Extended Data Fig. 6: Dissociated visual, auditory and motor-related activity using a regression model: overview of findings.
From: Triple dissociation of visual, auditory and motor processing in mouse primary visual cortex
(a) Explained variance of V1 firing for an increasing number of video PCs. The video PCs beyond the number of selected video PCs for analysis (n = 34) contributed little to V1 variance. Mean ± SEM. (b) Distribution across neurons of explained variance of V1 firing rate by each subset of predictors using the regression model. X and Y axis positions indicate the neuron’s rank order across all V1 neurons in the amount of variance explained by visual, auditory or motor variables. Each dot is a neuron. Left, absence of a correlation between visual and motor predictors (r = 0.00, p = 0.94, Spearman rank correlation) indicates that the strength by which vision and behavior modulate V1 firing rates are close to unrelated. Middle, auditory and visual EV are unrelated across neurons (r = −0.07, p = 0.06), while auditory and motor coding were moderately correlated (r = 0.30, p = 2.52*10−17). Neurons with exclusive visual, auditory or motor coding are colored (motor - green; visual - blue; auditory - red). (c) Same as Fig. 3f, but for each of the task cohorts separately and auditory trials only. Auditory-related activity was present in all three cohorts. Sound-evoked motor-related activity was larger in the MST cohort, quantified in (Fig. 3g). (d) The absolute sound volume used in a given session (x-axis) was not related to the amount of variance in V1 firing rates explained by auditory stimulus variables in the regression model (Spearman rank correlation between sound volume and auditory-related EV, r = −0.045, p = 0.178). Each dot is a neuron. (e) Same as d, but for motor-related variance. No significant correlation was found (Spearman rank correlation between sound volume and motor-related EV, r = 0.048, p = 0.152). The results shown in panels d and e jointly suggest that the volumes we employed are too low to induce a startle response, that would have been otherwise accompanied by an increase in either auditory- or motor-related EV at volumes higher than a certain threshold. (f) Same as Fig. 3g in the main text, but now for subsamples of 108 neurons from each cohort to account for the unequal number of V1 neurons sampled across the cohorts (NE: n = 108, UST: n = 112, MST: n = 570). Note how results are consistent with those presented in the main figure. Center line, box edges, and whiskers demarcate 5th, 25th, 50th, 75th and 95th percentile of the mean auditory and motor-related sound-evoked response based on 1000 subsamples. **p < 0.01. Neurons from all cohorts were combined for panels a, b, d, e.
