When animals move, they induce changes in the relative position of objects in their environment. These changes are perceived as visual motion signals, known as optic flow. Fast and efficient interpretation of the global structure of optic flow is thought to be critical for maintaining stability and guiding the animal's course during navigation. In flies, the lobula plate tangential cells (LPTCs), a highly interconnected network of ~60 neurons, are known to be specialised optic flow detectors. Bridging the sensory periphery with motor output structures, LPTCs are ideally positioned for course control and are therefore commonly referred to as the 'cockpit of the fly'. Although their physiological response properties have been extensively characterised in different fly species and their activity has been linked to simple turning responses, it remains a mystery why the fly requires such a sophisticated network of cells for such simple tasks. Here, we decided to address this question by using stochastic neuronal manipulations to mimic an extended repertoire of LPTC network states. This allows us to explore combinations of activation motifs similar to those resulting from different visual optic flow stimuli. Our current work recapitulates the known properties of horizontal motion-sensitive LPTCs in yaw movements and extends the role of vertical motion-sensitive LPTCs in controlling the roll axis, e.g. through fine control of wing movements. Our framework allows us to deconstruct the functional coordination of the LPTC network and provides insights into how the population activity of a relatively simple neural network controls complex behaviours.