Fig. 7: Onset of report-related activity in task A and drop in noise correlations predict effects of late silencing. | Nature Communications

Fig. 7: Onset of report-related activity in task A and drop in noise correlations predict effects of late silencing.

From: Multisensory task demands temporally extend the causal requirement for visual cortex in perception

Fig. 7

a Average spiking rate for all orientation-selective neurons for preferred and non-preferred orientations is split by hits and misses (UST and MST neurons combined; task A). b Orientation decoding performance over time. Right panel: decoding performance increased post-stim (0 to +500 ms) versus pre-stim (−500 to 0 ms; n = 11 sessions, cohorts combined, ANOVA, F(1,17) = 44.76, p = 4.1 × 10−6) and increased in individual sessions from all cohorts (colored dots). c Change in noise correlation (NC) relative to baseline (200 to 1000 ms compared to baseline −500 to 0 ms) for visual trials split by choice and cohort (for auditory trials, see Supplementary Fig. 9a). Boxplots show the median and interquartile range (box limits) and 0.5 × interquartile range (whiskers). Noise correlations decreased only during hits in UST and MST mice (ANOVA, UST, n = 1930 pairs, F(1,3856) = 82.44, p < 1 × 10−19; MST n = 13972, F(1,28188) = 142.96, p < 1 × 10−33). Misses in NE mice were associated with a slight increase in noise correlations (n = 2904 pairs, F(1,5805) = 14.67, p < 0.001). d Reaction time distributions for visual hits in UST and MST cohorts and tertile ranges. e1 Noise correlations over time with respect to baseline, either aligned to stimulus change (left) or first lick (right). Horizontal dashed lines indicate for each tertile the threshold for each tertile for the onset of the drop in NCs (below 2 standard deviations of the baseline; −500 to 0 ms) and this onset is highlighted with colored arrows. Note how noise correlations (aligned to stimulus change) drop first in fast trials, and progressively later in medium and slow trials. Right panels show that, when aligning to lick onset, the drop in noise correlations precedes reaction times by a similar lag, independent of reaction time tertile. e2 Same as e1, but for MST mice. f Reaction time and moment of decorrelation were significantly correlated (Pearson correlation, n = 6, r = 0.960, p = 0.002). Scatterplot shows median reaction time and earliest time point of decorrelation for each tertile in the two visually trained cohorts. g Average Z-scored firing rates just before photostimulation (100–200 ms) were higher if the trial resulted in a visual hit rather than a miss in UST mice (Thr: F(1,156) = 10.16, p = 0.002; Max: F(1,152) = 5.66, p = 0.019; ANOVA). h Same as g, but for MST mice. Firing rates just before photostimulation were higher for hits than misses only for threshold visual changes (F(1,268) = 13.19, p = 0.001), but not maximal changes (F(1,254) = 0.59, p = 0.44). i Noise correlations for visual hit and miss trials before photostimulation onset (grouped across UST and MST cohorts and saliency levels). Black bar on top indicates time bins with significantly different NCs between hits and misses (p < 0.05). Throughout the figure, lines and shading are mean ± SEM.

Back to article page