Holger Lyre (Magdeburg) gives a talk at the MCMP Colloquium (6 May, 2015) titled "Structures, Mechanisms and Dynamics in Theoretical Neuroscience". Abstract: Proponents of mechanistic explanations have recently proclaimed that all explanations in the neurosciences appeal to mechanisms – including computational and dynamical explanations. The purpose of the talk is to critically assess these statements. I shall defend an understanding of both dynamical and computational explanations according to which they focus on the explanatorily relevant spatiotemporal-cum-causal structures in the target domain. This has impact on at least three important issues: reductionism, multi-realizability, and explanatory relevance. A variety of examples from the theoretical neurosciences shall be used to show that very often the explanatory relevance, burden, and advantage in view of law-like generalizability lies in picking out the relevant structure rather than characterizing mechanisms in all details.