Non-homogeneous time-varying correlated spikes trains

[Elliot Imler]

Hi Romain et al.,

I was wondering if there was a way to create correlated spike trains
from a population which is time varying. Specifically I have a set of
highly correlated Poisson neurons with a baseline firing rate and then
a much higher stimulus response rate which follows an alpha function
(so I cannot simply break the responses down into time epochs) with
each neuron showing a different maximal response. Using your 2009
paper I was easily able to implement a custom mixture method when the
neurons were time-varying but homogeneous but I am wondering if there
is an algorithm for the general case.

Cheers,

Elliot

[Romain Brette]

Hi Elliot,

Do you mean that the firing rate is modulated in the same way for all
neurons by an alpha function? Then I suppose you could simply modulate
the firing rate of source spike trains?

Cheers,
Romain

Le 21/06/2011 23:20, Elliot Imler a �crit :

[Elliot Imler]

That works when all neurons have the same firing rate, you pull from a
common source train as mentioned in the paper. However when the
neurons have different rates and also vary over time I'm wondering if
there's a way to mix them. In my case the normalized modulation of
each neuron is the same (a set alpha function) but the rate multiplier
for each is different, which might simplify the problem a little.

THanks,

Elliot