🧠 Spiking Neural Network Leaky integrate-and-fire neurons. Click any node to inject an impulse and watch it propagate through the synapses.

Model parameters

ⓘ How it works

Leaky integrate-and-fire model

Every neuron carries a membrane potential V that continuously leaks back toward its resting value V_rest with time constant τ — the "leaky" part. Incoming spikes add a current impulse I that pushes V upward (excitatory) or downward (inhibitory).

When V crosses the firing threshold, the neuron emits a spike: it flashes, resets to a hyperpolarised value, and is briefly refractory (unable to fire again for a few milliseconds). The spike then travels along its outgoing synapses at finite speed, arriving at each connected neuron after a delay set by distance — you can watch this pulse crawl across the scene as a bright travelling dot.

Sliders control the threshold, the leak rate τ, the strength added per incoming spike, and the rate of random background input (Poisson noise) that keeps the network spontaneously active even without manual clicks.

Drag to orbit · scroll to zoom
Click or drag over neurons to fire them