Residual correlations gave no clear Ccn1 group difference in eight baseline mouse V1 recordings

We tested whether shared visual activity after astrocyte Ccn1 deletion reflected stimulus responses or residual trial fluctuations. We used eight baseline mouse V1 recordings, with four mice per genotype. The panel had 2,095 regions labeled as neurons by the source and 289,852 distinct within-mouse pairs. Each mouse had equal weight. Mean residual correlation was 0.1463 in WT and 0.1294 in cKO mice. The difference was -0.0169, with a central 95% mouse resampling range from -0.0742 to +0.0386. The exact two-sided label-reference probability was 0.7143. Checks of cell score, pair medians, time bins, trends, and repeat splits also gave ranges across zero. Removing one mouse could reverse the contrast. The source used a narrower cell panel, so this is a new analysis of the released population. A separate code audit found an allocated final trial that stayed zero in a released loop. Setting the final real trial to zero raised mean total correlation by 0.0080 to 0.0209 in all eight mice. This check does not establish the processing history of the source paper's final figures. Four mice per group, uneven sex counts, and calcium processing limit biological inference. These results do not show equal function or measure anatomical connections. The package preserves compact responses, row identities, source hashes, all results, code, figures, and repeat instructions. Direct checks matched 1,620 source response windows. A fresh numerical repeat matched all result files.

astrocytes · CCN1 · mouse visual cortex · calcium imaging · residual correlation · stimulus covariance · data extraction · reproducibility