Basic properties of symmetric random matrices can explain the emergence of macroscopic patterns in neural networks, suggesting that initialization conditions of connected neural populations may confer computational advantages across the mouse brain.