Conflict as Catalyst: Mining Disagreement Among Emergence Detectors for Theory Discovery

Recent work has focused on developing machine learning methods for detecting or predicting emergence (Bouakrif et al., 2023; Dahia & Szabo, 2024), but there is rarely consensus on what constitutes "emergent" in complex, high-dimensional settings. Rather than seeking a single best detector, this research proposes running a diverse ensemble of emergence detectors (e.g., fuzzy logic, supervised learning, causal inference models) on the same system and systematically analyzing instances where their predictions diverge. These "conflict zones" in the parameter or behavior space are then mined to generate new theoretical insights: What kind of emergent behaviors cause such disagreement? Are there overlooked, intermediate forms of emergence? This meta-analytic approach transforms the ambiguity and conflict in automated detection into a tool for theory generation, pushing beyond current efforts that treat emergence detection as a settled classification problem.

References:

1. A Trained Fuzzy Expert System to Detect Emergent Behavior. Fatima Zohra Bouakrif, Ali Boukehila, Nora Taleb (2023). 2023 International Conference on Decision Aid Sciences and Applications (DASA).
2. Detecting Emergent Behavior in Complex Systems: A Machine Learning Approach. Simranjeet Singh Dahia, Claudia Szabo (2024). SIGSIM Principles of Advanced Discrete Simulation.