From Neurons to Neutrons: A Case Study in Interpretability