Geometry Before Wiring

Royer, Paquola, Rodriguez-Cruces and colleagues, writing in Nature Communications (December 2025), built a coordinate system for human cortical dynamics out of intracranial EEG — electrodes inside the skull, not sensors reading through it — and then asked which structural facts about the cortex predict where a region lands in that space. The candidates were geometry (how far apart regions sit), microstructure (how similar their cellular composition is), and the connectome (who is wired to whom).

The answer is unusually clean for this literature. Across most of the cortex, the pattern of dynamic similarity is largely explained by inter-regional distance alone. Microstructural similarity and connectivity add real explanatory power in one place: transmodal association cortex, the regions that coordinate across systems rather than serving a single sense. The relative weight of local and macroscale constraints varies systematically along the cortical sheet, and it varies along the same axis that separates unimodal from association territory. The result holds when the electrophysiology and the imaging come from different large-scale atlases, and again when both come from the same individuals.

Why this is worth a page here

This project has been running a figure — brain waves that look like a trout stream, cortex as a continuous medium carrying travelling disturbances — with an explicit warning attached to it. The warning, set out in the trout-fishing essay, is that a shared picture fixes vocabulary and not mechanism. Two continuous media can look alike without sharing any physics.

The Royer result does something the resemblance could not do on its own: it puts a number on the geometric term. Distance is not a metaphor in that analysis, it is the leading predictor, measured against two named competitors on the same data. That is the difference between borrowing geometry as a description and finding it doing work. The distinction is the whole argument of Volume 27, which is about the difference between geometry that is native to a system and geometry that has been imported to talk about it. The volume now carries this material formally in Chapter 49, The Cortical Manifold, which states the geometric object, catalogues the legitimate borrowings, prices the trout-stream resemblance, and ends with the experiment that would separate the pure-geometry regime from the geometry-plus-network regime.

What it does not say

• It does not make the cortex a fluid. Distance-dependence is what you expect from conduction delays, local coupling, and shared afferents. Any medium with finite propagation speed produces it. The finding is compatible with the stream figure and equally compatible with accounts that owe the figure nothing.
• It does not settle whether phase carries information. Explaining where a region sits in a dynamic-similarity space is not the same as showing that its rhythms route anything. The open question logged as relay entry #070 stands untouched: an intervention on phase alone, with spike rates held fixed, that changes behaviour.
• It does not license curvature talk. "Geometry" here means Euclidean distance along and through the cortical sheet. It is not a Riemannian metric with intrinsic curvature, and nothing in the paper invites the gravitational vocabulary this corpus uses elsewhere.
• It is not a causal result. Every claim is variance explained across regions, not a manipulation. Distance is fixed at birth and cannot be randomised.

The honest reading

Where the cortex is doing one job — vision, touch, hearing — its rhythms look like those of its neighbours, and how far away a neighbour is accounts for most of the resemblance. Where the cortex is coordinating across jobs, the wiring diagram and the cellular makeup start to matter, and proximity stops being sufficient. The association regions are, in this specific and testable sense, the places where the brain stops being explainable by its shape.

That is the part worth keeping. It is also the part most likely to be deleted in transit: the shareable version of this paper is "brain rhythms are set by geometry," which is the finding with its exception removed — and the exception is where the interesting cortex lives. This is the same failure the corpus tracks under density is not proof: a bounded result travels further once its qualifier has been stripped, and the stripping is invisible to the reader who never saw the original.