Fixed voltage windows gave no clear correlation increase after spikes in 35 mouse V1 inhibitory output pairs

We tested whether two targets of one inhibitory neuron have more trial correlation after a source spike than before it. We used Znamenskiy and colleagues' mouse V1 data, Figshare version 1, and a prior SCI trial identity benchmark. Exact source waveform matches identified simultaneous exposures. The fixed test retained 35 output pairs from 54 detected connections, 23 PV source neurons and 18 mice. It used fixed voltage windows and removed a linear trial trend. Pairs were averaged within source, sources within mouse, and mice equally. Mean correlation was 0.0048 after the spike and -0.0031 before it. The increase was 0.0079. Its central 95% range from 9,999 complete mouse resamples was -0.1075 to 0.1300. All seven planned checks had ranges across zero. An earlier baseline check, added after trace review, gave 0.0004 with a range from -0.1501 to 0.1341. Independent calculations reproduced all 70 main pair correlations and 10,444 window measurements. A repeated raw extraction and compact replay matched by hash. The package preserves 1,492 short voltage traces, matched trial identities, code, source hashes, figures, all results and repeat steps. This test concerns joint voltage noise on a selected local connection graph. It does not establish independent transmitter release or predict intact-brain activity.

inhibitory connections · divergent outputs · trial correlation · parvalbumin · mouse V1 · shared fluctuations · reproducible benchmark