Spiral waves over the touch map

In mice, brain activity organises into rotating waves that pivot on the somatosensory cortex and sweep across the body maps, one spiral per hemisphere, mirrored. Nobody has shown human S1 does the same. This page draws the idea anyway, on the real group-average human brain, and labels it for exactly what it is: a schematic riding real anatomy.

Schematic wave, not a measurement Loading the labels that anchor it.
Mirrored handedness
left CCW · right CW
centres from areas 3b · 1 · 2

Loading the surface

View
Wave

What is measured, and in whom

The mouse result. Ye and colleagues imaged calcium activity across the whole dorsal cortex of mice and found large rotating waves whose pivots sit on the somatosensory cortex, whose fronts sweep across the somatotopic body maps in topographic order, and whose rotation is mirrored between the two hemispheres (Ye et al., Science 392, eadx1369, 2026, doi:10.1126/science.adx1369; data CC BY 4.0). Those waves turn in fractions of a second. The authors are explicit that whether human somatosensory cortex hosts the same waves is undetermined.

The human result. Human cortex does host measured spiral waves, at a very different speed: Xu, Long, Feng and Gong found interacting spiral patterns in the phase of resting and task fMRI from 100 Human Connectome Project adults, band-passed 0.01 to 0.1 Hz (Nature Human Behaviour 7, 2023, doi:10.1038/s41562-023-01626-5). Those spirals rotate over tens of seconds, not milliseconds, and they wander across association cortex rather than pinning to S1. The underlying HCP timeseries sit behind ConnectomeDB data-use terms and cannot be redistributed here.

What this page draws, exactly

The brain is real: the HCP S1200 group-average cortex, the same meshes every surface page on this site loads. The anchor is real: at load time the page reads the HCP-MMP1 atlas labels it already ships and puts each hemisphere's rotation centre at the centroid of the S1 core strips, areas 3b, 1 and 2. The dim paint under the wave is the same atlas paint the sensorimotor map uses, washed down.

The wave is not real. It is a one-armed rotating front computed in each vertex's angle about that S1 centroid, curled slightly with radius so it reads as a spiral, and run with opposite handedness in the two hemispheres because that is what the mouse paper reports. It turns about once every six seconds, a speed chosen to be watchable, which matches neither the mouse waves (much faster) nor the human fMRI spirals (slower). In the site's evidence vocabulary this layer is schematic, the same class as the dotted teaching routes on the circuits page, and the amber chip on the canvas says so at all times.

The honest path to a measured human layer

The Xu et al. recipe is reproducible in principle on this site's own surface: band-pass the BOLD signal, take its phase, find the phase singularities, and accumulate a group spiral-centre density per vertex. scripts/build_spiral_density.py is the skeleton of exactly that pipeline, with the checks written first. It refuses to run without the HCP timeseries, and its refusal message is the documentation of what is needed: a ConnectomeDB account, acceptance of the HCP data-use terms, and the resting-state dtseries files. The blockers and the cheap unlocks, including simply asking the authors for a shareable derived map, are written up in docs/SPIRALS.md.

The house rule, applied. Never make the visualization more certain than the science. Rotating somatosensory waves: measured in mice. Human spiral waves: measured, slow, elsewhere on the cortex, data not redistributable. A rotating wave on human S1: undetermined, and therefore drawn here only as a labelled schematic. When a measured human layer exists, it replaces this one.