(A) Avalanche size probability densities for LATE (orange) significantly deviate from the power law observed during BASE (black/green) and EARLY (purple) for left (left panel), but not right trials (right panel). Area: 95% confidence interval. (B) Size distributions normalized by a power law with exponent −1.5 reveal good agreement with theory (horizontal curve with a sharp cut-off), except during LATE for left trials, where a large deviation can be observed. Inset: Area under the curve of the normalized distributions for avalanches of size > 20 quantify differences between epochs. Error bars: SD from 1000 bootstraps. (C) Same as in A, but for adaptive binning, which collapses size distributions for BASE, EARLY and LATE. (D) Adaptive binning leads to a decrease in the area difference between LATE and the other epochs. (E) Distributions of the difference between the mean avalanche size of LATE and BASE for each trial (positive values indicate avalanches in LATE were larger than those in BASE for a given trial) for fixed (grey) and adaptive binning (black/green). Significant differences between distributions are marked by an asterisk (Kolmogorov-Smirnov test; ). For left trials, LATE is significantly biased toward smaller avalanches for adaptive binning compared to fixed binning.