Figure 5.

Validation of binary model analysis with the LIF spiking model. (A) The input currents are dissected into background I_c(t) (blue) and inhibitory synaptic I_g(t) (orange) components: I_total(t) = I_c(t) + I_g(t) (black). A single realization for a pair of E-cells is shown with V = 1. (B) The covariance of the input currents, averaged across pairs, has the following features: the input covariance from the background is constant (blue), the inhibitory synaptic input’s covariance increases with firing rate (orange), but together the total input covariance decreases with firing rate (black) – cf. Figure 4Cii. (C) The covariance susceptibility function S for the LIF spiking model increases with firing rate. (D) The input covariance for a low (dashed, V = 0.1) and high velocity (solid, V = 1) whisker input in time decreases by different amounts upon whisker stimulation. (E) The susceptibility function in time is determined by the instantaneous firing rate [interpolating the curve in (C)]; low velocities have smaller S and higher velocities have larger S. (F) The approximation of the spike count covariance in green, obtained by multiplying corresponding black and red curves, is small like the actual Cov_EE (cf. Figure 4Ciii). The dashed-lines correspond to low velocity, solid lines correspond to high velocity.