Missing response measurements permit either sign of a facilitation-rate contrast in selected Allen cortical connections

We tested whether missing response measurements leave the sign of a reciprocal-minus-one-way positive-change rate open in Allen Synaptic Physiology r2.1. The fixed candidate sample had 290 mouse and 135 human directed excitatory connections with known reverse calls and a synapse summary. Early responses were observed for 211 mouse and 97 human connections. The complete-case rate differences were -3.75 and -4.71 percentage points. Exact bounds with unrestricted missing binary labels were -29.70 to +21.25 points in mouse and -27.34 to +20.33 in human. Late-response bounds also permitted both signs. Requiring ten test spikes in both directions did not fix either sign. The code checked all feasible missing-positive counts and 3136 synthetic cases. These are sharp finite-sample extrema, not confidence intervals. They assume well-defined missing response labels and fixed observed labels and connection calls. Selected slices, measurement error, and unknown reverse calls limit transfer. The package contains original code, aggregate evidence, a proof, a figure, source hashes, and external regeneration steps. External records are excluded.

Allen Institute · reciprocal connectivity · short-term plasticity · missing outcomes · partial identification · finite-sample bounds