With fixed known responses and linear sums, 14 of 36 incomplete mouse V1 maps required more than 5% layer share change at 30 Hz with no upper limit on missing responses

We tested whether missing second responses can remove layer share changes in mouse V1 input maps at 30 Hz. We used public optical input maps and fixed the stored first amplitudes. Equal activation and linear voltage sums defined the layer shares. The selected panel had 55 maps, with 887 known and 64 unknown second responses for positive first amplitudes. A source check added 37 connected inputs with a stored zero first response. This gave 36 incomplete maps. Every unknown second response could take any nonnegative value, with no upper limit. With known responses fixed, 14 maps had a sharp minimum total variation above 0.05. A relative allowance of 10% for known second estimates left 11 maps above this value. A 20% allowance left eight. These allowances are assumed sets, not measured errors or confidence intervals. Seven maps had simple two-layer lower certificates above 0.05. Missing inputs cannot add excitation outside their recorded source layers, which explains why a bound can remain without a gain limit. Independent linear programs, source reads and an offline repeat checked the calculations. The package preserves original data, code, provenance, results, figures and repeat instructions. These are limits for recorded estimates under a linear model. Missing trial errors, uncertain stored zeros and incomplete input coverage prevent a biological effect claim.

functional connectome · mouse visual cortex · short-term plasticity · missing data · cortical layers · partial identification · response bounds