All hypothesis tests used non-parametric randomization methods that do not rely on specific assumptions about data distributions (Manly, 2007). Each session was treated as an observation, and randomizations were performed across sessions. All randomization statistics were resampled 10,000 times and evaluated in a two-tailed fashion. To test whether a mean value differed significantly from baseline, we used a randomized sign test in which a t-statistic was computed on both the observed data, and on data where the sign of each baseline-centered observation was randomly flipped.
