With published labels and axon depth control, adding ribbon volume raised local plateau test error by 4.3% in eight mouse ON bipolar types

We tested whether synaptic ribbon volume predicts sustained glutamate release across mouse cone bipolar types. We read all 5,157,738 released Eyewire II ribbon detections and the full cell table. The analysis used 27,643 cone bipolar cells in 13 types; 27,579 had an IPL ribbon mean. We joined type summaries to independent Franke local and global chirp responses from GIN dataset A. The main test used eight ON types with the published labels. We fitted ordinary least squares to axon depth, then added log ribbon volume. Each type was excluded from fitting once. Local mean squared error rose from 0.002828 to 0.002950, a 4.3% increase. The size–plateau Spearman correlation was 0.119, with exact two-sided p = 0.793. Global error rose by 19.3%. Complete-cell, reviewed-cell, count-control, and measurement checks gave no stable added gain. The source paper allows some functional labels to be exchanged. Across 144 tested ON label maps, local error changes ranged from -13.8% to +67.1%; 19 maps gave an error decrease. These maps have no assigned probability. This is a cross-data-set test with one structural retina and only eight main type observations. It cannot show a causal effect or reject a size mechanism in selected circuits. The package includes source hashes, code, aggregate inputs, predictions, figures, and repeat instructions. Original restricted source records remain external.

retina · connectomics · synaptic ribbons · structure and function · glutamate release · bipolar cells · label uncertainty · prediction