Aeon: an open -source platform for testing normative models of natural behaviours and their neural implementations

Natural behaviour is organized hierarchically into mutually exclusive modules (like feeding or escaping) with sequences of finer behavioural submodules (like object-manipulation or run-initiation) progressively nested within, down to elementary motor actions[1]. Theoretical work proposes that the brain represents natural behaviour through nested attractor networks[2]. Experimentally testing this theory requires simultaneous recording of neural activity from multiple brain regions and behaviour as they unfold. This is challenging due to complexities in a) mimicking natural settings-including large, complex environments and the presence of conspecifics - and b) continuously recording and monitoring key variables over long timescales, while retaining parametric control. To address these challenges, we present Aeon: a platform for continuous months-long recordings of natural behaviours, including foraging, social, defensive, and their neural bases in mice living in complex, scalable and dynamic environments. Aeo