Dendrite elongation gave no clear gain in a fixed linear test of motion axis responses in 315 cells from one mouse retina

We tested whether the long axis of a ganglion cell arbor predicts its measured response to a moving bar beyond cell type and image position. We used matched electron microscopy and calcium records from one mouse retina: 315 typed cells, 47 types, nine recordings, eight directions, and four retained repeats. We computed exact cable covariance and represented shape and response with second angular moments. Each fixed ridge test excluded a whole recording. One shared shape coefficient raised total test error by 0.074%. The central 95% recording resampling range for error decrease was -0.730% to +0.549%, with predictions held fixed. A free linear map, inner-disc cable, and cells clear of the volume edge also gave no clear gain. Other response repeats reduced error by 24.5-24.7%. A later raw-trace check with separate repeat-group scales gave 25.7-25.9%. Thus, repeatable measured axis signals were present, but this simple shape readout added little when type and position were known. The data come from one retina and one fixed stimulus order. These results do not establish a causal mechanism or a population effect across mice. The package has new compact facts, code, all predictions, figures, source hashes, and repeat instructions.

retina · structure and function · dendritic geometry · motion axis · calcium imaging · connectomics · prediction · reproducible methods