Biology Reference
In-Depth Information
signed number, it turns on. We assume that there is some level of stim-
ulus in the nest that needs to be regulated or eliminated, for example,
heat. If the temperature is too hot, the insects might fan their wings to
cool the nest, a task regularly performed by honey bees. h e model as-
sumes that some level
S
of stimulus is experienced by the model insects,
represented by the elements. When an individual is βon,β it decreases the
stimulus (cools the nest) by some constant amount, 1/
S.
h e number of
elements that are on is called the
density (D)
of the network. h e current
stimulus level
(S
t
)
experienced by the elements (our network model in-
sects) is equal to the total initial stimulus level (
S
0
) minus the number
of individuals that are currently on, so
S
t
=
S
0
D.
h is is equivalent to
a directed-graph network with all nodes connected to all other nodes,
as in Figure 2.3. On the basis of a predetermined method of sampling
the elements in the network, sampled elements i rst turn of (stop per-
forming some task), which then increases the stimulus level,
S
t
,
by one
unit. h e element then checks the current stimulus level,
S
t
(an indirect
way of determining the states of all other elements in the network rep-
resented by
D
) and makes a decision to be on or of on the basis of its
threshold function. If it turns on,
S
t
is decreased by one unit.
h is is the stone-soup model: binary behavior, Boolean logic, and
random assignment of decision functions. We looked at ensembles of
these models. With the ensemble-modeling approach, we generated
many models by varying
N, K, {F},
and
S
0
over their plausible ranges.
We set
N
= 100 or 1,000 and
K
=
N
and varied the threshold set with
respect to its distribution. We looked at general (global) properties of
the system rather than specii c outcomes, such as the ability of dif erent
combinations to regulate the stimulus level at some value (equilib-
rium), how long it took the network system to reach the regulated equi-
librium, and the activity (number of elements that turned on or of ) of
the network at equilibrium. We also incrementally increased
S
to test
the ability of the network to respond and regulate the increasing stimu-
lus level (equivalent to turning up the heat in our example) and varied
the way in which the elements sampled the environment: we assigned
elements numbers, randomly selected elements one at a time, sampled
them in sequential order, or sampled all simultaneously. h e global
β