Assumed 1 ms relative timing variation raised the minimum model residual in 57 selected Allen mouse input pairs

We tested a model of convergent excitation and inhibition with fixed gain and trial-to-trial timing variation. We used passed current-clamp fits from the same -55 mV holding bin in Allen Synaptic Physiology r2.1. The selected mouse visual-cortex sample had 57 input pairs at 47 targets from 43 inferred donors. We optimized mean relative delay and one inhibitory gain for each pair. Gaussian relative-delay variation with a standard deviation of 1 ms increased mean residual energy from 0.2641 to 0.2919 of excitatory-input energy. The increase was 0.0279, with a conditional 95% donor bootstrap interval of 0.0194 to 0.0373. Distance and interior-fit checks kept the increase above a chosen 0.01 reporting threshold. Gaussian timing variation cannot improve this optimized expected-energy score under the stated model; the data set its size. Timing variation was assumed, not measured. Linear voltage addition, fixed kernels, inferred donors, missing fits and selected slice circuits limit the result. This extends a published finite-rise benchmark in the same cohort. It is not independent biological validation.

Allen Institute · excitation and inhibition · relative timing variation · fixed gain · conditional model · convergent inputs