At 08:38 PM 1/13/2007 +0100, Virginia wrote:
>Hi Rafael, Maarten, all,
>I agree with all you say. The point is to use cognitive agents in
>simulations. As there are several theories of human cognition,
snip (interesting discussion)
I have never worked in social simulation, but was familiar with system
dynamics from the early 1970s and with agents since the first Artificial
Life conference in 1986. I don't have a bias toward one or the other; both
are attempting to model reality and to the extent they succeed, the results
of each approach should be similar within error bounds.
Something you might consider is modeling agents as the result of a deeper
layer of agents, namely genes. This is the evolutionary psychology
approach which states that all human psychological traits are either the
direct result of selection or a side effect of some trait that was directly
selected. (Keeping inclusive fitness in mind.)
My experiences with a cult ( http://en.wikipedia.org/wiki/Keith_Henson )
eventually led me to writing a paper on how wars happen through the
intermediary step of xenophobic meme amplification.
Not considering perturbations, rising population in hunter gather bands
eventually made the future game and berry prospects look grim. That
activated evolved brain mechanisms to turn up the gain on circulating
xenophobic memes. Those xenophobic memes eventually synched up the
warriors for a do or die attack on neighbors.
Win or lose, there were fewer mouths to feed. In the worse case, all the
males of one band were killed and the band's young women became booty. So
the genes for these traits went on even when the band was wiped out.
Anyway, I think simulations based on humans viewed as having a long stone
age evolution would be a really interesting approach.