Scanning and active sampling behaviours emerge from conserved insect neural circuits

Navigating insects often pause and rotate to sample their surroundings, behaviours termed scanning. These and other active sampling behaviours embody navigational uncertainty and are key for spatial learning, yet their neural basis remains unclear and existing models impose scanning behaviours rather than explaining its emergence. Here, we show that desert ants’ scanning dynamics can emerge spontaneously from the same conserved neural circuits used for goal-directed navigation, without requiring a specialised scanning module. We built a biologically grounded model combining central complex steering and lateral accessory lobe oscillators and added a downstream stochastic inhibition of forward speed. This minimal system produced diverse, realistic scan dynamics; saccades, fixations, and reversals, whose features were qualitatively compared to high-speed video recordings of Melophorus bagoti scanning. Detailed analysis of these natural scans confirmed model predictions, including how scan structure depends on oscillator phase, goal-heading deviation, and navigational uncertainty. Furthermore, the model reveals that simple modulation of forward speed unifies a broad range of behaviours across ant species, from dashes to smooth oscillatory trajectories to pirouettes and voltes. Crucially, this model suggests a distributed control principle where forward speed acts as a single adjustable parameter, for both individuals and through evolution, to regulate the balance between goal-driven exploitation and information-seeking exploration.