Visual similarity gave no clear gain in fixed ridge predictions of second responses at 30 Hz for 44 mouse V1 inhibitory links

We tested whether visual response similarity helps predict the second voltage response in a 30 Hz train. We used matched visual and slice data from mouse primary visual cortex, called V1. The main sample had 44 inhibitory PV-to-pyramidal connections from 17 mice. A fixed ridge rule used first voltage, distance, baseline voltage, and visual signal fractions. Adding visual response cosine increased mean test error by 5.61%. The central 95% whole-mouse resample range for gain was -34.34% to +13.14%. Each test mouse stayed outside fitting. Separate trial halves and all specified inhibitory checks also gave higher point error. A secondary test of 28 excitatory links gave no clear gain. A simpler first-voltage rule gave the same pattern. The source paper reported no correlation with paired-pulse ratios. We add a test of prediction in excluded mice and a checked matched benchmark. The data do not establish absence of a visual effect on synaptic dynamics. Voltage tails, full-record selection, and the small selected sample limit the result. The package has code, 2,600 trial rows, 127 visual mean arrays, source hashes, all predictions, figures, and repeat steps.

visual cortex · synaptic dynamics · structure and function · cross-validation · parvalbumin · paired-pulse response · reproducible benchmark