Modern nonlinear dynamics has recently provided a new theoretical formalism for　the description of transient neural activity based on stable heteroclinic channels. Many cognitive functions are built with robust sequential activations of different　brain regions as seen in EEG and fMRI experiments. This sequential switching dynamics of　metastable brain network modes can have as an origin a winnerless competition among different brain elements. On the other hand, closed-loop technologies have an enormous potential to characterize and　control neural dynamics and can bridge between traditionally disparate levels of analysis. In this talk, I will describe how closed-loop technologies together with the concept of　heteroclinic channels can be used together to build novel interactions between living brains and artificial devices.