Three-pulse Allen mouse slice data leave the mean first-to-third linear residual link unresolved

We tested first-to-third response association after removal of the middle response in 255 selected Allen mouse visual cortex connections from 82 inferred mice. Verified three-pulse sufficient statistics came from a prior signed pulse audit. The mouse-weighted mean Fisher partial correlation was -0.01640, with nominal 95 percent interval -0.04745 to 0.01507. Rest and trial-count checks also left the pooled mean unresolved. A secondary three-response sum-variance score was -3.89 percent. Under a fitted Gaussian one-step covariance model, its simulated mean was -1.91 percent and its central 95 percent range included the observed score. Thus, a negative score alone need not show a biological covariance defect. The result does not establish a one-step synaptic law or absence of memory. Few trials, selected sources, fit error, missing raw trial order and strong simulation assumptions limit inference. Original code, aggregate evidence, figures, input hashes and verified retrieval steps support reproduction. External record tables remain outside the bundle.

Allen Institute · synaptic dynamics · partial correlation · temporal memory · sampling bias · null result