With ±20 µV amplitude bounds, a nonnegative linear model fixed the 50 Hz excitation share direction in 11 of 29 mouse V1 maps for all fixed positive input weights

We tested whether unknown fixed input weights can change an excitation share change during a 50 Hz train. We read all 123,506 pair records in Allen Synaptic Physiology r2.1. The selected mouse V1 cohort had 65 measured inputs to 29 targets from distinct animal keys. We used separate first and late voltage estimates. A prior audit showed that the released pulse vectors cannot preserve pulse order. Our model sums voltage magnitudes and sets negative late estimates to zero. Each input has an unknown positive weight that stays fixed across responses. We give ratio bounds that test every such weight at once. At the point estimates, 21 maps had a rise for all weights and six had a fall. With independent errors of up to ±20 microvolts on every amplitude, eight rises and three falls remained fixed. The other 18 maps allowed either direction. Equal weights gave a mean rise of 9.15 percentage points. Its 95% animal resample range was -1.03 to +19.35 points. Removing negative late estimates gave a positive mean range in 27 maps. Larger first-amplitude floors and independent source class labels weakened that result. These are conditional model bounds for sampled inputs. They do not estimate conductance balance or the response of an intact brain. The package preserves code, source hashes, derived ratios, error bounds, opposite-direction cases, figures, and repeat steps.

synaptic dynamics · excitation and inhibition · linear input models · error bounds · mouse visual cortex · Allen Synaptic Physiology