Fig. 4.

Sustained activation of VTA single units before stimulus onset. A: the largest proportion of units (n = 29 of 123 total) was activated during correct trials, whereas the proportion of significantly activated units was reduced during incorrect and omission trials (**P < 0.01). Significance was determined using Z score (see materials and methods for details). B: mean normalized firing rate across units that were significantly activated before the stimulus is plotted (250-ms bins). The shading indicates SE. Stimulus onset is at t = 0 s. The magnitude of change was largest during correct trials and reduced during the other trial types (P < 0.0001). During omission trials, the firing rate did not change. The inset shows that nonnormalized firing rate was highly variable across single units. The x-axis shows the mean rate from the prestimulus period (−6 to 0 s), and the y-axis shows the number of units. C: mean normalized firing rate of the remaining nonresponsive units (n = 94) is plotted with shading indicating SE from the mean. These units did not respond during the prestimulus period. However, the population responded after stimulus onset. D–F: units were split into groups with different firing rates to reduce variability in the population mean perievent time histograms. Top: mean and SE of firing rate (Hz) in 250-ms bins. Bottom: individual units on the y-axis and time around stimulus onset on the x-axis. Units were activated for a sustained period of multiple seconds before stimulus onset. Note that 1 unit with a mean firing rate of ∼20 Hz was excluded from the plots to reduce variability but is included in the statistical tests (ANOVA) reported in the results. VTA unit activation was observed during both correct and incorrect trials, but a change in firing rate did not occur during omission trials.