The brain is organized on multiple levels. The lowest meaningful one pertains to the molecular realm, followed by subcellular structures like the synapses, by cells like the neurons, and by microcircuits, mesocircuits and large-scale circuit assemblies. This stratified structure has so far hampered the interpretation of brain functions in terms of elementary electrochemical events occurring in the membranes of neurons and synapses. Each organization level is governed by emerging rules that do not simply account for the summation of events at the lower levels but require the understanding of highly non-linear interactions occurring in complex feed-forward and feed-back loops. Moreover, various forms of plasticity can persistently modify the neural circuits and their connections depending on the interactions of the organism with the environment. The brain appears thus to operate as a complex adaptive dynamical system and interpreting its function requires understanding the time-dependent evolution of multiple local activities and their rewiring during behaviour. While experimental evidence is instrumental to any further consideration on how the brain might operate, interpreting its multiscale organization in mechanistic terms requires the development of appropriate models. In this work we will illustrate how low-level representations of neuronal activity, intermediate level large-field networks and high-level connectomics can be used to explain how ensemble brain functions might emerge from elementary neuronal components.