A variety of techniques have been developed to noninvasively image human brain function that are central to research and clinical applications endeavoring to understand how the brain works and to detect pathology (e.g. epilepsy, schizophrenia, etc.). Current methods can be broadly divided into those that rely on hemodynamic responses as indicators of neural activity (e.g. fMRI, optical, and PET) and methods that measure neural activity directly (e.g. MEG and EEG). The approaches all suffer from poor temporal resolution, poor spatial localization, or indirectly measuring neural activity. It has been suggested that the proton spin population will be altered by neural activity resulting in a measurable effect on the NMR signal that can be imaged by MRI methods. We present here the physical basis and experimental evidence for the resonant interaction between magnetic fields such as those arising from neural activity, with the spin population in ultra-low field (microT) NMR experiments. We demonstrate through the use of current phantoms that, in the case of correlated zero-mean current distributions such as those one might expect to result from neural activity, resonant interactions will produce larger changes in the observed NMR signal than dephasing. The observed resonant interactions reported here might one day form the foundation of a new functional neuroimaging modality ultimately capable of simultaneous direct neural activity and brain anatomy tomography.