Bandwidth matching gave no clear connection prediction gain in selected Allen mouse cortical slices

We tested whether similar stored chirp response bandwidths help predict detected chemical connections. We used Allen Synaptic Physiology r2.1 data and held out whole inferred donors. The main sample had 2418 mouse visual cortex directions, 236 detected connections and 197 donors. Models included cell labels, geometry, probe counts, passive properties and separate bandwidth terms. The gain from adding absolute log2 bandwidth difference was -0.321% (positive means less error). The conditional 95% donor bootstrap interval was -0.701% to 0.042%. A separate human temporal cortex fit used 921 directions and gave -0.569% with interval -1.069% to -0.146%. The primary mouse test does not resolve an added gain from this matching term. A limited 1–15 Hz measurement window, selected slice recordings, inferred donor labels and fixed-prediction intervals limit the result. Code, source hashes, selected measurements, predictions and checks are supplied.

Allen Institute · chirp bandwidth · chemical connectivity · donor validation · null result