our findings suggest that the selection of behaviors occurs not in specialized centers but rather through a distributed process involving different regions of the nervous system. Furthermore, in order to investigate the functional contributions of individual circuit elements, we identify the neurotransmitter identity of key neurons (to determine whether they are excitatory and inhibitory. This combined with tracing on connectivity in electron microscopy and functional studies will reveal circuits motifs involved in competitive interactions between actions. In addition, the five behaviors are often organized in a sequence where the individual actions need to be ordered one after the other. We thus further develop a framework for using our dataset for studying the neural basis of sequence transitions during larval response to an air-puff and identify circuit elements involved in sequence generation. In addition to a peer-reviewed publication (Masson et al, 2020), results from this were also presented at an International conference and will be used to prepare two additional manuscripts.