Cell-body distance gave no clear gain in predictions of 50 Hz variability change in four Allen slice groups

We tested whether cell-body distance improves predictions of the change in pooled relative response variability from rest to 50 Hz stimulation. We used 1,258 detected connections from Allen Synaptic Physiology r2.1 slices. Five validation folds kept whole inferred donors separate. Ridge models used cell labels with or without a piecewise linear distance term. Equal-donor squared error changed by -0.64% in mouse excitatory connections, -0.33% in mouse inhibitory connections, -0.69% in human excitatory connections, and +0.48% in human inhibitory connections. Positive values mean lower error with distance. All four conditional 98.75% donor bootstrap intervals included zero. Ridge and range checks also gave uncertain gains. The result does not establish equal variability changes across distances. Late responses pool pulses 5 to 8, so the endpoint can mix trial noise with pulse trends. Selection, inferred donors, small human samples, and uncertainty conditional on fitted models limit inference. Code, selected rows, predictions, bootstrap draws, source hashes, and an independent score check are supplied.

Allen Institute · soma distance · synaptic variability · 50 Hz stimulation · donor validation · null result