Noise had additive and multiplicative effects on f–I relationships in pyramidal neurons. A, f–I relationships of a layer 2/3 pyramidal cell with noise [f4(I), at σ = 4] and without noise [f0(I), at σ = 0], showing the fitting procedure used to measure the A and the ΔG. The solid line shows the fit of f4(I) with a copy of f0(I) that was multiplied by 1 + ΔG and shifted upward by A. B, Bar graph showing the mean additive effect of noise in layer 2/3 neurons (n = 14) and layer 5 neurons (n = 29). The additive effect did not differ significantly between layer 2/3 and layer 5. C, Bar graph showing the mean gain increase by noise in layer 2/3 and layer 5. The average gain increase was greater in layer 2/3 than in layer 5. D, Plots showing the mean additive effect versus noise amplitude in layer 2/3 neurons (n = 5) and layer 5 neurons (n = 7). The additive effect increased with noise amplitude in both layers. E, Mean gain change versus noise amplitude. The gain change increased with noise amplitude in both layers. F, Mean additive effect of noise versus stimulus duration. The additive effect decreased with longer stimulus duration. G, Mean gain change versus stimulus duration. The gain increase grew with longer stimuli. H, Bar graph showing the maximal steady-state firing rate attained before depolarization block occurred (taken from 2–3 s after stimulus onset) with and without noise (σ = 4) in layer 2/3 and layer 5. Noise raised the maximal steady-state firing rate in both layers. I, Scatter plot showing the relationship between percent gain change caused by noise (σ = 4) and percent change in the maximal (Chg in Max) firing rate. The solid line is the regression line for all data points. A significant correlation was observed, suggesting that a common mechanism might underlie both effects of noise. Asterisks indicate significant differences (p < 0.05) here and in subsequent figures. L2/3, Layer 2/3; L5, layer 5; Stim, stimulus.
