Networks of Neurons - Computational Explorations in Cognitive Neuroscience - page 94

to the inhibitory conductance of a neuron instead of its
excitatory conductance. We have also set one of the ac-
tivation parameters to be different for these inhibitory
neurons, as discussed below.
Let's begin as usual by viewing the weights of the
network.

1−

0.8−

0.6−

0.4−

Select r.wt and click on some of the hidden layer
and inhib layer units.
Most of the weights are random, except for those
from the inhibitory units, which are fixed at a constant
value of .5. Notice also that the hidden layer excita-
tory units receive from the input and inhibitory units,
while the inhibitory units receive feedforward connec-
tions from the input layer, and feedback connections
from the excitatory hidden units, as well as inhibitory
connections from themselves.
Now, we will run the network. First, we can open a
graph log that will record the overall levels of activation
(average activation) in the hidden and inhibitory units.

inhib

0.2−

, !

hidden

0−

0 102030405060708090100

Figure 3.21: Plot of average hidden (excitatory — red line
in the actual simulation) and inhibitory (orange line) unit ac-
tivities, showing how inhibition anticipates excitation.

To toggle the display off, click on the Display but-
ton in the upper left-hand corner of the NetView window.

Hit View in the control panel and select

Strength of Inhibitory Conductances

GRAPH_LOG .
The avg_act_Hidd line (in red) plots the average
hidden activations, and the avg_act_Inhi line (in
orange) plots the average inhibitory activations.

, !

Let's start by manipulating the maximal conduc-
tance for the inhibitory current into the excitatory
units, g_bar_i.hidden (note that the . means that
hidden is a subfield of the overall g_bar_i field that
occupies one line in the control panel), which multiplies
the level of inhibition (considered to be a proportion of
the maximal value) coming into the hidden layer (ex-
citatory) neurons. Clearly, one would predict that this
plays an important role.

Now, select act to view activations in the network
window, and press Run in the control panel.
You will see the input units activated by a random
activity pattern, and after several cycles of activation
updating, the hidden and inhibitory units will become
active. The activation appears quite controlled, as the
inhibition counterbalances the excitation from the input
layer. Note that the level of the leak current g_bar_l
is very small at .01, so that virtually all of the counter-
balancing of excitation is being performed by the inhibi-
tion, not by the leak current. From the average activity
plotted in the graph window (figure 3.21), you should
see that the hidden layer (red line) has around 10 per-
cent activation.
In the next sections, we manipulate some of the pa-
rameters in the control panel to get a better sense of
the principles underlying the inhibitory dynamics in the
network. Because we will be running the network many
times, you may want to toggle the network display off
to speed up the settling process (the graph log contains
the relevant information anyway).

, !

Decrease g_bar_i.hidden from 5 to 3 and press
Run . Then increase g_bar_i.hidden to 7 and press
Run .

, !

Question 3.10 (a) What effect does decreasing
g_bar_i.hidden have on the average level of ex-
citation of the hidden units and of the inhibitory units?
(b) What effect does increasing g_bar_i.hidden
have on the average level of excitation of the hidden
units and of the inhibitory units? (c) Explain this pat-
tern of results.

Set g_bar_i.hidden back to 5.
Now, let's see what happens when we manipulate the
corresponding parameter for the inhibition coming into
the inhibitory neurons, g_bar_i.inhib . You might

, !