How exploration rewires what your brain learns

Researchers built an AI agent navigating a maze that could either explore widely or exploit known rewards. Agents that explored developed clean, spatially-organized internal maps; reward-focused agents created messier representations. When they compared this to real mice doing the same task, exploratory mice had brain-like representations matching exploratory agents, while mice with restricted movement patterns resembled reward-driven agents. Active exploration appears to be how brains organize knowledge into generalizable mental maps rather than narrow, task-specific patterns.