Extended Data Fig. 9: Laminar distributions of potentiated and depressed pairs.

(a) Noise correlation difference (laser minus control) for pairs of units with broad waveforms with potentiation index (‘PI’) > 1 (potentiated, light green) or < 1 (depressed, dark green). Asterisks indicate trial blocks in which the change in correlations was significantly different from zero (Wilcoxon signed rank test, two-sided, false discovery rate corrected, P ≤ 0.0059). Vertical dotted line shows trial block 4, where the laser induced increase in correlated variability begins to decline between broad pairs. (b) Same as panel A, but for pairs of narrow and broad wave- form pairs. Asterisks indicate significantly different from zero trial blocks (Wilcoxon signed rank test, two-sided, false discovery rate corrected, P ≤ 0.0068). Vertical dotted line shows trial block 4, where correlated variability is first significantly increased between broad-narrow cell pairs with a PI > 1. The common inflection point between panels A and B (dotted vertical line at trial block 4) suggests that broad cells become more coupled with narrow cells and simultaneously less coupled with other broad cells, consistent with the idea that I–E connections are potentiated leading to the overall decrease in correlated variability across trials. (c) Pairs were assigned to layers (supragranular ‘SG’, granular ‘G’, and infragranular ‘IG’) using the current source density method (21, 23). Distributions show potentiation indices across broad wave- form pairs in each layer combination (dark green indicates PI < 1, light green indicates PI > 1). Distributions are only shown for layer combinations in which the median PI was significantly different from unity (two-sided Wilcoxon signed rank test, P values as shown). (d) Same as panel c, but for narrow–broad waveform pairs. (e) Summary of results in panels c–d. Broad–narrow cell pairs were potentiated across SG–G and SG–IG, while broad pairs were potentiated within SG but depressed between SG–G and within G.