State-dependent correlation between excitatory and inhibitory firing is explained by opposing correlating input with anti-correlating neural interactions

Cyrus Omar, Carnegie Mellon University


In neurophysiological studies we have found that populations of putative interneurons (I) and pyramidal cells (E) in layer 2/3 of the rat barrel cortex are positively correlated with each other in the spontaneous state. Upon whisker deflection E-I correlation decreases to near zero. These results are inconsistent with models of uncoupled populations where an increase in response accompanies an increase in correlation (de la Rocha et al., 2007), suggesting neuronal interactions mediate this effect. We study a simple stochastic mean-field model of coupled E and I populations. The amount of correlation and time course of background activity is set on the basis of local field potential recordings in layer 2/3. In order to fully explain both spontaneous and whisker-evoked data, we find that the following features are required: strongly correlated background activity, sufficiently strong feedforward inhibition, and a non-linear input-to-firing rate transfer function. Feedforward inhibition anti-correlates E and I responses. During spontaneous conditions, the transfer function non-linearity diminishes inhibitory gain and hence this anti-correlating influence, leaving a residual positive E-I correlation. Whisker deflection transiently shifts network dynamics towards the high-gain region of the transfer function. This amplifies the anti-correlating effects of feedforward inhibition, canceling the correlating background and yielding an uncorrelated E-I whisker response. Global firing synchrony likely degrades stimulus discrimination. Our model provides a mechanism by which E-I responses can be de-correlated in a state-dependent manner within a strongly coupled network subject to highly correlated background input.



de la Rocha, J., Doiron, B., Shea-Brown, E., KreŇ°imir, J., Reyes, A. Correlation between neural spike trains increases with firing rate. Nature (2007) vol. 448 pp. 802-806

Abstract Author(s): Cyrus Omar, Jason Middleton, Dan Simons, Brent Doiron