The predominance of decreases versus increases in firing rate using this method was then also tested using a Binomial test. For burst analysis, a burst was defined by the following criteria: maximum interval to start

burst (i.e., the first interspike interval within a burst) must be = 4 ms; maximum interval to end burst, 10 ms; minimum interval between bursts, 100 ms; minimum duration of burst, 4 ms; and minimum number of spikes within a burst, two. Phase-locking of MD units to mPFC LFPs was accomplished by fist filtering the field potentials in either the theta (4–12 Hz), beta (13–30 Hz), or gamma (40–60 Hz) range using a zero phase delay filter and computing phase using a Hilbert transform. Each unit spike was assigned a phase based on its simultaneous field potential sample. The magnitude of phase-locking was quantified mean resultant length (MRL)

of the sum of the unit vectors representing MK8776 the phases at which each spike occurred, divided by the number of spikes. The MRL is sensitive to the number of spikes used in the analysis. Therefore, to compare phase-locking strength by condition we computed the MRL for multiple (1,000) subsamples of 50 spikes per condition and averaged across subsamples for each condition, and for each unit. Units that fired fewer than 50 spikes in each condition were not analyzed. The statistical significance of phase-locking was assessed using the Rayleigh test for circular uniformity. Data were pooled for task independent and task dependent behavior whatever sessions. Cells with fewer than AP24534 supplier 600 spikes during the entire recording session were not analyzed. To determine the temporal relationship between unit activity and beta oscillations in the mFPC,

phase-locking was calculated for 50 different temporal offsets from the mPFC LFP for each unit recording. Only units with significant Bonferroni-corrected phase-locking in at least one of the 50 shifts are shown in Figure 5C. For coherence and behavior across learning, recordings were binned into early (first 5 trials), middle (trials 25–30), and late (last 5 trials on the day the animal achieved criterion performance) trials. Coherence of the field potentials was computed using the multitaper method (MATLAB routines provided by K. Harris). Field potential samples for the trials in each bin were concatenated and then divided into 1,000 ms segments (800 ms overlap). The Fourier transform of each segment was computed after being multiplied by two orthogonal data tapers. Coherence was computed by averaging the cross-spectral densities of two field potential signals across data windows and tapers and normalizing to the power spectral densities of each signal. Beta coherence was computed as the mean coherence in the 13–30 Hz range.