Regions & stimulus

The four cortical regions

Every neuron belongs to one of four regions, modeled as a Rust enum:

rust/brain_sim/src/lib.rs

enum RegionKey { Visual, Auditory, Language, Attention }

These match the TypeScript RegionKey type so telemetry round-trips cleanly:

src/sim/types.ts

export type RegionKey = 'visual' | 'auditory' | 'language' | 'attention'

Region classification

Region is assigned at construction by classify_region, which normalizes a neuron's position within the mesh bounding box and applies a set of spatial rules:

rust/brain_sim/src/lib.rs

fn classify_region(p: Vec3, min: Vec3, max: Vec3) -> RegionKey {
    // nx, ny, nz are normalized 0..1; lateral is distance from the midline
    if nz > 0.66 { return RegionKey::Visual; }
    if lateral > 0.34 && ny > 0.28 && ny < 0.68 { return RegionKey::Auditory; }
    if nx < 0.44 && nz > 0.34 && nz < 0.72 { return RegionKey::Language; }
    if ny > 0.58 || nz < 0.28 { return RegionKey::Attention; }
    if lateral > 0.24 { RegionKey::Auditory } else { RegionKey::Visual }
}

In rough anatomical terms: the back of the mesh (high Z) is visual, the lateral mid-band is auditory, the left-front is language, and the top or front is attention. The fall-through keeps every neuron assigned.

Injecting stimulus

The UI injects stimulus by calling Brain.injectStimulus, which forwards to the WASM inject_stimulus:

src/brain.ts

injectStimulus(opts: { type?: string; intensity?: number; durationMs?: number } = {}): void {
  this.world.inject_stimulus(opts)
}

On the Rust side, inject_stimulus:

1. Records a StimulusPulse (region, intensity, duration, expiry timestamp) into telemetry so the response curves can react.
2. Computes how many signals to release from the intensity: releases = round(2.0 + intensity * 14.0), at least 1.
3. Releases those signals onto neurons in the targeted region (or spread across all regions if no region was specified), picking neurons from the region's neuron bucket.

rust/brain_sim/src/lib.rs

let releases = (2.0 + parsed.intensity * 14.0).round().max(1.0) as usize;

So a higher intensity produces a larger burst of new signals, and the pulse record drives the onset/sustained response telemetry while it is active. If the target region is omitted, the burst is distributed round-robin across all four regions.

See Telemetry for how pulses shape the response index.