Nasal breathing can entrain fast oscillations in the prefrontal cortex and limbic systems, as experiments with rodents and intracranial EEG recordings in patients have shown. Recently, it was demonstrated that the activity of the amygdala and hippocampus can also be studied non-invasively, using functional magnetic resonance imaging (fMRI). When simultaneously recording BOLD signals, respiration and cardiac RR interval (RRI) time courses, and applying a multivariate autoregressive (MVAR) model combined with Granger causality analysis, it becomes possible to assess directed coupling, or information flow, between brain structures and the body organs (i.e., heart and lung). Klimesch's binary hierarchy model links fast neural oscillations in the gamma and beta range with medium-range rhythms of the heart and respiration, as well as with infra-slow BOLD oscillations (<0.1 Hz), all aimed to minimize the brain's energy demands. This model predicts three preferred breathing frequencies, 0.32 Hz, 0.16 Hz, and 0.08 Hz, as well as an infra-slow oscillation (ISO) at 0.02 Hz. Remarkably, the preferred breathing frequency of 0.32 Hz (∼20 cycles per minute) appears to be a universal marker of anxiety, which is observed in patients and healthy participants during MRI sessions.