Dear colleagues,
In my opinion, most of the discussion is more about human behaviour and
human minds (AI) than about the simulation of social systems as evolving
coordination mechanisms. Only if we understand the latter, we can perhaps
model the reflection of the developments at the supra-individual level in
the minds of participants and the possible consequences of these (social
developments + individual reflections) for human behaviour.
First things first.
With best wishes,
Loet
________________________________
Loet Leydesdorff
Amsterdam School of Communications Research (ASCoR)
Kloveniersburgwal 48, 1012 CX Amsterdam
Tel.: +31-20- 525 6598; fax: +31-20- 525 3681
[log in to unmask] ; http://www.leydesdorff.net/
> -----Original Message-----
> From: News and discussion about computer simulation in the
> social sciences [mailto:[log in to unmask]] On Behalf Of
> Michael Sellers
> Sent: Saturday, January 20, 2007 2:51 AM
> To: [log in to unmask]
> Subject: Re: [SIMSOC] working on emergence of norms and beliefs
>
> Keith wrote:
> > >Placing our architectural emphasis on emotions and their attendant
> > >physiological, neurological, and quasi-cognitive aspects
> means that we
> > are
> > >not trapped by rules or similar structures, but we are
> still able to use
> > >them as our "islands." And among other things, this, like
> Damasio's
> > work,
> > >points out the inadequacy of fictional views of AI such as
> Hal or Data:
> > AI
> > >without emotions is ultimately AI without cognition.
> >
> > I agree, though building an AI with human type emotions is
> a task to be
> > considered *very* carefully.
>
> No doubt. It hasn't stopped us though. :) We have a highly
> multivariate,
> layered, nuanced emotional system. This isn't a variant of
> the OCC or any
> other state-driven model; we can derive states, but they
> aren't mutually
> exclusive. Also, in our architecture, emotions don't drive
> action, they
> accompany it -- correlative, not causal. That is, you don't
> hit someone
> because you're angry; you have the internal state which you
> experience as
> anger and which also induces you to strike out.
>
> > >There are a lot of different levels at which you can cut
> off the modeling
> > >and say "from here we just use a rule/variable/equation."
> > Psychologically
> > >it's turtles all the way down, so you need to choose what your
> > >turtle-resolution will be. Where that level of cut-off is
> depends on
> > >whether you're more interested in en masse interactions or
> individual
> > ones.
> >
> > I have to ask how close you have come? I.e., how
> predictive are your
> > models?
>
> Very difficult question to answer. How predictive is *any*
> psychological or
> psychosocial model? Our agents have at least the appearance of
> predictability, in that the agents act (and emote) in predictable,
> understandable ways from a human POV (or, they are as
> inscrutable as humans,
> which perversely sometimes amounts to the same thing). We've
> been working
> with DARPA on this; our sponsor there said, "all models are
> wrong; some
> models are useful." So, I'm sure our model is wrong in some
> respects; it
> may also be useful in simulating emotive, relational
> characters acting in an
> environment.
>
> Mike Sellers
>
|