Verdant-Minds is a research-grade cognitive architecture that processes text through a structured nine-stage pipeline (sensory → pattern → memory → communication → reasoning → ethics → action → language → learning). The system maintains a symbolic MemoryWeb (graph memory) coupled bidirectionally to an Extended Cognitive Wave Function (ECWF), a compact, configurable wave-state model whose entropy, magnitude, and phase influence memory reinforcement, decay, and concept formation. A governance layer ("Three Kings") provides oversight across data quality, ethics, and action selection. We report evidence from cultivation runs that emergent concepts—concepts created autonomously by the system's wave dynamics rather than seeded by designers—not only appear, but preferentially link to earlier emergent concepts in the memory backbone. In a representative run (119 concepts, 37 emergent nodes), emergent→emergent edges in a sparsified backbone exhibit a strong timing bias toward linking newer emergents to older emergents (observed earlier-share 0.807; shuffle baseline ~0.499±0.083; z ≈ 3.71), suggesting temporal scaffolding during concept formation.