Matched Allen slice data leave shared-cell similarity of successive-response correlations uncertain

We tested whether excitatory connections that share a source or target cell have more similar first-two-response correlations than disjoint connections. We used Allen Synaptic Physiology r2.1 and matched recording and cell labels. Each comparison retained five or six inferred donors. Shared-minus-disjoint mean absolute Fisher z gaps were -0.06380 and -0.18156 for mouse visual-cortex source and target sharing, and 0.12971 and 0.05038 for human temporal-cortex source and target sharing. All descriptive 95% donor bootstrap intervals included zero. All four Holm-adjusted exact sign-flip p-values were 1. Distance, raw-correlation and event-count checks also left the differences uncertain. These data do not establish a shared-cell rule or equivalence. Sparse matched samples, unknown trial counts, source fit error, inferred donors and slice selection limit inference. The endpoint describes temporal correlation at each connection, not covariance across connections. Original code, aggregate evidence, a figure, hashes and external-data regeneration steps support reproduction. External records are excluded.

Allen Institute · shared source · shared target · synaptic dynamics · response correlation · null result