Variation within movie repeats supplied 83% of the fitted common input slope in one mouse V1 calcium panel

We tested which part of response variation carries a fitted anatomical common input relation in one mouse V1 panel. We used 193 cells, 6,416 selected pairs and eight repeats of a two-minute movie. An exact covariance split kept all response parts on the same scale. All parts used the same cell, geometry, direct connection and mean response controls. The total event coefficient was 0.006681 per doubling of one plus the common source root count. Repeat offsets supplied 0.001152, or 17.2%. Variation within repeats supplied 0.005529, or 82.8%. Its central 95% range from 1,999 joint repeat and movie-block draws was -0.000003 to 0.011566. The paired range for the within-minus-offset coefficient also included zero. Fluorescence change gave an observed within repeat share of 80.0%. The percentages describe fitted coefficients in this panel. They are not causal shares or population estimates. Repeat offsets carried 2.73% of mean event covariance and a median 1.05% of cell residual variance. One mouse, eight repeats, source segments and unmeasured state limit inference. The package preserves source graphs, trial arrays, provenance, every response part, all draws, figures, code and offline repeat instructions.

connectome · visual cortex · shared input · noise correlation · covariance · calcium imaging · temporal scale · reproducibility