To assay for spiking activity changes from baseline in either direction (increases or decreases), we took the absolute value of the measured z-scores for each multi-unit (Figure S1A). Significance was evaluated using paired-observation tests relative to baseline. We also measured delay activity separately for subpopulations of multi-units with responses above or below baseline (Figure S1B). Since PFC units often have complex dynamics, this segregation into subpopulations was performed independently for each unit and time point. To avoid circularity in the analysis, segregation into subpopulations and computation of mean activity within each subpopulation were performed on odd and even-numbered trials, respectively.
