Input signal accumulation capability of the FitzHugh-Nagumo neuron.

Abstract

We present numerical results on the effects of two presynaptic FitzHugh-Nagumo neurons on a postsynaptic neuron under unidirectional electrical coupling. The presynaptic neurons affect the postsynaptic neuron not simultaneously but with a certain time shift. We consider cases where the amplitudes of the presynaptic spikes can be both higher and lower than the excitation threshold level. The latter case receives the main attention in our work. We carefully examine the conditions under which the postsynaptic neuron is excited by the two asynchronous external spikes. With arbitrarily chosen parameters, the FitzHugh-Nagumo neuron is almost incapable of accumulating the energy of external signals, unlike, for example, the leaky integrate-and-fire neuron. In this case, the postsynaptic neuron only excites with a very short time delay between external impulses. However, we have discovered, for the first time, a parameter region where neuron excitation is possible even with significant time delays between presynaptic impulses with subthreshold amplitudes. We explain this effect in detail and describe the mechanism behind its occurrence. We identify the boundaries of this region in the parameter plane of time delay and coupling coefficient by varying the control parameter values of the neurons. The FitzHugh-Nagumo neuron has not previously been used as a node in spiking neural networks for training via spike-timing-dependent plasticity due to the lack of an integrate-and-fire effect. However, the detection of a certain range of parameters makes the potential application of this neuron for STDP training possible.