Spike adaptation gave no clear added gain in a ridge test of initial synaptic change in 422 mouse cortical connections

We tested whether spike adaptation in two connected cells improves prediction of short-term synaptic change. We used Allen Synaptic Physiology r2.1. The primary panel had 422 confirmed chemical connections from 185 mice. A fixed ridge rule used cell class, layer, distance, age, depth, and passive membrane properties. All data from each animal stayed in one test fold. Adding source and target adaptation indices increased mean test error by 1.76%. The central 95% animal resample range for error decrease was -7.48% to +3.64%, with model refitting. Planned checks also gave no clear added benefit. The pooled source rank correlation was 0.298. It fell to 0.019 after directed cell-class adjustment. Later checks used other penalties and a smaller baseline; all gave negative added-gain point estimates. A separate human panel had 94 connections from 30 donors and gave uncertain results. Selection, measurement noise, and model limits restrict the result. The test does not show that adaptation has no role in cortical function. The package gives code, source hashes, all predictions, uncertainty draws, figures, and repeat steps. Original Allen data remain external.

connectome · synaptic dynamics · spike adaptation · cortical circuits · multipatch · prediction · animal groups · short-term plasticity