Modeling Cortex input as Poisson Process

[dperakis]

I am modeling the basal ganglia and I need an input from the cortex.
I don't want to built separate neurons for the cortex but just 10 driving stimulators that follow a Poisson process.
When I use the NetStim command, I can't obtain (see in a graph) the output result of the 'cortex spiking activity'.

Code: Select all

f=20  // the desired mean frequency
objectvar CTXcells[10]
for i=0, 9 {
	CTXcells[i]=new NetStim(0)
	CTXcells[i].interval=1000/f
	CTXcells[i].number=(tstop*f)/1000
	CTXcells[i].start=0
	CTXcells[i].noise=0.5
}

Which is the most efficient way to have these stimulators and also visualize their additive output (for example the input to an STN neuron from 10 cortical neurons) ?
Thanks

[ted]

First a hint: instead of using arrays of objects, it is much easier to use lists to manage
collections of things.

When I use the NetStim command, I can't obtain (see in a graph) the output result of the 'cortex spiking activity'.

See
Raster plot of spikes in network
https://www.neuron.yale.edu/phpBB2/viewtopic.php?t=785
NetStims are artificial spiking cells, so the syntax for setting up the NetCon changes from this form
section netcon = new NetCon(&v(x), target)
to the simpler form
netcon = new NetCon(source, target)
That is, replace the statement
cells.object(i).soma tobj = new NetCon(&v(0.5), nil)
with
tobj = new NetCon(cells.object(i), nil)

and also visualize their additive output (for example the input to an STN neuron from 10 cortical neurons) ?

Please define what you mean by "their additive output"

[dperakis]

I mean the total impact or else the summation of the output of all these cortical neurons

[ted]

total impact or else the summation of the output of all these cortical neurons

Way too vague, especailly "impact" and "output." Please define this in nonmetaphorical
terms, i.e. in terms of experimentally observable variables.

[pjthomas]

Ted, dperakis: If you have N poisson processes each with mean firing rate alpha, and you aggregate all the events (i.e. spikes) into a single list, it will be as if you had a single poisson process with mean firing rate N*alpha. The Poisson process is unusual in this regard. Why not take advantage of this property and forget about having multiple inputs that you have to combine somehow?

[ted]

Excellent point, Peter, and a good way to simplify the implementation as long an individual event has an identical postsynaptic effect regardless of which input stream delivered it. Wouldn't work if weights or time courses were different.