Supplementary Figure 1 Lick response during the delayed Go versus No-Go task. Trial-averaged lick rate was averaged across all mice used for pyramidal cell imaging (n = 9). Different colors denote different trial types (blue, Hit; light blue, Miss; magenta, Correct Rejection (CR); orange, False Alarm (FA)). Colored shadings, ± s.e.m.
Supplementary Figure 2 Pupil sizes in Go and No-Go trials. The averaged pupil sizes during the delay period were not significantly different between Go and No-Go trials (n = 11 sessions from 5 mice, P = 0.76; Wilcoxon signed-rank test). Gray lines, individual sessions; black, mean ± s.e.m.
Supplementary Figure 3 Fractions of significant Go- and NG-preferring cells vs. distance from pia at two different DV levels. (a) Schematic showing estimated bottom positions of near-vertically implanted lenses at two different DV levels (red lines; 650 and 2150 m from the dorsal surface). (b) At both DV levels, deep layers had higher fractions of significant NG-preferring cells (P < 0.005, one-way ANOVA). All error bars, ± s.e.m.
Supplementary Figure 4 Decoding of neural activity in pyramidal cells. A decoding analysis was performed using pyramidal cell activity measured in the imaging experiments and a correlation coefficientbased classifier. Go probability was computed as the proportion of each type of trials classified as Go. Colored shadings, ± s.e.m.
Supplementary Figure 5 Histology. (a) Optogenetic stimulation site in an example SST-ChR2 mouse. Yellow arrowhead, position of optic fiber tip. (b,c,d) similar to (a) but for PV-, VIP-, CaMKII -ChR2 mice. (e) Recording site in an example SST-ChR2 mouse. White, Dil labelling. Yellow arrowhead, position of electrode tip. (f) same as (e) but for a VIP-ChR2 mouse. Scale bars, 500 m.
Supplementary Figure 6 Spike sorting in an example recording session. (a) Spike waveform projections in feature space. Each axis denotes a principal component (PC) derived from the spike waveforms. Note that clustering was performed using additional PCs to those shown here; Unit 2 (red) was well separated from the unsorted spikes (gray) in other projections. (b) Left, spike waveform of each unit (gray, individual spikes; colored, average). Right, auto-correlograms.
Supplementary Figure 7 Effect of SST neuron activation on dmpfc activity. (a) Activity of a significant Go-preferring (upper) and a significant NG-preferring cell (lower) with or without SST stimulation. (b) Laser stimulation of SST neurons significantly reduced the Euclidean distance between Go and NoGo activity. The black horizontal bars on the top indicate the analysis periods including baseline (1 s before the beginning of sample), sample, delay, and post-delay (0.5 s after the end of delay) periods. **P < 0.0001 for sample, delay, and post-delay periods; P = 0.49 for the baseline period; bootstrap. Shadings and error bars, 95% confidence intervals (bootstrap).
Supplementary Figure 8 Behavioral effects of optogenetic activation of SST and PV neurons in the vibrissa primary somatosensory cortex (vs1). (a,b) Whole-trial stimulation of SST (a) or PV (b) neurons in the vs1 did not affect the behavioral performance (SST: P = 0.91, 1.0, and 0.91 for correct response, Hit, and False-Alarm (FA) rates, respectively; PV: P = 0.81, 0.38, and 0.81 for correct response, Hit, and FA rates, respectively; Wilcoxon signed-rank test). (c,d) SST (c) or PV (d) neuron activation in the vs1 during the early delay period did not change the behavioral performance (SST: P = 0.81, 0.81, and 1.0 for correct response, Hit, and FA rates, respectively; PV: P = 0.44, 1.0, and 0.31 for correct response, Hit, and FA rates, respectively). Error bars, ± s.e.m.
Supplementary Figure 9 Behavioral effect of optogenetic activation of VIP neurons in the vibrissa primary somatosensory cortex (vs1). (a) Whole-trial activation of VIP neurons in the vs1 did not affect the behavioral performance (P = 0.50, 0.25, and 0.57 for correct response, Hit, and FA rates, respectively; Wilcoxon signed-rank test). (b) Activation of VIP neurons in the vs1 during the early delay period did not change the performance (P = 0.64, 0.55, and 1.0 for correct response, Hit, and FA rates, respectively). Error bars, ± s.e.m.
Supplementary Figure 10 Behavioral effects of optogenetic silencing of VIP, SST or PV neurons in the vs1. (a) Arch-mediated silencing of VIP neurons in the vs1 did not change the performance (P = 0.58, 0.58, and 0.69 for correct response, Hit, and FA rates, respectively). (b) Arch-mediated silencing of SST neurons in the vs1 did not change the behavioral performance (P = 0.31, 0.31, and 0.31 for correct response, Hit, and FA rates, respectively). (c) Arch- or halorhodopsin (Halo)-mediated silencing of PV neurons in the vs1 did not change the performance (P = 0.38, 1.0, and 0.38 for correct response, Hit, and FA rates, respectively). Error bars, ± s.e.m.
Supplementary Figure 11 Behavioral effect of optogenetic activating CaMKII -expressing pyramidal neurons in the dmpfc. ChR2-mediated activation of pyramidal cells impaired the behavioral performance (P = *0.039, **0.0078, and 0.64 for correct response, Hit, and FA rates, respectively; Wilcoxon signed-rank test). Error bars, ± s.e.m.
Supplementary Figure 12 Distribution of firing rate changes induced by VIP neuron activation. For each neuron group (Go-preferring, NG-preferring, and unmodulated cells), the firing rate difference between laser-on and -off trials divided by their sum was calculated in Go and No-Go trials. Arrowhead, mean of the population. *P < 0.05; ** P < 0.001; *** P < 0.0001, paired t-test.