Partial Information Decomposition in the Frequency Domain: Application to Control Mechanisms of Heart Rate Variability at Rest and during Postural Stress

Abstract

We exploit a recently proposed framework for assessing causal influences in the frequency domain to construct the partial information decomposition (PID) for informational circuits of three variables, thus obtaining the spectral decomposition of redundancy, synergy and unique information. The approach is applied to heart period (HP), systolic pressure (SP) and respiration (RESP) variability series measured in healthy subjects in baseline and head up tilt conditions. Integrating the informational quantities in the respiratory band, the total influence from RESP to HP does not change in the two conditions. However, we find that in baseline RESP causes HP mostly through the direct pathway describing central autonomic effects, whilst in head up tilt condition the direct influence decreases and becomes comparable to the information pathway mediated by SP describing baroreflex effects. Our results show the usefulness of the spectral decomposition of PID.