The scarcity of detailed experimental information is often a
severe limit to designing accurate and realistic mechanistic models;
the intrinsic flexibility of a MAS-based simulation strategy, how-
ever, may often smooth the problem. At our knowledge, such a
strategy has not been systematically attempted as yet to simulation
of regulatory phenomena where local interactions between couples
of nodes (neurons) induce the emergence of a global behavior
pointing to a homeostatic equilibrium.
A direct comparison with an alternative strategy based on dif-
ferential equations would probably be unfavorable to our approach
in terms of speed and efficiency, but not in terms of straightforward,
intuitive correspondence between in vivo and in silico events.
The basic experimental reference typical of any simulation of
functionally correlated brain areas is depicted in Fig.6: from such
information a functional connectivity matrix [20] is derived and
henceforth used as a compact reference to the number of nodes in
the network and to their reciprocal interaction.
The simulation strategies designed in the two following sec-
tions deal with the so-calleddefaultcondition of brain activity,
which is in a resting state characterized by the absence of anyFig. 5Oscillating activity waves of brain neurons simulated by a MAS. Panels fromItoVIhave been recorded
in sequence at about the same time interval (105 s) from each other. The program used in the simulation is
described in [17]
Multi-agent Simulations of Population Behavior... 315