Simulation · culture explorer
A neuron culture, alive in your browser.
Thousands of simulated neurons grow on a 60-electrode array. Left alone they stay quiet, then fire together in bursts that sweep across the dish. Click an electrode to stimulate it.
excitatory inhibitory electrode, click to stimulate
Spike raster
last 3 s · one row per neuron, sorted left → right across the dishPopulation rate
spikes per neuron per second, 10 ms binsWhat you're watching
Each point is a neuron, and each flash is a spike. Neurons connect mostly to their neighbours, so activity spreads as a wave. A burst ends when the synapses run out of resources, and the culture stays quiet for a few seconds while they recover. Real cultures on multi-electrode arrays show the same rhythm. Slow the speed down to 0.1× to watch a burst travel.
The model
- Izhikevich neurons: 80% excitatory (regular spiking and chattering), 20% fast-spiking inhibitory.
- 100 synapses each, mostly within ~150 µm, with delays set by distance at 150 µm/ms.
- Short-term synaptic depression (Tsodyks–Markram), which shapes the bursts.
- Sparse random input at 0.1 Hz per neuron, like spontaneous release.
- Deterministic: a seed always grows the same culture.
The Pong experiment
The same cultures can also play Pong through their electrodes, following the DishBrain protocol: predictable feedback when they hit the ball, noise when they miss. A control culture with random feedback plays next to them. Run the experiment.
Izhikevich (2003), Simple model of spiking neurons. Tsodyks & Markram (1997). Kagan et al. (2022), In vitro neurons learn and exhibit sentience when embodied in a simulated game-world. Sinapayen, Masumori & Ikegami (2017), Learning by stimulation avoidance.