5281/zenodo.19812200 Modern intelligent systems and human decision makers often operate in environmentscharacterized by conflicting information, competing narratives, and incomplete signals.Classical probabilistic approaches interpret decision making as aggregation of independentevidence, which typically leads to unstable or diffuse conclusions when signals contradicteach other.This paper introduces a coherence-based framework for reasoning under narrativeconflict. The key idea is to reinterpret probability not as a measure of truth, but asan allocation of of limited computational attention across competing interpretations, determinedby their internal structural consistency. We define a Heuristic Decision Field(HDF) where each alternative hypothesis corresponds to a configuration of signals, andits coherence determines its prominence in the decision landscape.The model naturally captures conflict, ambiguity, and the emergence of multiple competingnarratives. We compare HDF with softmax, logistic regression with pairwise interactions,and a conflict-aware Bayesian network (CBN) that explicitly models signaldependencies and includes a penalty term for inconsistent configurations. Experimentalresults demonstrate that coherence-based reasoning yields more stable and interpretableoutcomes, especially under high-conflict conditions. The proposed framework providesa foundation for cognitive decision making, narrative analysis, and artificial intelligencesystems operating under uncertainty. Keywords: narrative conflict, coherence, decision making, attention allocation, heuristicdecision field, cognitive science, artificial intelligence.
Sergey Dzhumaev (Sun,) studied this question.