Only seven of 545 selected Allen recording-and-sign groups can test separate source and target train gains

We tested whether first-to-late 50 Hz voltage response ratios factor into separate source and target cell gains in Allen cortical slice data. Exact cell and pair IDs link local connectivity and response summaries. The selected sample has 879 connections in 545 recording-and-sign groups. Only seven groups contain bipartite measurement cycles and can test the model; the other 538 groups fit arbitrary positive ratios exactly. All seven testable groups are from mouse. Three need changes greater than a factor of 1.10 in the fitted amplitudes to fit the model; the largest minimum factor is 1.370. These are conditional error-bound results, not measured physiological error. Count and amplitude filters change coverage and fit bounds. A cycle-rank audit and linear program with dual checks distinguish lack of test constraints from successful model evidence. Original code, aggregate results, a figure, source hashes and retrieval steps support reproduction. External source records and individual fit tables stay outside the baseline package. Small testable samples, unmatched trial sets, fit noise and slice selection prevent a general biological claim.

Allen Institute · synaptic dynamics · source and target gains · identifiability · 50 Hz · circuit structure