The effects were largest for the L5 pyramidal cell population whereas the check details LFP amplitude from L4 stellate cell population was largely unaffected ( Figures 4E1 and 4F1). It also depended on the spatial distribution of synapses: there were pronounced effects for either apical or basal input, but only a modest effect

for homogeneous synaptic distributions ( Figures 4E1 and 4F1). To explore these differences further we computed the mean pairwise correlation cϕcϕ (see Experimental Procedures and Supplemental Equation 18) between single-cell LFP contributions as a function of input correlation cξcξ for the different cell types and input scenarios (Figure 4G). This provided

an explanation for why the effect of correlations was found to be so different for the different cell types and synaptic distributions: for example, LFP contributions are more correlated for L5 pyramidal Bcl-2 cleavage cells than the other cell types, and apical input gives higher correlations than basal or homogenous input. Thus, the extent to which input correlations have an effect on the reach of LFP depends on how reliably input correlations cξcξ are translated to correlations between LFP contributions cϕcϕ. Replotting the LFP reach and amplitude as function of the LFP correlations further supported this second interpretation as all simulation results then collapsed onto the same curve (Figures 4E2 and 4F2). This clearly demonstrates the importance of the level of correlation between individual LFP contributions in determining both the reach and amplitude of the LFP. The results depicted in Figure 4 demonstrate the key role played by synaptic correlations in determining the LFP amplitude. From the analytical formulas in (1) and (2), we further see that the contribution from correlated neuronal sources scales differently with the density of sources (g1(R)∼ρ2)(g1(R)∼ρ2) than for uncorrelated sources (g0(R)∼ρ)(g0(R)∼ρ). Thus the correlated contributions to the LFP will

generally dominate the uncorrelated contributions when the correlation coefficient cϕcϕ and/or the source density ρ are large. This is illustrated for particular examples in Figure S1, available online. Until now, we have implicitly assumed that the synaptic input to different neurons are equally correlated throughout the whole population. How will the results change if the level of correlation between LFP contributions is dependent on the radial distance to the electrode? We studied a simple case where LFP contributions were assumed to be homogeneously correlated only within a certain radius Rc