Two inferred mice gave most of the mean score in a depletion test of 231 selected Allen V1 slice connections

We tested one necessary moment identity of a simple vesicle-depletion model in selected Allen mouse V1 slice response fits. The sample had 231 detected connections, 110 source cells, 78 inferred mice, and 7,026 first-two-response trial pairs. An unbiased product estimate with a separate amplitude scale gave a mean dimensionless score of -2.07651. Its nominal 95 percent mouse bootstrap interval was -6.14765 to 0.05484. Two mouse groups contributed -2.10400 to the mean. After their post hoc omission, the other 76 groups had mean 0.02821. Raw fits, trial trimming, detrending, and a 40-trial rule all gave intervals that included zero. Baseline subtraction gave nonpositive first-response variance estimates in 40 connections. Exact probability checks, independent CSV calculations, and simulations passed. The test remains uncertain. It does not establish model adequacy or identify a release mechanism. Selected slice circuits, few trials, large fits, assumed initial occupancy, and unverified error matching limit transfer. Code, proof, source hashes, sufficient statistics, uncertainty draws, and retrieval steps support inspection and reproduction.

Allen Institute · synaptic depression · response covariance · moment test · vesicle depletion · circuit reliability · mouse V1 · null result