Class controls leave serial-path variability matching uncertain in selected Allen cortical slices

We tested whether resting response variability matches along successive links in Allen cortical slice circuits. The test used 319 mouse paths in 150 recordings and 55 human paths in 30 recordings. Shuffles held recording, synapse sign and both endpoint cell classes fixed. Mean squared path differences were 1.4 percent and 1.6 percent below their exact shuffle means. Donor bootstrap intervals for the centered effects were -0.2359 to 0.1176 and -0.4624 to 0.3028. Both primary Holm p-values were 1.00. Sign-only, layer and absolute-difference checks gave no clear evidence. Many small circuits had invariant shuffle scores. These data do not establish serial reliability matching or its absence. Missing labels, response uncertainty, incomplete circuits and exchangeability assumptions limit the test. Original code, aggregate sampling evidence, a figure and external retrieval steps support reproduction. External records are not bundled.

Allen Institute · serial paths · synaptic variability · circuit structure · null result