As Jay has pointed out, I think that an important test case for a general
theory of emergent phenomena is the work done in simulating artificial
society. The general concern of this approach is to model society - and
social phenomena - from the bottom up and I think that his concerns about
the physicist/biologist framework reveal the key point of the problem that
is:
* does social reality (and its cultural effects) play a role in the agent
behavior or is it just the effect of that behavior? Can (social) order be
fully explained without referring to a top-down process?
When Jay refers to biology as a model for complex system theory, I think
that the suggestion is to consider as more prominent those models that
share a biological inspiration such as Artificial Life. AL is the most
paradigmatic example of a biology inspired approach and I think is the
candidate model to better consider all the processual aspects of cognition
(interaction with the environment, with other agents and in time). What
makes it "biological" is that it adopts connectionism (neural nets) in
evolutionary terms (genetic algorithms) in order to let systems evolve. In
this model of agency, the top down perspective is considered in a temporal
sense. There are effects in the opportunity of reproduction (fitness)
between a population of agents relative to the adaptation to a specific
state of the environment. These effects are translated in a modification
of the genetic patrimony and, consequently, influence the behaviors - and
adaptation - of future generations.
The same model is also used to explain cultural reality where a culture is
seen as a set of behaviors shared by a group of people where these people
have learned the behaviors from each other. Artefacts are one of the means
of cultural transmission because behavior, ideas and goals can be acquired
interacting with them. So culture records past experiences (as solutions
to frequently encountered problems) and offers them to future generations
of agents. In general, cultural transmission is presented in term of
imitation of the fittests in the last generation by the new one.
But what about influence? Does the imitation process tell the whole story
about socio-cultural interaction?
In the xmca community I think that this problem seems relatively plain.
However I don't have a clear idea of the concept of "influence" developed
in the artificial society tradition. Jay says that the top down
elements are the constraints which are necessary to the emergence of a
specific order (not only a new one). Between different levels of
organization, institutions well represent this "enabling constraint" which
order social relationships.
But how are really implemented these institutions in the mind of the
agents? If an institution emerges from the behavior of multiple agents as
a solution to a coordination/cooperation problem (as in game theoretic
models), how does it persist its functioning?
I think that to account for this problem we need to introduce
in the AL agents a sort of cognitive layer (intentionality?) which must be
conceptualized as the output process of other levels (as in the CHAT
tradition). Otherwise we will find that the spectrum of methodological
individualism will be defended again also in the complex systems approach
(I agree with Keith on this and I would be very interested in his paper
on the argument).
Can the process of internalization be seen as a sort of downward
causation? Is the agent in Artificial Life able to implement this level of
(secondary or cognitive) emergence? What does it need?
Luca
___________________________________________________
Luca Tummolini
PHD candidate in "Cognitive Science"
University of Siena - Dep. of Communication Sciences
"Cognitive Science" TEl. +39-0577- 2350-79 or -98
Mobile: +39 335 66 08 983
E-mail: tummoli@media.unisi.it
This archive was generated by hypermail 2b29 : Mon Jul 01 2002 - 01:00:06 PDT