Kresimir Josic, Univ of Houston and Valentin Dragoi, Univ of Texas at Houston

In experimental studies post-adaptation increases in spike-field gamma coherence (essentially the synchrony between spikes in single cell, and the underlying rhythm in the 40-80Hz range) are consistent with an increase in neurons’ coding performance. We have used a model network of integrate and fire neurons to see what could explain this phenomenon and the ability of the network to represent information. The simulated neurons were assumed to be part of a larger population exhibiting oscillations in the gamma range. Adaptation was assumed to induce short term synaptic depression (weakening of connections) and was modeled by a decrease in the synaptic weights across the network. This resulted in increased spike-field synchrony, a decrease in response variability, as well as firing rate. Neuronal discriminability (a measure of how well two stimuli, such as two bars with a slightly different orientation, can be distinguished by observing only the neuronal response) increased with adaptation strength. This increase was driven by decrease in the variability of the response. Our preliminary analysis suggests that the reduction of variability observed numerically is a result of a type of network stochastic resonance. This phenomenon has previously been studied on a single neuron level using linear response theory. The goal of this project is to test these predictions and their robustness on more realistic networks that include inhibitory interneurons. We will also see whether linear response theory can be used to explain the observations analytically. Kresimir Josic