frequently in a specific (CA1) hyppocampal area, followed by the
neo-cortex, and it remains a most interesting phenomenon of
neural synchronization.2.2.2 The Model The facilitated tuning of the reciprocal interactions between agents
in a MAS has always produced a most impressive output whenever
such interactions are endowed with a rythmic trend reflected by a
set of regular/periodic oscillations.
Brain neurons may be represented by agents having an activity-
cycle of the same duration and able to switch from a resting to an
active state at a given instant within its activity-cycle. Activation of
each single neuron occurs at a randomically chosen instant within
its activity-cycle; it often occurs, however, that synchronous activa-
tion patterns emerge as a result of an autocatalytic process in which
more and more neurons become active at the same time. The initial
random activation becomes synchronized since each neuron tries
and rearranges its own activity cycle in order to match the activity
cycle of other neurons in the neighborhood. It is allowed to do
that, however, only if the neuron senses a minimum number of
active neurons within defined spatial and temporal windows. In
order to reproduce periodic synchronization we made the activity
status of each neuron depending upon its metabolic energy so that,
below a given energy threshold, the neuron is unable to synchro-
nize with other neurons. In addition, the metabolic energy level
influences the maximal distance at which the neuron may sense
other neurons. Under resting conditions (random activation), the
metabolic energy increases at a relatively slow rate; during synchro-
nous activity, however, the decrease in energy proceeds at a much
higher rate. In other words, the synchronous activation regime is
characterized by a high metabolic cost, which cannot be sustained
for a long time. Whenever a lower energy threshold is reached, the
synchronization mechanism is stopped and the random activation
(resting condition) restored. Thus, the energy supplies can be
replenished making a further synchronization event possible.
2.2.3 Some Results Figure4 shows the activity patterns observed in the area represent-
ing a coronal section of the human brain, by means of a simulation
device described elsewhere [17]. The six panels in the figure refer to
the neurons activity distribution at various times (proportional to
the ticks number in each panel) from an initial fully randomic
distribution (see the inset) to the synchronous ring of neurons
clustering in different and alternating regions of the brain.
As compared to other programmable tools specialized for neu-
ronal systems, like Neuron™[18],the MAS approach appears
much more flexible, although probably less powerful in terms of
manageable models size. As an example, by a minimum amount of
programming effort in the Netlogo environment, it was relatively
straightforward to work out simulations based upon completely
Multi-agent Simulations of Population Behavior... 313