Table 5. Summary of results for proxies and suggested adaption of results to other situations.

REFERENCE TABLE FOR LFP PROXY USE
For the following reference conditions:
• Layer 2/3 pyramidal cell (two neighboring dendritic bushes, (Fig 1B, left)
• Synaptic distribution with GABA synapses in lower (basal) bush and AMPA synapses in both (basal+apical) bushes (Fig 9A, central panel)
• Current-based synapses
• 1.5 spikes/ms input
• Electrode recording from the center of the 3D network at depth = 100 μm relative to inversion point
The optimal proxy (RWS) is
$LF{P}_{RWS}\left(r,d,t\right)=f_{RWS}\left(r,d\right)*Norm\left[{\displaystyle \sum _{pyr}}AMPA\left(t-6ms\right)-1.65\left({\displaystyle \sum _{pyr}}GABA\left(t\right)\right)\right]$
The proxy is, however, robust or easily adaptable to a variety of conditions as described below:
Aim	Action required
Use a simpler proxy	Model temporal part of LFP as sum of AMPA and GABA currents or simply as GABA currents (Fig 4F and 4G)
Consider a different recording depth	Use the RWS (Fig 5C) unless you are within 50 μm from the inversion point, in which the LFP largely cancels out (Fig 2A and 2B)
Simulate LFP recorded with electrode radially displaced from the center of the 3D network	Use the RWS and determine amplitude fRWS(r,d) by means of Fig 2B
Vary rate of external synaptic input (input intensity)	Use the RWS (as long as the synaptic input is sufficiently strong to generate a sizable LFP, cf. Fig 6C)
Include/remove LFP contribution from transmembrane currents of stellate interneurons	Use the same RWS, since interneurons have a negligible effect on the LFP (Fig 3)
Simulate neurons with morphologies different from pyramidal cell of layer 2/3	Use the RWS for all morphologies in which two dendritic bushes are vertically more distant than 150 μm (i.e., for all cells except stellate cells that do not contribute to LFP), see Fig 8E.
Simulate neurons with synaptic distributions different from our reference case	As shown in Fig 9F and 9G:
	• Both synapses in both bushes: no appreciable LFP and no good proxy available
	• AMPA synapses only in upper bush: RWS performs well (R2 = 0.81), but to get better results increase relative weight of AMPA currents as follows:
	LFP RWS(r, d, t) = f RWS(r, d)
	$*Norm\left[{\displaystyle \sum _{pyr}}AMPA\left(t-6ms\right)-0.69\left({\displaystyle \sum _{pyr}}GABA\left(t\right)\right)\right]$
	• AMPA synapses only in lower bush: RWS performs well (R2 = 0.77)
Simulate neurons with conductance synapses	Change AMPA and GABA coefficients as a function of input rate as indicated in Fig 10C