Modelling of the high firing variability of real cortical neurons with the temporal noisy-leaky integrator neuron model

Abstract

Using the Temporal Noisy-Leaky Integrator (TNLI) neuron model with reset, we observed that high firing variability can be achieved for certain input parameter values which results from the temporal summation of noise in the dendrites and the use of random synaptic inputs (0-pRAMs). This was done by calculating the rate normalised Coefficient of Variation (Cv) of the interspike interval distribution. Recent experimental observations have shown that firing in real cortical neurons is consistent with a near-random process (e.g., Cv ≈ 1), but there has been concern that neuron models which use the temporal integration of random EPSP's can only produce very low firing variability (Cv ≪ 1), (Softky and Koch, 1993). We also show in this paper, that for bursting behaviour large Cv's (Cv ≫ 1) can indeed be achieved, which results from the use of random synapses (0-pRAMs) and distal inputs. For non-bursting behaviour, the irregularity of the output firing of the TNLI increases when inhibitory inputs are added.