J. Xing et Gl. Gerstein, NETWORKS WITH LATERAL CONNECTIVITY .1. DYNAMIC PROPERTIES MEDIATED BYTHE BALANCE OF INTRINSIC EXCITATION AND INHIBITION, Journal of neurophysiology, 75(1), 1996, pp. 184-199
1. We studied the rapid dynamic changes of neuron response properties
in the somatosensory cortex by the use of computer simulations. The mo
del consists of three feedforward layers of spiking neurons, correspon
ding to skin, subcortex, and cortex structures. Measurements and analy
sis of model activity throughout this work are similar to those used i
n neurophysiological experiments. 2. The effects of various parameters
on response properties of model neurons were investigated. The most i
mportant parameters were the lateral excitation and inhibition in the
simulated cortical network. 3. The balance between excitation and inhi
bition is a key factor in determining the stability of the network mod
el. There is a large excitation-inhibition (E-I) parameter region with
in which the model can stably respond to inputs. 4. The input-output r
elations and receptive field (RF) sizes of simulated neurons are modif
iable by the E-I balance. The shapes of RFs are determined by both fee
dforward projections and the spatial distribution of lateral connectio
ns.5. We simulated changes in temporal and spatial properties of neuro
ns in response to manipulations that mimic bicuculine methiodide or gl
utamate application to the cortex. Simulation results agreed well with
experimental data, suggesting that cortical transmitter levels play a
n important role in the dynamic responses of the neural net through th
eir effects on E-I balance. 6. With parameters of the model set to an
inhibition-dominant scheme, the model was able to reproduce experiment
ally observed rapid RF expansions that follow cortical lesion or input
denervation. Simulation results also suggested that spontaneous input
s to a sensory system can serve as a source of tonic inhibition in the
cortex. 7. We conclude that lateral connections could produce and mai
ntain a cortical network having dynamic properties without the need to
invoke synaptic plasticity. Individual neuron properties could be mod
ified by changing the balance of cortical layer excitation and inhibit
ion. In a real brain, this could be achieved either by changing levels
of cortical transmitter (gamma-aminobutyric acid, for example) or by
changing tonic background input to the cortical network.