Effect of Noise on the Coding Properties of Two Fundamental Types of Neurons

It is known that from a dynamics-systems point of view there are two fundamental classes of spiking neurons: type-I and II. Here we analyse the effect of noise on basic properties of these two classes (f-I-curves, latencies, phase-resetting curves). We also contrast the relative information coding capabilities of each type for band-limited Gaussian signals. With increasing noise, the f-I-curves of both classes become similar. However, the two classes are still distinguishable based on latency, phase resetting and spike clustering measurements. The unexpected result of the information study is that the "resonator" type-II neuron encodes at least as well as the type-I neuron