Measurement randomization methods. Left: Three sample binarized time series with blocks of cell activity indicated by shaded squares. Right: Spaghetti plots of collective events simulated in NetLogo. (A) The original dataset. (B) Positions of entire trajectories are shuffled between cells while preserving the original dynamics. (C) Time series are independently and randomly shifted back and forth in time while preserving positions of entire tracks. The original dynamics are preserved, although the onset changes. (D) Blocks of activity are shuffled in time independently for every trajectory while preserving positions of entire tracks. (E) Individual time points are shuffled independently for every trajectory while preserving positions of entire tracks. (F) The activity is shuffled between cells independently in every time frame. Positions of entire tracks remain intact.