Monotone recovery needs changes above 0.10 in 518 of 797 exploratory Allen connection summaries

We tested whether the absolute paired response change after a 50 Hz train decreases as the pause grows in Allen cortical slice data. No connection met the planned rule of 20 paired trials at each of three delays. An added exploratory five-trial screen retained 797 connections from mouse visual and human temporal cortex. In 518, the stored median residuals needed an additive change above 0.10 normalized units to agree with any nonincreasing curve. With ten trials, only 47 connections remained and 15 exceeded 0.10. We give a sharp tolerance formula and check it by linear programming. Low trial counts, different trial sets, fit error and missing protocols prevent a biological recovery-failure claim. This is a screening method and a coverage result. Code, aggregate results, provenance and retrieval instructions support reproduction. External records are excluded.

Allen Institute · synaptic recovery · short-term plasticity · monotone regression · measurement tolerance