Shared targets gave no clear extra halfwidth similarity in 108 connection pairs from 16 human cortical donors

We tested whether inputs to the same human cortical neuron have similar excitatory postsynaptic potential halfwidths. We used Planert et al. source tables, Figshare 28184126 version 1. Selection required usable cell types, intrinsic-quality flags, geometry, SNR at least 3, and a reported halfwidth within the 50 ms stimulus interval. Models used 330 connections from 21 donors. Each fit excluded its test donor. The primary test used 108 shared-target edge pairs from 41 recording panels and 16 donors. Their squared residual difference ratio was 0.787 against the within-panel label reference. The 95% donor bootstrap interval was 0.351 to 1.160, with all models refit. The lower-tail label probability was 0.1907. Thus, the primary test did not resolve extra similarity. A specified secondary completion rule used other measured inputs to the target. A fixed half blend with panel calibration reduced squared log error by 22.9% in 163 edges from 16 donors. Its full-refit interval was 2.0% to 43.3%. An added target-type baseline gave a 21.1% gain. However, the SNR-at-least-10 subset gave a 1.9% error increase. An added donor sign-flip test gave p = 0.0619. These findings concern selected measured panels. They do not establish a common biological filter, independent edge durations, or prediction from anatomy alone. The package preserves new code, source hashes, model states, predictions, all random draws, figures, checks, and repeat steps. The original source data remain external under their stated terms.

human cortex · synaptic kinetics · EPSP halfwidth · shared targets · functional connectivity · response completion · donor bootstrap