A little air bubble was left between the tube and the syringe. The bubble was carefully monitored throughout the experiment to PLX-4720 in vitro control for changes in pressure inside the tube or leaks. The cannula was lowered into the cortex with a microdrive that had an electrode attached at 1 mm distance. The neuronal activity recorded from this electrode

was an indication that the cannula was in close proximity to active neurons. At the time of the injection, the pump was manually turned on. Injections consisted of 3 μl of saline or 30 μg of SCH23390 at an infusion rate of 0.3 μl/min. The concentration and volume were chosen according to previous studies (Sawaguchi and Goldman-Rakic, 1991 and Sawaguchi and Goldman-Rakic, 1994; Nakamura and Hikosaka, 2006). During the injection, we carefully monitored any changes in firing rate of the neurons. Typically, one or two neurons (of 15–30) showed changes in spiking that elapsed once the injection had finished. In a few sessions, we could not keep track of one or two neurons after the injection, and these neurons were not included in the analysis. In 20 sessions, large amplitude deflections were observed in the LFP signals shortly after the injection of SCH23390,

an indication that the drug had been successfully delivered into the brain. Altogether, the consistent movement of the air bubble inside the tube and the changes of activity and/or LFP signals during or shortly after JAK/stat pathway the infusions were good references for successful injections. We did not perform any histology to measure the exact spread of SCH23390

inside the brain. The analysis of the data suggests that it had spread at least 2 mm (∼4 mm3), because large negative deflections were observed in electrodes located 2 mm away from the injection site. A previous study in rats (Granon et al., 2000) reported that 0.5 μl of a radioligand of SCH23390 infused in the rat PFC, at a similar infusion rate as ours, spreads up to 6 mm3 but with substantial dilution. This suggests that in case SCH23390 had spread outside the lateral PFC, its effective concentration would have been compromised. We examined whether the injections altered the quality of waveform sorting. Offline Sorter (Plexon Technologies) was used to separate spikes from noise and to sort spikes from different neurons recorded from the same electrode. Histamine H2 receptor The waveforms of each neuron were manually classified in different clusters using principal component analysis (PCA). Sessions were then divided in segments of 15 min (roughly the duration of a block), and sorting quality statistics were performed segment by segment. The degree to which the unit clusters were separated in 2D and 3D space was determined by a multivariate analysis of variance test (MANOVA) in Offline Sorter. Small p values (<0.05) in this test indicate that each of the unit clusters has a statistically different location in 2D/3D space.