Source peers predicted paired-pulse ratios better than target peers in 22 human cortical donors

We tested whether other connections from the same source cell or into the same target cell predict a connection's paired-pulse ratio. We used public human layer 2–3 recordings at 20 Hz. The matched test had 577 connections from 22 donors. After adjustment for amplitude, cell types, soma distance, and depth, source peers gave a mean squared log-ratio error of 0.1025. Target peers gave 0.1397. The difference was 0.0372, with a 95% donor bootstrap interval of 0.0129 to 0.0646. The recording mean gave lower error, 0.0901. Thus, source peers outperformed target peers but did not improve on the recording mean. Sparse peer counts can limit direct use of shared response properties. This is an exploratory connection-completion test in one processed dataset. It does not establish a release mechanism or transfer to intact healthy cortex, other species, or behavior.

human cortex · short-term plasticity · paired-pulse ratio · functional connectivity · connection completion