With cell type given, contact concentration raised error by 13% in a fixed linear test of motion selectivity strength in 27 On-Off cells from one mouse retina

We tested whether the directional concentration of starburst contacts predicts motion selectivity strength within known cell types. We used all 27 matched On-Off direction-selective ganglion cells in four types from one mouse retina. Both strength measures were normalized first circular moments. After type means were removed, the correlation was 0.139. Its one-sided within-type assignment p value was 0.27195. In an excluded-cell linear test, adding contact concentration raised mean squared error from 0.011434 to 0.012938, or 13.16%. The conditional cell resample range for error reduction was -24.16% to -3.15%, with predictions fixed. Contact angles still gave median motion angle error of 14.64 degrees. Response strength agreed across repeat halves, with correlation 0.642 after type means were removed. Planned position and coverage controls gave a 17.76% point gain, with range -14.29% to 41.15%. An expanded 42-cell panel also gave an uncertain gain. A later nested test selected zero contact contribution for all 27 primary predictions. The result limits this fixed linear rule. It does not show that synaptic inhibition has no role in motion selectivity. Physical contacts, one-animal sampling, supplied type labels and fixed stimulus order limit inference. The package preserves code, derived cell data, all predictions, source hashes, figures and repeat steps. A fresh source rebuild matched all numerical files and both PNG figures.

retina · starburst amacrine cells · direction selectivity · contact concentration · connectome · calcium imaging · structure and function · prediction