Extended Data Fig. 2: Dynamics of putative fast-spiking interneurons in the basic task and in the presence of distractors.

Activity of putative fast-spiking interneurons (Methods) recorded in vS1 (n = 92 cells) and ALM (left ALM n = 184; right ALM n = 96 cells). a, Example cells and grand population average (labels as in Fig. 1d-g). b-c, Hierarchical clustering of trial-averaged PSTHs of putative interneurons (labels as in Extended Data. Figure 1a-b, bars indicate mean ± s.e.m across cells). Statistical significance was assessed by Wilcoxon rank sum test (two-sided). Putative interneurons in vS1 and ALM show a diversity of response profiles, suggesting that their activity does not simply reflect average activity of excitatory neurons. d, Session-averaged projections of neural activity of putative fast-spiking interneurons in left ALM on Stimulus, Choice, and Ramping modes (n = 104 cells, distractor trained mice). Red, lick-left trials; blue: lick-right trials. e-f, Trials with distractors projected on Stimulus mode (e, left) and Choice mode (f). Color-code as in Fig. 3 f,h. There was no suppression of distractors in the Stimulus mode during delay (e, right), whereas Choice mode exhibited robustness to distractors at late delay. Error bars, mean ± s.e.m., across sessions. Response size was assessed by repeated measures ANOVA, and also verified by fitting linear regression model. Statistical significance between response size to early- and late-delay distractors was done using two-sided Student t-test with Tukey-Kramer correction for multiple comparison. g, Endpoints of ramping mode of simultaneously recorded putative pyramidal cells versus putative fast-spiking interneurons. Each dot represents the Ramping mode endpoint on a single trial computed for the last 0.2 s of the delay epoch. Endpoints of ramping mode of simultaneously recorded putative fast-spiking (F.S.) interneurons and putative pyramidal neurons were correlated on single trials (r represents the Pearson correlation coefficient).