A fixed rule for separate ON and OFF contact directions gave no clear gain over one direction per cell in 27 cells from one mouse retina

We asked if separate contact directions in two retinal layers explain the difference between responses to bright and dark edges. We used published anatomy and calcium recordings from one mouse retina. The test included all 27 matched ganglion cells in four ON–OFF direction-selective anatomical types. A fixed rule used the ON-minus-OFF starburst contact direction to predict the ON-minus-OFF motion preference. A common-direction rule predicted zero difference. Separate directions reduced the main mean squared error by 12.8%. The 95% cell-resample range extended from a 37.6% loss to a 39.0% gain. With the source authors' fitted responses, separate directions increased error by 5.4%. In 24 cells with more precise response directions, the main gain was 0.5%. A secondary comparison with exchanged layers gave a larger apparent gain than comparison with a shared contact axis. The ON–OFF direction differences agreed weakly between repeat halves. This test gave no clear added prediction from separate layers. It does not establish equal circuit function. The main limits are one retina, one bar speed, fixed direction order, slow calcium signals, and contacts that are not verified synapses. We provide the paired features, all tested definitions, figures, code, source checksums, and two reproduction routes.

retina · connectome · structure and function · direction selectivity · starburst amacrine cells · ON and OFF pathways · calcium imaging · paired comparison · reproducible research