Cortical wave obstacle

Cortical wave obstacle#

This experiment launches a spiking activity wave across an excitatory and inhibitory sheet, places a silent circular obstacle in its path, and maps whether the wave crosses, bends, splits, or dies.

Prompt#

Create a sheet of excitatory and inhibitory neurons and trigger a brief spark at its left edge. Show the activity wave crossing the sheet, then place a silent circular patch in its path and reveal whether the wave bends around it, splits, or dies. Sweep the patch size and inhibition strength and summarize the outcomes in a phase map.

Agent decision path#

  1. Classify the model as a spatial point-neuron E/I network.

  2. Route neural dynamics to brainpy-state and sparse spikes to brainevent.

  3. Place one excitatory and one inhibitory LIF neuron at each sheet position.

  4. Build explicit local CSR projections and launch a 3 ms left-edge current pulse.

  5. Silence the obstacle with both a hyperpolarizing clamp and spike mask.

  6. Map seven lesion radii and six inhibitory gains to independent state lanes.

  7. Define crossing, bending, splitting, and death from far-edge reach and bypass-corridor activation.

  8. Save the storyboard, quantitative phase map, outcomes, and focused checks.

Result#

The 42-condition sweep produces 4 crossings, 16 splits, 4 one-sided bends, and 18 deaths. Each phase-map label is paired with the measured right-edge reach fraction.

Time-ordered cortical sheet snapshots compare an intact propagating wave with a wave encountering a silent circular obstacle. Activity routes around the obstacle rather than passing through it.

Fig. 5 Matched snapshots make the spatial perturbation visible.#

Phase map across obstacle radius and inhibitory gain with cells labeled crossing, splitting, bending, or dying and annotated by right-edge reach.

Fig. 6 The phase map connects each qualitative outcome to the continuous reach measurement used to classify it.#

With-BrainX skill/Without skill comparison#

With BrainX skill

Source captured · benchmark pending

Measure the same 42 conditions with compilation, execution, memory, and nonblank-artifact checks reported separately.

Without BrainX skill

Matched run not collected

Preserve topology, grid size, obstacle intervention, and phase definitions before comparing code or performance.