Facilitating the science in computational neuroscience

'Computational neuroscience' means different things to different people, but to me, a defining feature of the computational approach is that the two-way bridge between data and theory is emphasized from the beginning. All science, of course, depends on a symbiosis between observation and interpretation, but achieving the right balance has been particularly challenging for neuroscience. Here I discuss some of the difficulties facing the field, and suggest how they might be overcome.