www.neuron.yale.edu

How can i implement a gaussian noise in a network of intfire1 cells that can generate spontaneus activity and subthreshold oscillations?

The implementation of the IntFire1 class assumes that the state variable m (analogous to membrane potential) is entirely deterministic, affected only by synaptic inputs or the firing of the cell (which happens if m > 1). The implementation does not allow for endogenous fluctuations of m or of spike threshold. Noise can be introduced by driving the cell with events that occur at random times or that have random fluctuations of weight; would either of those be suitable to your purpose?

Is it possible to modify the IntFire1 mod file, create a new artificial cell with a noise that affect m inside?

Yes, at the cost of reducing its computational efficiency. The main reason for IntFire1's efficiency is the fact that calculations are done only when it receives a new event. A model that includes "noise" may impose more computational overhead. There are many possible ways to introduce "noise"--what did you have in mind? perturbation of m itself, or of the spike threshold, or what? Or do you merely need "synaptic noise"?

I was thinking about a perturbation of m but i'm also interested at any other, more efficient method that allow me to introduce noise in my network...

The easiest noise to add is synaptic noise. Doesn't require any NMODL code to be written at all. This can be done with one or more NetStims (read about this class's noise parameter). With a pair of NetStims--one that attaches to the target with a NetCon with weight > 0, and another that attaches with weight < 0--you could even construct noisy synaptic input in which the excitatory and inhibitory inputs cancel out "on the average" (i.e. the mean effect on m would be 0, yet m would definitely show as much variance as you like).

Ok, i will use NetStims pairs, thanks for your help.