Source width gave lower test delay error than target width in selected Allen inhibitory slice samples

We compared source and target spike widths as predictors of synaptic onset delay. We linked cell and connection records from Allen Synaptic Physiology r2.1. Distance and cell labels were included in both models. Ten repeated five-fold tests kept inferred donors separate. In 466 mouse inhibitory connections from 198 inferred animals, target-model error minus source-model error was 0.01212 ms squared. The conditional 95% donor bootstrap interval was 0.00259 to 0.02201. The primary interval favors source width. The wider four-group interval includes zero in mice. The small human inhibitory sample also favors source width, but does not resolve a gain over the baseline. Missing widths, selected slice connections, inferred donor groups and uncertainty without model refitting limit this test. These data do not establish equal roles for source and target cells or a causal timing mechanism.

Allen Institute · synaptic delay · spike width · source and target · grouped validation · inhibitory connections