A new hierarchical framework integrates Hebbian neuroplasticity, graph neural networks, and meta-learning to manage autonomous UAV swarms. The system mimics biological reflexes, skills, and reasoning to optimize search and rescue operations. By utilizing a digital twin for strategic decisions, the architecture improves tactical coordination. This provides a structured template for scaling multi-agent swarm intelligence.