Computational neuroscience has become effectively divided into three branches that study either the structure, dynamics or computation of neural systems. It is argued here that the field as a whole would benefit from the creation of mathematical links between these three branches. The four papers published during the course of this Ph.D. are presented, each dealing with different ways of analysing neuroimaging data in the context of graph theory (structure), percolation theory (dynamics) and information theory (computation). Future directions are discussed, highlighting ways in which first steps may be taken towards the formulation of a unified mathematical foundation of neural systems.