How a Cat in a Box in 1898 Laid the Groundwork for Modern AI Learning

Between 1898 and 1949, a loosely connected group of psychologists studying animal behavior unknowingly reverse-engineered the core loop behind modern reinforcement learning. Edward Thorndike's 1898 puzzle-box experiments with hungry cats showed that animals learn not through reasoning but through gradual trial-and-error reinforcement of useful behaviors. His observations produced what became known as the Law of Effect: actions followed by positive outcomes are more likely to be repeated. The key concepts that emerged from this era — agent, state, action, reward, policy, and value — were all derived from animal studies, decades before computers existed. These biological foundations now underpin every reinforcement learning system in use today.