A Kernel-Spectral Index for Single-Neuron Input-Output Complexity

The Spectral Complexity Index (SCI) separates human versus mouse cortical neuron models across 35 biophysically detailed cells, with strong agreement to a DNN-based complexity metric (Spearman rho=0.797).

• Why it matters: Surrogate-based single-neuron complexity measures are hard to interpret and can be computationally expensive, limiting reproducible comparisons across cells and species.
• What they did: SCI is a white-box kernel-spectral measure that encodes presynaptic histories and counts ordered spectral modes needed to explain somatic response energy across 35 human and mouse cortical neuron models.
• The result: SCI remains stable under sampling and kernel perturbations, matches DNN-based Functional Complexity Index (rho=0.797), and reduces computational cost versus FCI to about 1/10 or less.

AI summary of the abstract · may contain errors · not medical advice · read the paper.