Predictive control for spike pattern modulation of a two-compartment neuron model

Modulating the spike patterns of single neuron is an important content of closed-loop cellular electrophysiology. This paper investigates the spike pattern control of a physiologically accurate two-compartment Pinsky-Rinzel model with a fast and local optimized generalized predictive control algorithm. In order to simulate the electrophysiological experiments more realistically, uncertainties including model parameter variations, noises with arbitrary intensities and changes of inner states are introduced in the model. Simulation results demonstrate the robustness of the proposed algorithm on different uncertainties. What is more, the control performances rely on the value of the geometric parameter that characterizes the ratio between the soma area and the dendrite area while has little to do with the internal coupling conductance. The derived theoretical simulations represent a new direction towards the design of online optimal stimulus waveforms in real cellular electrophysiological experiments, which can effectively cope with the physiological uncertainties.