The issue of the data architecture of a brain is one of only a few key points in which the present viewpoint differs from traditional science. In traditional neuroscience and neural models, real-time continuous processing is obtained by aggregating many discrete components (like neurons, or dendritic spines). I take exactly the opposite view: that real-time continuous processing is the elemental process requiring a made-to-order continuous data substrate, with discrete components sub-sampling that continuum into discrete events for transmission elsewhere. In my view, the bottom-most layer is analog, not digital. That, in a nutshell, describes this model of the brain, and even of several brains in concert.
Those issues are discussed at length in Elastic Nanocomputation.