Figure 2.

Schematic showing the construction of the generative model for birdsongs. This comprises two Lorenz attractors where the higher attractor delivers two control parameters (grey circles) to a lower level attractor, which, in turn, delivers two control parameters to a synthetic syrinx to produce amplitude- and frequency-modulated stimuli ((a) neuronal hierarchy and (b) syrinx). This stimulus is represented as a sonogram in (c). The equations represent the hierarchical dynamical model in the form of equation (2.6).