An orientation interaction gave no clear connection prediction gain in selected Allen cortical slices

We tested whether cell-pair orientation changes the prediction value of a known reverse chemical connection in Allen Synaptic Physiology r2.1 slice data. We used 16,856 mouse visual-cortex directions from 746 inferred donors and 1,546 human temporal-cortex directions from 51 donors. Three donor-fold allocations compared logistic models with distance, orientation, cell labels, layers and probing counts. A reverse call gave a clear prediction gain in mouse. The human reverse-call interval included zero. An added reverse-call times orientation term gave mouse log-loss gain -0.000031 nats per direction, with a conditional 97.5% donor interval from -0.000343 to 0.000292. Human gain was -0.000315, with interval from -0.003146 to 0.001916. Neither interval excluded zero. The package supplies code, paired evidence, predictions, plans, source hashes and a figure. This result applies to one pooled linear interaction and one penalty. It does not prove equal orientation effects or a causal wiring rule. Slice selection, inferred donor labels, coordinate error, the small human sample and uncertainty conditional on fitted models limit the result.

Allen Institute · reciprocal connectivity · cortical geometry · donor validation · null result