The brain is characterized by the presence of architectural structures – neurons, minicolumns, cortical columns – across a hierarchy of spatial scales. In addition, the activity of the brain is expressed – through action potentials, EEG oscillations, diurnal rhythms – across a multitude of temporal scales. We propose that bipolar disorder arises as a biological disturbance at a very fine spatial and temporal scale, within transmembrane dynamics, which then cascades across scales to be expressed at the slower scales of symptoms, episodes and ultimately the illness across the life span. This proposal is embedded within a hierarchical model of neocortical activity. Innovative data analysis methods, allowing the investigation of EEG and functional magnetic resonance imaging data from such a multiscale perspective, are presented. We hence propose a set of functional neuroscience experiments that would allow this ‘multiscale hypothesis’ of bipolar disorder to be tested.