The reverse PSP reduced squared log amplitude test error by 29% in 88 mouse V1 cell pairs with each test mouse excluded from fitting

We tested whether one measured postsynaptic potential (PSP) helps predict the opposite direction in a reciprocal cell pair. We used the public mouse V1 data of Znamenskiy et al. The source already reported correlated strengths. We rebuilt responses from the first 20 accepted traces and recovered 88 pairs from 29 mice, with 58 PV cells and 79 pyramidal cells. Each test mouse was excluded from fitting. A fixed linear model used distance, mean signed recording depth and age. Adding log excitatory PSP reduced squared log inhibitory PSP error from 0.109431 to 0.077175, a 29.48% reduction. The central 95% range from 2,000 mouse resamples with full refitting was 7.24% to 47.50%. All sensitivity point gains were positive. A later control for both mean recording voltages retained a 28.95% gain. Partner shuffles within each PV cell gave a mean gain of 13.56%; four of 999 reached the recorded gain. These are descriptive assignment controls. We also recovered the source visual-response counts and rounded correlations. In 31 reciprocal pairs with visual responses, the further gain after visual similarity was 25.61%, but its resample range was -35.68% to 55.87%. This smaller test was inconclusive. The main result supports a scoped prediction rule for detected reciprocal pairs in one study. It does not measure conductance or establish a plasticity mechanism or transfer to another laboratory. The package preserves code, provenance, all derived input tables, predictions, fold coefficients, resamples, figures and repeat instructions. Independent trace and solver checks passed.

reciprocal connections · synaptic strength · structure and function · visual cortex · parvalbumin · prediction · reproducibility