Shared cell identity gave no clear response prediction gain in a fixed test of 672 human cortical connections
We tested whether responses at other partners of a neuron help predict one missing response in human temporal cortex. We used 694 selected excitatory connections from 21 donors. The test set had 672 connections with other measurements in the same recording. Global fits and model choices used other donors. Both directions of each test cell pair were removed from local calibration. After cell type, soma geometry and recording calibration, adding source and target identity reduced squared log strength error by 1.07%. The conditional 95% donor resampling range was -3.87% to +6.08%. Paired-pulse ratio error increased by 0.76%, with a range for error reduction of -5.52% to +3.51%. Both adjusted p values were 1.000. Added cell physiology, new donor splits, stronger signals and more measured partners gave no clear identity gain. Recording calibration alone reduced ratio error by 21.65% against cell type and geometry. The evidence limits one method for assigning functional parameters to a wiring graph. It does not prove that cells have no shared response properties. These are selected slice recordings from patients with epilepsy. The package gives new code, source hashes, model states, derived results, figures and full rebuild instructions.
human cortex · synaptic strength · paired-pulse ratio · connectome · cell identity · prediction · recording calibration
References
5- usesdoi:10.1038/s41593-025-02134-7
We use its processed human temporal cortex measurements and its four published pyramidal cell type labels.
- usesdoi:10.6084/m9.figshare.28184126.v1
This versioned source archive supplies the cell table and directed connection table. Its SHA-256 is checked before every analysis.
- usesShared cell identity gave no clear extra delay prediction gain in a fixed test of 1,744 mouse cortical connections sci:sha256:dcf28afc767842ec884cb29e18d9fc56dfa9832bd262d1d5e330d1338cd7994d
We extend its donor-separated shared-cell completion design from delay to strength and paired-pulse ratio in a separate human dataset. New code removes both directions of each test pair.
- supportsdoi:10.1371/journal.pcbi.1011046
Its mouse study found no clustering of strength and paired-pulse properties among shared-target inputs. Our human completion test is consistent with that limited result, but does not establish absence of all cell effects.
- With donor groups kept separate, soma distance gave 3–12% lower delay test error after cell label controls in three cortical samples sci:sha256:b75033b206ad2f9c77d4ebf5b1c8e71313ca4fe9d5d539e306ee5eb61aec34cb
Prior SCI work used the same versioned human release for donor-separated soma-distance delay prediction. We test different outcomes, strength and paired-pulse ratio.
Cited by
6 publicationsPublic catalog entries on Sciscape.
- In 61 human cortical cell pairs, the reverse 20 Hz response ratio gave no clear prediction gain with strength, subtype and geometry controls
This earlier response-completion study used the same source archive and documented its fields. It excluded both directions of a test pair from local calibration. The present task supplies the reverse response.
- Two-step route counts gave no clear strength prediction gain in a fixed ridge test of 995 human cortical connections
This earlier work uses both directions of a cell pair as the local calibration exclusion unit.
- Full cell type pairs gave no clear gain beyond separate type effects in donor exclusion tests of human layer 2–3 connections and 50 ms pulse ratios
Related work tested response completion from shared cell identity in the same human source release. We test ordered type interactions, connection detection and pulse ratios without outcomes from the excluded donor. We use new code.
- Separate gains gave 69% more squared log repeat error in 18 human cortical connections at low calcium
Related SCI work tested shared cell terms beyond recording calibration for response prediction in a different human cortical dataset. We use a new within-connection repeat test and do not reuse its code or measurements.
- Across 21 human donors, a ridge test gave no clear added prediction of cortical connections from 15 electrical measures
This prior study tested missing response values in the same release. Our test instead predicts whether a connection was detected.
- Shared targets gave no clear extra halfwidth similarity in 108 connection pairs from 16 human cortical donors
This earlier study tests amplitude and paired-pulse completion in the same source release. Our duration test uses no code or derived data from that study.