Biomedical Engineering Reference
In-Depth Information
peak in the cross-correlogram is also small, indicating that the neurons fired nearly
independently. The peak is
narrow because all synaptic time constants were set to
500 ms
a
b
c
30
0
1.2
1.5
2.5
0.8
0.7
0
−200
0 200
−200
0 200
−200
0 200
Time shift (ms)
d
e
30
0
1.5
1.3
0.7
0.9
−200
0 200
−400
0 400
Time shift (ms)
Figure 12.2
Spike trains correlated by local network interactions. Same format as in
Figure 12.1
.
Neurons were modeled as integrate-and-fire units receiving two types of inputs: a
background synaptic drive that was independent across neurons, and recurrent synap-
tic input from other units in the network. The full network consisted of 100 excitatory
and 25 inhibitory model neurons. Synaptic connections were all-to-all, with conduc-
tances chosen randomly (uniformly) between 0 and a maximum value
g
max
. Note the
different y-axes for cross-correlation histograms.
3 ms. Figures 12.2b and 12.2c show what happens when the connections are made
progressivly stronger. The central peak becomes much taller (notice the different
y-axes), and secondary peaks, indicating oscillatory activity, become apparent. In
contrast to the rest of the cross-correlograms, the one in Figure 12.2c was generated
from a short segment of data. This enhanced the secondary peaks, which practically
disappeared when longer stretches of data were used (not shown). This is because
the frequency of the oscillatory activity is not constant so, over a long time, many
Search WWH ::
Custom Search