Neural Networks

Democracy is written mainly with a neural network. Not the classic computer science setup of a NN. (I dont even know how that works) but my own impression of how neurons connect.
It works great, its cool to visualise, and very effective. You have 2 neurons with a neural effect between them (a dendrite I think its called in neuroscience) and tht effect transmits some modulated value from neuron A to B.
THIS IS EASY.
But this next bit isn't...
I add smoothing to the neural effect result over time.
So it maintains a history, and outputs a smoothed result.
At some point I have to do the periodic smoothing.
this is where I get into catch 22.
I need to do the smoothing at the end of processing, so the current effect is up to date.
But...
The smoothing changes the output effect, and thus the value of the next neuron. Which itself affects the dendrites flowing out from it.
OMFG.
Still if it was easy, everyone would be doing it right?
Doesnt help that Jerry McGuire is on TV tonite.
SHOW ME THE NEURONS!!!!