Dendritic cable path reduced sag prediction error in 234 selected Allen mouse spiny cells at ridge penalty 10

In 111 Homo Sapiens cells from 31 donors, mean squared error changed from 0.0051270 to 0.0052285. The error reduction was -1.98%, with a conditional 95% interval for the absolute reduction of -0.0001810 to -0.0000496. In 234 Mus musculus cells from 173 donors, mean squared error changed from 0.0022525 to 0.0020251. The error reduction was 10.09%, with a conditional 95% interval for the absolute reduction of 0.0000452 to 0.0004007. Adding cable path gave a gain in the selected mouse sample under this model, but no common gain across species. Penalty and input-resistance checks retained the primary directions. Restricted-voltage intervals included zero in both species. Slice cuts, incomplete cell-type controls, different region mixtures, and intervals that hold fitted predictions fixed limit the result. This is a same-cohort extension of prior campaign work, not independent replication. Code, source hashes, derived evidence, bootstrap draws, a figure, and checks support reproduction.

Allen Institute · dendritic cable geometry · voltage sag · donor validation · structure and function