The Nooscope · atlas
Every issued reading, placed on the thing it was taken from and lit by the value it actually carries. Drag to spin. Choose a state, then play the response and watch the perturbation travel — or fail to.
The condition the subject was in. Readings taken in other states dim.
Milliseconds after the jolt. Every issued reading measures 15–300 ms; the track runs to 450 so the response can be watched dying.
Node color is the reading's value on its own scale. Size is its interval width.
The contrast
The cortex is real; the placement on it is not. The surface is the fsaverage5 pial mesh, both hemispheres, lit by its own sulcal-depth values — so the gyri and sulci you see are measured anatomy. None of the deposits behind these readings publishes electrode coordinates, so a node still sits where that recording geometry would sit, and the cerebellum and brainstem are modelled rather than measured. The values, intervals, states and counts are real and come straight from the register. The propagating wavefront is driven by each reading's own value; it is not a replay of recorded signal, and no reading in the register contains one.
Watching a brain idle tells you little: noise looks busy. The Nooscope perturbs the system and measures how hard the echo is to compress. A rich, structured echo scores high; a short stereotyped one scores low.
PCIst runs unbounded, roughly 5 to 60. PCI-LZ and sPCI run 0 to 1. Putting them on one axis would be a category error, so the atlas colors each node against its own scale and says which.
NS-0012 and NS-0013 are the same seven patients on the same implanted grid, awake and under propofol. They occupy one location because they are one array. Switching state is the contrast the register exists to detect.
A Type F reading is not a place. It is the ratio between two issued readings, computed pairwise on the units both share, so it is drawn as an arc joining its two parents rather than as a node of its own.