EPSP weights raised the mean fraction of stable gains that permit brief growth in 54 human cortical panel models with equal cell time constants

We tested how measured synaptic strengths change a connection map model. The source was a fixed release of human cortical paired recordings. The main scope had 54 cyclic panels from 19 epilepsy surgery donors. The linear model used equal cell time constants and positive mean EPSPs as edge weights. F was the fraction of the stable gain interval that allowed some population pulse to grow briefly before decay. With equal donor weights, mean F rose from 0.238 with equal edge weights to 0.443 with EPSP weights. The paired increase was 0.205. Its central 95% donor bootstrap interval was 0.138 to 0.278. Seventeen donor means increased and two decreased. Random weight placement on the same edges gave a comparable mean of 0.431, with a 95% reference range of 0.373 to 0.487 and a two-sided rank of 0.689. Source and target controls also gave no clear added placement effect. The result concerns a model gain interval, not measured pulse growth in a living circuit. It assumes equal cell units and omits inhibition outside the small recorded panels. The package includes source tables, provenance, all model matrices, code, results, figures, numerical checks, and repeat instructions.

human cortex · synaptic strength · transient amplification · linear dynamics · paired recordings · connectome · EPSP · model benchmark