Reverse excitation gave no clear inhibitory delay prediction gain in three fixed ridge tests of 221 reciprocal mouse V1 cell pairs

We tested whether stronger excitation predicts an earlier inhibitory response in the reverse direction of a recorded cell pair. We used two Allen Synaptic Physiology mouse V1 samples and raw PV/pyramidal recordings from another laboratory. Fixed ridge models predicted log inhibitory onset from own inhibitory strength, soma distance, excitatory onset and, in Allen data, cell and recording features. One complete mouse was withheld per test. Adding reverse excitation gave error decreases of -0.71% in 60 Allen coarse pairs, +1.24% in 96 Allen patch-seq pairs and -1.78% in 65 independent pairs. Every 95% whole-mouse resample range included zero gain. A small negative adjusted slope in patch-seq did not give a clear test-error gain. A secondary tested-edge comparison also found no clear added delay gain from recorded graph reciprocity. The independent onset pipeline recovered 88 reciprocal source pairs; 65 passed fixed fit rules, with 22 inhibitory fits at the lower onset bound. The package supplies all fits, mean traces, source hashes, code, resample results and figures. These are selected physiological slice graphs, not complete anatomical connectomes. Manual onset annotation, waveform assumptions, detection and missing fast fits limit inference. Unitary onset does not include the time needed to recruit an inhibitory spike.

feedback inhibition · synaptic latency · reciprocal connectivity · mouse visual cortex · structure and function · paired recordings · prediction · reproducible analysis