Decomposition of neural systems with nonlinear feedback using stimulus-response data

Abstract The need often arises in many modeling studies of physiological systems with feedback, for separate characterization of the feedthrough and feedback components using stimulus–response data from the entire system. For instance, the hippocampal formation consists of multiple feedback connections which are difficult to identify experimentally. This paper investigates the application of adaptive estimation techniques in the context of the Volterra–Wiener approach to decompose unobservable subsystems from the overall feedback system data. Computer simulation studies have demonstrated its effectiveness over the traditional approach which employs nonlinear systems analysis in the frequency domain. This approach can be used to indirectly characterize the unobservable feedback basket cells in the hippocampus utilizing experimental stimulus–response data from dentate granule cells.