Conversely, a value of zero indicates that responses to the distracter were always greater than responses to the target. The auROC values for individual neurons were calculated for a 10 ms window from 0 ms to 600 ms from color-change onset slid in 1 ms increments along the spike train. When calculating auROC values for different distances, we corrected

for different number of trials in the conditions through a randomization procedure (see above). For each unit, the auROC values were then plotted as a function of time to describe the time course of neuronal choice probability. The latency with which neurons could distinguish the target from the distracter was defined as the time from color-change onset when the auROC time series reached the criterion value of 0.64. This value is lower than the one used in FEF studies of target selection (0.75) (Thompson et al., 1996) but is substantially higher than 0.5, which is the chance level. The latter learn more has been used in studies of dlPFC neurons’ ability to encode rules (Bongard and Nieder, 2010). However, when increasing our threshold ZD1839 cell line to 0.75 or lowering

it to 0.5, the number of neurons reaching the threshold for each distance decreased or increased, respectively, but the relative proportion across distances remained similar. The accuracy of the neuronal decision was the maximum amplitude of the time series. To obtain the estimates of latency and amplitude for the sample of neurons, we excluded cells that failed to yield latency values for at least one of the three distances. To quantify changes in response to targets and distracters during time periods following stimulus onset and around the color-change onset, we computed an MI [MI = below (Rbin − Rbl)/(Rbin + Rbl)] during both task and fixation trials. Rbin are responses during the period from stimulus onset to 400 ms after or from 200 ms prior to color-change onset to 400 ms after, computed in bins of 10 ms and increments of 1 ms. Rbl was defined as the mean

activity within a 300 ms time period immediately preceding the stimulus onset (baseline period). This allowed us to track the MIs across time and compare modulation in the main task with fixation. MI values of zero indicate similar responses; indices between 0 and 1 indicate an enhancement relative to the prestimulus-onset baseline period, whereas values between 0 and −1 point to a relative decrease. We computed Student’s t tests in bins of 50 ms and increments of 1 ms while correcting for different number of trials, and tested for significant differences between responses and baseline (evaluated at Bonferroni-corrected p < 0.05/number of comparisons across time; Figure 5, lower panels). This work was supported by grants to J.M.-T. from the CIHR, NSERC, EJLB Foundation, and Canada Research Chair program. The CIHR Canada Graduate Vanier Scholarship supported T.L. We acknowledge Mr. Walter Kucharski and Mr.