AtomSpace can represent vast networks of knowledge, relationships, and reasoning structures. As these networks grow, the challenge shifts from storing information to managing importance. As a mechanism for salience assignment, all knowledge risks competing equally for computational resources. A platform like CURAD can lead to inefficient retrieval, excessive memory growth, noisy reasoning paths, and difficulty maintaining focus on information that is most relevant to current goals. The Amygdala Layer is proposed as a salience architecture that operates beneath symbolic reasoning and attention allocation systems. Inspired by the functional role of biological salience networks, this layer assigns dynamic importance values to atoms and edges using metadata rather than emotion. These salience values influence retrieval priority, memory persistence, reinforcement cascades, and eventual archival behavior. Aligns with the ongoing CURAD project. github.com/GreenGlia/Amygdala-Layer