Dear Glen and Scott,
I found your discussion quite interesting. I do not want enter the
technical comparison between SWARM and SDML (I do not feel competent on
this). I would simply add some consideration on the theoretical aspects of
your debate.
Swarm -as far as I understand it- is in fact quite open to any kind of
approach and of agents (included the cognitive ones). However it seems to
me more leaning to implement and model "swarm intelligence". And in general
it is the product of that important trend in current paradigmatic conflict,
based on complexity theory, emergence, evolution, etc. A quite
anti-cognitive trend at least as it is proposed by its creators and
followers (I recognise the same background view in Glen's scepticism about
modelling cognition for modelling social behaviour). I'm personally very
interested in theorising and modelling emergent and swarm intelligence,
cooperation, etc., but I believe that the current opposition between
cognition (explicit symbolic mental representations) and emergence is a
scientific mistake. My claim is that a major scientific challenge of the
first part of the century will precisely be the construction of a new
"synthetic" paradigm: a paradigm that reconciles, in a principled and
non-eclectic way, cognition and emergence, information processing and
self-organisation, reactivity and intentionality, situatedness and
planning, etc. (Castelf 97; 98).
The challenge is that of modelling emergent unaware, functional social
phenomena (ex. unaware cooperation, non-orchestrated problem solving, and
swarm intelligence) also among cognitive and planning agents. In fact, for
a theory of cooperation and society among intelligent agents, mind is not
enough. We have to explain how collective phenomena emerge from individual
action and intelligence, and how a collaborative plan can be only partially
represented in the minds of the participants, and some part represented in
no mind at all.
In other words, emergent intelligence and cooperation do not pertain only
to reactive agents. Mind cannot understand, predict, and dominate all the
global and compound effects of actions at the collective level. Some of
these effects are self-reinforcing and self-organising. There are forms of
cooperation which are not based on knowledge, mutual beliefs, reasoning and
constructed social structure and agreements.
Glen said:
>Good answer. And I suppose it indirectly addresses what I really want
>to know, which is whether the foci of the social scientists are primarily
>directed at understanding the individual or primarily directed at group
>behavior. What you're telling me is that social science is probably intent
>on providing a *map* between psychology social processes. And that I
>can believe.
In fact, the social scientists are primarily directed at understanding the group
behavior. The problem is what does it mean 'understanding' the group
behavior unless understanding the causal mechanisms that lie behind it and
generate it. I would like to have not only a descriptive science, or a
science simply based on correlations or external laws, or on wordy
theories, but I would like to arrive to operational 'explanatory' models.
Now, precisely computer simulation can provide the social sciences (SS)
with the methodological and theoretical apparatus to eventually cope with
the problem of the invisible hand and of the micro-macro link and to
achieve an explanatory level. In other words, agent-based social simulation
could precisely explain 'group behaviour'(and cognition) in terms of
'individual behavior' (and cognition), and vice versa.
In the previous discussion on the list (soc simulation and mas) very few
people stressed that social AI is not only engineering aimed at
constructing some practical system and interested in importing ideas,
notions and theories from the social sciences, but it is a science itself
entitled to provide new theories and models and to challenge those
furnished by the SS. A few people stressed how soc sim is important not
only
a) for simulating complex social phenomena and for exploring policies,
b) for testing theories and models provided by the SS.
Soc sim not only has an instrumental and subordinate role relative to the
SS; it is important (especially if combined with complex agent modelling):
c) for deeply modifying the current scientific apparatus (concepts,
theories, models) of the social sciences.
This is precisely what will happen. As in the '60s and '70s the impact of
the sciences of the artificial (information theory, cybernetics, AI) deeply
changed psychology, philosophy, linguistics, logics, etc. giving origin to
Cognitive Science; so, now social simulation, AI, ALife will deeply
transform our social theory.
>It does seem to contrast with the little I know about anthropology, which,
>if I were to study it, would be an attempt to prohibit assumptions about the
>internal processes of the individuals, gather as much data as possible
>about the group dynamics and then use that data to bootstrap a testable
>theory about the generative causes of those dynamics (which would most
>likely turn out to be a psychology for the individuals). I guess the question
>bears more on "What, exactly, is being studied, rather than how is it
>being studied." And I imagine the answer to that question varies wildly
>between disciplines.
First, there are also strong 'cognitive' traditions in anthropology, in
sociology, in linguistics, etc.
Second, testing a
>theory about the generative causes of those dynamics (which would most
>likely turn out to be a psychology for the individuals), as Glen says,
seems difficult without formulating hypotheses and explicit models about
these 'generative causes'. You will have a model of group dynamics but not
of their generative mechanisms that must pass through the 'control system'
of the agents: 'minds' with cognitive agents.
In those agents, the behavior is oriented and controlled by internal
anticipatory representations. Of course not all group behaviour and
dynamics are intentional or contractual. Not at all! A lot of them are
unintended and emerging. But if we do not model intentionality and
cognition (ex. expectations, beliefs) we miss precisely what should be
explained: as I said,
the real issue is precisely the fact that the intentional actions of the
agents (not only their unintentional behaviours) give rise to functional,
unaware collective phenomena (ex. the division of labour). How to build
unaware functions and cooperation on top of intentional actions and
intended effects? How is it possible that positive results -thanks to their
advantages- reinforce and reproduce the actions of intentional agents, and
self-organise and reproduce themselves, without becoming simple intentions?
Von Hayek claimed this (spontaneous order among intentional agents)
"... doubtless <to be> THE core theoretical problem of the whole social
science" (von Hayek, Knowledge, Market, Planning). I believe that SS cannot
solve 'THE core theoretical problem' without soc sim with cognitive agents.
====================================================================
Cristiano Castelfranchi
National Research Council - Institute of Psychology
Division of "Artificial Intelligence, Cognitive and Interaction Modelling"
Viale Marx, 15 - 00137 Roma - ITALY tel +39 06 860 90 518 fax +39 06 82 47 37
University of Siena - Communication Sciences
"Cognitive Psychology"
"Theories and Systems of Artificial Intelligence"
E-mail [log in to unmask] or [log in to unmask]
===========================================================================
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|