Abstract: Neuroimaging has allowed us to map the correlations between brain activation, and external stimuli or behaviour. Yet these correlations can only hint at the function of the brain regions involved. In order to more casually investigate these relationships between brain and behaviour, we must perturb the brain, and see what changes this brings about in behaviour. I will provide a framework for doing so through covert neurofeedback. This technique allows us to perturb brain networks by reinforcing desired network states directly, through a reward orthogonal to the networks being trained. Yet a prerequisite for such a test of function and causality, is a strong hypothesis regarding the purported link between a specific network and behaviour. We must therefore also develop better behavioural tools, in order to establish such links.