A visual explanation of reinforcement learning
Summary
Aatish Bhatia and Arkarup Banerjee visualize reinforcement learning by building a tiny four-neuron 'spark' that learns to seek light through trial-and-error and reward. The article explains how simple brain wiring, a step-by-step learning algorithm (REINFORCE-like), and multiple trials produce emergent, expert behavior, with biologically inspired examples and visualization of learning paths.