MorphoAgent: Developmental Biology as Computational Architecture for Self-Organizing LLM-Based Digital Organisms

We present MorphoAgent, a digital artificial life system that uses real Drosophila melanogaster developmental biology as its computational architecture for LLM-based agents. Each agent carries a 25-gene genome spanning 8 signaling pathways (Hedgehog, Wnt, BMP, Notch, Hippo, Insulin, Toll/Imd, EGFR). Agents self-organize through Notch lateral inhibition for role differentiation, Hippo pathway for density-dependent population control, and morphogen diffusion through an extracellular matrix for distributed communication. The LLM serves as a ribosome—a universal translation machine—rather than the system's intelligence. Simulation experiments demonstrate emergent role diversity (entropy 0.861 vs 0.256 baseline), stable population dynamics, and heritable phenotypic evolution across 50 generations. Open-source implementation included.