banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Statistics of subthreshold neuronal voltage fluctuations due to conductance-based synaptic shot noise
Rent this article for
View: Figures


Image of FIG. 1.
FIG. 1.

Comparison of the shot-noise and diffusion-approximation voltage dynamics. (a)–(c) A case demonstrating the boundary error of the diffusion approximation: a neuron receiving both excitatory and strong inhibitory drive bringing it close to the inhibitory potential , with a leak potential also set at . (a) Shot-noise simulation of the dynamics. Between pulses the neuronal voltage relaxes to . (b) Diffusion simulation using Eq. (11) . Note that the voltage erroneously crosses the inhibitory reversal potential, which is not possible under the correct dynamics. (c) The steady-state shot-noise and diffusion-level voltage distributions. The diffusion approximation [simulation, Eq. (11) ; exact solution, Eq. (13) ] gives a significant weight to the forbidden region below . (d)–(f) A case demonstrating the skew error of the diffusion approximation: a neuron receiving excitatory drive only with a leak potential . (d) Shot-noise simulation and (e) diffusion-level simulation. The average voltage is in both cases (dotted lines). (f) The shot-noise distribution is positively skewed (peak left of the average, tail to the right) whereas its diffusion approximation predicts a negative skew with the peak to the right of the average voltage. The parameters used for the drive were as follows: (a)–(c) {0.5, 0.01, 10, 0.05}, (d)–(f) {0.25, 0.04, 0, 0} with the rates given in kHz.

Image of FIG. 2.
FIG. 2.

Comparison of the voltage moments calculated for the shot-noise and diffusion-level dynamics. Three cases are considered for which is held constant (marked on panels) while is varied over its full range {0, 1} in the absence of inhibition. In the upper panels, for all cases (a)–(c), it is seen that the diffusion-level mean, Eq. (10) , and variance, Eq. (15) (here plotted in the form of the standard deviation), agree with the corresponding shot-noise moments derived from Eq. (19) . The diffusion-level skew, Eq. (16) , however, can be seen to be incorrect in comparison with the shot-noise form, Eq. (21) , even in its expected range of validity for small large. (a) Lower panel: the true skew is positive, whereas the diffusion approximation predicts a negative value. (c) Lower panel: both the shot-noise and diffusion-approximation skew are negative, but with different magnitudes. (b) Lower panel: the intermediate case. In all examples and .

Image of FIG. 3.
FIG. 3.

Comparison of shot-noise, diffusion-level, and Gaussian-approximation voltage densities for neurons receiving synapses with a distribution of amplitudes. The upper panels show example voltage trajectories for each case, where from (a) to (c), the dynamics becomes more diffusive and less shot-like. (a) Large-amplitude shot noise; , giving mean pulse amplitudes of from the leak potential . The Gaussian approximation gives a slightly better account of the full distribution, which is positively skewed, than the diffusion approximation does, which itself is negatively skewed. (b) Moderate-amplitude shot noise; , giving mean pulse amplitudes of from the leak potential . Again the full distribution shows a significant positive skew. As can be seen in the inset, the diffusion approximation, and to a lesser extent, the Gaussian approximation underestimate the probability density for . Despite being at the tail of this distribution, this difference could potentially cause a significant error for estimations of the firing rate of neurons in the presence of a threshold for the generation of outgoing action potentials, which is typically set in the range . (c) Low-amplitude shot noise; , giving mean pulse amplitudes of from the reversal leak . In this high-rate case corresponding to the diffusion limit, both the diffusion approximation and the full shot-noise distribution approach the Gaussian approximation. For all cases, the shot-noise distribution was calculated using the numerical scheme given in Eq. (37) with the leak potential set at and with .


Article metrics loading...


Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Statistics of subthreshold neuronal voltage fluctuations due to conductance-based synaptic shot noise