Proceedings of the Annual Meeting of the Cognitive Science Society

Understanding the cognitive structure of explanations— andthe cognitive processes that assemble them— is a milestonefor understanding how people learn and communicate. Re-cent research on explanatory coexistence suggests that peo-ple’s causal beliefs are less globally coherent than previouslythought: people use seemingly-competing supernatural and bi-ological causes to explain different aspects of the same phe-nomenon, or they assemble supernatural and biological causesinto single, coherent explanations (Legare & Gelman, 2008;Legare & Shtulman, 2018; Shtulman & Lombrozo, 2016).This coexistence— and unexpected coherence— of diversecausal mechanisms poses interesting questions about the roleof coherence and fragmentation in people’s mental models andexplanations. This paper presents a computational model ofexplanatory coherence in the well-characterized domain of dis-ease transmission, extending a previous cognitive model ofexplanation-based conceptual change (Friedman, Forbus, &Sherin, 2018). Our approach (1) retrieves diverse causal modelfragments based on the phenomenon to explain, (2) assem-bles coherent causal models using relevance-directed abduc-tive reasoning, and (3) selects explanatory paths that supportwithin-explanation and within-scenario coherence. Our modelsimulates the three different types of explanatory coexistencedetailed in the literature.