Matched tests left delay similarity uncertain for 77 mouse V1 connection pairs with a shared source

We tested whether two cortical connections have more similar onset delays when they share a source or target neuron. We used Allen Synaptic Physiology r2.1 and fixed each experiment and directed cell-class pair. The eligible mouse V1 panel had 807 connections. Only 77 shared-source edge pairs from 42 experiments and 67 shared-target pairs from 39 experiments allowed an informative comparison. The squared log-delay difference ratio was 0.907 for the source role and 1.076 for the target role. One is the random-assignment expectation. The 95% slice resample intervals were 0.609–1.143 and 0.876–1.236. Both Holm p-values were 1.000. Geometry, fit mode, and sign checks left wide uncertainty. All 2,665 stored scalar delays matched their accepted fit summaries exactly. Simulations on the fixed graph detected a source effect in 26.0% of panels when it supplied one quarter of log-delay variance, and in 89.1% when it supplied one half. These rates assume independent Gaussian source effects and edge noise. The biological result does not establish equal or independent delays. We provide checked code, aggregate results, simulations, figures, source hashes, and repeat instructions.

synaptic delay · cortical circuits · multipatch · cell class · matched randomization · statistical power