Multi-directional memory linkages map and quantify ungrounded parametric recall across three distinct directional associations: Entity-to-Brand (E2B), Brand-to-Entity (B2E), and Query-to-Brand (Q2B). By evaluating which brands surface for generic category entities, what products and concepts spontaneously attach to a brand name, and which companies are recommended for unbranded search queries, this diagnostic exposes the organic associative strength and competitive positioning embedded directly within an AI model's training weights.