A fitted strength slope cut squared log error for mean pulse 2 amplitudes by 7.9% at 30 Hz in 991 mouse V1 links with each test mouse excluded

We tested whether a fitted first response strength slope improves prediction of the mean second response in mouse visual cortex. We used Chou and colleagues' optomapping data at 30 Hz. All 1,057 figure records matched the source map files. The main panel had 991 detected excitatory connections, 69 targets, and 39 mice aged 17 to 25 days. Each outer test excluded one entire mouse. Inner mouse groups selected the context penalty. Context included cell type, layers, location, age, and sex. Equal mouse squared log errors were 1.0094 for context alone, 0.5637 for proportional first strength, and 0.5191 for a fitted strength slope. The fitted rule cut error by 48.6% versus context and 7.90% versus proportional strength. The fixed-prediction mouse resample range for the latter gain was 2.16 to 13.33%. A later check repeated all fits in 499 mouse resamples and gave a central range of 0.87 to 16.47%. All main fits reached the original penalty bound. A later wider grid gave a 9.09% gain. Separate target-type tests gave a clear gain for basket targets and uncertain gains for pyramidal and Martinotti targets. The complete-data strength slope was 0.761. The test uses processed mean responses from the same stimulation repeats. It does not identify a release mechanism, predict individual trials, or test anatomical synapse counts. Measurement error, response selection, and the slice protocol limit interpretation. The package contains the unchanged CC0 source archive, a reusable extraction, predictions, a forecast tool, all resamples, figures, source hashes, independent solver checks, and repeat instructions.

mouse visual cortex · optomapping · short-term plasticity · synaptic strength · grouped prediction · paired pulses