Center surround has never robustly been shown in any modelling approach

[Helveg]

@claudiacasellato I went over ALL different ways the CS has ever been shown, and I prove mathematically that they ALL introduce FAKE center surround effects. Here is a plot of constant tonic inhibition, and a constant difference between gabazine and control condition. And even though they are under the effect of perfect constant inhibition all used methods introduce a center surround shape:

import numpy as np
import plotly.graph_objs as go
from plotly.subplots import make_subplots

fig = make_subplots(rows=3, cols=1, shared_xaxes=True)
fig.update_layout(title_text="Complete absence of center surround in E/I")
one_side = np.concatenate((np.zeros(10000), np.linspace(0, 4, 1000)))
gabazine = np.concatenate((one_side, np.flip(one_side)))
fig.add_scatter(y=gabazine, name="gabazine")
one_side_inh = np.concatenate((np.zeros(9000), np.ones(2000) * 1))
inh = np.concatenate((one_side_inh, np.flip(one_side_inh)))
control = np.maximum(gabazine - inh, 0)
fig.add_scatter(y=inh, row=2, col=1, name="true_inhibition")
fig.add_scatter(y=control, name="control")
fig.update_yaxes(range=[-4, 4])

# Calculation methods

E = control
nE = E / abs(E).max()
I = gabazine - control
nI = I / abs(I).max()
pE = nE
pI = (ng := gabazine / abs(gabazine).max()) - (nc := control / abs(control).max())

fig.add_scatter(y=ng, row=2, col=1, name="gaba norm")
fig.add_scatter(y=nc, row=2, col=1, name="c norm")

# B = (E - I) / E
fig.add_scatter(y=(E - I) / E, row=3, col=1, name="classical balance")
# B = (E - I) / (E + 1)
fig.add_scatter(y=(E - I) / (E + 1), row=3, col=1, name="divzero-corr balance")
# B = (E - I)
fig.add_scatter(y=(E - I), row=3, col=1, name="EI diff")
# B = (nE - nI)
fig.add_scatter(y=(nE - nI), row=3, col=1, name="afternorm. EI diff")
fig.add_scatter(y=(pE - pI), row=3, col=1, name="prenorm. EI diff")

fig.update_yaxes(range=[-10, 10], row=3, col=1)
fig.show()

[Helveg]

This is the raster plot of granules cells sorted by distance from the (150, 100) point; Their first spike after stimulus colored red, second spike enlarged and colored purple, third spike lime and enlarged further.

There is no significant increase of 2nd spikes in the surround in gabazine conditions. The reverse is actually true: the amount of secondary spike increases in the CENTER, not in the surround

[Helveg]

Not only is every center surround ever calculated incorrectly, I found evidence of fraud in Mapelli 2007, the original paper:

This figure supposedly portraits the famous E/I balance: B = (E - I) / E, it says so in the paper, but it is a handdrawn fraudulent plot:

1. Where E and I intersect (E - I) / E = 0 this is not the case in the plot
2. Near the edges where the limit of E goes to 0, you can only converge to 0 (as is drawn) if the limit of (E - I) monotonically converges to 0 as well, which is not the case on either side of the plots and is highly unlikely (impossible) in any kind of experimental or simulated data.
3. On the left I > E (black pixels on the bottom even at the moment the black line merges into the axis) yet there is no asymptotic behavior
4. On the right side th difference between E and I is not monotonic and crosses over a 2nd time, yet there is no sign of this additional bump

Here is pixel proof of (1) and what I think the original plot would have looked like for (2), (3) and (4)

The original proof of the center surround balance calculation is a fraudulent hand drawn plot

I also think that the 3d maps are doctored to look better and a lot of "favorable" smoothing and amplifying of the data will have been going on.

[Helveg]

This is the real plot:

If we assume a friendly "errata", and don't divide by E we get this plot:

That looks like there was a whole lot less fraud comitted! BUT! They decided to hand draw the plot anyway hahahahaha, enlarging the inhibition, and smoothing out the edges:

(I overlap the images in GIMP and you can see that the input lines perfectly overlap, but the result is hand drawn)